{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999380267724343,
  "eval_steps": 500,
  "global_step": 32270,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 5.5003252029418945,
      "learning_rate": 2.063983488132095e-08,
      "loss": 1.4433,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.714099884033203,
      "learning_rate": 4.12796697626419e-08,
      "loss": 1.7894,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.425031661987305,
      "learning_rate": 6.191950464396286e-08,
      "loss": 2.3674,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.665287017822266,
      "learning_rate": 8.25593395252838e-08,
      "loss": 1.3795,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 102.36792755126953,
      "learning_rate": 1.0319917440660475e-07,
      "loss": 5.4067,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.612777233123779,
      "learning_rate": 1.2383900928792572e-07,
      "loss": 1.419,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.844910621643066,
      "learning_rate": 1.4447884416924666e-07,
      "loss": 1.2927,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 109.35301208496094,
      "learning_rate": 1.651186790505676e-07,
      "loss": 5.2585,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 113.8509521484375,
      "learning_rate": 1.8575851393188855e-07,
      "loss": 5.8918,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 136.58035278320312,
      "learning_rate": 2.063983488132095e-07,
      "loss": 5.9173,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.09292221069336,
      "learning_rate": 2.2703818369453046e-07,
      "loss": 1.9264,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.943973541259766,
      "learning_rate": 2.4767801857585145e-07,
      "loss": 1.4245,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 116.83894348144531,
      "learning_rate": 2.683178534571723e-07,
      "loss": 6.1004,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 107.78886413574219,
      "learning_rate": 2.889576883384933e-07,
      "loss": 5.2359,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.372946739196777,
      "learning_rate": 3.0959752321981424e-07,
      "loss": 1.6223,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.237632751464844,
      "learning_rate": 3.302373581011352e-07,
      "loss": 1.7519,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 98.89286804199219,
      "learning_rate": 3.5087719298245616e-07,
      "loss": 4.5225,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 106.96707153320312,
      "learning_rate": 3.715170278637771e-07,
      "loss": 5.2484,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 114.63630676269531,
      "learning_rate": 3.921568627450981e-07,
      "loss": 5.803,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.574666023254395,
      "learning_rate": 4.12796697626419e-07,
      "loss": 1.4659,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 114.06707000732422,
      "learning_rate": 4.3343653250774e-07,
      "loss": 5.5937,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.854344367980957,
      "learning_rate": 4.540763673890609e-07,
      "loss": 1.4991,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 113.8956069946289,
      "learning_rate": 4.747162022703819e-07,
      "loss": 5.393,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 99.88960266113281,
      "learning_rate": 4.953560371517029e-07,
      "loss": 4.942,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.62392807006836,
      "learning_rate": 5.159958720330238e-07,
      "loss": 1.9181,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.635759353637695,
      "learning_rate": 5.366357069143447e-07,
      "loss": 1.5825,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.672242641448975,
      "learning_rate": 5.572755417956656e-07,
      "loss": 1.38,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.302175045013428,
      "learning_rate": 5.779153766769866e-07,
      "loss": 1.5798,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 83.12200927734375,
      "learning_rate": 5.985552115583076e-07,
      "loss": 4.7125,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 66.9862289428711,
      "learning_rate": 6.191950464396285e-07,
      "loss": 4.3205,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.917734146118164,
      "learning_rate": 6.398348813209495e-07,
      "loss": 1.4812,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.460594177246094,
      "learning_rate": 6.604747162022703e-07,
      "loss": 1.9869,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.07828712463379,
      "learning_rate": 6.811145510835913e-07,
      "loss": 1.2484,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 72.02684020996094,
      "learning_rate": 7.017543859649123e-07,
      "loss": 4.6785,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.396835327148438,
      "learning_rate": 7.223942208462333e-07,
      "loss": 1.3162,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 49.478824615478516,
      "learning_rate": 7.430340557275542e-07,
      "loss": 3.7256,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 47.18341064453125,
      "learning_rate": 7.636738906088752e-07,
      "loss": 3.4609,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 50.98810577392578,
      "learning_rate": 7.843137254901962e-07,
      "loss": 3.7725,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.12209129333496,
      "learning_rate": 8.049535603715171e-07,
      "loss": 1.9963,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 54.29254150390625,
      "learning_rate": 8.25593395252838e-07,
      "loss": 3.5669,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.02449607849121,
      "learning_rate": 8.46233230134159e-07,
      "loss": 1.5269,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 45.38787841796875,
      "learning_rate": 8.6687306501548e-07,
      "loss": 3.3593,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 38.80595016479492,
      "learning_rate": 8.87512899896801e-07,
      "loss": 3.1138,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 48.698463439941406,
      "learning_rate": 9.081527347781218e-07,
      "loss": 3.2422,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 42.8900260925293,
      "learning_rate": 9.287925696594428e-07,
      "loss": 3.3033,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 35.08656311035156,
      "learning_rate": 9.494324045407638e-07,
      "loss": 2.8568,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.060311317443848,
      "learning_rate": 9.700722394220848e-07,
      "loss": 1.3199,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.00004577636719,
      "learning_rate": 9.907120743034058e-07,
      "loss": 2.6519,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 40.63880920410156,
      "learning_rate": 1.0113519091847268e-06,
      "loss": 3.0969,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 30.872892379760742,
      "learning_rate": 1.0319917440660475e-06,
      "loss": 2.9478,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.41267204284668,
      "learning_rate": 1.0526315789473685e-06,
      "loss": 1.3697,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 50.90806198120117,
      "learning_rate": 1.0732714138286893e-06,
      "loss": 3.1617,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.898908615112305,
      "learning_rate": 1.0939112487100103e-06,
      "loss": 2.4543,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.130802154541016,
      "learning_rate": 1.1145510835913313e-06,
      "loss": 2.6161,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.290111541748047,
      "learning_rate": 1.1351909184726523e-06,
      "loss": 2.5627,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.50565528869629,
      "learning_rate": 1.1558307533539732e-06,
      "loss": 2.6029,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.437199592590332,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 1.2563,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.709607124328613,
      "learning_rate": 1.1971104231166152e-06,
      "loss": 0.9418,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.83498764038086,
      "learning_rate": 1.2177502579979362e-06,
      "loss": 2.2445,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.089672088623047,
      "learning_rate": 1.238390092879257e-06,
      "loss": 2.407,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.292842864990234,
      "learning_rate": 1.259029927760578e-06,
      "loss": 1.0684,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.49665355682373,
      "learning_rate": 1.279669762641899e-06,
      "loss": 1.3804,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.450586318969727,
      "learning_rate": 1.30030959752322e-06,
      "loss": 1.5274,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.550836563110352,
      "learning_rate": 1.3209494324045407e-06,
      "loss": 1.8588,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.456924438476562,
      "learning_rate": 1.3415892672858619e-06,
      "loss": 2.0501,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.057917594909668,
      "learning_rate": 1.3622291021671827e-06,
      "loss": 1.2637,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.85093116760254,
      "learning_rate": 1.3828689370485039e-06,
      "loss": 1.8316,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.342580795288086,
      "learning_rate": 1.4035087719298246e-06,
      "loss": 2.1815,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.769253730773926,
      "learning_rate": 1.4241486068111456e-06,
      "loss": 1.6567,
      "step": 69
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.563549995422363,
      "learning_rate": 1.4447884416924666e-06,
      "loss": 1.2281,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.641158103942871,
      "learning_rate": 1.4654282765737876e-06,
      "loss": 1.4094,
      "step": 71
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.329587936401367,
      "learning_rate": 1.4860681114551084e-06,
      "loss": 1.0154,
      "step": 72
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.570966720581055,
      "learning_rate": 1.5067079463364296e-06,
      "loss": 1.2241,
      "step": 73
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.956279754638672,
      "learning_rate": 1.5273477812177503e-06,
      "loss": 1.8056,
      "step": 74
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.90291976928711,
      "learning_rate": 1.5479876160990715e-06,
      "loss": 1.7803,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 21.39967918395996,
      "learning_rate": 1.5686274509803923e-06,
      "loss": 1.976,
      "step": 76
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.510459899902344,
      "learning_rate": 1.5892672858617133e-06,
      "loss": 1.8955,
      "step": 77
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.897970199584961,
      "learning_rate": 1.6099071207430343e-06,
      "loss": 1.7718,
      "step": 78
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.060806274414062,
      "learning_rate": 1.6305469556243553e-06,
      "loss": 1.7228,
      "step": 79
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.426802635192871,
      "learning_rate": 1.651186790505676e-06,
      "loss": 1.8793,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.96075439453125,
      "learning_rate": 1.671826625386997e-06,
      "loss": 1.17,
      "step": 81
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.220455169677734,
      "learning_rate": 1.692466460268318e-06,
      "loss": 1.8519,
      "step": 82
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.870081901550293,
      "learning_rate": 1.7131062951496388e-06,
      "loss": 1.6996,
      "step": 83
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.70217514038086,
      "learning_rate": 1.73374613003096e-06,
      "loss": 1.6718,
      "step": 84
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.117308616638184,
      "learning_rate": 1.7543859649122807e-06,
      "loss": 1.5028,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.296212196350098,
      "learning_rate": 1.775025799793602e-06,
      "loss": 2.0271,
      "step": 86
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.907048225402832,
      "learning_rate": 1.7956656346749227e-06,
      "loss": 1.4581,
      "step": 87
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.691845893859863,
      "learning_rate": 1.8163054695562437e-06,
      "loss": 1.1344,
      "step": 88
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.547973871231079,
      "learning_rate": 1.8369453044375647e-06,
      "loss": 1.2686,
      "step": 89
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.268970489501953,
      "learning_rate": 1.8575851393188857e-06,
      "loss": 1.4954,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8655571937561035,
      "learning_rate": 1.8782249742002064e-06,
      "loss": 1.0741,
      "step": 91
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.22858715057373,
      "learning_rate": 1.8988648090815276e-06,
      "loss": 2.1765,
      "step": 92
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.000706672668457,
      "learning_rate": 1.9195046439628486e-06,
      "loss": 1.5064,
      "step": 93
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.815850257873535,
      "learning_rate": 1.9401444788441696e-06,
      "loss": 1.3169,
      "step": 94
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.922850608825684,
      "learning_rate": 1.96078431372549e-06,
      "loss": 1.3956,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.31998062133789,
      "learning_rate": 1.9814241486068116e-06,
      "loss": 1.3194,
      "step": 96
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.949653387069702,
      "learning_rate": 2.002063983488132e-06,
      "loss": 1.3195,
      "step": 97
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.9020414352417,
      "learning_rate": 2.0227038183694535e-06,
      "loss": 1.3114,
      "step": 98
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.260583877563477,
      "learning_rate": 2.043343653250774e-06,
      "loss": 1.3835,
      "step": 99
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.205341339111328,
      "learning_rate": 2.063983488132095e-06,
      "loss": 1.119,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.296168327331543,
      "learning_rate": 2.084623323013416e-06,
      "loss": 0.214,
      "step": 101
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.865838050842285,
      "learning_rate": 2.105263157894737e-06,
      "loss": 1.1635,
      "step": 102
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.549725532531738,
      "learning_rate": 2.125902992776058e-06,
      "loss": 1.2748,
      "step": 103
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.251093864440918,
      "learning_rate": 2.1465428276573786e-06,
      "loss": 1.4317,
      "step": 104
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.383101463317871,
      "learning_rate": 2.1671826625387e-06,
      "loss": 1.2082,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.603120803833008,
      "learning_rate": 2.1878224974200206e-06,
      "loss": 1.3428,
      "step": 106
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.8437862396240234,
      "learning_rate": 2.208462332301342e-06,
      "loss": 1.2601,
      "step": 107
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.075748443603516,
      "learning_rate": 2.2291021671826625e-06,
      "loss": 1.1596,
      "step": 108
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.020939826965332,
      "learning_rate": 2.249742002063984e-06,
      "loss": 1.2814,
      "step": 109
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.235538482666016,
      "learning_rate": 2.2703818369453045e-06,
      "loss": 1.2509,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.05907154083252,
      "learning_rate": 2.2910216718266255e-06,
      "loss": 0.9975,
      "step": 111
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.21036434173584,
      "learning_rate": 2.3116615067079465e-06,
      "loss": 1.0522,
      "step": 112
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.227554321289062,
      "learning_rate": 2.3323013415892675e-06,
      "loss": 1.1603,
      "step": 113
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.955459594726562,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 1.0211,
      "step": 114
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.552540302276611,
      "learning_rate": 2.3735810113519094e-06,
      "loss": 0.9748,
      "step": 115
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.963927268981934,
      "learning_rate": 2.3942208462332304e-06,
      "loss": 0.9302,
      "step": 116
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.91703987121582,
      "learning_rate": 2.4148606811145514e-06,
      "loss": 0.9729,
      "step": 117
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.402760028839111,
      "learning_rate": 2.4355005159958724e-06,
      "loss": 1.1665,
      "step": 118
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.60770845413208,
      "learning_rate": 2.456140350877193e-06,
      "loss": 1.378,
      "step": 119
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.844650745391846,
      "learning_rate": 2.476780185758514e-06,
      "loss": 1.1611,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.025947570800781,
      "learning_rate": 2.497420020639835e-06,
      "loss": 1.1109,
      "step": 121
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.748651027679443,
      "learning_rate": 2.518059855521156e-06,
      "loss": 0.9338,
      "step": 122
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.096957683563232,
      "learning_rate": 2.538699690402477e-06,
      "loss": 1.0764,
      "step": 123
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.581175804138184,
      "learning_rate": 2.559339525283798e-06,
      "loss": 1.9406,
      "step": 124
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.335753440856934,
      "learning_rate": 2.579979360165119e-06,
      "loss": 1.2273,
      "step": 125
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.6319899559021,
      "learning_rate": 2.60061919504644e-06,
      "loss": 1.0335,
      "step": 126
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.930748462677002,
      "learning_rate": 2.621259029927761e-06,
      "loss": 0.9317,
      "step": 127
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.926667213439941,
      "learning_rate": 2.6418988648090814e-06,
      "loss": 0.9102,
      "step": 128
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.303267955780029,
      "learning_rate": 2.662538699690403e-06,
      "loss": 1.0773,
      "step": 129
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7645421028137207,
      "learning_rate": 2.6831785345717238e-06,
      "loss": 1.4497,
      "step": 130
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.069040775299072,
      "learning_rate": 2.7038183694530443e-06,
      "loss": 0.9757,
      "step": 131
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.866879463195801,
      "learning_rate": 2.7244582043343653e-06,
      "loss": 1.0235,
      "step": 132
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.428579807281494,
      "learning_rate": 2.7450980392156867e-06,
      "loss": 0.9324,
      "step": 133
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.043857097625732,
      "learning_rate": 2.7657378740970077e-06,
      "loss": 1.0261,
      "step": 134
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.800389289855957,
      "learning_rate": 2.7863777089783283e-06,
      "loss": 1.3667,
      "step": 135
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.459311008453369,
      "learning_rate": 2.8070175438596493e-06,
      "loss": 0.8669,
      "step": 136
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5616519451141357,
      "learning_rate": 2.8276573787409707e-06,
      "loss": 1.306,
      "step": 137
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.422781229019165,
      "learning_rate": 2.8482972136222912e-06,
      "loss": 0.2445,
      "step": 138
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.951080560684204,
      "learning_rate": 2.8689370485036122e-06,
      "loss": 1.1339,
      "step": 139
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.004542827606201,
      "learning_rate": 2.889576883384933e-06,
      "loss": 0.8923,
      "step": 140
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.640811443328857,
      "learning_rate": 2.9102167182662538e-06,
      "loss": 1.5505,
      "step": 141
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.231103897094727,
      "learning_rate": 2.930856553147575e-06,
      "loss": 1.4462,
      "step": 142
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4667102098464966,
      "learning_rate": 2.951496388028896e-06,
      "loss": 0.1937,
      "step": 143
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.350134372711182,
      "learning_rate": 2.9721362229102167e-06,
      "loss": 1.3869,
      "step": 144
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.48054838180542,
      "learning_rate": 2.9927760577915377e-06,
      "loss": 0.8459,
      "step": 145
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.804610252380371,
      "learning_rate": 3.013415892672859e-06,
      "loss": 0.7771,
      "step": 146
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.31807017326355,
      "learning_rate": 3.0340557275541797e-06,
      "loss": 0.7663,
      "step": 147
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.911772727966309,
      "learning_rate": 3.0546955624355007e-06,
      "loss": 1.1221,
      "step": 148
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.20274019241333,
      "learning_rate": 3.0753353973168216e-06,
      "loss": 0.8808,
      "step": 149
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.349485397338867,
      "learning_rate": 3.095975232198143e-06,
      "loss": 0.8446,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.358565807342529,
      "learning_rate": 3.1166150670794636e-06,
      "loss": 1.0014,
      "step": 151
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1005990505218506,
      "learning_rate": 3.1372549019607846e-06,
      "loss": 0.8576,
      "step": 152
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.641711235046387,
      "learning_rate": 3.157894736842105e-06,
      "loss": 1.0109,
      "step": 153
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.811860084533691,
      "learning_rate": 3.1785345717234266e-06,
      "loss": 1.2155,
      "step": 154
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.322361469268799,
      "learning_rate": 3.1991744066047476e-06,
      "loss": 0.9539,
      "step": 155
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.9809648990631104,
      "learning_rate": 3.2198142414860685e-06,
      "loss": 0.8534,
      "step": 156
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.166919231414795,
      "learning_rate": 3.240454076367389e-06,
      "loss": 0.9431,
      "step": 157
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3409974575042725,
      "learning_rate": 3.2610939112487105e-06,
      "loss": 0.8793,
      "step": 158
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7103872299194336,
      "learning_rate": 3.2817337461300315e-06,
      "loss": 0.1821,
      "step": 159
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.791818618774414,
      "learning_rate": 3.302373581011352e-06,
      "loss": 1.0162,
      "step": 160
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.245203018188477,
      "learning_rate": 3.323013415892673e-06,
      "loss": 1.1535,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0914018154144287,
      "learning_rate": 3.343653250773994e-06,
      "loss": 0.9095,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.451181411743164,
      "learning_rate": 3.364293085655315e-06,
      "loss": 1.4022,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.714071750640869,
      "learning_rate": 3.384932920536636e-06,
      "loss": 0.9273,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7420856952667236,
      "learning_rate": 3.405572755417957e-06,
      "loss": 1.01,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.411999702453613,
      "learning_rate": 3.4262125902992775e-06,
      "loss": 1.4142,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.66733980178833,
      "learning_rate": 3.446852425180599e-06,
      "loss": 0.8678,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.329855918884277,
      "learning_rate": 3.46749226006192e-06,
      "loss": 2.0358,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.532449960708618,
      "learning_rate": 3.4881320949432405e-06,
      "loss": 1.2191,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.166656255722046,
      "learning_rate": 3.5087719298245615e-06,
      "loss": 0.8419,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6001129150390625,
      "learning_rate": 3.529411764705883e-06,
      "loss": 0.9922,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4403557777404785,
      "learning_rate": 3.550051599587204e-06,
      "loss": 1.0786,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.126995086669922,
      "learning_rate": 3.5706914344685244e-06,
      "loss": 0.9683,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7295210361480713,
      "learning_rate": 3.5913312693498454e-06,
      "loss": 0.9131,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.292628765106201,
      "learning_rate": 3.611971104231167e-06,
      "loss": 0.8896,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.158903121948242,
      "learning_rate": 3.6326109391124874e-06,
      "loss": 0.9261,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.734338760375977,
      "learning_rate": 3.6532507739938084e-06,
      "loss": 1.508,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.148423433303833,
      "learning_rate": 3.6738906088751294e-06,
      "loss": 1.3157,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.810297966003418,
      "learning_rate": 3.69453044375645e-06,
      "loss": 0.7709,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2125327587127686,
      "learning_rate": 3.7151702786377713e-06,
      "loss": 0.8274,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.194484233856201,
      "learning_rate": 3.7358101135190923e-06,
      "loss": 1.1376,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9719680547714233,
      "learning_rate": 3.756449948400413e-06,
      "loss": 0.8504,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2703850269317627,
      "learning_rate": 3.777089783281734e-06,
      "loss": 0.9895,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3759653568267822,
      "learning_rate": 3.7977296181630553e-06,
      "loss": 0.8748,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9249353408813477,
      "learning_rate": 3.818369453044376e-06,
      "loss": 0.8256,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.91081166267395,
      "learning_rate": 3.839009287925697e-06,
      "loss": 0.8407,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.335546493530273,
      "learning_rate": 3.859649122807018e-06,
      "loss": 1.5301,
      "step": 187
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0702273845672607,
      "learning_rate": 3.880288957688339e-06,
      "loss": 0.8234,
      "step": 188
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.679457187652588,
      "learning_rate": 3.90092879256966e-06,
      "loss": 1.0787,
      "step": 189
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.532447338104248,
      "learning_rate": 3.92156862745098e-06,
      "loss": 0.8003,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.894765377044678,
      "learning_rate": 3.942208462332302e-06,
      "loss": 1.1026,
      "step": 191
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7737975120544434,
      "learning_rate": 3.962848297213623e-06,
      "loss": 0.8627,
      "step": 192
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.156681060791016,
      "learning_rate": 3.983488132094944e-06,
      "loss": 1.0625,
      "step": 193
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.322453498840332,
      "learning_rate": 4.004127966976264e-06,
      "loss": 1.2584,
      "step": 194
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4439775943756104,
      "learning_rate": 4.024767801857586e-06,
      "loss": 0.8815,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.264427185058594,
      "learning_rate": 4.045407636738907e-06,
      "loss": 1.6976,
      "step": 196
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.298707485198975,
      "learning_rate": 4.066047471620228e-06,
      "loss": 1.1706,
      "step": 197
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4422543048858643,
      "learning_rate": 4.086687306501548e-06,
      "loss": 0.8125,
      "step": 198
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3644297122955322,
      "learning_rate": 4.107327141382869e-06,
      "loss": 0.9224,
      "step": 199
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0315279960632324,
      "learning_rate": 4.12796697626419e-06,
      "loss": 1.2661,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.472780466079712,
      "learning_rate": 4.148606811145512e-06,
      "loss": 0.2093,
      "step": 201
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.49786376953125,
      "learning_rate": 4.169246646026832e-06,
      "loss": 0.9411,
      "step": 202
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.092754602432251,
      "learning_rate": 4.189886480908153e-06,
      "loss": 0.8948,
      "step": 203
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2472598552703857,
      "learning_rate": 4.210526315789474e-06,
      "loss": 0.96,
      "step": 204
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.479743480682373,
      "learning_rate": 4.2311661506707955e-06,
      "loss": 1.1443,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.868480682373047,
      "learning_rate": 4.251805985552116e-06,
      "loss": 0.9903,
      "step": 206
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3956146240234375,
      "learning_rate": 4.272445820433437e-06,
      "loss": 1.3741,
      "step": 207
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1167409420013428,
      "learning_rate": 4.293085655314757e-06,
      "loss": 0.8755,
      "step": 208
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.434950113296509,
      "learning_rate": 4.313725490196079e-06,
      "loss": 0.8366,
      "step": 209
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1728010177612305,
      "learning_rate": 4.3343653250774e-06,
      "loss": 0.7873,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.689771890640259,
      "learning_rate": 4.355005159958721e-06,
      "loss": 0.9486,
      "step": 211
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3442420959472656,
      "learning_rate": 4.375644994840041e-06,
      "loss": 0.9676,
      "step": 212
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.354705572128296,
      "learning_rate": 4.3962848297213626e-06,
      "loss": 1.4547,
      "step": 213
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5237979888916016,
      "learning_rate": 4.416924664602684e-06,
      "loss": 0.854,
      "step": 214
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4126832485198975,
      "learning_rate": 4.4375644994840045e-06,
      "loss": 0.8285,
      "step": 215
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8544695377349854,
      "learning_rate": 4.458204334365325e-06,
      "loss": 0.8568,
      "step": 216
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2815592288970947,
      "learning_rate": 4.4788441692466465e-06,
      "loss": 0.7846,
      "step": 217
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.511366367340088,
      "learning_rate": 4.499484004127968e-06,
      "loss": 0.8701,
      "step": 218
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.76605224609375,
      "learning_rate": 4.5201238390092885e-06,
      "loss": 1.05,
      "step": 219
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.480175733566284,
      "learning_rate": 4.540763673890609e-06,
      "loss": 1.3295,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0367016792297363,
      "learning_rate": 4.56140350877193e-06,
      "loss": 1.3931,
      "step": 221
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.752288579940796,
      "learning_rate": 4.582043343653251e-06,
      "loss": 0.9519,
      "step": 222
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.235337734222412,
      "learning_rate": 4.602683178534572e-06,
      "loss": 0.8816,
      "step": 223
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.730571746826172,
      "learning_rate": 4.623323013415893e-06,
      "loss": 0.9514,
      "step": 224
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8050954341888428,
      "learning_rate": 4.6439628482972135e-06,
      "loss": 0.9063,
      "step": 225
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.622385501861572,
      "learning_rate": 4.664602683178535e-06,
      "loss": 0.9881,
      "step": 226
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.52246356010437,
      "learning_rate": 4.685242518059856e-06,
      "loss": 0.9023,
      "step": 227
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2295572757720947,
      "learning_rate": 4.705882352941177e-06,
      "loss": 0.9,
      "step": 228
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1268625259399414,
      "learning_rate": 4.7265221878224975e-06,
      "loss": 0.8807,
      "step": 229
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3187460899353027,
      "learning_rate": 4.747162022703819e-06,
      "loss": 1.1527,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8927950859069824,
      "learning_rate": 4.7678018575851394e-06,
      "loss": 0.8014,
      "step": 231
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.41295862197876,
      "learning_rate": 4.788441692466461e-06,
      "loss": 0.8883,
      "step": 232
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.467696189880371,
      "learning_rate": 4.809081527347781e-06,
      "loss": 1.6395,
      "step": 233
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.642843246459961,
      "learning_rate": 4.829721362229103e-06,
      "loss": 1.7892,
      "step": 234
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.270721912384033,
      "learning_rate": 4.850361197110423e-06,
      "loss": 1.0559,
      "step": 235
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.371065855026245,
      "learning_rate": 4.871001031991745e-06,
      "loss": 1.2958,
      "step": 236
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6573071479797363,
      "learning_rate": 4.891640866873065e-06,
      "loss": 0.8934,
      "step": 237
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2050814628601074,
      "learning_rate": 4.912280701754386e-06,
      "loss": 1.0963,
      "step": 238
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.204519748687744,
      "learning_rate": 4.932920536635707e-06,
      "loss": 0.8031,
      "step": 239
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4568469524383545,
      "learning_rate": 4.953560371517028e-06,
      "loss": 0.8675,
      "step": 240
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.188784122467041,
      "learning_rate": 4.974200206398349e-06,
      "loss": 0.8136,
      "step": 241
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4234769344329834,
      "learning_rate": 4.99484004127967e-06,
      "loss": 1.2518,
      "step": 242
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.490062952041626,
      "learning_rate": 5.015479876160991e-06,
      "loss": 1.0024,
      "step": 243
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.811455011367798,
      "learning_rate": 5.036119711042312e-06,
      "loss": 1.0147,
      "step": 244
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.7155537605285645,
      "learning_rate": 5.056759545923633e-06,
      "loss": 0.8772,
      "step": 245
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.835154056549072,
      "learning_rate": 5.077399380804954e-06,
      "loss": 1.1691,
      "step": 246
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3302066326141357,
      "learning_rate": 5.098039215686274e-06,
      "loss": 0.9653,
      "step": 247
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4761104583740234,
      "learning_rate": 5.118679050567596e-06,
      "loss": 0.8361,
      "step": 248
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.855759859085083,
      "learning_rate": 5.139318885448917e-06,
      "loss": 0.9035,
      "step": 249
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.308495044708252,
      "learning_rate": 5.159958720330238e-06,
      "loss": 1.2238,
      "step": 250
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.489437103271484,
      "learning_rate": 5.180598555211559e-06,
      "loss": 1.4046,
      "step": 251
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.547394037246704,
      "learning_rate": 5.20123839009288e-06,
      "loss": 0.9935,
      "step": 252
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8860406875610352,
      "learning_rate": 5.2218782249742e-06,
      "loss": 0.7879,
      "step": 253
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3788418769836426,
      "learning_rate": 5.242518059855522e-06,
      "loss": 0.851,
      "step": 254
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.303330421447754,
      "learning_rate": 5.263157894736842e-06,
      "loss": 1.1384,
      "step": 255
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.373776912689209,
      "learning_rate": 5.283797729618163e-06,
      "loss": 0.8974,
      "step": 256
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.290968894958496,
      "learning_rate": 5.304437564499485e-06,
      "loss": 0.8264,
      "step": 257
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2738277912139893,
      "learning_rate": 5.325077399380806e-06,
      "loss": 1.0145,
      "step": 258
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4710803031921387,
      "learning_rate": 5.345717234262127e-06,
      "loss": 0.8641,
      "step": 259
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.85501766204834,
      "learning_rate": 5.3663570691434476e-06,
      "loss": 1.6008,
      "step": 260
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0004756450653076,
      "learning_rate": 5.386996904024768e-06,
      "loss": 0.7417,
      "step": 261
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.206768989562988,
      "learning_rate": 5.407636738906089e-06,
      "loss": 1.4839,
      "step": 262
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0550179481506348,
      "learning_rate": 5.42827657378741e-06,
      "loss": 0.826,
      "step": 263
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.25710916519165,
      "learning_rate": 5.448916408668731e-06,
      "loss": 1.2867,
      "step": 264
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.070608139038086,
      "learning_rate": 5.469556243550051e-06,
      "loss": 1.204,
      "step": 265
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.48609447479248,
      "learning_rate": 5.4901960784313735e-06,
      "loss": 1.983,
      "step": 266
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.174840927124023,
      "learning_rate": 5.510835913312694e-06,
      "loss": 1.0407,
      "step": 267
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5282928943634033,
      "learning_rate": 5.5314757481940154e-06,
      "loss": 0.8572,
      "step": 268
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.362003326416016,
      "learning_rate": 5.552115583075336e-06,
      "loss": 1.1792,
      "step": 269
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4138717651367188,
      "learning_rate": 5.5727554179566566e-06,
      "loss": 0.8776,
      "step": 270
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0345277786254883,
      "learning_rate": 5.593395252837978e-06,
      "loss": 0.8123,
      "step": 271
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.986610770225525,
      "learning_rate": 5.6140350877192985e-06,
      "loss": 0.8779,
      "step": 272
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.698418140411377,
      "learning_rate": 5.634674922600619e-06,
      "loss": 0.9895,
      "step": 273
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.243988037109375,
      "learning_rate": 5.655314757481941e-06,
      "loss": 0.815,
      "step": 274
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.330624580383301,
      "learning_rate": 5.675954592363262e-06,
      "loss": 1.0395,
      "step": 275
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2893364429473877,
      "learning_rate": 5.6965944272445825e-06,
      "loss": 1.3761,
      "step": 276
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.2848896980285645,
      "learning_rate": 5.717234262125904e-06,
      "loss": 1.3561,
      "step": 277
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.290637731552124,
      "learning_rate": 5.7378740970072245e-06,
      "loss": 0.8269,
      "step": 278
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.731445789337158,
      "learning_rate": 5.758513931888545e-06,
      "loss": 1.0698,
      "step": 279
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.565122127532959,
      "learning_rate": 5.779153766769866e-06,
      "loss": 1.4359,
      "step": 280
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.208116054534912,
      "learning_rate": 5.799793601651187e-06,
      "loss": 0.8865,
      "step": 281
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.818727731704712,
      "learning_rate": 5.8204334365325075e-06,
      "loss": 0.9927,
      "step": 282
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.312882423400879,
      "learning_rate": 5.84107327141383e-06,
      "loss": 1.3863,
      "step": 283
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2942044734954834,
      "learning_rate": 5.86171310629515e-06,
      "loss": 0.9303,
      "step": 284
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.467881441116333,
      "learning_rate": 5.882352941176471e-06,
      "loss": 1.0068,
      "step": 285
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4276177883148193,
      "learning_rate": 5.902992776057792e-06,
      "loss": 0.7071,
      "step": 286
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8502180576324463,
      "learning_rate": 5.923632610939113e-06,
      "loss": 0.9376,
      "step": 287
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.6037750244140625,
      "learning_rate": 5.9442724458204335e-06,
      "loss": 1.0359,
      "step": 288
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.3249664306640625,
      "learning_rate": 5.964912280701755e-06,
      "loss": 0.88,
      "step": 289
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.62144947052002,
      "learning_rate": 5.985552115583075e-06,
      "loss": 1.5435,
      "step": 290
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.08443546295166,
      "learning_rate": 6.006191950464398e-06,
      "loss": 0.7781,
      "step": 291
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.309182167053223,
      "learning_rate": 6.026831785345718e-06,
      "loss": 1.0599,
      "step": 292
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.035701036453247,
      "learning_rate": 6.047471620227039e-06,
      "loss": 0.8737,
      "step": 293
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3540802001953125,
      "learning_rate": 6.068111455108359e-06,
      "loss": 0.8375,
      "step": 294
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1782968044281006,
      "learning_rate": 6.088751289989681e-06,
      "loss": 0.7441,
      "step": 295
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.09958553314209,
      "learning_rate": 6.109391124871001e-06,
      "loss": 1.0904,
      "step": 296
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.827800750732422,
      "learning_rate": 6.130030959752322e-06,
      "loss": 0.8878,
      "step": 297
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9068882465362549,
      "learning_rate": 6.150670794633643e-06,
      "loss": 0.7677,
      "step": 298
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.131204843521118,
      "learning_rate": 6.171310629514964e-06,
      "loss": 0.8416,
      "step": 299
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4731783866882324,
      "learning_rate": 6.191950464396286e-06,
      "loss": 0.9799,
      "step": 300
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2314236164093018,
      "learning_rate": 6.212590299277607e-06,
      "loss": 0.8919,
      "step": 301
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.29278564453125,
      "learning_rate": 6.233230134158927e-06,
      "loss": 0.9652,
      "step": 302
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1969964504241943,
      "learning_rate": 6.253869969040249e-06,
      "loss": 0.934,
      "step": 303
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.760963439941406,
      "learning_rate": 6.274509803921569e-06,
      "loss": 1.0732,
      "step": 304
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.375929832458496,
      "learning_rate": 6.29514963880289e-06,
      "loss": 1.2437,
      "step": 305
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.861569404602051,
      "learning_rate": 6.31578947368421e-06,
      "loss": 1.03,
      "step": 306
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.488997459411621,
      "learning_rate": 6.336429308565532e-06,
      "loss": 0.876,
      "step": 307
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6058530807495117,
      "learning_rate": 6.357069143446853e-06,
      "loss": 0.9527,
      "step": 308
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9431750774383545,
      "learning_rate": 6.3777089783281746e-06,
      "loss": 0.9036,
      "step": 309
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3260626792907715,
      "learning_rate": 6.398348813209495e-06,
      "loss": 0.9331,
      "step": 310
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7434581518173218,
      "learning_rate": 6.418988648090816e-06,
      "loss": 0.7903,
      "step": 311
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0144248008728027,
      "learning_rate": 6.439628482972137e-06,
      "loss": 0.6606,
      "step": 312
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3015334606170654,
      "learning_rate": 6.460268317853458e-06,
      "loss": 0.2505,
      "step": 313
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.108041286468506,
      "learning_rate": 6.480908152734778e-06,
      "loss": 0.8045,
      "step": 314
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.189596176147461,
      "learning_rate": 6.501547987616099e-06,
      "loss": 0.9239,
      "step": 315
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5117337703704834,
      "learning_rate": 6.522187822497421e-06,
      "loss": 1.0081,
      "step": 316
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9573718309402466,
      "learning_rate": 6.542827657378742e-06,
      "loss": 0.7331,
      "step": 317
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.282013177871704,
      "learning_rate": 6.563467492260063e-06,
      "loss": 1.3607,
      "step": 318
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8821285963058472,
      "learning_rate": 6.5841073271413836e-06,
      "loss": 0.8848,
      "step": 319
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5456414222717285,
      "learning_rate": 6.604747162022704e-06,
      "loss": 0.9462,
      "step": 320
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3612148761749268,
      "learning_rate": 6.6253869969040255e-06,
      "loss": 1.2431,
      "step": 321
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1678802967071533,
      "learning_rate": 6.646026831785346e-06,
      "loss": 0.9339,
      "step": 322
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.077140808105469,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.2657,
      "step": 323
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.425755500793457,
      "learning_rate": 6.687306501547988e-06,
      "loss": 1.4406,
      "step": 324
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.23679256439209,
      "learning_rate": 6.7079463364293095e-06,
      "loss": 0.802,
      "step": 325
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.015200614929199,
      "learning_rate": 6.72858617131063e-06,
      "loss": 1.2286,
      "step": 326
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0655832290649414,
      "learning_rate": 6.7492260061919514e-06,
      "loss": 0.8474,
      "step": 327
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6255805492401123,
      "learning_rate": 6.769865841073272e-06,
      "loss": 0.8572,
      "step": 328
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.77984356880188,
      "learning_rate": 6.7905056759545926e-06,
      "loss": 0.9642,
      "step": 329
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.902426242828369,
      "learning_rate": 6.811145510835914e-06,
      "loss": 1.1251,
      "step": 330
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3367494344711304,
      "learning_rate": 6.8317853457172345e-06,
      "loss": 0.2243,
      "step": 331
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.06036114692688,
      "learning_rate": 6.852425180598555e-06,
      "loss": 0.8425,
      "step": 332
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2773269414901733,
      "learning_rate": 6.873065015479877e-06,
      "loss": 0.2889,
      "step": 333
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5169479846954346,
      "learning_rate": 6.893704850361198e-06,
      "loss": 1.0254,
      "step": 334
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.235578775405884,
      "learning_rate": 6.914344685242519e-06,
      "loss": 0.881,
      "step": 335
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2792482376098633,
      "learning_rate": 6.93498452012384e-06,
      "loss": 0.9369,
      "step": 336
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2300132513046265,
      "learning_rate": 6.9556243550051604e-06,
      "loss": 0.2749,
      "step": 337
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9619669914245605,
      "learning_rate": 6.976264189886481e-06,
      "loss": 1.1383,
      "step": 338
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3334288597106934,
      "learning_rate": 6.996904024767802e-06,
      "loss": 0.9469,
      "step": 339
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9630320072174072,
      "learning_rate": 7.017543859649123e-06,
      "loss": 1.3471,
      "step": 340
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1983423233032227,
      "learning_rate": 7.0381836945304435e-06,
      "loss": 0.8463,
      "step": 341
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3795864582061768,
      "learning_rate": 7.058823529411766e-06,
      "loss": 0.927,
      "step": 342
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.163703203201294,
      "learning_rate": 7.079463364293086e-06,
      "loss": 0.9644,
      "step": 343
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7443466186523438,
      "learning_rate": 7.100103199174408e-06,
      "loss": 0.869,
      "step": 344
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.100193977355957,
      "learning_rate": 7.120743034055728e-06,
      "loss": 0.5197,
      "step": 345
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2442145347595215,
      "learning_rate": 7.141382868937049e-06,
      "loss": 0.8867,
      "step": 346
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2126035690307617,
      "learning_rate": 7.1620227038183694e-06,
      "loss": 0.8389,
      "step": 347
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.924902439117432,
      "learning_rate": 7.182662538699691e-06,
      "loss": 1.2376,
      "step": 348
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5713391304016113,
      "learning_rate": 7.203302373581011e-06,
      "loss": 0.9434,
      "step": 349
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3302741050720215,
      "learning_rate": 7.223942208462334e-06,
      "loss": 0.9002,
      "step": 350
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.1899824142456055,
      "learning_rate": 7.244582043343654e-06,
      "loss": 0.8839,
      "step": 351
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.0494279861450195,
      "learning_rate": 7.265221878224975e-06,
      "loss": 1.0508,
      "step": 352
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.384840488433838,
      "learning_rate": 7.285861713106296e-06,
      "loss": 0.9381,
      "step": 353
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.132380723953247,
      "learning_rate": 7.306501547987617e-06,
      "loss": 0.8267,
      "step": 354
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9253169298171997,
      "learning_rate": 7.327141382868937e-06,
      "loss": 0.8652,
      "step": 355
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9077160358428955,
      "learning_rate": 7.347781217750259e-06,
      "loss": 0.8404,
      "step": 356
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.6819748878479,
      "learning_rate": 7.368421052631579e-06,
      "loss": 0.9659,
      "step": 357
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5425972938537598,
      "learning_rate": 7.3890608875129e-06,
      "loss": 0.8271,
      "step": 358
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2720842361450195,
      "learning_rate": 7.409700722394222e-06,
      "loss": 0.8416,
      "step": 359
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.158013343811035,
      "learning_rate": 7.430340557275543e-06,
      "loss": 1.0641,
      "step": 360
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.572980880737305,
      "learning_rate": 7.450980392156863e-06,
      "loss": 0.8206,
      "step": 361
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.07082200050354,
      "learning_rate": 7.471620227038185e-06,
      "loss": 0.7653,
      "step": 362
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.265717029571533,
      "learning_rate": 7.492260061919505e-06,
      "loss": 0.8364,
      "step": 363
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.053001403808594,
      "learning_rate": 7.512899896800826e-06,
      "loss": 0.9908,
      "step": 364
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.728774070739746,
      "learning_rate": 7.533539731682147e-06,
      "loss": 1.4005,
      "step": 365
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.484060764312744,
      "learning_rate": 7.554179566563468e-06,
      "loss": 0.9385,
      "step": 366
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3029589653015137,
      "learning_rate": 7.57481940144479e-06,
      "loss": 0.7965,
      "step": 367
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.382920265197754,
      "learning_rate": 7.5954592363261105e-06,
      "loss": 0.6099,
      "step": 368
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0114738941192627,
      "learning_rate": 7.616099071207431e-06,
      "loss": 0.8875,
      "step": 369
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.021771192550659,
      "learning_rate": 7.636738906088752e-06,
      "loss": 1.2095,
      "step": 370
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.062788486480713,
      "learning_rate": 7.657378740970072e-06,
      "loss": 0.7563,
      "step": 371
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.478100299835205,
      "learning_rate": 7.678018575851394e-06,
      "loss": 0.8963,
      "step": 372
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.038093090057373,
      "learning_rate": 7.698658410732715e-06,
      "loss": 0.9001,
      "step": 373
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5237808227539062,
      "learning_rate": 7.719298245614036e-06,
      "loss": 0.735,
      "step": 374
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.254703998565674,
      "learning_rate": 7.739938080495358e-06,
      "loss": 0.8512,
      "step": 375
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.693696022033691,
      "learning_rate": 7.760577915376678e-06,
      "loss": 1.7353,
      "step": 376
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6659436225891113,
      "learning_rate": 7.781217750257999e-06,
      "loss": 0.9656,
      "step": 377
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.462066650390625,
      "learning_rate": 7.80185758513932e-06,
      "loss": 0.8435,
      "step": 378
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0737125873565674,
      "learning_rate": 7.82249742002064e-06,
      "loss": 0.7673,
      "step": 379
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.560333251953125,
      "learning_rate": 7.84313725490196e-06,
      "loss": 0.8822,
      "step": 380
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2976932525634766,
      "learning_rate": 7.863777089783283e-06,
      "loss": 0.8735,
      "step": 381
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0386619567871094,
      "learning_rate": 7.884416924664603e-06,
      "loss": 0.8458,
      "step": 382
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9179277420043945,
      "learning_rate": 7.905056759545924e-06,
      "loss": 0.851,
      "step": 383
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3947057723999023,
      "learning_rate": 7.925696594427246e-06,
      "loss": 0.8851,
      "step": 384
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9864063262939453,
      "learning_rate": 7.946336429308567e-06,
      "loss": 0.8578,
      "step": 385
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7051559686660767,
      "learning_rate": 7.966976264189887e-06,
      "loss": 0.7431,
      "step": 386
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.832138776779175,
      "learning_rate": 7.987616099071208e-06,
      "loss": 0.9594,
      "step": 387
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9499704837799072,
      "learning_rate": 8.008255933952529e-06,
      "loss": 0.7854,
      "step": 388
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.717242240905762,
      "learning_rate": 8.028895768833849e-06,
      "loss": 1.2027,
      "step": 389
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.507983446121216,
      "learning_rate": 8.049535603715171e-06,
      "loss": 1.045,
      "step": 390
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2987607717514038,
      "learning_rate": 8.070175438596492e-06,
      "loss": 0.1996,
      "step": 391
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.285400390625,
      "learning_rate": 8.090815273477814e-06,
      "loss": 0.8179,
      "step": 392
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2222065925598145,
      "learning_rate": 8.111455108359135e-06,
      "loss": 1.3082,
      "step": 393
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7509393692016602,
      "learning_rate": 8.132094943240455e-06,
      "loss": 0.6657,
      "step": 394
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.214837074279785,
      "learning_rate": 8.152734778121776e-06,
      "loss": 1.0336,
      "step": 395
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3009932041168213,
      "learning_rate": 8.173374613003096e-06,
      "loss": 0.946,
      "step": 396
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.168182134628296,
      "learning_rate": 8.194014447884417e-06,
      "loss": 0.906,
      "step": 397
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9929451942443848,
      "learning_rate": 8.214654282765738e-06,
      "loss": 0.7822,
      "step": 398
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.025161266326904,
      "learning_rate": 8.23529411764706e-06,
      "loss": 1.3595,
      "step": 399
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0429751873016357,
      "learning_rate": 8.25593395252838e-06,
      "loss": 0.7981,
      "step": 400
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.955232620239258,
      "learning_rate": 8.276573787409703e-06,
      "loss": 0.9553,
      "step": 401
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.416951894760132,
      "learning_rate": 8.297213622291023e-06,
      "loss": 0.9466,
      "step": 402
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0307815074920654,
      "learning_rate": 8.317853457172344e-06,
      "loss": 0.8354,
      "step": 403
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.311262130737305,
      "learning_rate": 8.338493292053664e-06,
      "loss": 1.2822,
      "step": 404
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4545369148254395,
      "learning_rate": 8.359133126934985e-06,
      "loss": 0.8024,
      "step": 405
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9260571002960205,
      "learning_rate": 8.379772961816305e-06,
      "loss": 0.778,
      "step": 406
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6735445261001587,
      "learning_rate": 8.400412796697626e-06,
      "loss": 0.7624,
      "step": 407
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3167941570281982,
      "learning_rate": 8.421052631578948e-06,
      "loss": 1.0899,
      "step": 408
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2326719760894775,
      "learning_rate": 8.441692466460269e-06,
      "loss": 0.8186,
      "step": 409
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.099907159805298,
      "learning_rate": 8.462332301341591e-06,
      "loss": 0.7799,
      "step": 410
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9985566139221191,
      "learning_rate": 8.482972136222912e-06,
      "loss": 0.7904,
      "step": 411
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9661548137664795,
      "learning_rate": 8.503611971104232e-06,
      "loss": 0.8284,
      "step": 412
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0828280448913574,
      "learning_rate": 8.524251805985553e-06,
      "loss": 0.9942,
      "step": 413
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.051852226257324,
      "learning_rate": 8.544891640866873e-06,
      "loss": 1.1277,
      "step": 414
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.039681911468506,
      "learning_rate": 8.565531475748194e-06,
      "loss": 1.3854,
      "step": 415
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2055606842041016,
      "learning_rate": 8.586171310629514e-06,
      "loss": 0.883,
      "step": 416
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8504616022109985,
      "learning_rate": 8.606811145510837e-06,
      "loss": 0.8508,
      "step": 417
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2073280811309814,
      "learning_rate": 8.627450980392157e-06,
      "loss": 0.2087,
      "step": 418
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0111279487609863,
      "learning_rate": 8.64809081527348e-06,
      "loss": 0.8322,
      "step": 419
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6424567699432373,
      "learning_rate": 8.6687306501548e-06,
      "loss": 0.879,
      "step": 420
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.133761167526245,
      "learning_rate": 8.68937048503612e-06,
      "loss": 0.8406,
      "step": 421
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.496655225753784,
      "learning_rate": 8.710010319917441e-06,
      "loss": 0.8414,
      "step": 422
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.989524841308594,
      "learning_rate": 8.730650154798762e-06,
      "loss": 0.8361,
      "step": 423
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.11848783493042,
      "learning_rate": 8.751289989680082e-06,
      "loss": 0.8208,
      "step": 424
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9416618347167969,
      "learning_rate": 8.771929824561405e-06,
      "loss": 0.8544,
      "step": 425
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.966656684875488,
      "learning_rate": 8.792569659442725e-06,
      "loss": 1.2459,
      "step": 426
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9632823467254639,
      "learning_rate": 8.813209494324046e-06,
      "loss": 0.8014,
      "step": 427
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1590057611465454,
      "learning_rate": 8.833849329205368e-06,
      "loss": 0.2026,
      "step": 428
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.934181809425354,
      "learning_rate": 8.854489164086688e-06,
      "loss": 0.7931,
      "step": 429
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.027122974395752,
      "learning_rate": 8.875128998968009e-06,
      "loss": 1.2264,
      "step": 430
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5689454078674316,
      "learning_rate": 8.89576883384933e-06,
      "loss": 0.8684,
      "step": 431
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.021468162536621,
      "learning_rate": 8.91640866873065e-06,
      "loss": 1.1612,
      "step": 432
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.6894354820251465,
      "learning_rate": 8.93704850361197e-06,
      "loss": 1.1578,
      "step": 433
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.030055999755859,
      "learning_rate": 8.957688338493293e-06,
      "loss": 0.9502,
      "step": 434
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2254951000213623,
      "learning_rate": 8.978328173374614e-06,
      "loss": 0.8855,
      "step": 435
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.327990770339966,
      "learning_rate": 8.998968008255936e-06,
      "loss": 0.8564,
      "step": 436
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.035968780517578,
      "learning_rate": 9.019607843137256e-06,
      "loss": 0.8592,
      "step": 437
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.820474624633789,
      "learning_rate": 9.040247678018577e-06,
      "loss": 1.4671,
      "step": 438
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.429450035095215,
      "learning_rate": 9.060887512899897e-06,
      "loss": 0.8295,
      "step": 439
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4079408645629883,
      "learning_rate": 9.081527347781218e-06,
      "loss": 0.7937,
      "step": 440
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9770573377609253,
      "learning_rate": 9.102167182662539e-06,
      "loss": 0.7417,
      "step": 441
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.513748168945312,
      "learning_rate": 9.12280701754386e-06,
      "loss": 1.5767,
      "step": 442
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.997674584388733,
      "learning_rate": 9.143446852425181e-06,
      "loss": 0.8076,
      "step": 443
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.064607858657837,
      "learning_rate": 9.164086687306502e-06,
      "loss": 0.8409,
      "step": 444
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.088892459869385,
      "learning_rate": 9.184726522187824e-06,
      "loss": 1.3532,
      "step": 445
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1655526161193848,
      "learning_rate": 9.205366357069145e-06,
      "loss": 0.7735,
      "step": 446
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7030978202819824,
      "learning_rate": 9.226006191950465e-06,
      "loss": 0.8891,
      "step": 447
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.711034297943115,
      "learning_rate": 9.246646026831786e-06,
      "loss": 0.8833,
      "step": 448
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2394826412200928,
      "learning_rate": 9.267285861713106e-06,
      "loss": 0.815,
      "step": 449
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.663512706756592,
      "learning_rate": 9.287925696594427e-06,
      "loss": 0.8156,
      "step": 450
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.43387770652771,
      "learning_rate": 9.30856553147575e-06,
      "loss": 0.9134,
      "step": 451
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.18117094039917,
      "learning_rate": 9.32920536635707e-06,
      "loss": 0.8789,
      "step": 452
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.947688102722168,
      "learning_rate": 9.34984520123839e-06,
      "loss": 1.0815,
      "step": 453
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.946306586265564,
      "learning_rate": 9.370485036119713e-06,
      "loss": 0.767,
      "step": 454
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3249911069869995,
      "learning_rate": 9.391124871001033e-06,
      "loss": 0.1875,
      "step": 455
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5682454109191895,
      "learning_rate": 9.411764705882354e-06,
      "loss": 0.7209,
      "step": 456
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2556345462799072,
      "learning_rate": 9.432404540763674e-06,
      "loss": 0.8171,
      "step": 457
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.595491409301758,
      "learning_rate": 9.453044375644995e-06,
      "loss": 0.8852,
      "step": 458
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.028874397277832,
      "learning_rate": 9.473684210526315e-06,
      "loss": 0.9975,
      "step": 459
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.062005043029785,
      "learning_rate": 9.494324045407638e-06,
      "loss": 0.8402,
      "step": 460
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.454856872558594,
      "learning_rate": 9.514963880288958e-06,
      "loss": 1.1188,
      "step": 461
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.105121612548828,
      "learning_rate": 9.535603715170279e-06,
      "loss": 1.6014,
      "step": 462
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2930235862731934,
      "learning_rate": 9.556243550051601e-06,
      "loss": 0.8361,
      "step": 463
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.007819175720215,
      "learning_rate": 9.576883384932922e-06,
      "loss": 1.0109,
      "step": 464
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.708876132965088,
      "learning_rate": 9.597523219814242e-06,
      "loss": 1.1862,
      "step": 465
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5368728637695312,
      "learning_rate": 9.618163054695563e-06,
      "loss": 0.9218,
      "step": 466
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.245659828186035,
      "learning_rate": 9.638802889576883e-06,
      "loss": 0.8073,
      "step": 467
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5120251178741455,
      "learning_rate": 9.659442724458206e-06,
      "loss": 0.9137,
      "step": 468
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.918213129043579,
      "learning_rate": 9.680082559339526e-06,
      "loss": 1.2233,
      "step": 469
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.192105770111084,
      "learning_rate": 9.700722394220847e-06,
      "loss": 0.904,
      "step": 470
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0452046394348145,
      "learning_rate": 9.721362229102167e-06,
      "loss": 0.8209,
      "step": 471
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4177074432373047,
      "learning_rate": 9.74200206398349e-06,
      "loss": 0.9903,
      "step": 472
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.126894950866699,
      "learning_rate": 9.76264189886481e-06,
      "loss": 0.8859,
      "step": 473
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2687416076660156,
      "learning_rate": 9.78328173374613e-06,
      "loss": 1.2794,
      "step": 474
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.729794979095459,
      "learning_rate": 9.803921568627451e-06,
      "loss": 0.7877,
      "step": 475
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9461151361465454,
      "learning_rate": 9.824561403508772e-06,
      "loss": 0.8302,
      "step": 476
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.161412477493286,
      "learning_rate": 9.845201238390094e-06,
      "loss": 0.7282,
      "step": 477
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2456846237182617,
      "learning_rate": 9.865841073271415e-06,
      "loss": 0.7534,
      "step": 478
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0521328449249268,
      "learning_rate": 9.886480908152735e-06,
      "loss": 1.0994,
      "step": 479
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5295674800872803,
      "learning_rate": 9.907120743034056e-06,
      "loss": 0.8032,
      "step": 480
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.964946746826172,
      "learning_rate": 9.927760577915378e-06,
      "loss": 1.0871,
      "step": 481
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.246405601501465,
      "learning_rate": 9.948400412796699e-06,
      "loss": 0.8927,
      "step": 482
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2223660945892334,
      "learning_rate": 9.969040247678019e-06,
      "loss": 0.8315,
      "step": 483
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.398897171020508,
      "learning_rate": 9.98968008255934e-06,
      "loss": 1.0705,
      "step": 484
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.4833059310913086,
      "learning_rate": 1.001031991744066e-05,
      "loss": 1.0725,
      "step": 485
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4536335468292236,
      "learning_rate": 1.0030959752321983e-05,
      "loss": 0.7804,
      "step": 486
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.166163921356201,
      "learning_rate": 1.0051599587203303e-05,
      "loss": 0.924,
      "step": 487
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.238274335861206,
      "learning_rate": 1.0072239422084624e-05,
      "loss": 0.8161,
      "step": 488
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.412026882171631,
      "learning_rate": 1.0092879256965946e-05,
      "loss": 0.8488,
      "step": 489
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8876681327819824,
      "learning_rate": 1.0113519091847266e-05,
      "loss": 1.1661,
      "step": 490
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.626690149307251,
      "learning_rate": 1.0134158926728587e-05,
      "loss": 1.0145,
      "step": 491
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.63667631149292,
      "learning_rate": 1.0154798761609908e-05,
      "loss": 1.1167,
      "step": 492
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.954810857772827,
      "learning_rate": 1.017543859649123e-05,
      "loss": 0.9265,
      "step": 493
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.602902889251709,
      "learning_rate": 1.0196078431372549e-05,
      "loss": 0.8423,
      "step": 494
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0218653678894043,
      "learning_rate": 1.0216718266253871e-05,
      "loss": 0.886,
      "step": 495
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.221170663833618,
      "learning_rate": 1.0237358101135192e-05,
      "loss": 0.7851,
      "step": 496
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8215280771255493,
      "learning_rate": 1.0257997936016512e-05,
      "loss": 0.7786,
      "step": 497
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.654120922088623,
      "learning_rate": 1.0278637770897834e-05,
      "loss": 1.1123,
      "step": 498
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2662657499313354,
      "learning_rate": 1.0299277605779155e-05,
      "loss": 0.3369,
      "step": 499
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.702775955200195,
      "learning_rate": 1.0319917440660475e-05,
      "loss": 1.0371,
      "step": 500
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.29379940032959,
      "learning_rate": 1.0340557275541796e-05,
      "loss": 0.7836,
      "step": 501
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.193016529083252,
      "learning_rate": 1.0361197110423118e-05,
      "loss": 1.1053,
      "step": 502
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.522302627563477,
      "learning_rate": 1.0381836945304437e-05,
      "loss": 0.8395,
      "step": 503
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.237677812576294,
      "learning_rate": 1.040247678018576e-05,
      "loss": 0.9078,
      "step": 504
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.281116485595703,
      "learning_rate": 1.042311661506708e-05,
      "loss": 0.7845,
      "step": 505
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7226152420043945,
      "learning_rate": 1.04437564499484e-05,
      "loss": 1.0539,
      "step": 506
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9174081087112427,
      "learning_rate": 1.0464396284829723e-05,
      "loss": 0.9279,
      "step": 507
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0828797817230225,
      "learning_rate": 1.0485036119711043e-05,
      "loss": 0.9289,
      "step": 508
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7258973121643066,
      "learning_rate": 1.0505675954592366e-05,
      "loss": 0.9154,
      "step": 509
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.7252037525177,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 0.863,
      "step": 510
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.301940441131592,
      "learning_rate": 1.0546955624355007e-05,
      "loss": 0.5637,
      "step": 511
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2585761547088623,
      "learning_rate": 1.0567595459236326e-05,
      "loss": 0.9813,
      "step": 512
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.921789169311523,
      "learning_rate": 1.0588235294117648e-05,
      "loss": 0.9736,
      "step": 513
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.904989719390869,
      "learning_rate": 1.060887512899897e-05,
      "loss": 1.0517,
      "step": 514
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4612571001052856,
      "learning_rate": 1.0629514963880289e-05,
      "loss": 0.2095,
      "step": 515
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.579632759094238,
      "learning_rate": 1.0650154798761611e-05,
      "loss": 0.8695,
      "step": 516
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.301539421081543,
      "learning_rate": 1.0670794633642932e-05,
      "loss": 0.9027,
      "step": 517
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8422813415527344,
      "learning_rate": 1.0691434468524254e-05,
      "loss": 0.7855,
      "step": 518
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.682985305786133,
      "learning_rate": 1.0712074303405573e-05,
      "loss": 0.9787,
      "step": 519
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.002133846282959,
      "learning_rate": 1.0732714138286895e-05,
      "loss": 0.836,
      "step": 520
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1815173625946045,
      "learning_rate": 1.0753353973168214e-05,
      "loss": 0.9645,
      "step": 521
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4090397357940674,
      "learning_rate": 1.0773993808049536e-05,
      "loss": 0.8355,
      "step": 522
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5252296924591064,
      "learning_rate": 1.0794633642930859e-05,
      "loss": 0.8601,
      "step": 523
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9689760208129883,
      "learning_rate": 1.0815273477812177e-05,
      "loss": 0.8978,
      "step": 524
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3681801557540894,
      "learning_rate": 1.08359133126935e-05,
      "loss": 0.282,
      "step": 525
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9549200534820557,
      "learning_rate": 1.085655314757482e-05,
      "loss": 1.025,
      "step": 526
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3207664489746094,
      "learning_rate": 1.0877192982456142e-05,
      "loss": 0.9025,
      "step": 527
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.229623317718506,
      "learning_rate": 1.0897832817337461e-05,
      "loss": 1.6304,
      "step": 528
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9299676418304443,
      "learning_rate": 1.0918472652218784e-05,
      "loss": 0.7761,
      "step": 529
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.462193727493286,
      "learning_rate": 1.0939112487100102e-05,
      "loss": 0.8148,
      "step": 530
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2263103723526,
      "learning_rate": 1.0959752321981425e-05,
      "loss": 0.2202,
      "step": 531
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.437952756881714,
      "learning_rate": 1.0980392156862747e-05,
      "loss": 0.7681,
      "step": 532
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9940085411071777,
      "learning_rate": 1.1001031991744066e-05,
      "loss": 0.9448,
      "step": 533
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2840688228607178,
      "learning_rate": 1.1021671826625388e-05,
      "loss": 0.8925,
      "step": 534
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2060329914093018,
      "learning_rate": 1.1042311661506709e-05,
      "loss": 0.8327,
      "step": 535
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.333324432373047,
      "learning_rate": 1.1062951496388031e-05,
      "loss": 1.3513,
      "step": 536
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.961103439331055,
      "learning_rate": 1.108359133126935e-05,
      "loss": 0.9162,
      "step": 537
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2090816497802734,
      "learning_rate": 1.1104231166150672e-05,
      "loss": 0.6963,
      "step": 538
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7619309425354004,
      "learning_rate": 1.1124871001031991e-05,
      "loss": 0.861,
      "step": 539
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.271238803863525,
      "learning_rate": 1.1145510835913313e-05,
      "loss": 1.4724,
      "step": 540
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.899474143981934,
      "learning_rate": 1.1166150670794635e-05,
      "loss": 1.3385,
      "step": 541
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.440961837768555,
      "learning_rate": 1.1186790505675956e-05,
      "loss": 1.5643,
      "step": 542
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.402846336364746,
      "learning_rate": 1.1207430340557277e-05,
      "loss": 1.057,
      "step": 543
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.750955820083618,
      "learning_rate": 1.1228070175438597e-05,
      "loss": 0.9669,
      "step": 544
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.435425043106079,
      "learning_rate": 1.124871001031992e-05,
      "loss": 0.2702,
      "step": 545
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.208491802215576,
      "learning_rate": 1.1269349845201238e-05,
      "loss": 0.8007,
      "step": 546
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.008247375488281,
      "learning_rate": 1.128998968008256e-05,
      "loss": 1.324,
      "step": 547
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9660873413085938,
      "learning_rate": 1.1310629514963883e-05,
      "loss": 0.8764,
      "step": 548
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.504286050796509,
      "learning_rate": 1.1331269349845202e-05,
      "loss": 0.8152,
      "step": 549
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3568620681762695,
      "learning_rate": 1.1351909184726524e-05,
      "loss": 0.6238,
      "step": 550
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.815100908279419,
      "learning_rate": 1.1372549019607844e-05,
      "loss": 1.2317,
      "step": 551
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.569467067718506,
      "learning_rate": 1.1393188854489165e-05,
      "loss": 1.2647,
      "step": 552
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9206604957580566,
      "learning_rate": 1.1413828689370486e-05,
      "loss": 0.7987,
      "step": 553
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.4559504985809326,
      "learning_rate": 1.1434468524251808e-05,
      "loss": 0.8114,
      "step": 554
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0717923641204834,
      "learning_rate": 1.1455108359133127e-05,
      "loss": 0.8623,
      "step": 555
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.085383892059326,
      "learning_rate": 1.1475748194014449e-05,
      "loss": 0.8314,
      "step": 556
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.099616765975952,
      "learning_rate": 1.1496388028895771e-05,
      "loss": 0.8533,
      "step": 557
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8819007873535156,
      "learning_rate": 1.151702786377709e-05,
      "loss": 0.8929,
      "step": 558
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8509879112243652,
      "learning_rate": 1.1537667698658412e-05,
      "loss": 0.9518,
      "step": 559
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4528279304504395,
      "learning_rate": 1.1558307533539733e-05,
      "loss": 0.9729,
      "step": 560
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.794097423553467,
      "learning_rate": 1.1578947368421053e-05,
      "loss": 1.5098,
      "step": 561
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3197436332702637,
      "learning_rate": 1.1599587203302374e-05,
      "loss": 1.0699,
      "step": 562
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0615828037261963,
      "learning_rate": 1.1620227038183696e-05,
      "loss": 1.3573,
      "step": 563
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7535152435302734,
      "learning_rate": 1.1640866873065015e-05,
      "loss": 0.9882,
      "step": 564
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.763131618499756,
      "learning_rate": 1.1661506707946337e-05,
      "loss": 0.7727,
      "step": 565
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0765397548675537,
      "learning_rate": 1.168214654282766e-05,
      "loss": 1.2391,
      "step": 566
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.050936222076416,
      "learning_rate": 1.1702786377708978e-05,
      "loss": 0.8043,
      "step": 567
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1477434635162354,
      "learning_rate": 1.17234262125903e-05,
      "loss": 0.7694,
      "step": 568
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1560885906219482,
      "learning_rate": 1.1744066047471621e-05,
      "loss": 0.8314,
      "step": 569
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9780583381652832,
      "learning_rate": 1.1764705882352942e-05,
      "loss": 0.8117,
      "step": 570
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.105690956115723,
      "learning_rate": 1.1785345717234262e-05,
      "loss": 1.0769,
      "step": 571
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.72727370262146,
      "learning_rate": 1.1805985552115585e-05,
      "loss": 0.882,
      "step": 572
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.390698432922363,
      "learning_rate": 1.1826625386996907e-05,
      "loss": 1.2592,
      "step": 573
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.330127716064453,
      "learning_rate": 1.1847265221878226e-05,
      "loss": 0.9543,
      "step": 574
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.335740566253662,
      "learning_rate": 1.1867905056759548e-05,
      "loss": 0.7355,
      "step": 575
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.685586929321289,
      "learning_rate": 1.1888544891640867e-05,
      "loss": 1.1209,
      "step": 576
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0079174041748047,
      "learning_rate": 1.190918472652219e-05,
      "loss": 0.8824,
      "step": 577
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.334710121154785,
      "learning_rate": 1.192982456140351e-05,
      "loss": 0.947,
      "step": 578
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.059558153152466,
      "learning_rate": 1.195046439628483e-05,
      "loss": 0.9419,
      "step": 579
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.694844245910645,
      "learning_rate": 1.197110423116615e-05,
      "loss": 1.639,
      "step": 580
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7198328971862793,
      "learning_rate": 1.1991744066047473e-05,
      "loss": 1.0404,
      "step": 581
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.874852180480957,
      "learning_rate": 1.2012383900928795e-05,
      "loss": 1.344,
      "step": 582
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.290665626525879,
      "learning_rate": 1.2033023735810114e-05,
      "loss": 0.9368,
      "step": 583
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.964677095413208,
      "learning_rate": 1.2053663570691436e-05,
      "loss": 0.9907,
      "step": 584
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.6431379318237305,
      "learning_rate": 1.2074303405572755e-05,
      "loss": 0.9966,
      "step": 585
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8766043186187744,
      "learning_rate": 1.2094943240454078e-05,
      "loss": 1.0478,
      "step": 586
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1363134384155273,
      "learning_rate": 1.2115583075335398e-05,
      "loss": 0.8001,
      "step": 587
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.8629021644592285,
      "learning_rate": 1.2136222910216719e-05,
      "loss": 1.3593,
      "step": 588
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.479402542114258,
      "learning_rate": 1.215686274509804e-05,
      "loss": 1.5566,
      "step": 589
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.059048652648926,
      "learning_rate": 1.2177502579979362e-05,
      "loss": 0.8048,
      "step": 590
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.632474184036255,
      "learning_rate": 1.2198142414860684e-05,
      "loss": 0.9712,
      "step": 591
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2955644130706787,
      "learning_rate": 1.2218782249742003e-05,
      "loss": 0.799,
      "step": 592
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.434901237487793,
      "learning_rate": 1.2239422084623325e-05,
      "loss": 1.7145,
      "step": 593
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.485126256942749,
      "learning_rate": 1.2260061919504644e-05,
      "loss": 1.0507,
      "step": 594
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.790374279022217,
      "learning_rate": 1.2280701754385966e-05,
      "loss": 0.8965,
      "step": 595
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.355168104171753,
      "learning_rate": 1.2301341589267287e-05,
      "loss": 0.8786,
      "step": 596
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.412199974060059,
      "learning_rate": 1.2321981424148607e-05,
      "loss": 0.8904,
      "step": 597
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.140639066696167,
      "learning_rate": 1.2342621259029928e-05,
      "loss": 0.9817,
      "step": 598
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3510262966156006,
      "learning_rate": 1.236326109391125e-05,
      "loss": 0.9384,
      "step": 599
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.58213472366333,
      "learning_rate": 1.2383900928792572e-05,
      "loss": 0.9245,
      "step": 600
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1175804138183594,
      "learning_rate": 1.2404540763673891e-05,
      "loss": 0.8734,
      "step": 601
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.02022933959961,
      "learning_rate": 1.2425180598555213e-05,
      "loss": 1.7758,
      "step": 602
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.176136493682861,
      "learning_rate": 1.2445820433436532e-05,
      "loss": 1.1956,
      "step": 603
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9297196865081787,
      "learning_rate": 1.2466460268317854e-05,
      "loss": 0.9013,
      "step": 604
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8722550868988037,
      "learning_rate": 1.2487100103199175e-05,
      "loss": 0.9208,
      "step": 605
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0798001289367676,
      "learning_rate": 1.2507739938080497e-05,
      "loss": 0.7908,
      "step": 606
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0911556482315063,
      "learning_rate": 1.2528379772961818e-05,
      "loss": 0.1787,
      "step": 607
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2343039512634277,
      "learning_rate": 1.2549019607843138e-05,
      "loss": 0.9019,
      "step": 608
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4921445846557617,
      "learning_rate": 1.256965944272446e-05,
      "loss": 0.8191,
      "step": 609
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.972806692123413,
      "learning_rate": 1.259029927760578e-05,
      "loss": 0.854,
      "step": 610
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2201857566833496,
      "learning_rate": 1.2610939112487102e-05,
      "loss": 0.8449,
      "step": 611
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0410780906677246,
      "learning_rate": 1.263157894736842e-05,
      "loss": 1.1397,
      "step": 612
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.177655220031738,
      "learning_rate": 1.2652218782249743e-05,
      "loss": 0.978,
      "step": 613
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.195432186126709,
      "learning_rate": 1.2672858617131063e-05,
      "loss": 0.9398,
      "step": 614
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.723642110824585,
      "learning_rate": 1.2693498452012386e-05,
      "loss": 1.0082,
      "step": 615
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5743868350982666,
      "learning_rate": 1.2714138286893706e-05,
      "loss": 0.8332,
      "step": 616
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.871006965637207,
      "learning_rate": 1.2734778121775027e-05,
      "loss": 0.7221,
      "step": 617
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9153144359588623,
      "learning_rate": 1.2755417956656349e-05,
      "loss": 1.1983,
      "step": 618
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9911627769470215,
      "learning_rate": 1.2776057791537668e-05,
      "loss": 0.7773,
      "step": 619
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3505163192749023,
      "learning_rate": 1.279669762641899e-05,
      "loss": 0.2597,
      "step": 620
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.952637195587158,
      "learning_rate": 1.2817337461300309e-05,
      "loss": 1.3993,
      "step": 621
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5087759494781494,
      "learning_rate": 1.2837977296181631e-05,
      "loss": 0.8161,
      "step": 622
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.296115398406982,
      "learning_rate": 1.2858617131062952e-05,
      "loss": 0.9159,
      "step": 623
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.040927410125732,
      "learning_rate": 1.2879256965944274e-05,
      "loss": 1.2147,
      "step": 624
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9400643110275269,
      "learning_rate": 1.2899896800825595e-05,
      "loss": 0.7712,
      "step": 625
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7022438049316406,
      "learning_rate": 1.2920536635706915e-05,
      "loss": 0.9269,
      "step": 626
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8375781774520874,
      "learning_rate": 1.2941176470588238e-05,
      "loss": 0.7924,
      "step": 627
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.17718505859375,
      "learning_rate": 1.2961816305469556e-05,
      "loss": 0.9542,
      "step": 628
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0840229988098145,
      "learning_rate": 1.2982456140350879e-05,
      "loss": 0.7977,
      "step": 629
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.049139976501465,
      "learning_rate": 1.3003095975232198e-05,
      "loss": 0.7564,
      "step": 630
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7735207080841064,
      "learning_rate": 1.302373581011352e-05,
      "loss": 1.2734,
      "step": 631
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.438393592834473,
      "learning_rate": 1.3044375644994842e-05,
      "loss": 1.0803,
      "step": 632
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.197047710418701,
      "learning_rate": 1.3065015479876163e-05,
      "loss": 0.9068,
      "step": 633
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.051522254943848,
      "learning_rate": 1.3085655314757483e-05,
      "loss": 1.0637,
      "step": 634
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.439009666442871,
      "learning_rate": 1.3106295149638804e-05,
      "loss": 0.8903,
      "step": 635
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0076725482940674,
      "learning_rate": 1.3126934984520126e-05,
      "loss": 0.7949,
      "step": 636
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.301661729812622,
      "learning_rate": 1.3147574819401445e-05,
      "loss": 0.9443,
      "step": 637
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.990489959716797,
      "learning_rate": 1.3168214654282767e-05,
      "loss": 0.9104,
      "step": 638
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.974132776260376,
      "learning_rate": 1.3188854489164088e-05,
      "loss": 0.7562,
      "step": 639
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3714828491210938,
      "learning_rate": 1.3209494324045408e-05,
      "loss": 0.9412,
      "step": 640
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9513351917266846,
      "learning_rate": 1.323013415892673e-05,
      "loss": 0.7646,
      "step": 641
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.303224086761475,
      "learning_rate": 1.3250773993808051e-05,
      "loss": 1.4076,
      "step": 642
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.464000701904297,
      "learning_rate": 1.3271413828689372e-05,
      "loss": 0.82,
      "step": 643
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.11598014831543,
      "learning_rate": 1.3292053663570692e-05,
      "loss": 1.1502,
      "step": 644
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0008656978607178,
      "learning_rate": 1.3312693498452014e-05,
      "loss": 1.232,
      "step": 645
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.955205202102661,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.9087,
      "step": 646
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.485811948776245,
      "learning_rate": 1.3353973168214656e-05,
      "loss": 0.8905,
      "step": 647
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9894049167633057,
      "learning_rate": 1.3374613003095976e-05,
      "loss": 0.9125,
      "step": 648
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.886642336845398,
      "learning_rate": 1.3395252837977297e-05,
      "loss": 0.8837,
      "step": 649
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9319517612457275,
      "learning_rate": 1.3415892672858619e-05,
      "loss": 0.7103,
      "step": 650
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.083390712738037,
      "learning_rate": 1.343653250773994e-05,
      "loss": 1.2107,
      "step": 651
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9312684535980225,
      "learning_rate": 1.345717234262126e-05,
      "loss": 0.745,
      "step": 652
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2080614566802979,
      "learning_rate": 1.347781217750258e-05,
      "loss": 0.301,
      "step": 653
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.6073553562164307,
      "learning_rate": 1.3498452012383903e-05,
      "loss": 1.0439,
      "step": 654
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.728792905807495,
      "learning_rate": 1.3519091847265222e-05,
      "loss": 0.907,
      "step": 655
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.7948248386383057,
      "learning_rate": 1.3539731682146544e-05,
      "loss": 1.1711,
      "step": 656
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.648977041244507,
      "learning_rate": 1.3560371517027865e-05,
      "loss": 0.9065,
      "step": 657
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.050502300262451,
      "learning_rate": 1.3581011351909185e-05,
      "loss": 0.6748,
      "step": 658
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5828847885131836,
      "learning_rate": 1.3601651186790507e-05,
      "loss": 0.858,
      "step": 659
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5186398029327393,
      "learning_rate": 1.3622291021671828e-05,
      "loss": 0.9828,
      "step": 660
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.637180805206299,
      "learning_rate": 1.3642930856553148e-05,
      "loss": 1.0181,
      "step": 661
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5330822467803955,
      "learning_rate": 1.3663570691434469e-05,
      "loss": 0.9099,
      "step": 662
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2518559694290161,
      "learning_rate": 1.3684210526315791e-05,
      "loss": 0.3428,
      "step": 663
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.209096670150757,
      "learning_rate": 1.370485036119711e-05,
      "loss": 0.8911,
      "step": 664
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3078516721725464,
      "learning_rate": 1.3725490196078432e-05,
      "loss": 0.2286,
      "step": 665
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2753353118896484,
      "learning_rate": 1.3746130030959755e-05,
      "loss": 0.9397,
      "step": 666
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9966657161712646,
      "learning_rate": 1.3766769865841074e-05,
      "loss": 0.947,
      "step": 667
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.285130262374878,
      "learning_rate": 1.3787409700722396e-05,
      "loss": 0.9944,
      "step": 668
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2331318855285645,
      "learning_rate": 1.3808049535603716e-05,
      "loss": 1.0809,
      "step": 669
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0396547317504883,
      "learning_rate": 1.3828689370485039e-05,
      "loss": 0.8798,
      "step": 670
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5811963081359863,
      "learning_rate": 1.3849329205366357e-05,
      "loss": 0.9516,
      "step": 671
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.214770793914795,
      "learning_rate": 1.386996904024768e-05,
      "loss": 0.8903,
      "step": 672
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.75118637084961,
      "learning_rate": 1.3890608875128999e-05,
      "loss": 1.413,
      "step": 673
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1497151851654053,
      "learning_rate": 1.3911248710010321e-05,
      "loss": 0.8934,
      "step": 674
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.696061134338379,
      "learning_rate": 1.3931888544891643e-05,
      "loss": 0.9529,
      "step": 675
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.972203731536865,
      "learning_rate": 1.3952528379772962e-05,
      "loss": 1.3121,
      "step": 676
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.987114429473877,
      "learning_rate": 1.3973168214654284e-05,
      "loss": 0.9041,
      "step": 677
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1096885204315186,
      "learning_rate": 1.3993808049535605e-05,
      "loss": 0.9035,
      "step": 678
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4208306074142456,
      "learning_rate": 1.4014447884416927e-05,
      "loss": 0.2481,
      "step": 679
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.866555213928223,
      "learning_rate": 1.4035087719298246e-05,
      "loss": 1.8943,
      "step": 680
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9045910835266113,
      "learning_rate": 1.4055727554179568e-05,
      "loss": 0.9295,
      "step": 681
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.042476177215576,
      "learning_rate": 1.4076367389060887e-05,
      "loss": 0.8376,
      "step": 682
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7910802364349365,
      "learning_rate": 1.409700722394221e-05,
      "loss": 1.2647,
      "step": 683
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.115025043487549,
      "learning_rate": 1.4117647058823532e-05,
      "loss": 0.8568,
      "step": 684
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9886552095413208,
      "learning_rate": 1.413828689370485e-05,
      "loss": 0.7766,
      "step": 685
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6024813652038574,
      "learning_rate": 1.4158926728586173e-05,
      "loss": 0.8037,
      "step": 686
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.741818904876709,
      "learning_rate": 1.4179566563467493e-05,
      "loss": 1.2264,
      "step": 687
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.411665439605713,
      "learning_rate": 1.4200206398348816e-05,
      "loss": 0.9719,
      "step": 688
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7651686668395996,
      "learning_rate": 1.4220846233230134e-05,
      "loss": 0.9297,
      "step": 689
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7055211067199707,
      "learning_rate": 1.4241486068111457e-05,
      "loss": 0.8242,
      "step": 690
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.9714789390563965,
      "learning_rate": 1.4262125902992779e-05,
      "loss": 1.2879,
      "step": 691
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.536702632904053,
      "learning_rate": 1.4282765737874098e-05,
      "loss": 0.8863,
      "step": 692
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.242483615875244,
      "learning_rate": 1.430340557275542e-05,
      "loss": 0.847,
      "step": 693
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8093516826629639,
      "learning_rate": 1.4324045407636739e-05,
      "loss": 0.6821,
      "step": 694
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7578562498092651,
      "learning_rate": 1.4344685242518061e-05,
      "loss": 0.8844,
      "step": 695
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.41190242767334,
      "learning_rate": 1.4365325077399382e-05,
      "loss": 0.8833,
      "step": 696
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.734397888183594,
      "learning_rate": 1.4385964912280704e-05,
      "loss": 0.9244,
      "step": 697
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0205118656158447,
      "learning_rate": 1.4406604747162023e-05,
      "loss": 0.8834,
      "step": 698
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7269811630249023,
      "learning_rate": 1.4427244582043345e-05,
      "loss": 1.2637,
      "step": 699
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1443581581115723,
      "learning_rate": 1.4447884416924667e-05,
      "loss": 1.3155,
      "step": 700
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2673909664154053,
      "learning_rate": 1.4468524251805986e-05,
      "loss": 0.8887,
      "step": 701
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.707179546356201,
      "learning_rate": 1.4489164086687308e-05,
      "loss": 0.9255,
      "step": 702
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.276535987854004,
      "learning_rate": 1.4509803921568629e-05,
      "loss": 1.1733,
      "step": 703
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.84492301940918,
      "learning_rate": 1.453044375644995e-05,
      "loss": 1.2137,
      "step": 704
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.351797580718994,
      "learning_rate": 1.455108359133127e-05,
      "loss": 0.9826,
      "step": 705
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.068830728530884,
      "learning_rate": 1.4571723426212592e-05,
      "loss": 1.1424,
      "step": 706
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9184956550598145,
      "learning_rate": 1.4592363261093911e-05,
      "loss": 1.2885,
      "step": 707
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2192530632019043,
      "learning_rate": 1.4613003095975234e-05,
      "loss": 0.8643,
      "step": 708
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.878883361816406,
      "learning_rate": 1.4633642930856556e-05,
      "loss": 1.0901,
      "step": 709
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.511321544647217,
      "learning_rate": 1.4654282765737875e-05,
      "loss": 0.9713,
      "step": 710
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9292006492614746,
      "learning_rate": 1.4674922600619197e-05,
      "loss": 0.8562,
      "step": 711
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7075300216674805,
      "learning_rate": 1.4695562435500517e-05,
      "loss": 1.0767,
      "step": 712
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.476599931716919,
      "learning_rate": 1.4716202270381838e-05,
      "loss": 0.1834,
      "step": 713
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6482491493225098,
      "learning_rate": 1.4736842105263159e-05,
      "loss": 1.1515,
      "step": 714
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2067039012908936,
      "learning_rate": 1.475748194014448e-05,
      "loss": 0.8591,
      "step": 715
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6442182064056396,
      "learning_rate": 1.47781217750258e-05,
      "loss": 1.1238,
      "step": 716
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.379202365875244,
      "learning_rate": 1.4798761609907122e-05,
      "loss": 1.4846,
      "step": 717
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.905818223953247,
      "learning_rate": 1.4819401444788444e-05,
      "loss": 0.7774,
      "step": 718
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.427229404449463,
      "learning_rate": 1.4840041279669763e-05,
      "loss": 1.2433,
      "step": 719
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9891692399978638,
      "learning_rate": 1.4860681114551085e-05,
      "loss": 0.8284,
      "step": 720
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.845991611480713,
      "learning_rate": 1.4881320949432406e-05,
      "loss": 0.8576,
      "step": 721
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1612300872802734,
      "learning_rate": 1.4901960784313726e-05,
      "loss": 0.8075,
      "step": 722
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.101815700531006,
      "learning_rate": 1.4922600619195047e-05,
      "loss": 0.9092,
      "step": 723
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0751209259033203,
      "learning_rate": 1.494324045407637e-05,
      "loss": 0.8198,
      "step": 724
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2238924503326416,
      "learning_rate": 1.496388028895769e-05,
      "loss": 1.005,
      "step": 725
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.082861423492432,
      "learning_rate": 1.498452012383901e-05,
      "loss": 0.85,
      "step": 726
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.500270843505859,
      "learning_rate": 1.5005159958720333e-05,
      "loss": 0.7975,
      "step": 727
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1490049362182617,
      "learning_rate": 1.5025799793601652e-05,
      "loss": 1.0322,
      "step": 728
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5326225757598877,
      "learning_rate": 1.5046439628482974e-05,
      "loss": 0.8956,
      "step": 729
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.514100432395935,
      "learning_rate": 1.5067079463364294e-05,
      "loss": 0.2191,
      "step": 730
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.386711835861206,
      "learning_rate": 1.5087719298245615e-05,
      "loss": 0.919,
      "step": 731
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9188940525054932,
      "learning_rate": 1.5108359133126935e-05,
      "loss": 0.8353,
      "step": 732
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.339253544807434,
      "learning_rate": 1.5128998968008258e-05,
      "loss": 0.1546,
      "step": 733
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8555716276168823,
      "learning_rate": 1.514963880288958e-05,
      "loss": 0.8112,
      "step": 734
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.241121530532837,
      "learning_rate": 1.5170278637770899e-05,
      "loss": 1.0165,
      "step": 735
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.439626693725586,
      "learning_rate": 1.5190918472652221e-05,
      "loss": 1.3415,
      "step": 736
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9610862731933594,
      "learning_rate": 1.521155830753354e-05,
      "loss": 1.1169,
      "step": 737
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5897128582000732,
      "learning_rate": 1.5232198142414862e-05,
      "loss": 0.9754,
      "step": 738
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.459921360015869,
      "learning_rate": 1.5252837977296183e-05,
      "loss": 0.6181,
      "step": 739
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.040696620941162,
      "learning_rate": 1.5273477812177503e-05,
      "loss": 0.6946,
      "step": 740
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.773125171661377,
      "learning_rate": 1.5294117647058822e-05,
      "loss": 1.2033,
      "step": 741
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1527156829833984,
      "learning_rate": 1.5314757481940144e-05,
      "loss": 0.7885,
      "step": 742
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6334121227264404,
      "learning_rate": 1.5335397316821467e-05,
      "loss": 0.9234,
      "step": 743
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.7344794273376465,
      "learning_rate": 1.535603715170279e-05,
      "loss": 1.3936,
      "step": 744
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1018638610839844,
      "learning_rate": 1.5376676986584108e-05,
      "loss": 0.9135,
      "step": 745
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1536662578582764,
      "learning_rate": 1.539731682146543e-05,
      "loss": 1.182,
      "step": 746
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3944091796875,
      "learning_rate": 1.5417956656346752e-05,
      "loss": 0.8622,
      "step": 747
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.99155855178833,
      "learning_rate": 1.543859649122807e-05,
      "loss": 1.2113,
      "step": 748
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.7808902263641357,
      "learning_rate": 1.5459236326109393e-05,
      "loss": 0.6178,
      "step": 749
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.293442487716675,
      "learning_rate": 1.5479876160990716e-05,
      "loss": 0.9459,
      "step": 750
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0923421382904053,
      "learning_rate": 1.5500515995872035e-05,
      "loss": 0.8641,
      "step": 751
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7287662029266357,
      "learning_rate": 1.5521155830753357e-05,
      "loss": 0.835,
      "step": 752
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.082916736602783,
      "learning_rate": 1.5541795665634676e-05,
      "loss": 1.3273,
      "step": 753
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.508984327316284,
      "learning_rate": 1.5562435500515998e-05,
      "loss": 0.9467,
      "step": 754
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9289511442184448,
      "learning_rate": 1.5583075335397317e-05,
      "loss": 0.8066,
      "step": 755
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.025780200958252,
      "learning_rate": 1.560371517027864e-05,
      "loss": 0.86,
      "step": 756
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9731948375701904,
      "learning_rate": 1.5624355005159958e-05,
      "loss": 0.7181,
      "step": 757
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.614958763122559,
      "learning_rate": 1.564499484004128e-05,
      "loss": 1.3544,
      "step": 758
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.90205979347229,
      "learning_rate": 1.5665634674922602e-05,
      "loss": 0.7689,
      "step": 759
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.274383068084717,
      "learning_rate": 1.568627450980392e-05,
      "loss": 0.9577,
      "step": 760
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4528098106384277,
      "learning_rate": 1.5706914344685244e-05,
      "loss": 0.9006,
      "step": 761
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2713289260864258,
      "learning_rate": 1.5727554179566566e-05,
      "loss": 0.1939,
      "step": 762
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3876373767852783,
      "learning_rate": 1.5748194014447885e-05,
      "loss": 0.9499,
      "step": 763
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.458173990249634,
      "learning_rate": 1.5768833849329207e-05,
      "loss": 1.395,
      "step": 764
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8740383386611938,
      "learning_rate": 1.578947368421053e-05,
      "loss": 0.8698,
      "step": 765
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9059534072875977,
      "learning_rate": 1.5810113519091848e-05,
      "loss": 0.9669,
      "step": 766
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.701801061630249,
      "learning_rate": 1.583075335397317e-05,
      "loss": 0.777,
      "step": 767
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0678870677948,
      "learning_rate": 1.5851393188854493e-05,
      "loss": 1.2915,
      "step": 768
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1532185077667236,
      "learning_rate": 1.587203302373581e-05,
      "loss": 0.9709,
      "step": 769
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.215928554534912,
      "learning_rate": 1.5892672858617134e-05,
      "loss": 0.887,
      "step": 770
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8328677415847778,
      "learning_rate": 1.5913312693498453e-05,
      "loss": 0.7822,
      "step": 771
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.628467082977295,
      "learning_rate": 1.5933952528379775e-05,
      "loss": 1.005,
      "step": 772
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.531232833862305,
      "learning_rate": 1.5954592363261094e-05,
      "loss": 1.181,
      "step": 773
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.300394058227539,
      "learning_rate": 1.5975232198142416e-05,
      "loss": 0.895,
      "step": 774
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0242631435394287,
      "learning_rate": 1.5995872033023735e-05,
      "loss": 0.9599,
      "step": 775
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.808718681335449,
      "learning_rate": 1.6016511867905057e-05,
      "loss": 0.7048,
      "step": 776
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.278226375579834,
      "learning_rate": 1.603715170278638e-05,
      "loss": 0.8623,
      "step": 777
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.020397424697876,
      "learning_rate": 1.6057791537667698e-05,
      "loss": 1.0014,
      "step": 778
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.095561981201172,
      "learning_rate": 1.607843137254902e-05,
      "loss": 0.8932,
      "step": 779
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2724335193634033,
      "learning_rate": 1.6099071207430343e-05,
      "loss": 1.0245,
      "step": 780
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9805166721343994,
      "learning_rate": 1.611971104231166e-05,
      "loss": 0.7566,
      "step": 781
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4771947860717773,
      "learning_rate": 1.6140350877192984e-05,
      "loss": 0.89,
      "step": 782
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2777161598205566,
      "learning_rate": 1.6160990712074306e-05,
      "loss": 0.8856,
      "step": 783
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.010645627975464,
      "learning_rate": 1.618163054695563e-05,
      "loss": 1.0116,
      "step": 784
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6804285049438477,
      "learning_rate": 1.6202270381836947e-05,
      "loss": 0.9361,
      "step": 785
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.500929594039917,
      "learning_rate": 1.622291021671827e-05,
      "loss": 1.1493,
      "step": 786
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4537482261657715,
      "learning_rate": 1.624355005159959e-05,
      "loss": 0.8997,
      "step": 787
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0544307231903076,
      "learning_rate": 1.626418988648091e-05,
      "loss": 0.9112,
      "step": 788
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3180643320083618,
      "learning_rate": 1.628482972136223e-05,
      "loss": 0.225,
      "step": 789
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8119345903396606,
      "learning_rate": 1.6305469556243552e-05,
      "loss": 0.7934,
      "step": 790
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2183244228363037,
      "learning_rate": 1.632610939112487e-05,
      "loss": 0.8268,
      "step": 791
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3415682315826416,
      "learning_rate": 1.6346749226006193e-05,
      "loss": 0.7908,
      "step": 792
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.382126808166504,
      "learning_rate": 1.6367389060887515e-05,
      "loss": 1.5953,
      "step": 793
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.74240255355835,
      "learning_rate": 1.6388028895768834e-05,
      "loss": 1.3294,
      "step": 794
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.693722248077393,
      "learning_rate": 1.6408668730650156e-05,
      "loss": 1.3209,
      "step": 795
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4867446422576904,
      "learning_rate": 1.6429308565531475e-05,
      "loss": 1.3063,
      "step": 796
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3194468021392822,
      "learning_rate": 1.6449948400412797e-05,
      "loss": 0.8803,
      "step": 797
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0583131313323975,
      "learning_rate": 1.647058823529412e-05,
      "loss": 0.8576,
      "step": 798
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3020801544189453,
      "learning_rate": 1.649122807017544e-05,
      "loss": 0.9026,
      "step": 799
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.88465690612793,
      "learning_rate": 1.651186790505676e-05,
      "loss": 1.1481,
      "step": 800
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.38336181640625,
      "learning_rate": 1.6532507739938083e-05,
      "loss": 1.7191,
      "step": 801
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1249916553497314,
      "learning_rate": 1.6553147574819405e-05,
      "loss": 0.7879,
      "step": 802
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6030354499816895,
      "learning_rate": 1.6573787409700724e-05,
      "loss": 0.7741,
      "step": 803
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.23799467086792,
      "learning_rate": 1.6594427244582046e-05,
      "loss": 0.6772,
      "step": 804
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.662739276885986,
      "learning_rate": 1.6615067079463365e-05,
      "loss": 1.6264,
      "step": 805
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2664172649383545,
      "learning_rate": 1.6635706914344687e-05,
      "loss": 0.977,
      "step": 806
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.895082473754883,
      "learning_rate": 1.6656346749226006e-05,
      "loss": 1.0526,
      "step": 807
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.082688331604004,
      "learning_rate": 1.667698658410733e-05,
      "loss": 1.6933,
      "step": 808
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0400142669677734,
      "learning_rate": 1.669762641898865e-05,
      "loss": 0.8868,
      "step": 809
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.449713945388794,
      "learning_rate": 1.671826625386997e-05,
      "loss": 0.747,
      "step": 810
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9117581844329834,
      "learning_rate": 1.6738906088751292e-05,
      "loss": 1.3582,
      "step": 811
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.36892032623291,
      "learning_rate": 1.675954592363261e-05,
      "loss": 1.2889,
      "step": 812
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.221977710723877,
      "learning_rate": 1.6780185758513933e-05,
      "loss": 1.5478,
      "step": 813
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.481227397918701,
      "learning_rate": 1.6800825593395252e-05,
      "loss": 1.0826,
      "step": 814
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.712165355682373,
      "learning_rate": 1.6821465428276574e-05,
      "loss": 1.1613,
      "step": 815
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.078705310821533,
      "learning_rate": 1.6842105263157896e-05,
      "loss": 0.8703,
      "step": 816
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1824427843093872,
      "learning_rate": 1.686274509803922e-05,
      "loss": 0.2095,
      "step": 817
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.007754325866699,
      "learning_rate": 1.6883384932920538e-05,
      "loss": 1.2807,
      "step": 818
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1731059551239014,
      "learning_rate": 1.690402476780186e-05,
      "loss": 0.2412,
      "step": 819
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.909427523612976,
      "learning_rate": 1.6924664602683182e-05,
      "loss": 0.8285,
      "step": 820
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.364019393920898,
      "learning_rate": 1.69453044375645e-05,
      "loss": 1.1937,
      "step": 821
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.1376914978027344,
      "learning_rate": 1.6965944272445823e-05,
      "loss": 0.9406,
      "step": 822
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1164138317108154,
      "learning_rate": 1.6986584107327142e-05,
      "loss": 0.8705,
      "step": 823
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.1106951236724854,
      "learning_rate": 1.7007223942208464e-05,
      "loss": 1.2699,
      "step": 824
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7784013748168945,
      "learning_rate": 1.7027863777089783e-05,
      "loss": 0.8226,
      "step": 825
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.073694944381714,
      "learning_rate": 1.7048503611971105e-05,
      "loss": 1.0933,
      "step": 826
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.154778242111206,
      "learning_rate": 1.7069143446852428e-05,
      "loss": 0.9311,
      "step": 827
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.574521780014038,
      "learning_rate": 1.7089783281733747e-05,
      "loss": 0.9306,
      "step": 828
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.370180368423462,
      "learning_rate": 1.711042311661507e-05,
      "loss": 1.1146,
      "step": 829
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3623881340026855,
      "learning_rate": 1.7131062951496388e-05,
      "loss": 0.7522,
      "step": 830
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.4133529663085938,
      "learning_rate": 1.715170278637771e-05,
      "loss": 0.8818,
      "step": 831
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.405074119567871,
      "learning_rate": 1.717234262125903e-05,
      "loss": 0.7834,
      "step": 832
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.835138201713562,
      "learning_rate": 1.719298245614035e-05,
      "loss": 0.8053,
      "step": 833
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0545475482940674,
      "learning_rate": 1.7213622291021673e-05,
      "loss": 1.0443,
      "step": 834
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.101170778274536,
      "learning_rate": 1.7234262125902996e-05,
      "loss": 1.1123,
      "step": 835
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0158796310424805,
      "learning_rate": 1.7254901960784314e-05,
      "loss": 0.8616,
      "step": 836
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1885790824890137,
      "learning_rate": 1.7275541795665637e-05,
      "loss": 0.9475,
      "step": 837
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3003180027008057,
      "learning_rate": 1.729618163054696e-05,
      "loss": 0.8815,
      "step": 838
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1454856395721436,
      "learning_rate": 1.7316821465428278e-05,
      "loss": 0.9693,
      "step": 839
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.746404647827148,
      "learning_rate": 1.73374613003096e-05,
      "loss": 1.564,
      "step": 840
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.585626125335693,
      "learning_rate": 1.735810113519092e-05,
      "loss": 1.5562,
      "step": 841
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.552372932434082,
      "learning_rate": 1.737874097007224e-05,
      "loss": 0.8521,
      "step": 842
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.231968879699707,
      "learning_rate": 1.7399380804953563e-05,
      "loss": 0.7471,
      "step": 843
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.773878574371338,
      "learning_rate": 1.7420020639834882e-05,
      "loss": 0.8515,
      "step": 844
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7279326915740967,
      "learning_rate": 1.7440660474716205e-05,
      "loss": 1.2562,
      "step": 845
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.0272674560546875,
      "learning_rate": 1.7461300309597523e-05,
      "loss": 1.445,
      "step": 846
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3361096382141113,
      "learning_rate": 1.7481940144478846e-05,
      "loss": 0.9045,
      "step": 847
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2217347621917725,
      "learning_rate": 1.7502579979360165e-05,
      "loss": 0.8532,
      "step": 848
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2896385192871094,
      "learning_rate": 1.7523219814241487e-05,
      "loss": 0.9473,
      "step": 849
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.410688877105713,
      "learning_rate": 1.754385964912281e-05,
      "loss": 0.9533,
      "step": 850
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3228580951690674,
      "learning_rate": 1.7564499484004128e-05,
      "loss": 1.0138,
      "step": 851
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4164535999298096,
      "learning_rate": 1.758513931888545e-05,
      "loss": 0.8754,
      "step": 852
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1215150356292725,
      "learning_rate": 1.7605779153766772e-05,
      "loss": 0.8195,
      "step": 853
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.597811222076416,
      "learning_rate": 1.762641898864809e-05,
      "loss": 0.7301,
      "step": 854
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3557934761047363,
      "learning_rate": 1.7647058823529414e-05,
      "loss": 0.8697,
      "step": 855
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.117879867553711,
      "learning_rate": 1.7667698658410736e-05,
      "loss": 0.8213,
      "step": 856
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5506739616394043,
      "learning_rate": 1.7688338493292055e-05,
      "loss": 0.8469,
      "step": 857
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.90605092048645,
      "learning_rate": 1.7708978328173377e-05,
      "loss": 1.1193,
      "step": 858
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.273343324661255,
      "learning_rate": 1.7729618163054696e-05,
      "loss": 0.8426,
      "step": 859
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1555352210998535,
      "learning_rate": 1.7750257997936018e-05,
      "loss": 0.8313,
      "step": 860
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.114813804626465,
      "learning_rate": 1.777089783281734e-05,
      "loss": 1.6571,
      "step": 861
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.893997073173523,
      "learning_rate": 1.779153766769866e-05,
      "loss": 0.768,
      "step": 862
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1128458976745605,
      "learning_rate": 1.781217750257998e-05,
      "loss": 0.759,
      "step": 863
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3025269508361816,
      "learning_rate": 1.78328173374613e-05,
      "loss": 1.2525,
      "step": 864
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.841341018676758,
      "learning_rate": 1.7853457172342623e-05,
      "loss": 0.9995,
      "step": 865
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.412636756896973,
      "learning_rate": 1.787409700722394e-05,
      "loss": 1.209,
      "step": 866
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.506025791168213,
      "learning_rate": 1.7894736842105264e-05,
      "loss": 0.9932,
      "step": 867
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.393709421157837,
      "learning_rate": 1.7915376676986586e-05,
      "loss": 1.2664,
      "step": 868
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6348812580108643,
      "learning_rate": 1.7936016511867905e-05,
      "loss": 0.9266,
      "step": 869
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6982309818267822,
      "learning_rate": 1.7956656346749227e-05,
      "loss": 0.8502,
      "step": 870
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.608398914337158,
      "learning_rate": 1.797729618163055e-05,
      "loss": 0.7982,
      "step": 871
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1839792728424072,
      "learning_rate": 1.799793601651187e-05,
      "loss": 0.8597,
      "step": 872
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3128461837768555,
      "learning_rate": 1.801857585139319e-05,
      "loss": 1.0388,
      "step": 873
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.536811351776123,
      "learning_rate": 1.8039215686274513e-05,
      "loss": 1.1476,
      "step": 874
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8447456359863281,
      "learning_rate": 1.805985552115583e-05,
      "loss": 0.8843,
      "step": 875
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.007814884185791,
      "learning_rate": 1.8080495356037154e-05,
      "loss": 0.901,
      "step": 876
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7982279062271118,
      "learning_rate": 1.8101135190918476e-05,
      "loss": 0.793,
      "step": 877
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.413174867630005,
      "learning_rate": 1.8121775025799795e-05,
      "loss": 1.1771,
      "step": 878
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.112526893615723,
      "learning_rate": 1.8142414860681117e-05,
      "loss": 1.1905,
      "step": 879
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.55914044380188,
      "learning_rate": 1.8163054695562436e-05,
      "loss": 1.3307,
      "step": 880
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.279138565063477,
      "learning_rate": 1.818369453044376e-05,
      "loss": 1.4393,
      "step": 881
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9494882822036743,
      "learning_rate": 1.8204334365325077e-05,
      "loss": 0.843,
      "step": 882
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6027424335479736,
      "learning_rate": 1.82249742002064e-05,
      "loss": 1.014,
      "step": 883
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.123661994934082,
      "learning_rate": 1.824561403508772e-05,
      "loss": 0.9716,
      "step": 884
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.595890760421753,
      "learning_rate": 1.826625386996904e-05,
      "loss": 0.8902,
      "step": 885
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.173978805541992,
      "learning_rate": 1.8286893704850363e-05,
      "loss": 1.3419,
      "step": 886
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.277070999145508,
      "learning_rate": 1.8307533539731682e-05,
      "loss": 1.2095,
      "step": 887
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.25092887878418,
      "learning_rate": 1.8328173374613004e-05,
      "loss": 1.2837,
      "step": 888
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.060940980911255,
      "learning_rate": 1.8348813209494326e-05,
      "loss": 0.9563,
      "step": 889
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.244100332260132,
      "learning_rate": 1.836945304437565e-05,
      "loss": 0.8732,
      "step": 890
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.025287628173828,
      "learning_rate": 1.8390092879256967e-05,
      "loss": 0.8091,
      "step": 891
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4755120277404785,
      "learning_rate": 1.841073271413829e-05,
      "loss": 1.0222,
      "step": 892
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5241122245788574,
      "learning_rate": 1.843137254901961e-05,
      "loss": 0.9325,
      "step": 893
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0517289638519287,
      "learning_rate": 1.845201238390093e-05,
      "loss": 0.763,
      "step": 894
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.180239200592041,
      "learning_rate": 1.8472652218782253e-05,
      "loss": 0.9355,
      "step": 895
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1919591426849365,
      "learning_rate": 1.8493292053663572e-05,
      "loss": 0.8375,
      "step": 896
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1862611770629883,
      "learning_rate": 1.8513931888544894e-05,
      "loss": 0.9086,
      "step": 897
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.731229066848755,
      "learning_rate": 1.8534571723426213e-05,
      "loss": 0.9001,
      "step": 898
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.982900619506836,
      "learning_rate": 1.8555211558307535e-05,
      "loss": 0.8987,
      "step": 899
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.109675168991089,
      "learning_rate": 1.8575851393188854e-05,
      "loss": 0.8942,
      "step": 900
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.006359100341797,
      "learning_rate": 1.8596491228070176e-05,
      "loss": 0.9704,
      "step": 901
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2035419940948486,
      "learning_rate": 1.86171310629515e-05,
      "loss": 0.8357,
      "step": 902
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6967506408691406,
      "learning_rate": 1.8637770897832817e-05,
      "loss": 0.8864,
      "step": 903
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.009337902069092,
      "learning_rate": 1.865841073271414e-05,
      "loss": 0.8166,
      "step": 904
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.465500831604004,
      "learning_rate": 1.8679050567595462e-05,
      "loss": 1.6079,
      "step": 905
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9864754676818848,
      "learning_rate": 1.869969040247678e-05,
      "loss": 1.1792,
      "step": 906
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2768142223358154,
      "learning_rate": 1.8720330237358103e-05,
      "loss": 0.9825,
      "step": 907
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.7305097579956055,
      "learning_rate": 1.8740970072239425e-05,
      "loss": 0.9699,
      "step": 908
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1380860805511475,
      "learning_rate": 1.8761609907120744e-05,
      "loss": 0.751,
      "step": 909
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5219448804855347,
      "learning_rate": 1.8782249742002066e-05,
      "loss": 0.2671,
      "step": 910
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9199776649475098,
      "learning_rate": 1.880288957688339e-05,
      "loss": 0.8662,
      "step": 911
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5571796894073486,
      "learning_rate": 1.8823529411764708e-05,
      "loss": 0.8044,
      "step": 912
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.048795461654663,
      "learning_rate": 1.884416924664603e-05,
      "loss": 0.8556,
      "step": 913
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.704789400100708,
      "learning_rate": 1.886480908152735e-05,
      "loss": 0.8452,
      "step": 914
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.67938232421875,
      "learning_rate": 1.888544891640867e-05,
      "loss": 1.0536,
      "step": 915
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.141230821609497,
      "learning_rate": 1.890608875128999e-05,
      "loss": 0.7771,
      "step": 916
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.839456558227539,
      "learning_rate": 1.8926728586171312e-05,
      "loss": 0.9987,
      "step": 917
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.584556818008423,
      "learning_rate": 1.894736842105263e-05,
      "loss": 0.9349,
      "step": 918
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6313791275024414,
      "learning_rate": 1.8968008255933953e-05,
      "loss": 0.3131,
      "step": 919
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.876047134399414,
      "learning_rate": 1.8988648090815275e-05,
      "loss": 1.2319,
      "step": 920
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.89080810546875,
      "learning_rate": 1.9009287925696594e-05,
      "loss": 1.316,
      "step": 921
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3241190910339355,
      "learning_rate": 1.9029927760577917e-05,
      "loss": 0.8873,
      "step": 922
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2694931030273438,
      "learning_rate": 1.905056759545924e-05,
      "loss": 0.9089,
      "step": 923
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4544498920440674,
      "learning_rate": 1.9071207430340558e-05,
      "loss": 0.2162,
      "step": 924
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.449764728546143,
      "learning_rate": 1.909184726522188e-05,
      "loss": 0.8894,
      "step": 925
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.183394432067871,
      "learning_rate": 1.9112487100103202e-05,
      "loss": 1.0724,
      "step": 926
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2838327884674072,
      "learning_rate": 1.913312693498452e-05,
      "loss": 0.9074,
      "step": 927
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.339951276779175,
      "learning_rate": 1.9153766769865843e-05,
      "loss": 0.936,
      "step": 928
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4395158290863037,
      "learning_rate": 1.9174406604747166e-05,
      "loss": 0.8478,
      "step": 929
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9054656028747559,
      "learning_rate": 1.9195046439628485e-05,
      "loss": 0.8165,
      "step": 930
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6614012718200684,
      "learning_rate": 1.9215686274509807e-05,
      "loss": 0.5478,
      "step": 931
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.941601037979126,
      "learning_rate": 1.9236326109391126e-05,
      "loss": 0.8614,
      "step": 932
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0323469638824463,
      "learning_rate": 1.9256965944272448e-05,
      "loss": 0.7912,
      "step": 933
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.822203159332275,
      "learning_rate": 1.9277605779153767e-05,
      "loss": 1.2003,
      "step": 934
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.015343189239502,
      "learning_rate": 1.929824561403509e-05,
      "loss": 0.91,
      "step": 935
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.1323184967041016,
      "learning_rate": 1.931888544891641e-05,
      "loss": 0.9213,
      "step": 936
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.580379962921143,
      "learning_rate": 1.933952528379773e-05,
      "loss": 1.1155,
      "step": 937
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0405306816101074,
      "learning_rate": 1.9360165118679052e-05,
      "loss": 0.8297,
      "step": 938
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.307424068450928,
      "learning_rate": 1.938080495356037e-05,
      "loss": 0.7382,
      "step": 939
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4632604122161865,
      "learning_rate": 1.9401444788441694e-05,
      "loss": 0.8839,
      "step": 940
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.542407274246216,
      "learning_rate": 1.9422084623323016e-05,
      "loss": 0.9829,
      "step": 941
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1135730743408203,
      "learning_rate": 1.9442724458204335e-05,
      "loss": 0.8677,
      "step": 942
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.931725263595581,
      "learning_rate": 1.9463364293085657e-05,
      "loss": 1.291,
      "step": 943
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.896773338317871,
      "learning_rate": 1.948400412796698e-05,
      "loss": 0.8968,
      "step": 944
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.719735860824585,
      "learning_rate": 1.95046439628483e-05,
      "loss": 0.8955,
      "step": 945
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9444618225097656,
      "learning_rate": 1.952528379772962e-05,
      "loss": 0.948,
      "step": 946
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6186962127685547,
      "learning_rate": 1.9545923632610943e-05,
      "loss": 0.9336,
      "step": 947
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.012291193008423,
      "learning_rate": 1.956656346749226e-05,
      "loss": 0.8927,
      "step": 948
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6866306066513062,
      "learning_rate": 1.9587203302373584e-05,
      "loss": 0.9415,
      "step": 949
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.875425100326538,
      "learning_rate": 1.9607843137254903e-05,
      "loss": 0.9028,
      "step": 950
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2891786098480225,
      "learning_rate": 1.9628482972136225e-05,
      "loss": 0.8192,
      "step": 951
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2158498764038086,
      "learning_rate": 1.9649122807017544e-05,
      "loss": 0.9488,
      "step": 952
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.45755672454834,
      "learning_rate": 1.9669762641898866e-05,
      "loss": 1.4203,
      "step": 953
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.462456703186035,
      "learning_rate": 1.9690402476780188e-05,
      "loss": 0.8872,
      "step": 954
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0470523834228516,
      "learning_rate": 1.9711042311661507e-05,
      "loss": 0.8648,
      "step": 955
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.633334159851074,
      "learning_rate": 1.973168214654283e-05,
      "loss": 1.0084,
      "step": 956
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.230492115020752,
      "learning_rate": 1.9752321981424148e-05,
      "loss": 0.784,
      "step": 957
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7443817853927612,
      "learning_rate": 1.977296181630547e-05,
      "loss": 0.8139,
      "step": 958
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5990638732910156,
      "learning_rate": 1.9793601651186793e-05,
      "loss": 0.8874,
      "step": 959
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.075067043304443,
      "learning_rate": 1.981424148606811e-05,
      "loss": 1.0621,
      "step": 960
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8026694059371948,
      "learning_rate": 1.9834881320949434e-05,
      "loss": 0.7939,
      "step": 961
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.004704475402832,
      "learning_rate": 1.9855521155830756e-05,
      "loss": 0.8696,
      "step": 962
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1410999298095703,
      "learning_rate": 1.9876160990712078e-05,
      "loss": 0.9226,
      "step": 963
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9863609075546265,
      "learning_rate": 1.9896800825593397e-05,
      "loss": 0.7361,
      "step": 964
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0464158058166504,
      "learning_rate": 1.991744066047472e-05,
      "loss": 0.891,
      "step": 965
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.116268634796143,
      "learning_rate": 1.9938080495356038e-05,
      "loss": 0.8257,
      "step": 966
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.466829538345337,
      "learning_rate": 1.995872033023736e-05,
      "loss": 1.055,
      "step": 967
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8928427696228027,
      "learning_rate": 1.997936016511868e-05,
      "loss": 1.3019,
      "step": 968
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.304253101348877,
      "learning_rate": 2e-05,
      "loss": 1.5346,
      "step": 969
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4848263263702393,
      "learning_rate": 1.999936108360221e-05,
      "loss": 0.3671,
      "step": 970
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.886800527572632,
      "learning_rate": 1.999872216720442e-05,
      "loss": 1.2888,
      "step": 971
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0771920680999756,
      "learning_rate": 1.9998083250806633e-05,
      "loss": 1.1258,
      "step": 972
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8076460361480713,
      "learning_rate": 1.9997444334408844e-05,
      "loss": 0.7771,
      "step": 973
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7667150497436523,
      "learning_rate": 1.9996805418011053e-05,
      "loss": 1.1547,
      "step": 974
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.556473970413208,
      "learning_rate": 1.9996166501613264e-05,
      "loss": 0.2325,
      "step": 975
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.022305727005005,
      "learning_rate": 1.9995527585215476e-05,
      "loss": 0.798,
      "step": 976
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0343635082244873,
      "learning_rate": 1.9994888668817687e-05,
      "loss": 0.8983,
      "step": 977
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6259570121765137,
      "learning_rate": 1.9994249752419895e-05,
      "loss": 1.0279,
      "step": 978
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4461593627929688,
      "learning_rate": 1.9993610836022107e-05,
      "loss": 0.8454,
      "step": 979
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.455113410949707,
      "learning_rate": 1.999297191962432e-05,
      "loss": 0.9809,
      "step": 980
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.697553038597107,
      "learning_rate": 1.999233300322653e-05,
      "loss": 0.8395,
      "step": 981
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3993921279907227,
      "learning_rate": 1.9991694086828738e-05,
      "loss": 1.0367,
      "step": 982
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9209116697311401,
      "learning_rate": 1.999105517043095e-05,
      "loss": 0.773,
      "step": 983
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.224452972412109,
      "learning_rate": 1.999041625403316e-05,
      "loss": 1.2028,
      "step": 984
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.880981683731079,
      "learning_rate": 1.9989777337635373e-05,
      "loss": 0.8364,
      "step": 985
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.011688709259033,
      "learning_rate": 1.998913842123758e-05,
      "loss": 0.9144,
      "step": 986
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.18118953704834,
      "learning_rate": 1.9988499504839793e-05,
      "loss": 1.0974,
      "step": 987
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.351951837539673,
      "learning_rate": 1.9987860588442004e-05,
      "loss": 0.8676,
      "step": 988
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.4265925884246826,
      "learning_rate": 1.9987221672044216e-05,
      "loss": 0.9576,
      "step": 989
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.029883861541748,
      "learning_rate": 1.9986582755646424e-05,
      "loss": 0.7883,
      "step": 990
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9426292181015015,
      "learning_rate": 1.9985943839248635e-05,
      "loss": 0.7891,
      "step": 991
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1115896701812744,
      "learning_rate": 1.9985304922850847e-05,
      "loss": 0.9137,
      "step": 992
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.612602472305298,
      "learning_rate": 1.998466600645306e-05,
      "loss": 1.2248,
      "step": 993
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.467275619506836,
      "learning_rate": 1.9984027090055267e-05,
      "loss": 0.986,
      "step": 994
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.835031986236572,
      "learning_rate": 1.9983388173657478e-05,
      "loss": 1.5267,
      "step": 995
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.891542911529541,
      "learning_rate": 1.998274925725969e-05,
      "loss": 0.9333,
      "step": 996
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9236042499542236,
      "learning_rate": 1.99821103408619e-05,
      "loss": 1.0165,
      "step": 997
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0620079040527344,
      "learning_rate": 1.998147142446411e-05,
      "loss": 0.9103,
      "step": 998
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.944850921630859,
      "learning_rate": 1.998083250806632e-05,
      "loss": 1.0004,
      "step": 999
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.674229621887207,
      "learning_rate": 1.998019359166853e-05,
      "loss": 0.8646,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0833897590637207,
      "learning_rate": 1.9979554675270744e-05,
      "loss": 0.946,
      "step": 1001
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6538105010986328,
      "learning_rate": 1.9978915758872952e-05,
      "loss": 0.2444,
      "step": 1002
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2773226499557495,
      "learning_rate": 1.9978276842475164e-05,
      "loss": 0.2089,
      "step": 1003
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6391780376434326,
      "learning_rate": 1.9977637926077372e-05,
      "loss": 0.9393,
      "step": 1004
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.9432172775268555,
      "learning_rate": 1.9976999009679587e-05,
      "loss": 1.443,
      "step": 1005
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.052863836288452,
      "learning_rate": 1.9976360093281795e-05,
      "loss": 0.8982,
      "step": 1006
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4648592472076416,
      "learning_rate": 1.9975721176884007e-05,
      "loss": 0.8791,
      "step": 1007
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2720811367034912,
      "learning_rate": 1.9975082260486215e-05,
      "loss": 0.237,
      "step": 1008
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2886152267456055,
      "learning_rate": 1.9974443344088426e-05,
      "loss": 0.8997,
      "step": 1009
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.2044801712036133,
      "learning_rate": 1.9973804427690638e-05,
      "loss": 1.0129,
      "step": 1010
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.190286874771118,
      "learning_rate": 1.997316551129285e-05,
      "loss": 0.9626,
      "step": 1011
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.998107433319092,
      "learning_rate": 1.9972526594895057e-05,
      "loss": 1.634,
      "step": 1012
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.685286045074463,
      "learning_rate": 1.997188767849727e-05,
      "loss": 0.8898,
      "step": 1013
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.2130703926086426,
      "learning_rate": 1.997124876209948e-05,
      "loss": 0.9992,
      "step": 1014
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3433027267456055,
      "learning_rate": 1.9970609845701692e-05,
      "loss": 1.1247,
      "step": 1015
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9481399059295654,
      "learning_rate": 1.99699709293039e-05,
      "loss": 0.8346,
      "step": 1016
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.0106072425842285,
      "learning_rate": 1.9969332012906112e-05,
      "loss": 1.2704,
      "step": 1017
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9337217807769775,
      "learning_rate": 1.9968693096508323e-05,
      "loss": 0.8013,
      "step": 1018
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.659430980682373,
      "learning_rate": 1.9968054180110535e-05,
      "loss": 0.9368,
      "step": 1019
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.314328670501709,
      "learning_rate": 1.9967415263712743e-05,
      "loss": 0.9783,
      "step": 1020
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0562005043029785,
      "learning_rate": 1.9966776347314955e-05,
      "loss": 0.5681,
      "step": 1021
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6720998287200928,
      "learning_rate": 1.9966137430917166e-05,
      "loss": 0.2621,
      "step": 1022
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1817374229431152,
      "learning_rate": 1.9965498514519378e-05,
      "loss": 0.9594,
      "step": 1023
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.788196086883545,
      "learning_rate": 1.9964859598121586e-05,
      "loss": 0.8735,
      "step": 1024
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6008706092834473,
      "learning_rate": 1.9964220681723797e-05,
      "loss": 0.9673,
      "step": 1025
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7474615573883057,
      "learning_rate": 1.996358176532601e-05,
      "loss": 0.8843,
      "step": 1026
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.267416477203369,
      "learning_rate": 1.996294284892822e-05,
      "loss": 0.8855,
      "step": 1027
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3631410598754883,
      "learning_rate": 1.996230393253043e-05,
      "loss": 0.9052,
      "step": 1028
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1404242515563965,
      "learning_rate": 1.996166501613264e-05,
      "loss": 0.9892,
      "step": 1029
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9569891691207886,
      "learning_rate": 1.9961026099734852e-05,
      "loss": 0.9142,
      "step": 1030
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.71665620803833,
      "learning_rate": 1.9960387183337063e-05,
      "loss": 0.9376,
      "step": 1031
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9962472915649414,
      "learning_rate": 1.995974826693927e-05,
      "loss": 0.8954,
      "step": 1032
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.931307315826416,
      "learning_rate": 1.9959109350541483e-05,
      "loss": 0.7954,
      "step": 1033
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.018752098083496,
      "learning_rate": 1.9958470434143695e-05,
      "loss": 0.8813,
      "step": 1034
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2621216773986816,
      "learning_rate": 1.9957831517745906e-05,
      "loss": 0.9002,
      "step": 1035
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.176887035369873,
      "learning_rate": 1.9957192601348114e-05,
      "loss": 0.8703,
      "step": 1036
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8759719133377075,
      "learning_rate": 1.9956553684950326e-05,
      "loss": 0.8925,
      "step": 1037
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9149177074432373,
      "learning_rate": 1.9955914768552534e-05,
      "loss": 1.0753,
      "step": 1038
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7076468467712402,
      "learning_rate": 1.995527585215475e-05,
      "loss": 1.2731,
      "step": 1039
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.073272228240967,
      "learning_rate": 1.9954636935756957e-05,
      "loss": 0.9659,
      "step": 1040
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.517648696899414,
      "learning_rate": 1.995399801935917e-05,
      "loss": 1.0362,
      "step": 1041
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.568938255310059,
      "learning_rate": 1.9953359102961377e-05,
      "loss": 0.9908,
      "step": 1042
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.019448757171631,
      "learning_rate": 1.995272018656359e-05,
      "loss": 1.4577,
      "step": 1043
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.14574933052063,
      "learning_rate": 1.99520812701658e-05,
      "loss": 0.8134,
      "step": 1044
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.659887313842773,
      "learning_rate": 1.995144235376801e-05,
      "loss": 1.2338,
      "step": 1045
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8483364582061768,
      "learning_rate": 1.995080343737022e-05,
      "loss": 0.8741,
      "step": 1046
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.974862813949585,
      "learning_rate": 1.995016452097243e-05,
      "loss": 0.815,
      "step": 1047
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.5793840885162354,
      "learning_rate": 1.9949525604574643e-05,
      "loss": 1.1375,
      "step": 1048
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.428741455078125,
      "learning_rate": 1.9948886688176854e-05,
      "loss": 1.0693,
      "step": 1049
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.021596670150757,
      "learning_rate": 1.9948247771779062e-05,
      "loss": 0.9232,
      "step": 1050
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.574223041534424,
      "learning_rate": 1.9947608855381274e-05,
      "loss": 0.9939,
      "step": 1051
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.891112804412842,
      "learning_rate": 1.9946969938983485e-05,
      "loss": 0.8804,
      "step": 1052
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.122378826141357,
      "learning_rate": 1.9946331022585697e-05,
      "loss": 1.6912,
      "step": 1053
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5461955070495605,
      "learning_rate": 1.9945692106187905e-05,
      "loss": 0.2512,
      "step": 1054
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.93086576461792,
      "learning_rate": 1.9945053189790117e-05,
      "loss": 1.4394,
      "step": 1055
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.172466993331909,
      "learning_rate": 1.9944414273392328e-05,
      "loss": 0.9487,
      "step": 1056
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.207985877990723,
      "learning_rate": 1.994377535699454e-05,
      "loss": 1.1186,
      "step": 1057
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4068379402160645,
      "learning_rate": 1.9943136440596748e-05,
      "loss": 0.9469,
      "step": 1058
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5402112007141113,
      "learning_rate": 1.994249752419896e-05,
      "loss": 0.9674,
      "step": 1059
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2840126752853394,
      "learning_rate": 1.994185860780117e-05,
      "loss": 0.2139,
      "step": 1060
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4219157695770264,
      "learning_rate": 1.9941219691403383e-05,
      "loss": 0.6936,
      "step": 1061
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.1031928062438965,
      "learning_rate": 1.994058077500559e-05,
      "loss": 1.199,
      "step": 1062
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3350794315338135,
      "learning_rate": 1.9939941858607802e-05,
      "loss": 0.8754,
      "step": 1063
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7704188823699951,
      "learning_rate": 1.9939302942210014e-05,
      "loss": 0.7965,
      "step": 1064
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3908374309539795,
      "learning_rate": 1.9938664025812225e-05,
      "loss": 0.8237,
      "step": 1065
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3034677505493164,
      "learning_rate": 1.9938025109414434e-05,
      "loss": 0.949,
      "step": 1066
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.11031174659729,
      "learning_rate": 1.9937386193016645e-05,
      "loss": 1.0319,
      "step": 1067
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.69677734375,
      "learning_rate": 1.9936747276618857e-05,
      "loss": 1.469,
      "step": 1068
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4103760719299316,
      "learning_rate": 1.9936108360221068e-05,
      "loss": 0.848,
      "step": 1069
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.489405870437622,
      "learning_rate": 1.9935469443823276e-05,
      "loss": 0.9255,
      "step": 1070
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7866729497909546,
      "learning_rate": 1.9934830527425488e-05,
      "loss": 0.8533,
      "step": 1071
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9116051197052,
      "learning_rate": 1.99341916110277e-05,
      "loss": 0.9115,
      "step": 1072
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.442702054977417,
      "learning_rate": 1.993355269462991e-05,
      "loss": 1.0612,
      "step": 1073
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.140232801437378,
      "learning_rate": 1.993291377823212e-05,
      "loss": 0.8845,
      "step": 1074
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1623497009277344,
      "learning_rate": 1.993227486183433e-05,
      "loss": 0.9506,
      "step": 1075
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7945443391799927,
      "learning_rate": 1.993163594543654e-05,
      "loss": 0.813,
      "step": 1076
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9633100032806396,
      "learning_rate": 1.9930997029038754e-05,
      "loss": 0.8361,
      "step": 1077
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.814467430114746,
      "learning_rate": 1.9930358112640962e-05,
      "loss": 1.1752,
      "step": 1078
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5159382820129395,
      "learning_rate": 1.9929719196243174e-05,
      "loss": 1.0208,
      "step": 1079
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.232361078262329,
      "learning_rate": 1.9929080279845382e-05,
      "loss": 1.0693,
      "step": 1080
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.468491792678833,
      "learning_rate": 1.9928441363447593e-05,
      "loss": 1.2051,
      "step": 1081
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9468070268630981,
      "learning_rate": 1.9927802447049805e-05,
      "loss": 0.8427,
      "step": 1082
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8259185552597046,
      "learning_rate": 1.9927163530652016e-05,
      "loss": 0.8565,
      "step": 1083
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3685483932495117,
      "learning_rate": 1.9926524614254225e-05,
      "loss": 0.2192,
      "step": 1084
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0667521953582764,
      "learning_rate": 1.9925885697856436e-05,
      "loss": 0.9076,
      "step": 1085
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.884813904762268,
      "learning_rate": 1.9925246781458648e-05,
      "loss": 0.9035,
      "step": 1086
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9715912342071533,
      "learning_rate": 1.992460786506086e-05,
      "loss": 0.8905,
      "step": 1087
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1496963500976562,
      "learning_rate": 1.9923968948663067e-05,
      "loss": 0.8782,
      "step": 1088
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.844049453735352,
      "learning_rate": 1.992333003226528e-05,
      "loss": 1.312,
      "step": 1089
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2252299785614014,
      "learning_rate": 1.992269111586749e-05,
      "loss": 0.8948,
      "step": 1090
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.6344361305236816,
      "learning_rate": 1.9922052199469702e-05,
      "loss": 0.9582,
      "step": 1091
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2723095417022705,
      "learning_rate": 1.992141328307191e-05,
      "loss": 0.8751,
      "step": 1092
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8073993921279907,
      "learning_rate": 1.992077436667412e-05,
      "loss": 0.8684,
      "step": 1093
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.280471086502075,
      "learning_rate": 1.9920135450276333e-05,
      "loss": 1.1446,
      "step": 1094
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0584652423858643,
      "learning_rate": 1.9919496533878545e-05,
      "loss": 1.3117,
      "step": 1095
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0404365062713623,
      "learning_rate": 1.9918857617480753e-05,
      "loss": 0.8461,
      "step": 1096
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5547919273376465,
      "learning_rate": 1.9918218701082964e-05,
      "loss": 1.1706,
      "step": 1097
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.358750820159912,
      "learning_rate": 1.9917579784685176e-05,
      "loss": 0.9114,
      "step": 1098
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.190530300140381,
      "learning_rate": 1.9916940868287388e-05,
      "loss": 0.8674,
      "step": 1099
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9901199340820312,
      "learning_rate": 1.9916301951889596e-05,
      "loss": 1.1206,
      "step": 1100
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9301748275756836,
      "learning_rate": 1.9915663035491807e-05,
      "loss": 1.0749,
      "step": 1101
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0998008251190186,
      "learning_rate": 1.991502411909402e-05,
      "loss": 0.8839,
      "step": 1102
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1153457164764404,
      "learning_rate": 1.991438520269623e-05,
      "loss": 0.9412,
      "step": 1103
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2345800399780273,
      "learning_rate": 1.991374628629844e-05,
      "loss": 0.9056,
      "step": 1104
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.973246455192566,
      "learning_rate": 1.991310736990065e-05,
      "loss": 0.9188,
      "step": 1105
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.7219038009643555,
      "learning_rate": 1.991246845350286e-05,
      "loss": 1.0162,
      "step": 1106
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9850891828536987,
      "learning_rate": 1.9911829537105073e-05,
      "loss": 1.2238,
      "step": 1107
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6960115432739258,
      "learning_rate": 1.991119062070728e-05,
      "loss": 0.8518,
      "step": 1108
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.502202272415161,
      "learning_rate": 1.9910551704309493e-05,
      "loss": 0.8865,
      "step": 1109
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7699934244155884,
      "learning_rate": 1.99099127879117e-05,
      "loss": 0.9989,
      "step": 1110
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9632651805877686,
      "learning_rate": 1.9909273871513916e-05,
      "loss": 1.2843,
      "step": 1111
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1487865447998047,
      "learning_rate": 1.9908634955116124e-05,
      "loss": 0.8583,
      "step": 1112
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.186727523803711,
      "learning_rate": 1.9907996038718336e-05,
      "loss": 1.0428,
      "step": 1113
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8667165040969849,
      "learning_rate": 1.9907357122320544e-05,
      "loss": 0.8698,
      "step": 1114
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.210991859436035,
      "learning_rate": 1.990671820592276e-05,
      "loss": 0.8838,
      "step": 1115
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.428396224975586,
      "learning_rate": 1.9906079289524967e-05,
      "loss": 1.0745,
      "step": 1116
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0416507720947266,
      "learning_rate": 1.990544037312718e-05,
      "loss": 0.7935,
      "step": 1117
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2813546657562256,
      "learning_rate": 1.9904801456729387e-05,
      "loss": 1.0096,
      "step": 1118
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7910680770874023,
      "learning_rate": 1.9904162540331598e-05,
      "loss": 0.831,
      "step": 1119
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9832797050476074,
      "learning_rate": 1.990352362393381e-05,
      "loss": 0.8213,
      "step": 1120
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2411911487579346,
      "learning_rate": 1.990288470753602e-05,
      "loss": 0.8407,
      "step": 1121
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.347659111022949,
      "learning_rate": 1.990224579113823e-05,
      "loss": 0.9017,
      "step": 1122
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5032784938812256,
      "learning_rate": 1.990160687474044e-05,
      "loss": 0.9128,
      "step": 1123
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2739763259887695,
      "learning_rate": 1.9900967958342653e-05,
      "loss": 1.1179,
      "step": 1124
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.858302116394043,
      "learning_rate": 1.9900329041944864e-05,
      "loss": 1.0397,
      "step": 1125
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.2642436027526855,
      "learning_rate": 1.9899690125547072e-05,
      "loss": 0.8385,
      "step": 1126
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7866331338882446,
      "learning_rate": 1.9899051209149284e-05,
      "loss": 0.7906,
      "step": 1127
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0439772605895996,
      "learning_rate": 1.9898412292751495e-05,
      "loss": 0.935,
      "step": 1128
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4016377925872803,
      "learning_rate": 1.9897773376353707e-05,
      "loss": 0.9854,
      "step": 1129
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.066676139831543,
      "learning_rate": 1.9897134459955915e-05,
      "loss": 0.8107,
      "step": 1130
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.067793369293213,
      "learning_rate": 1.9896495543558127e-05,
      "loss": 1.0609,
      "step": 1131
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.013837814331055,
      "learning_rate": 1.9895856627160338e-05,
      "loss": 1.1042,
      "step": 1132
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.694918155670166,
      "learning_rate": 1.989521771076255e-05,
      "loss": 0.6287,
      "step": 1133
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.877429008483887,
      "learning_rate": 1.9894578794364758e-05,
      "loss": 0.9751,
      "step": 1134
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1419804096221924,
      "learning_rate": 1.989393987796697e-05,
      "loss": 0.9234,
      "step": 1135
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.35927414894104,
      "learning_rate": 1.989330096156918e-05,
      "loss": 0.8489,
      "step": 1136
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2486572265625,
      "learning_rate": 1.9892662045171392e-05,
      "loss": 0.8694,
      "step": 1137
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1265687942504883,
      "learning_rate": 1.98920231287736e-05,
      "loss": 0.9316,
      "step": 1138
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.390369415283203,
      "learning_rate": 1.9891384212375812e-05,
      "loss": 0.8397,
      "step": 1139
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8663277626037598,
      "learning_rate": 1.9890745295978024e-05,
      "loss": 1.1896,
      "step": 1140
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1343705654144287,
      "learning_rate": 1.9890106379580235e-05,
      "loss": 0.8381,
      "step": 1141
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4907665252685547,
      "learning_rate": 1.9889467463182443e-05,
      "loss": 0.8344,
      "step": 1142
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0844850540161133,
      "learning_rate": 1.9888828546784655e-05,
      "loss": 0.7807,
      "step": 1143
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.410297155380249,
      "learning_rate": 1.9888189630386867e-05,
      "loss": 0.9261,
      "step": 1144
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9774106740951538,
      "learning_rate": 1.9887550713989078e-05,
      "loss": 0.8422,
      "step": 1145
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2684537172317505,
      "learning_rate": 1.9886911797591286e-05,
      "loss": 0.2538,
      "step": 1146
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.823617935180664,
      "learning_rate": 1.9886272881193498e-05,
      "loss": 0.8928,
      "step": 1147
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5775866508483887,
      "learning_rate": 1.9885633964795706e-05,
      "loss": 0.8895,
      "step": 1148
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.162088632583618,
      "learning_rate": 1.988499504839792e-05,
      "loss": 1.2035,
      "step": 1149
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5744786262512207,
      "learning_rate": 1.988435613200013e-05,
      "loss": 1.2231,
      "step": 1150
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9859652519226074,
      "learning_rate": 1.988371721560234e-05,
      "loss": 0.9973,
      "step": 1151
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8748703002929688,
      "learning_rate": 1.988307829920455e-05,
      "loss": 0.7275,
      "step": 1152
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.331425905227661,
      "learning_rate": 1.988243938280676e-05,
      "loss": 0.831,
      "step": 1153
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4098892211914062,
      "learning_rate": 1.9881800466408972e-05,
      "loss": 0.6796,
      "step": 1154
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.605154275894165,
      "learning_rate": 1.9881161550011183e-05,
      "loss": 0.9556,
      "step": 1155
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4450552463531494,
      "learning_rate": 1.988052263361339e-05,
      "loss": 0.6788,
      "step": 1156
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.544536590576172,
      "learning_rate": 1.9879883717215603e-05,
      "loss": 1.1879,
      "step": 1157
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.121704578399658,
      "learning_rate": 1.9879244800817815e-05,
      "loss": 1.0795,
      "step": 1158
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.013587713241577,
      "learning_rate": 1.9878605884420026e-05,
      "loss": 0.7762,
      "step": 1159
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.087172985076904,
      "learning_rate": 1.9877966968022234e-05,
      "loss": 1.1437,
      "step": 1160
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1560122966766357,
      "learning_rate": 1.9877328051624446e-05,
      "loss": 0.8211,
      "step": 1161
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9140461683273315,
      "learning_rate": 1.9876689135226657e-05,
      "loss": 0.8856,
      "step": 1162
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.066005229949951,
      "learning_rate": 1.987605021882887e-05,
      "loss": 1.1504,
      "step": 1163
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5657875537872314,
      "learning_rate": 1.9875411302431077e-05,
      "loss": 1.0275,
      "step": 1164
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.198387622833252,
      "learning_rate": 1.987477238603329e-05,
      "loss": 0.9506,
      "step": 1165
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9216283559799194,
      "learning_rate": 1.98741334696355e-05,
      "loss": 0.8627,
      "step": 1166
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.070709228515625,
      "learning_rate": 1.9873494553237712e-05,
      "loss": 0.9369,
      "step": 1167
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9520689249038696,
      "learning_rate": 1.987285563683992e-05,
      "loss": 0.8894,
      "step": 1168
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.338059425354004,
      "learning_rate": 1.987221672044213e-05,
      "loss": 0.3426,
      "step": 1169
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0062263011932373,
      "learning_rate": 1.9871577804044343e-05,
      "loss": 0.8661,
      "step": 1170
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.024712085723877,
      "learning_rate": 1.9870938887646555e-05,
      "loss": 1.5353,
      "step": 1171
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.712505578994751,
      "learning_rate": 1.9870299971248763e-05,
      "loss": 0.8268,
      "step": 1172
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.1798932552337646,
      "learning_rate": 1.9869661054850974e-05,
      "loss": 1.1737,
      "step": 1173
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.442825794219971,
      "learning_rate": 1.9869022138453186e-05,
      "loss": 0.9893,
      "step": 1174
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.085683345794678,
      "learning_rate": 1.9868383222055397e-05,
      "loss": 2.0205,
      "step": 1175
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4605251550674438,
      "learning_rate": 1.9867744305657606e-05,
      "loss": 0.2308,
      "step": 1176
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8983051776885986,
      "learning_rate": 1.9867105389259817e-05,
      "loss": 1.0656,
      "step": 1177
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2264814376831055,
      "learning_rate": 1.986646647286203e-05,
      "loss": 0.8686,
      "step": 1178
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9741642475128174,
      "learning_rate": 1.986582755646424e-05,
      "loss": 0.8811,
      "step": 1179
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.235365867614746,
      "learning_rate": 1.986518864006645e-05,
      "loss": 0.9828,
      "step": 1180
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6733317375183105,
      "learning_rate": 1.986454972366866e-05,
      "loss": 0.8344,
      "step": 1181
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.699023246765137,
      "learning_rate": 1.986391080727087e-05,
      "loss": 1.4097,
      "step": 1182
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7360268831253052,
      "learning_rate": 1.9863271890873083e-05,
      "loss": 0.8341,
      "step": 1183
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6385786533355713,
      "learning_rate": 1.986263297447529e-05,
      "loss": 1.0174,
      "step": 1184
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4276041984558105,
      "learning_rate": 1.9861994058077503e-05,
      "loss": 0.8426,
      "step": 1185
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2891409397125244,
      "learning_rate": 1.986135514167971e-05,
      "loss": 0.8833,
      "step": 1186
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1821978092193604,
      "learning_rate": 1.9860716225281926e-05,
      "loss": 0.8727,
      "step": 1187
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.111729145050049,
      "learning_rate": 1.9860077308884134e-05,
      "loss": 0.9489,
      "step": 1188
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0414366722106934,
      "learning_rate": 1.9859438392486345e-05,
      "loss": 0.9188,
      "step": 1189
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.926365375518799,
      "learning_rate": 1.9858799476088554e-05,
      "loss": 0.988,
      "step": 1190
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0451436042785645,
      "learning_rate": 1.9858160559690765e-05,
      "loss": 1.1382,
      "step": 1191
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.262140989303589,
      "learning_rate": 1.9857521643292977e-05,
      "loss": 0.8543,
      "step": 1192
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.780987024307251,
      "learning_rate": 1.9856882726895188e-05,
      "loss": 0.9191,
      "step": 1193
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9083267450332642,
      "learning_rate": 1.9856243810497396e-05,
      "loss": 0.86,
      "step": 1194
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4606246948242188,
      "learning_rate": 1.9855604894099608e-05,
      "loss": 0.8265,
      "step": 1195
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.092521905899048,
      "learning_rate": 1.985496597770182e-05,
      "loss": 0.8313,
      "step": 1196
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.101321220397949,
      "learning_rate": 1.985432706130403e-05,
      "loss": 0.8585,
      "step": 1197
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8640520572662354,
      "learning_rate": 1.985368814490624e-05,
      "loss": 0.8254,
      "step": 1198
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2429580688476562,
      "learning_rate": 1.985304922850845e-05,
      "loss": 0.8356,
      "step": 1199
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.253844738006592,
      "learning_rate": 1.9852410312110662e-05,
      "loss": 1.771,
      "step": 1200
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4679205417633057,
      "learning_rate": 1.9851771395712874e-05,
      "loss": 0.9008,
      "step": 1201
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7819736003875732,
      "learning_rate": 1.9851132479315082e-05,
      "loss": 0.8667,
      "step": 1202
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1856071949005127,
      "learning_rate": 1.9850493562917294e-05,
      "loss": 0.8174,
      "step": 1203
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3526554107666016,
      "learning_rate": 1.9849854646519505e-05,
      "loss": 0.9135,
      "step": 1204
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9950995445251465,
      "learning_rate": 1.9849215730121717e-05,
      "loss": 0.4066,
      "step": 1205
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.622030258178711,
      "learning_rate": 1.9848576813723925e-05,
      "loss": 1.4421,
      "step": 1206
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.662302017211914,
      "learning_rate": 1.9847937897326136e-05,
      "loss": 1.1404,
      "step": 1207
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6879258155822754,
      "learning_rate": 1.9847298980928348e-05,
      "loss": 1.2864,
      "step": 1208
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3447535037994385,
      "learning_rate": 1.984666006453056e-05,
      "loss": 1.0033,
      "step": 1209
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.137668609619141,
      "learning_rate": 1.9846021148132768e-05,
      "loss": 1.7255,
      "step": 1210
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.400848388671875,
      "learning_rate": 1.984538223173498e-05,
      "loss": 1.0036,
      "step": 1211
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1963346004486084,
      "learning_rate": 1.984474331533719e-05,
      "loss": 0.8999,
      "step": 1212
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5563948154449463,
      "learning_rate": 1.9844104398939402e-05,
      "loss": 0.9584,
      "step": 1213
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2327933311462402,
      "learning_rate": 1.984346548254161e-05,
      "loss": 0.9032,
      "step": 1214
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6154885292053223,
      "learning_rate": 1.9842826566143822e-05,
      "loss": 0.9821,
      "step": 1215
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.595482349395752,
      "learning_rate": 1.9842187649746034e-05,
      "loss": 0.7076,
      "step": 1216
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.914348125457764,
      "learning_rate": 1.9841548733348245e-05,
      "loss": 1.4635,
      "step": 1217
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.708934783935547,
      "learning_rate": 1.9840909816950453e-05,
      "loss": 1.3369,
      "step": 1218
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8871921300888062,
      "learning_rate": 1.9840270900552665e-05,
      "loss": 0.9262,
      "step": 1219
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8164371252059937,
      "learning_rate": 1.9839631984154873e-05,
      "loss": 0.8821,
      "step": 1220
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4724955558776855,
      "learning_rate": 1.9838993067757088e-05,
      "loss": 0.97,
      "step": 1221
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.972946643829346,
      "learning_rate": 1.9838354151359296e-05,
      "loss": 1.3388,
      "step": 1222
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8799099922180176,
      "learning_rate": 1.9837715234961508e-05,
      "loss": 1.1449,
      "step": 1223
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8477665185928345,
      "learning_rate": 1.9837076318563716e-05,
      "loss": 0.8784,
      "step": 1224
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8809276819229126,
      "learning_rate": 1.9836437402165927e-05,
      "loss": 0.8946,
      "step": 1225
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8942782878875732,
      "learning_rate": 1.983579848576814e-05,
      "loss": 0.8532,
      "step": 1226
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.661700963973999,
      "learning_rate": 1.983515956937035e-05,
      "loss": 1.1818,
      "step": 1227
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.060420274734497,
      "learning_rate": 1.983452065297256e-05,
      "loss": 0.9906,
      "step": 1228
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3925673961639404,
      "learning_rate": 1.983388173657477e-05,
      "loss": 1.0617,
      "step": 1229
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.576573848724365,
      "learning_rate": 1.983324282017698e-05,
      "loss": 1.379,
      "step": 1230
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7633429765701294,
      "learning_rate": 1.9832603903779193e-05,
      "loss": 0.7878,
      "step": 1231
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.666589617729187,
      "learning_rate": 1.98319649873814e-05,
      "loss": 0.3226,
      "step": 1232
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.29411506652832,
      "learning_rate": 1.9831326070983613e-05,
      "loss": 1.8352,
      "step": 1233
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5241448879241943,
      "learning_rate": 1.9830687154585824e-05,
      "loss": 0.8886,
      "step": 1234
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.381071090698242,
      "learning_rate": 1.9830048238188036e-05,
      "loss": 1.777,
      "step": 1235
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.38455867767334,
      "learning_rate": 1.9829409321790244e-05,
      "loss": 0.9753,
      "step": 1236
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.942826271057129,
      "learning_rate": 1.9828770405392456e-05,
      "loss": 1.3044,
      "step": 1237
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.884106159210205,
      "learning_rate": 1.9828131488994667e-05,
      "loss": 0.8886,
      "step": 1238
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.909729480743408,
      "learning_rate": 1.982749257259688e-05,
      "loss": 1.8482,
      "step": 1239
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.441164970397949,
      "learning_rate": 1.9826853656199087e-05,
      "loss": 1.155,
      "step": 1240
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.988142967224121,
      "learning_rate": 1.98262147398013e-05,
      "loss": 0.8916,
      "step": 1241
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7756644487380981,
      "learning_rate": 1.982557582340351e-05,
      "loss": 0.905,
      "step": 1242
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.502150535583496,
      "learning_rate": 1.982493690700572e-05,
      "loss": 1.0628,
      "step": 1243
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8136825561523438,
      "learning_rate": 1.982429799060793e-05,
      "loss": 0.9823,
      "step": 1244
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.223601818084717,
      "learning_rate": 1.982365907421014e-05,
      "loss": 1.195,
      "step": 1245
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.086151361465454,
      "learning_rate": 1.9823020157812353e-05,
      "loss": 0.9422,
      "step": 1246
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6465563774108887,
      "learning_rate": 1.9822381241414564e-05,
      "loss": 0.6836,
      "step": 1247
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.437141418457031,
      "learning_rate": 1.9821742325016773e-05,
      "loss": 1.2096,
      "step": 1248
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.323317527770996,
      "learning_rate": 1.9821103408618984e-05,
      "loss": 0.8194,
      "step": 1249
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2816388607025146,
      "learning_rate": 1.9820464492221196e-05,
      "loss": 0.3286,
      "step": 1250
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6244580745697021,
      "learning_rate": 1.9819825575823407e-05,
      "loss": 0.863,
      "step": 1251
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9776182174682617,
      "learning_rate": 1.9819186659425615e-05,
      "loss": 0.7782,
      "step": 1252
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1287877559661865,
      "learning_rate": 1.9818547743027827e-05,
      "loss": 0.819,
      "step": 1253
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9443957805633545,
      "learning_rate": 1.981790882663004e-05,
      "loss": 0.9925,
      "step": 1254
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3520658016204834,
      "learning_rate": 1.981726991023225e-05,
      "loss": 1.3504,
      "step": 1255
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.241909980773926,
      "learning_rate": 1.9816630993834458e-05,
      "loss": 0.8887,
      "step": 1256
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2529712915420532,
      "learning_rate": 1.981599207743667e-05,
      "loss": 0.384,
      "step": 1257
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.293728828430176,
      "learning_rate": 1.9815353161038878e-05,
      "loss": 1.2967,
      "step": 1258
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.244767904281616,
      "learning_rate": 1.9814714244641093e-05,
      "loss": 0.9786,
      "step": 1259
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4369518756866455,
      "learning_rate": 1.98140753282433e-05,
      "loss": 0.9809,
      "step": 1260
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7622766494750977,
      "learning_rate": 1.9813436411845512e-05,
      "loss": 0.6508,
      "step": 1261
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1483566761016846,
      "learning_rate": 1.981279749544772e-05,
      "loss": 0.8514,
      "step": 1262
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.274786472320557,
      "learning_rate": 1.9812158579049932e-05,
      "loss": 0.7014,
      "step": 1263
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.628084897994995,
      "learning_rate": 1.9811519662652144e-05,
      "loss": 1.2501,
      "step": 1264
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.596827983856201,
      "learning_rate": 1.9810880746254355e-05,
      "loss": 1.1584,
      "step": 1265
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.180609941482544,
      "learning_rate": 1.9810241829856563e-05,
      "loss": 0.8684,
      "step": 1266
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.771966934204102,
      "learning_rate": 1.9809602913458775e-05,
      "loss": 1.5137,
      "step": 1267
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9216457605361938,
      "learning_rate": 1.9808963997060987e-05,
      "loss": 0.8505,
      "step": 1268
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6706936359405518,
      "learning_rate": 1.9808325080663198e-05,
      "loss": 0.9209,
      "step": 1269
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.246652126312256,
      "learning_rate": 1.9807686164265406e-05,
      "loss": 1.1061,
      "step": 1270
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.387465715408325,
      "learning_rate": 1.9807047247867618e-05,
      "loss": 0.958,
      "step": 1271
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.616069793701172,
      "learning_rate": 1.980640833146983e-05,
      "loss": 1.3829,
      "step": 1272
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3469505310058594,
      "learning_rate": 1.980576941507204e-05,
      "loss": 0.8955,
      "step": 1273
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.582388877868652,
      "learning_rate": 1.980513049867425e-05,
      "loss": 1.8794,
      "step": 1274
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4652645587921143,
      "learning_rate": 1.980449158227646e-05,
      "loss": 0.8117,
      "step": 1275
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4328148365020752,
      "learning_rate": 1.9803852665878672e-05,
      "loss": 0.287,
      "step": 1276
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1388587951660156,
      "learning_rate": 1.9803213749480884e-05,
      "loss": 0.7806,
      "step": 1277
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.315118312835693,
      "learning_rate": 1.9802574833083092e-05,
      "loss": 0.9852,
      "step": 1278
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1047255992889404,
      "learning_rate": 1.9801935916685303e-05,
      "loss": 0.8635,
      "step": 1279
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1230990886688232,
      "learning_rate": 1.9801297000287515e-05,
      "loss": 0.8615,
      "step": 1280
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6248185634613037,
      "learning_rate": 1.9800658083889726e-05,
      "loss": 1.0345,
      "step": 1281
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.619817733764648,
      "learning_rate": 1.9800019167491935e-05,
      "loss": 1.19,
      "step": 1282
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3193247318267822,
      "learning_rate": 1.9799380251094146e-05,
      "loss": 1.2881,
      "step": 1283
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3238799571990967,
      "learning_rate": 1.9798741334696358e-05,
      "loss": 0.7588,
      "step": 1284
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.597584247589111,
      "learning_rate": 1.979810241829857e-05,
      "loss": 1.1414,
      "step": 1285
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6571152210235596,
      "learning_rate": 1.9797463501900777e-05,
      "loss": 0.9356,
      "step": 1286
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.1365151405334473,
      "learning_rate": 1.979682458550299e-05,
      "loss": 0.9655,
      "step": 1287
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7841544151306152,
      "learning_rate": 1.97961856691052e-05,
      "loss": 0.7777,
      "step": 1288
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.279428243637085,
      "learning_rate": 1.9795546752707412e-05,
      "loss": 0.9847,
      "step": 1289
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9230523109436035,
      "learning_rate": 1.979490783630962e-05,
      "loss": 0.9405,
      "step": 1290
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.312441349029541,
      "learning_rate": 1.9794268919911832e-05,
      "loss": 0.2399,
      "step": 1291
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.472478151321411,
      "learning_rate": 1.979363000351404e-05,
      "loss": 0.9005,
      "step": 1292
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6236836910247803,
      "learning_rate": 1.9792991087116255e-05,
      "loss": 0.8975,
      "step": 1293
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8113863468170166,
      "learning_rate": 1.9792352170718463e-05,
      "loss": 0.2534,
      "step": 1294
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4438223838806152,
      "learning_rate": 1.9791713254320675e-05,
      "loss": 0.9438,
      "step": 1295
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.589693307876587,
      "learning_rate": 1.9791074337922883e-05,
      "loss": 0.9339,
      "step": 1296
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.907613754272461,
      "learning_rate": 1.9790435421525098e-05,
      "loss": 0.7725,
      "step": 1297
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.314289569854736,
      "learning_rate": 1.9789796505127306e-05,
      "loss": 0.9921,
      "step": 1298
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4702541828155518,
      "learning_rate": 1.9789157588729517e-05,
      "loss": 1.0287,
      "step": 1299
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.006147861480713,
      "learning_rate": 1.9788518672331726e-05,
      "loss": 1.6076,
      "step": 1300
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.005297899246216,
      "learning_rate": 1.9787879755933937e-05,
      "loss": 0.9856,
      "step": 1301
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1986241340637207,
      "learning_rate": 1.978724083953615e-05,
      "loss": 0.9545,
      "step": 1302
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.274054527282715,
      "learning_rate": 1.978660192313836e-05,
      "loss": 0.9931,
      "step": 1303
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.512364625930786,
      "learning_rate": 1.978596300674057e-05,
      "loss": 0.887,
      "step": 1304
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2911531925201416,
      "learning_rate": 1.978532409034278e-05,
      "loss": 0.867,
      "step": 1305
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9385029077529907,
      "learning_rate": 1.978468517394499e-05,
      "loss": 0.9398,
      "step": 1306
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2947680950164795,
      "learning_rate": 1.9784046257547203e-05,
      "loss": 0.8131,
      "step": 1307
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9985542297363281,
      "learning_rate": 1.978340734114941e-05,
      "loss": 0.823,
      "step": 1308
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.912432670593262,
      "learning_rate": 1.9782768424751623e-05,
      "loss": 2.1309,
      "step": 1309
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.6220173835754395,
      "learning_rate": 1.9782129508353834e-05,
      "loss": 1.539,
      "step": 1310
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4271605014801025,
      "learning_rate": 1.9781490591956046e-05,
      "loss": 0.8459,
      "step": 1311
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.566188335418701,
      "learning_rate": 1.9780851675558254e-05,
      "loss": 0.8101,
      "step": 1312
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.837481498718262,
      "learning_rate": 1.9780212759160465e-05,
      "loss": 0.994,
      "step": 1313
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3608920574188232,
      "learning_rate": 1.9779573842762677e-05,
      "loss": 0.9061,
      "step": 1314
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.182598829269409,
      "learning_rate": 1.977893492636489e-05,
      "loss": 0.7581,
      "step": 1315
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.207989692687988,
      "learning_rate": 1.9778296009967097e-05,
      "loss": 1.7948,
      "step": 1316
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.576695442199707,
      "learning_rate": 1.9777657093569308e-05,
      "loss": 1.1977,
      "step": 1317
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9151201248168945,
      "learning_rate": 1.977701817717152e-05,
      "loss": 1.0249,
      "step": 1318
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4132208824157715,
      "learning_rate": 1.977637926077373e-05,
      "loss": 0.8984,
      "step": 1319
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.945712685585022,
      "learning_rate": 1.977574034437594e-05,
      "loss": 0.8588,
      "step": 1320
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.202361822128296,
      "learning_rate": 1.977510142797815e-05,
      "loss": 0.7942,
      "step": 1321
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.207347869873047,
      "learning_rate": 1.9774462511580363e-05,
      "loss": 0.9067,
      "step": 1322
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9290244579315186,
      "learning_rate": 1.9773823595182574e-05,
      "loss": 0.8199,
      "step": 1323
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.245739698410034,
      "learning_rate": 1.9773184678784782e-05,
      "loss": 0.8808,
      "step": 1324
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0301074981689453,
      "learning_rate": 1.9772545762386994e-05,
      "loss": 0.8036,
      "step": 1325
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.967079520225525,
      "learning_rate": 1.9771906845989205e-05,
      "loss": 0.893,
      "step": 1326
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5874974727630615,
      "learning_rate": 1.9771267929591417e-05,
      "loss": 0.8992,
      "step": 1327
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9899120330810547,
      "learning_rate": 1.9770629013193625e-05,
      "loss": 1.1298,
      "step": 1328
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7932802438735962,
      "learning_rate": 1.9769990096795837e-05,
      "loss": 0.9333,
      "step": 1329
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9492028951644897,
      "learning_rate": 1.9769351180398045e-05,
      "loss": 0.8283,
      "step": 1330
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.157676935195923,
      "learning_rate": 1.976871226400026e-05,
      "loss": 0.9441,
      "step": 1331
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.622736692428589,
      "learning_rate": 1.9768073347602468e-05,
      "loss": 0.8985,
      "step": 1332
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3581161499023438,
      "learning_rate": 1.976743443120468e-05,
      "loss": 0.8442,
      "step": 1333
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.265454649925232,
      "learning_rate": 1.9766795514806888e-05,
      "loss": 0.2039,
      "step": 1334
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.377545118331909,
      "learning_rate": 1.97661565984091e-05,
      "loss": 0.9518,
      "step": 1335
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4994454383850098,
      "learning_rate": 1.976551768201131e-05,
      "loss": 0.8757,
      "step": 1336
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9064321517944336,
      "learning_rate": 1.9764878765613522e-05,
      "loss": 0.9278,
      "step": 1337
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.072747230529785,
      "learning_rate": 1.976423984921573e-05,
      "loss": 1.1186,
      "step": 1338
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.100456237792969,
      "learning_rate": 1.9763600932817942e-05,
      "loss": 1.3404,
      "step": 1339
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.011589288711548,
      "learning_rate": 1.9762962016420154e-05,
      "loss": 0.8147,
      "step": 1340
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.173699855804443,
      "learning_rate": 1.9762323100022365e-05,
      "loss": 0.9879,
      "step": 1341
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8074464797973633,
      "learning_rate": 1.9761684183624573e-05,
      "loss": 0.9235,
      "step": 1342
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.341431140899658,
      "learning_rate": 1.9761045267226785e-05,
      "loss": 1.0283,
      "step": 1343
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3938775062561035,
      "learning_rate": 1.9760406350828996e-05,
      "loss": 0.8528,
      "step": 1344
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.504701852798462,
      "learning_rate": 1.9759767434431208e-05,
      "loss": 0.91,
      "step": 1345
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6094257831573486,
      "learning_rate": 1.9759128518033416e-05,
      "loss": 1.1041,
      "step": 1346
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.094193696975708,
      "learning_rate": 1.9758489601635628e-05,
      "loss": 0.8367,
      "step": 1347
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.461431503295898,
      "learning_rate": 1.975785068523784e-05,
      "loss": 2.0616,
      "step": 1348
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6400063037872314,
      "learning_rate": 1.975721176884005e-05,
      "loss": 0.9047,
      "step": 1349
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.362716197967529,
      "learning_rate": 1.975657285244226e-05,
      "loss": 1.0585,
      "step": 1350
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.829332947731018,
      "learning_rate": 1.975593393604447e-05,
      "loss": 0.3276,
      "step": 1351
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4211745262145996,
      "learning_rate": 1.9755295019646682e-05,
      "loss": 0.8685,
      "step": 1352
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.212749719619751,
      "learning_rate": 1.9754656103248893e-05,
      "loss": 0.9763,
      "step": 1353
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.2090847492218018,
      "learning_rate": 1.97540171868511e-05,
      "loss": 1.0552,
      "step": 1354
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.28576397895813,
      "learning_rate": 1.9753378270453313e-05,
      "loss": 1.2131,
      "step": 1355
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5539543628692627,
      "learning_rate": 1.9752739354055525e-05,
      "loss": 1.2679,
      "step": 1356
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.251633882522583,
      "learning_rate": 1.9752100437657733e-05,
      "loss": 0.9259,
      "step": 1357
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.151745319366455,
      "learning_rate": 1.9751461521259944e-05,
      "loss": 0.8484,
      "step": 1358
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.849780082702637,
      "learning_rate": 1.9750822604862156e-05,
      "loss": 1.1982,
      "step": 1359
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.260332107543945,
      "learning_rate": 1.9750183688464368e-05,
      "loss": 1.0861,
      "step": 1360
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.682803153991699,
      "learning_rate": 1.9749544772066576e-05,
      "loss": 1.4711,
      "step": 1361
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7916927337646484,
      "learning_rate": 1.9748905855668787e-05,
      "loss": 0.9578,
      "step": 1362
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0576868057250977,
      "learning_rate": 1.9748266939270995e-05,
      "loss": 0.8068,
      "step": 1363
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.137566328048706,
      "learning_rate": 1.974762802287321e-05,
      "loss": 0.8772,
      "step": 1364
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4613561630249023,
      "learning_rate": 1.974698910647542e-05,
      "loss": 1.0537,
      "step": 1365
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3666898012161255,
      "learning_rate": 1.974635019007763e-05,
      "loss": 0.268,
      "step": 1366
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2395427227020264,
      "learning_rate": 1.9745711273679838e-05,
      "loss": 0.9943,
      "step": 1367
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.349824905395508,
      "learning_rate": 1.974507235728205e-05,
      "loss": 1.0857,
      "step": 1368
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.285223484039307,
      "learning_rate": 1.974443344088426e-05,
      "loss": 1.0754,
      "step": 1369
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.1030354499816895,
      "learning_rate": 1.9743794524486473e-05,
      "loss": 0.7634,
      "step": 1370
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.924032688140869,
      "learning_rate": 1.974315560808868e-05,
      "loss": 0.9123,
      "step": 1371
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8223509788513184,
      "learning_rate": 1.9742516691690893e-05,
      "loss": 1.19,
      "step": 1372
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.260824203491211,
      "learning_rate": 1.9741877775293104e-05,
      "loss": 0.9299,
      "step": 1373
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3434970378875732,
      "learning_rate": 1.9741238858895316e-05,
      "loss": 1.0026,
      "step": 1374
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2354695796966553,
      "learning_rate": 1.9740599942497524e-05,
      "loss": 0.8416,
      "step": 1375
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9778947830200195,
      "learning_rate": 1.9739961026099735e-05,
      "loss": 0.899,
      "step": 1376
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8323067426681519,
      "learning_rate": 1.9739322109701947e-05,
      "loss": 0.8257,
      "step": 1377
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0401976108551025,
      "learning_rate": 1.973868319330416e-05,
      "loss": 0.8018,
      "step": 1378
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.83578360080719,
      "learning_rate": 1.9738044276906367e-05,
      "loss": 0.8681,
      "step": 1379
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3035266399383545,
      "learning_rate": 1.9737405360508578e-05,
      "loss": 1.0307,
      "step": 1380
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7847301959991455,
      "learning_rate": 1.973676644411079e-05,
      "loss": 0.8831,
      "step": 1381
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.296405553817749,
      "learning_rate": 1.9736127527713e-05,
      "loss": 0.7955,
      "step": 1382
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.5198750495910645,
      "learning_rate": 1.973548861131521e-05,
      "loss": 1.3742,
      "step": 1383
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7219762802124023,
      "learning_rate": 1.973484969491742e-05,
      "loss": 1.0913,
      "step": 1384
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.521392822265625,
      "learning_rate": 1.9734210778519632e-05,
      "loss": 0.7563,
      "step": 1385
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.285600423812866,
      "learning_rate": 1.9733571862121844e-05,
      "loss": 1.3859,
      "step": 1386
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2658593654632568,
      "learning_rate": 1.9732932945724052e-05,
      "loss": 0.3277,
      "step": 1387
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5373642444610596,
      "learning_rate": 1.9732294029326264e-05,
      "loss": 1.0268,
      "step": 1388
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4802207946777344,
      "learning_rate": 1.9731655112928475e-05,
      "loss": 0.971,
      "step": 1389
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7581801414489746,
      "learning_rate": 1.9731016196530687e-05,
      "loss": 0.7756,
      "step": 1390
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.317678928375244,
      "learning_rate": 1.9730377280132895e-05,
      "loss": 0.9346,
      "step": 1391
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2278802394866943,
      "learning_rate": 1.9729738363735107e-05,
      "loss": 0.9559,
      "step": 1392
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.55067777633667,
      "learning_rate": 1.9729099447337318e-05,
      "loss": 1.7519,
      "step": 1393
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.029666423797607,
      "learning_rate": 1.972846053093953e-05,
      "loss": 1.2841,
      "step": 1394
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.397387742996216,
      "learning_rate": 1.9727821614541738e-05,
      "loss": 1.2693,
      "step": 1395
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.802762508392334,
      "learning_rate": 1.972718269814395e-05,
      "loss": 0.9522,
      "step": 1396
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.441673994064331,
      "learning_rate": 1.9726543781746158e-05,
      "loss": 0.9478,
      "step": 1397
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.046999454498291,
      "learning_rate": 1.9725904865348372e-05,
      "loss": 1.0351,
      "step": 1398
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9910154342651367,
      "learning_rate": 1.972526594895058e-05,
      "loss": 0.7938,
      "step": 1399
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3287620544433594,
      "learning_rate": 1.9724627032552792e-05,
      "loss": 0.8692,
      "step": 1400
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3643336296081543,
      "learning_rate": 1.9723988116155e-05,
      "loss": 1.0637,
      "step": 1401
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.480466365814209,
      "learning_rate": 1.9723349199757212e-05,
      "loss": 1.2351,
      "step": 1402
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.004699230194092,
      "learning_rate": 1.9722710283359423e-05,
      "loss": 1.2049,
      "step": 1403
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0369157791137695,
      "learning_rate": 1.9722071366961635e-05,
      "loss": 0.9151,
      "step": 1404
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7266058921813965,
      "learning_rate": 1.9721432450563843e-05,
      "loss": 0.9523,
      "step": 1405
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.695828437805176,
      "learning_rate": 1.9720793534166055e-05,
      "loss": 1.0403,
      "step": 1406
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.580074310302734,
      "learning_rate": 1.9720154617768266e-05,
      "loss": 1.1946,
      "step": 1407
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.58383321762085,
      "learning_rate": 1.9719515701370478e-05,
      "loss": 1.6337,
      "step": 1408
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.308137893676758,
      "learning_rate": 1.9718876784972686e-05,
      "loss": 0.9531,
      "step": 1409
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7379796504974365,
      "learning_rate": 1.9718237868574897e-05,
      "loss": 1.0535,
      "step": 1410
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0137274265289307,
      "learning_rate": 1.971759895217711e-05,
      "loss": 0.9666,
      "step": 1411
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.690236568450928,
      "learning_rate": 1.971696003577932e-05,
      "loss": 1.0341,
      "step": 1412
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.931986331939697,
      "learning_rate": 1.971632111938153e-05,
      "loss": 1.007,
      "step": 1413
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.568826913833618,
      "learning_rate": 1.971568220298374e-05,
      "loss": 0.9738,
      "step": 1414
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.464833736419678,
      "learning_rate": 1.9715043286585952e-05,
      "loss": 0.9485,
      "step": 1415
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4891233444213867,
      "learning_rate": 1.9714404370188163e-05,
      "loss": 0.7153,
      "step": 1416
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1243786811828613,
      "learning_rate": 1.971376545379037e-05,
      "loss": 0.9569,
      "step": 1417
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1884756088256836,
      "learning_rate": 1.9713126537392583e-05,
      "loss": 0.8942,
      "step": 1418
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.739658832550049,
      "learning_rate": 1.9712487620994795e-05,
      "loss": 1.4497,
      "step": 1419
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.02050518989563,
      "learning_rate": 1.9711848704597006e-05,
      "loss": 0.8906,
      "step": 1420
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8636202812194824,
      "learning_rate": 1.9711209788199214e-05,
      "loss": 0.8597,
      "step": 1421
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.129124402999878,
      "learning_rate": 1.9710570871801426e-05,
      "loss": 0.8764,
      "step": 1422
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.945908308029175,
      "learning_rate": 1.9709931955403637e-05,
      "loss": 0.8018,
      "step": 1423
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4775450229644775,
      "learning_rate": 1.970929303900585e-05,
      "loss": 0.8911,
      "step": 1424
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.079346179962158,
      "learning_rate": 1.9708654122608057e-05,
      "loss": 0.8794,
      "step": 1425
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7894206047058105,
      "learning_rate": 1.970801520621027e-05,
      "loss": 1.0934,
      "step": 1426
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.575944423675537,
      "learning_rate": 1.970737628981248e-05,
      "loss": 0.8005,
      "step": 1427
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.542691230773926,
      "learning_rate": 1.9706737373414692e-05,
      "loss": 1.1298,
      "step": 1428
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3138222694396973,
      "learning_rate": 1.97060984570169e-05,
      "loss": 0.8619,
      "step": 1429
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3099966049194336,
      "learning_rate": 1.970545954061911e-05,
      "loss": 0.8537,
      "step": 1430
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9829639196395874,
      "learning_rate": 1.9704820624221323e-05,
      "loss": 0.8448,
      "step": 1431
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2991149425506592,
      "learning_rate": 1.9704181707823535e-05,
      "loss": 0.339,
      "step": 1432
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1184778213500977,
      "learning_rate": 1.9703542791425743e-05,
      "loss": 0.8767,
      "step": 1433
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.264078140258789,
      "learning_rate": 1.9702903875027954e-05,
      "loss": 1.407,
      "step": 1434
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.24450421333313,
      "learning_rate": 1.9702264958630162e-05,
      "loss": 0.8424,
      "step": 1435
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.027451276779175,
      "learning_rate": 1.9701626042232377e-05,
      "loss": 0.8524,
      "step": 1436
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.694553852081299,
      "learning_rate": 1.9700987125834585e-05,
      "loss": 1.2324,
      "step": 1437
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.513376235961914,
      "learning_rate": 1.9700348209436797e-05,
      "loss": 0.816,
      "step": 1438
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8468525409698486,
      "learning_rate": 1.9699709293039005e-05,
      "loss": 0.7886,
      "step": 1439
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4318056106567383,
      "learning_rate": 1.9699070376641217e-05,
      "loss": 0.876,
      "step": 1440
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.460547924041748,
      "learning_rate": 1.9698431460243428e-05,
      "loss": 1.0357,
      "step": 1441
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.092761516571045,
      "learning_rate": 1.969779254384564e-05,
      "loss": 0.8803,
      "step": 1442
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6966885328292847,
      "learning_rate": 1.9697153627447848e-05,
      "loss": 0.7738,
      "step": 1443
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6615915298461914,
      "learning_rate": 1.969651471105006e-05,
      "loss": 1.0722,
      "step": 1444
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.2521283626556396,
      "learning_rate": 1.969587579465227e-05,
      "loss": 1.0075,
      "step": 1445
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0698025226593018,
      "learning_rate": 1.9695236878254483e-05,
      "loss": 0.8952,
      "step": 1446
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1987249851226807,
      "learning_rate": 1.969459796185669e-05,
      "loss": 1.3447,
      "step": 1447
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8244071006774902,
      "learning_rate": 1.9693959045458902e-05,
      "loss": 0.8773,
      "step": 1448
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.126870632171631,
      "learning_rate": 1.9693320129061114e-05,
      "loss": 0.2513,
      "step": 1449
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.724263668060303,
      "learning_rate": 1.9692681212663325e-05,
      "loss": 0.9029,
      "step": 1450
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0488429069519043,
      "learning_rate": 1.9692042296265534e-05,
      "loss": 0.8388,
      "step": 1451
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0348687171936035,
      "learning_rate": 1.9691403379867745e-05,
      "loss": 0.9119,
      "step": 1452
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.2710747718811035,
      "learning_rate": 1.9690764463469957e-05,
      "loss": 1.7757,
      "step": 1453
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2728793621063232,
      "learning_rate": 1.9690125547072168e-05,
      "loss": 1.0467,
      "step": 1454
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.1988677978515625,
      "learning_rate": 1.9689486630674376e-05,
      "loss": 1.1393,
      "step": 1455
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.80509352684021,
      "learning_rate": 1.9688847714276588e-05,
      "loss": 0.9916,
      "step": 1456
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8889009952545166,
      "learning_rate": 1.96882087978788e-05,
      "loss": 0.8535,
      "step": 1457
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.3671793937683105,
      "learning_rate": 1.968756988148101e-05,
      "loss": 1.0822,
      "step": 1458
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.633622884750366,
      "learning_rate": 1.968693096508322e-05,
      "loss": 0.953,
      "step": 1459
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.978630304336548,
      "learning_rate": 1.968629204868543e-05,
      "loss": 0.9,
      "step": 1460
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8640953302383423,
      "learning_rate": 1.9685653132287642e-05,
      "loss": 0.9336,
      "step": 1461
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8255741596221924,
      "learning_rate": 1.9685014215889854e-05,
      "loss": 1.072,
      "step": 1462
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.306039333343506,
      "learning_rate": 1.9684375299492062e-05,
      "loss": 0.8972,
      "step": 1463
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1980698108673096,
      "learning_rate": 1.9683736383094274e-05,
      "loss": 0.8618,
      "step": 1464
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.835718154907227,
      "learning_rate": 1.9683097466696485e-05,
      "loss": 1.3092,
      "step": 1465
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.240163564682007,
      "learning_rate": 1.9682458550298697e-05,
      "loss": 0.9414,
      "step": 1466
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4317984580993652,
      "learning_rate": 1.9681819633900905e-05,
      "loss": 1.3578,
      "step": 1467
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5087358951568604,
      "learning_rate": 1.9681180717503116e-05,
      "loss": 1.2013,
      "step": 1468
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2971882820129395,
      "learning_rate": 1.9680541801105325e-05,
      "loss": 0.9486,
      "step": 1469
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.042153835296631,
      "learning_rate": 1.967990288470754e-05,
      "loss": 0.8294,
      "step": 1470
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.034128189086914,
      "learning_rate": 1.9679263968309748e-05,
      "loss": 0.82,
      "step": 1471
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3896167278289795,
      "learning_rate": 1.967862505191196e-05,
      "loss": 0.8729,
      "step": 1472
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.427447557449341,
      "learning_rate": 1.9677986135514167e-05,
      "loss": 0.9174,
      "step": 1473
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.71742844581604,
      "learning_rate": 1.9677347219116382e-05,
      "loss": 1.3697,
      "step": 1474
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2084364891052246,
      "learning_rate": 1.967670830271859e-05,
      "loss": 0.8824,
      "step": 1475
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4609845876693726,
      "learning_rate": 1.9676069386320802e-05,
      "loss": 0.252,
      "step": 1476
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3815882205963135,
      "learning_rate": 1.967543046992301e-05,
      "loss": 0.9913,
      "step": 1477
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.801792621612549,
      "learning_rate": 1.967479155352522e-05,
      "loss": 0.9922,
      "step": 1478
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.389023303985596,
      "learning_rate": 1.9674152637127433e-05,
      "loss": 1.2158,
      "step": 1479
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.375673294067383,
      "learning_rate": 1.9673513720729645e-05,
      "loss": 1.0456,
      "step": 1480
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.440507411956787,
      "learning_rate": 1.9672874804331853e-05,
      "loss": 1.2415,
      "step": 1481
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4373362064361572,
      "learning_rate": 1.9672235887934064e-05,
      "loss": 1.0884,
      "step": 1482
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1201670169830322,
      "learning_rate": 1.9671596971536276e-05,
      "loss": 0.9655,
      "step": 1483
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.8235445022583,
      "learning_rate": 1.9670958055138488e-05,
      "loss": 1.0268,
      "step": 1484
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4103903770446777,
      "learning_rate": 1.9670319138740696e-05,
      "loss": 0.9,
      "step": 1485
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.402499198913574,
      "learning_rate": 1.9669680222342907e-05,
      "loss": 0.9195,
      "step": 1486
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8270161151885986,
      "learning_rate": 1.966904130594512e-05,
      "loss": 0.8976,
      "step": 1487
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.445386290550232,
      "learning_rate": 1.966840238954733e-05,
      "loss": 0.3246,
      "step": 1488
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9550397396087646,
      "learning_rate": 1.966776347314954e-05,
      "loss": 0.8445,
      "step": 1489
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4971861839294434,
      "learning_rate": 1.966712455675175e-05,
      "loss": 0.865,
      "step": 1490
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2263031005859375,
      "learning_rate": 1.966648564035396e-05,
      "loss": 0.9837,
      "step": 1491
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3732194900512695,
      "learning_rate": 1.9665846723956173e-05,
      "loss": 0.9786,
      "step": 1492
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.142939329147339,
      "learning_rate": 1.966520780755838e-05,
      "loss": 0.8701,
      "step": 1493
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7044360637664795,
      "learning_rate": 1.9664568891160593e-05,
      "loss": 1.2032,
      "step": 1494
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9477543830871582,
      "learning_rate": 1.9663929974762804e-05,
      "loss": 0.8702,
      "step": 1495
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8378499746322632,
      "learning_rate": 1.9663291058365016e-05,
      "loss": 0.9123,
      "step": 1496
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1190526485443115,
      "learning_rate": 1.9662652141967224e-05,
      "loss": 0.924,
      "step": 1497
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4799742698669434,
      "learning_rate": 1.9662013225569436e-05,
      "loss": 0.9917,
      "step": 1498
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.255274295806885,
      "learning_rate": 1.9661374309171647e-05,
      "loss": 1.3006,
      "step": 1499
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4533650875091553,
      "learning_rate": 1.966073539277386e-05,
      "loss": 0.8864,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3909106254577637,
      "learning_rate": 1.9660096476376067e-05,
      "loss": 0.2601,
      "step": 1501
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9407470226287842,
      "learning_rate": 1.965945755997828e-05,
      "loss": 0.9895,
      "step": 1502
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8835687637329102,
      "learning_rate": 1.965881864358049e-05,
      "loss": 0.7605,
      "step": 1503
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.565187454223633,
      "learning_rate": 1.96581797271827e-05,
      "loss": 0.8894,
      "step": 1504
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4084603786468506,
      "learning_rate": 1.965754081078491e-05,
      "loss": 0.9627,
      "step": 1505
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.055267333984375,
      "learning_rate": 1.965690189438712e-05,
      "loss": 0.8076,
      "step": 1506
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3316853046417236,
      "learning_rate": 1.965626297798933e-05,
      "loss": 0.909,
      "step": 1507
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.326824188232422,
      "learning_rate": 1.9655624061591544e-05,
      "loss": 1.5315,
      "step": 1508
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.36887264251709,
      "learning_rate": 1.9654985145193753e-05,
      "loss": 0.9746,
      "step": 1509
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.199766159057617,
      "learning_rate": 1.9654346228795964e-05,
      "loss": 1.6673,
      "step": 1510
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.735211133956909,
      "learning_rate": 1.9653707312398172e-05,
      "loss": 0.9931,
      "step": 1511
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5572612285614014,
      "learning_rate": 1.9653068396000384e-05,
      "loss": 0.8007,
      "step": 1512
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1183769702911377,
      "learning_rate": 1.9652429479602595e-05,
      "loss": 0.8804,
      "step": 1513
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8595448732376099,
      "learning_rate": 1.9651790563204807e-05,
      "loss": 0.771,
      "step": 1514
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1974751949310303,
      "learning_rate": 1.9651151646807015e-05,
      "loss": 0.8089,
      "step": 1515
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.164168357849121,
      "learning_rate": 1.9650512730409227e-05,
      "loss": 0.8441,
      "step": 1516
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4444024562835693,
      "learning_rate": 1.9649873814011438e-05,
      "loss": 0.7412,
      "step": 1517
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0244572162628174,
      "learning_rate": 1.964923489761365e-05,
      "loss": 0.9237,
      "step": 1518
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.229496479034424,
      "learning_rate": 1.9648595981215858e-05,
      "loss": 1.7333,
      "step": 1519
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.673903465270996,
      "learning_rate": 1.964795706481807e-05,
      "loss": 1.027,
      "step": 1520
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.574796676635742,
      "learning_rate": 1.964731814842028e-05,
      "loss": 1.0319,
      "step": 1521
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.988382339477539,
      "learning_rate": 1.9646679232022492e-05,
      "loss": 0.855,
      "step": 1522
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.131941795349121,
      "learning_rate": 1.96460403156247e-05,
      "loss": 0.9753,
      "step": 1523
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.939652442932129,
      "learning_rate": 1.9645401399226912e-05,
      "loss": 0.8299,
      "step": 1524
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2229249477386475,
      "learning_rate": 1.9644762482829124e-05,
      "loss": 0.9596,
      "step": 1525
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.526024341583252,
      "learning_rate": 1.9644123566431335e-05,
      "loss": 1.3375,
      "step": 1526
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.781466245651245,
      "learning_rate": 1.9643484650033543e-05,
      "loss": 0.9949,
      "step": 1527
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.019471168518066,
      "learning_rate": 1.9642845733635755e-05,
      "loss": 0.9459,
      "step": 1528
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5218889713287354,
      "learning_rate": 1.9642206817237967e-05,
      "loss": 0.9798,
      "step": 1529
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2715165615081787,
      "learning_rate": 1.9641567900840178e-05,
      "loss": 0.9476,
      "step": 1530
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2732255458831787,
      "learning_rate": 1.9640928984442386e-05,
      "loss": 0.9382,
      "step": 1531
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.995832920074463,
      "learning_rate": 1.9640290068044598e-05,
      "loss": 0.7753,
      "step": 1532
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4159340858459473,
      "learning_rate": 1.963965115164681e-05,
      "loss": 0.2566,
      "step": 1533
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5230679512023926,
      "learning_rate": 1.963901223524902e-05,
      "loss": 0.9803,
      "step": 1534
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7142817974090576,
      "learning_rate": 1.963837331885123e-05,
      "loss": 1.1567,
      "step": 1535
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4485995769500732,
      "learning_rate": 1.963773440245344e-05,
      "loss": 0.914,
      "step": 1536
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6205856800079346,
      "learning_rate": 1.9637095486055652e-05,
      "loss": 0.8293,
      "step": 1537
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.0421342849731445,
      "learning_rate": 1.9636456569657864e-05,
      "loss": 0.9281,
      "step": 1538
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.242337465286255,
      "learning_rate": 1.9635817653260072e-05,
      "loss": 0.946,
      "step": 1539
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2220466136932373,
      "learning_rate": 1.9635178736862283e-05,
      "loss": 0.7362,
      "step": 1540
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6620428562164307,
      "learning_rate": 1.9634539820464495e-05,
      "loss": 0.8987,
      "step": 1541
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.068220376968384,
      "learning_rate": 1.9633900904066706e-05,
      "loss": 0.8968,
      "step": 1542
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2894182205200195,
      "learning_rate": 1.9633261987668915e-05,
      "loss": 0.8542,
      "step": 1543
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0301084518432617,
      "learning_rate": 1.9632623071271126e-05,
      "loss": 0.9226,
      "step": 1544
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8703051805496216,
      "learning_rate": 1.9631984154873334e-05,
      "loss": 0.8507,
      "step": 1545
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5460784435272217,
      "learning_rate": 1.963134523847555e-05,
      "loss": 0.9283,
      "step": 1546
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9338111877441406,
      "learning_rate": 1.9630706322077757e-05,
      "loss": 0.8356,
      "step": 1547
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3631792068481445,
      "learning_rate": 1.963006740567997e-05,
      "loss": 0.9016,
      "step": 1548
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.654624938964844,
      "learning_rate": 1.9629428489282177e-05,
      "loss": 1.8558,
      "step": 1549
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.655564785003662,
      "learning_rate": 1.962878957288439e-05,
      "loss": 0.9857,
      "step": 1550
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.988049030303955,
      "learning_rate": 1.96281506564866e-05,
      "loss": 0.8251,
      "step": 1551
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.1084208488464355,
      "learning_rate": 1.9627511740088812e-05,
      "loss": 1.4699,
      "step": 1552
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.101222515106201,
      "learning_rate": 1.962687282369102e-05,
      "loss": 1.3182,
      "step": 1553
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.162531852722168,
      "learning_rate": 1.962623390729323e-05,
      "loss": 1.0133,
      "step": 1554
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.694164752960205,
      "learning_rate": 1.9625594990895443e-05,
      "loss": 0.942,
      "step": 1555
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.244415760040283,
      "learning_rate": 1.9624956074497655e-05,
      "loss": 0.8325,
      "step": 1556
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.869112730026245,
      "learning_rate": 1.9624317158099863e-05,
      "loss": 0.9367,
      "step": 1557
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.9734036922454834,
      "learning_rate": 1.9623678241702074e-05,
      "loss": 1.3432,
      "step": 1558
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9975173473358154,
      "learning_rate": 1.9623039325304286e-05,
      "loss": 0.923,
      "step": 1559
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1486334800720215,
      "learning_rate": 1.9622400408906497e-05,
      "loss": 0.9578,
      "step": 1560
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5990500450134277,
      "learning_rate": 1.9621761492508706e-05,
      "loss": 0.9886,
      "step": 1561
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.498854160308838,
      "learning_rate": 1.9621122576110917e-05,
      "loss": 1.2875,
      "step": 1562
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6905107498168945,
      "learning_rate": 1.962048365971313e-05,
      "loss": 0.8222,
      "step": 1563
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7255311012268066,
      "learning_rate": 1.961984474331534e-05,
      "loss": 0.9871,
      "step": 1564
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7727837562561035,
      "learning_rate": 1.961920582691755e-05,
      "loss": 0.8176,
      "step": 1565
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.116112232208252,
      "learning_rate": 1.961856691051976e-05,
      "loss": 0.8449,
      "step": 1566
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.254271030426025,
      "learning_rate": 1.961792799412197e-05,
      "loss": 0.9941,
      "step": 1567
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.784253120422363,
      "learning_rate": 1.9617289077724183e-05,
      "loss": 1.5369,
      "step": 1568
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3866584300994873,
      "learning_rate": 1.961665016132639e-05,
      "loss": 0.9555,
      "step": 1569
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3488588333129883,
      "learning_rate": 1.9616011244928603e-05,
      "loss": 0.857,
      "step": 1570
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.205313682556152,
      "learning_rate": 1.9615372328530814e-05,
      "loss": 1.2742,
      "step": 1571
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.546192169189453,
      "learning_rate": 1.9614733412133026e-05,
      "loss": 1.2289,
      "step": 1572
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.392914295196533,
      "learning_rate": 1.9614094495735234e-05,
      "loss": 0.9221,
      "step": 1573
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4541186094284058,
      "learning_rate": 1.9613455579337445e-05,
      "loss": 0.2562,
      "step": 1574
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.322169303894043,
      "learning_rate": 1.9612816662939657e-05,
      "loss": 0.9351,
      "step": 1575
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.401289701461792,
      "learning_rate": 1.961217774654187e-05,
      "loss": 0.9918,
      "step": 1576
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5050534009933472,
      "learning_rate": 1.9611538830144077e-05,
      "loss": 0.2927,
      "step": 1577
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.790146827697754,
      "learning_rate": 1.9610899913746288e-05,
      "loss": 1.0738,
      "step": 1578
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0207648277282715,
      "learning_rate": 1.9610260997348496e-05,
      "loss": 0.8399,
      "step": 1579
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.00114107131958,
      "learning_rate": 1.960962208095071e-05,
      "loss": 0.825,
      "step": 1580
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1417016983032227,
      "learning_rate": 1.960898316455292e-05,
      "loss": 0.8753,
      "step": 1581
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.425787448883057,
      "learning_rate": 1.960834424815513e-05,
      "loss": 0.9478,
      "step": 1582
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4724459648132324,
      "learning_rate": 1.960770533175734e-05,
      "loss": 1.428,
      "step": 1583
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.151578426361084,
      "learning_rate": 1.960706641535955e-05,
      "loss": 0.912,
      "step": 1584
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9510958194732666,
      "learning_rate": 1.9606427498961762e-05,
      "loss": 1.0515,
      "step": 1585
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.380077362060547,
      "learning_rate": 1.9605788582563974e-05,
      "loss": 0.845,
      "step": 1586
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.336334705352783,
      "learning_rate": 1.9605149666166182e-05,
      "loss": 1.3047,
      "step": 1587
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1634604930877686,
      "learning_rate": 1.9604510749768394e-05,
      "loss": 0.9139,
      "step": 1588
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.998147487640381,
      "learning_rate": 1.9603871833370605e-05,
      "loss": 1.1211,
      "step": 1589
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7845929861068726,
      "learning_rate": 1.9603232916972817e-05,
      "loss": 0.879,
      "step": 1590
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9918431043624878,
      "learning_rate": 1.9602594000575025e-05,
      "loss": 0.8052,
      "step": 1591
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.207181930541992,
      "learning_rate": 1.9601955084177236e-05,
      "loss": 1.1396,
      "step": 1592
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4799964427948,
      "learning_rate": 1.9601316167779448e-05,
      "loss": 0.8418,
      "step": 1593
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1306822299957275,
      "learning_rate": 1.960067725138166e-05,
      "loss": 0.9852,
      "step": 1594
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.607753038406372,
      "learning_rate": 1.9600038334983868e-05,
      "loss": 1.1137,
      "step": 1595
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.718904972076416,
      "learning_rate": 1.959939941858608e-05,
      "loss": 1.2385,
      "step": 1596
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0107905864715576,
      "learning_rate": 1.959876050218829e-05,
      "loss": 0.8947,
      "step": 1597
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2962868213653564,
      "learning_rate": 1.9598121585790502e-05,
      "loss": 0.8649,
      "step": 1598
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9861419200897217,
      "learning_rate": 1.959748266939271e-05,
      "loss": 0.8772,
      "step": 1599
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.401103973388672,
      "learning_rate": 1.9596843752994922e-05,
      "loss": 1.0107,
      "step": 1600
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.807682514190674,
      "learning_rate": 1.9596204836597134e-05,
      "loss": 1.9412,
      "step": 1601
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1325595378875732,
      "learning_rate": 1.9595565920199345e-05,
      "loss": 0.9511,
      "step": 1602
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.838742971420288,
      "learning_rate": 1.9594927003801553e-05,
      "loss": 0.9806,
      "step": 1603
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.333221435546875,
      "learning_rate": 1.9594288087403765e-05,
      "loss": 0.8947,
      "step": 1604
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.886991500854492,
      "learning_rate": 1.9593649171005976e-05,
      "loss": 1.2542,
      "step": 1605
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.167876958847046,
      "learning_rate": 1.9593010254608188e-05,
      "loss": 0.8404,
      "step": 1606
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.749706745147705,
      "learning_rate": 1.9592371338210396e-05,
      "loss": 1.8157,
      "step": 1607
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2618632316589355,
      "learning_rate": 1.9591732421812608e-05,
      "loss": 0.8596,
      "step": 1608
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9432125091552734,
      "learning_rate": 1.959109350541482e-05,
      "loss": 0.9633,
      "step": 1609
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1187119483947754,
      "learning_rate": 1.959045458901703e-05,
      "loss": 0.8529,
      "step": 1610
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9933393001556396,
      "learning_rate": 1.958981567261924e-05,
      "loss": 0.9003,
      "step": 1611
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8059020042419434,
      "learning_rate": 1.958917675622145e-05,
      "loss": 1.0518,
      "step": 1612
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.760107517242432,
      "learning_rate": 1.9588537839823662e-05,
      "loss": 1.4391,
      "step": 1613
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.565420627593994,
      "learning_rate": 1.9587898923425873e-05,
      "loss": 0.6062,
      "step": 1614
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.272092819213867,
      "learning_rate": 1.958726000702808e-05,
      "loss": 0.9958,
      "step": 1615
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.209132194519043,
      "learning_rate": 1.9586621090630293e-05,
      "loss": 0.927,
      "step": 1616
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0836381912231445,
      "learning_rate": 1.95859821742325e-05,
      "loss": 0.9097,
      "step": 1617
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9318815469741821,
      "learning_rate": 1.9585343257834716e-05,
      "loss": 0.9182,
      "step": 1618
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0482676029205322,
      "learning_rate": 1.9584704341436924e-05,
      "loss": 0.8965,
      "step": 1619
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.482767105102539,
      "learning_rate": 1.9584065425039136e-05,
      "loss": 1.2501,
      "step": 1620
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6524884700775146,
      "learning_rate": 1.9583426508641344e-05,
      "loss": 0.8922,
      "step": 1621
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8132591247558594,
      "learning_rate": 1.9582787592243556e-05,
      "loss": 0.2505,
      "step": 1622
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4436116218566895,
      "learning_rate": 1.9582148675845767e-05,
      "loss": 0.9555,
      "step": 1623
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.260650634765625,
      "learning_rate": 1.958150975944798e-05,
      "loss": 0.8907,
      "step": 1624
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2849619388580322,
      "learning_rate": 1.9580870843050187e-05,
      "loss": 0.9021,
      "step": 1625
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.213973045349121,
      "learning_rate": 1.95802319266524e-05,
      "loss": 0.9161,
      "step": 1626
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8365018367767334,
      "learning_rate": 1.957959301025461e-05,
      "loss": 1.103,
      "step": 1627
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6209933757781982,
      "learning_rate": 1.957895409385682e-05,
      "loss": 1.0288,
      "step": 1628
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5083630084991455,
      "learning_rate": 1.957831517745903e-05,
      "loss": 0.8669,
      "step": 1629
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8972810506820679,
      "learning_rate": 1.957767626106124e-05,
      "loss": 0.8846,
      "step": 1630
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1925323009490967,
      "learning_rate": 1.9577037344663453e-05,
      "loss": 0.8765,
      "step": 1631
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.379424571990967,
      "learning_rate": 1.9576398428265664e-05,
      "loss": 0.8365,
      "step": 1632
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.754136562347412,
      "learning_rate": 1.9575759511867873e-05,
      "loss": 0.8575,
      "step": 1633
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7246668338775635,
      "learning_rate": 1.9575120595470084e-05,
      "loss": 1.3094,
      "step": 1634
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3405776023864746,
      "learning_rate": 1.9574481679072296e-05,
      "loss": 0.906,
      "step": 1635
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.183255910873413,
      "learning_rate": 1.9573842762674507e-05,
      "loss": 0.9586,
      "step": 1636
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3046646118164062,
      "learning_rate": 1.9573203846276715e-05,
      "loss": 0.8689,
      "step": 1637
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.629612922668457,
      "learning_rate": 1.9572564929878927e-05,
      "loss": 1.0897,
      "step": 1638
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5301427841186523,
      "learning_rate": 1.957192601348114e-05,
      "loss": 0.8846,
      "step": 1639
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.639415740966797,
      "learning_rate": 1.957128709708335e-05,
      "loss": 1.6227,
      "step": 1640
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.771056652069092,
      "learning_rate": 1.9570648180685558e-05,
      "loss": 1.2777,
      "step": 1641
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8265572786331177,
      "learning_rate": 1.957000926428777e-05,
      "loss": 0.8624,
      "step": 1642
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.095107078552246,
      "learning_rate": 1.956937034788998e-05,
      "loss": 1.2252,
      "step": 1643
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1807761192321777,
      "learning_rate": 1.9568731431492193e-05,
      "loss": 1.0073,
      "step": 1644
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.998547315597534,
      "learning_rate": 1.95680925150944e-05,
      "loss": 0.9704,
      "step": 1645
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.105874061584473,
      "learning_rate": 1.9567453598696612e-05,
      "loss": 1.1319,
      "step": 1646
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.516467332839966,
      "learning_rate": 1.9566814682298824e-05,
      "loss": 1.1766,
      "step": 1647
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.476054668426514,
      "learning_rate": 1.9566175765901036e-05,
      "loss": 1.4728,
      "step": 1648
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5071375370025635,
      "learning_rate": 1.9565536849503244e-05,
      "loss": 0.9964,
      "step": 1649
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1410233974456787,
      "learning_rate": 1.9564897933105455e-05,
      "loss": 0.9822,
      "step": 1650
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1970250606536865,
      "learning_rate": 1.9564259016707667e-05,
      "loss": 0.8478,
      "step": 1651
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.644478797912598,
      "learning_rate": 1.956362010030988e-05,
      "loss": 1.1118,
      "step": 1652
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.694169044494629,
      "learning_rate": 1.9562981183912087e-05,
      "loss": 1.6931,
      "step": 1653
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9954077005386353,
      "learning_rate": 1.9562342267514298e-05,
      "loss": 1.0359,
      "step": 1654
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.150074481964111,
      "learning_rate": 1.9561703351116506e-05,
      "loss": 1.224,
      "step": 1655
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.104336977005005,
      "learning_rate": 1.956106443471872e-05,
      "loss": 1.0027,
      "step": 1656
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1971778869628906,
      "learning_rate": 1.956042551832093e-05,
      "loss": 0.9045,
      "step": 1657
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1256752014160156,
      "learning_rate": 1.955978660192314e-05,
      "loss": 0.9561,
      "step": 1658
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.348708152770996,
      "learning_rate": 1.955914768552535e-05,
      "loss": 0.9277,
      "step": 1659
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0111072063446045,
      "learning_rate": 1.955850876912756e-05,
      "loss": 0.9469,
      "step": 1660
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.792049884796143,
      "learning_rate": 1.9557869852729772e-05,
      "loss": 1.0201,
      "step": 1661
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7530579566955566,
      "learning_rate": 1.9557230936331984e-05,
      "loss": 1.0246,
      "step": 1662
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1559407711029053,
      "learning_rate": 1.9556592019934192e-05,
      "loss": 1.0132,
      "step": 1663
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.584566354751587,
      "learning_rate": 1.9555953103536403e-05,
      "loss": 0.2486,
      "step": 1664
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7182681560516357,
      "learning_rate": 1.9555314187138615e-05,
      "loss": 1.138,
      "step": 1665
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.5291569232940674,
      "learning_rate": 1.9554675270740826e-05,
      "loss": 0.9139,
      "step": 1666
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.516005516052246,
      "learning_rate": 1.9554036354343035e-05,
      "loss": 1.0015,
      "step": 1667
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.302997350692749,
      "learning_rate": 1.9553397437945246e-05,
      "loss": 0.9287,
      "step": 1668
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.405195474624634,
      "learning_rate": 1.9552758521547458e-05,
      "loss": 0.7874,
      "step": 1669
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.702743649482727,
      "learning_rate": 1.955211960514967e-05,
      "loss": 0.8043,
      "step": 1670
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5147132873535156,
      "learning_rate": 1.9551480688751877e-05,
      "loss": 1.2798,
      "step": 1671
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.76985239982605,
      "learning_rate": 1.955084177235409e-05,
      "loss": 1.0519,
      "step": 1672
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.658514142036438,
      "learning_rate": 1.95502028559563e-05,
      "loss": 0.8844,
      "step": 1673
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5007729530334473,
      "learning_rate": 1.9549563939558512e-05,
      "loss": 1.2042,
      "step": 1674
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.8330769538879395,
      "learning_rate": 1.954892502316072e-05,
      "loss": 0.891,
      "step": 1675
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.035097599029541,
      "learning_rate": 1.9548286106762932e-05,
      "loss": 1.6432,
      "step": 1676
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7364659309387207,
      "learning_rate": 1.9547647190365143e-05,
      "loss": 0.6728,
      "step": 1677
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3316469192504883,
      "learning_rate": 1.9547008273967355e-05,
      "loss": 0.8785,
      "step": 1678
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2995758056640625,
      "learning_rate": 1.9546369357569563e-05,
      "loss": 0.9323,
      "step": 1679
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.575315237045288,
      "learning_rate": 1.9545730441171775e-05,
      "loss": 0.9106,
      "step": 1680
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.257739543914795,
      "learning_rate": 1.9545091524773986e-05,
      "loss": 0.9818,
      "step": 1681
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.67669939994812,
      "learning_rate": 1.9544452608376198e-05,
      "loss": 1.1054,
      "step": 1682
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0074994564056396,
      "learning_rate": 1.9543813691978406e-05,
      "loss": 0.8002,
      "step": 1683
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.821863889694214,
      "learning_rate": 1.9543174775580617e-05,
      "loss": 0.9579,
      "step": 1684
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.572709560394287,
      "learning_rate": 1.954253585918283e-05,
      "loss": 0.8896,
      "step": 1685
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.364169120788574,
      "learning_rate": 1.954189694278504e-05,
      "loss": 0.8469,
      "step": 1686
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7065563201904297,
      "learning_rate": 1.954125802638725e-05,
      "loss": 0.8798,
      "step": 1687
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.534282684326172,
      "learning_rate": 1.954061910998946e-05,
      "loss": 0.8197,
      "step": 1688
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6741325855255127,
      "learning_rate": 1.953998019359167e-05,
      "loss": 1.0028,
      "step": 1689
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7125680446624756,
      "learning_rate": 1.9539341277193883e-05,
      "loss": 0.9075,
      "step": 1690
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.593686819076538,
      "learning_rate": 1.953870236079609e-05,
      "loss": 0.9278,
      "step": 1691
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.8694024085998535,
      "learning_rate": 1.9538063444398303e-05,
      "loss": 0.9965,
      "step": 1692
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.5307538509368896,
      "learning_rate": 1.953742452800051e-05,
      "loss": 0.8292,
      "step": 1693
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.196516752243042,
      "learning_rate": 1.9536785611602723e-05,
      "loss": 1.0068,
      "step": 1694
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.969469666481018,
      "learning_rate": 1.9536146695204934e-05,
      "loss": 0.7711,
      "step": 1695
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.8294427394866943,
      "learning_rate": 1.9535507778807146e-05,
      "loss": 0.604,
      "step": 1696
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7018914222717285,
      "learning_rate": 1.9534868862409354e-05,
      "loss": 0.9191,
      "step": 1697
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9240920543670654,
      "learning_rate": 1.9534229946011565e-05,
      "loss": 1.22,
      "step": 1698
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0307624340057373,
      "learning_rate": 1.9533591029613777e-05,
      "loss": 0.8745,
      "step": 1699
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3712637424468994,
      "learning_rate": 1.953295211321599e-05,
      "loss": 0.8602,
      "step": 1700
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4617974758148193,
      "learning_rate": 1.9532313196818197e-05,
      "loss": 0.8962,
      "step": 1701
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4069302082061768,
      "learning_rate": 1.9531674280420408e-05,
      "loss": 1.0228,
      "step": 1702
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.168468475341797,
      "learning_rate": 1.953103536402262e-05,
      "loss": 0.8814,
      "step": 1703
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2700161933898926,
      "learning_rate": 1.953039644762483e-05,
      "loss": 0.8683,
      "step": 1704
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.738289713859558,
      "learning_rate": 1.952975753122704e-05,
      "loss": 0.864,
      "step": 1705
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0899620056152344,
      "learning_rate": 1.952911861482925e-05,
      "loss": 0.8876,
      "step": 1706
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6214003562927246,
      "learning_rate": 1.9528479698431463e-05,
      "loss": 0.9207,
      "step": 1707
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.357577323913574,
      "learning_rate": 1.9527840782033674e-05,
      "loss": 0.9033,
      "step": 1708
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1577200889587402,
      "learning_rate": 1.9527201865635882e-05,
      "loss": 1.0379,
      "step": 1709
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.149538516998291,
      "learning_rate": 1.9526562949238094e-05,
      "loss": 0.8547,
      "step": 1710
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6602208614349365,
      "learning_rate": 1.9525924032840305e-05,
      "loss": 0.9766,
      "step": 1711
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7783327102661133,
      "learning_rate": 1.9525285116442517e-05,
      "loss": 0.9305,
      "step": 1712
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.635292649269104,
      "learning_rate": 1.9524646200044725e-05,
      "loss": 0.3521,
      "step": 1713
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9499846696853638,
      "learning_rate": 1.9524007283646937e-05,
      "loss": 0.8707,
      "step": 1714
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.704010963439941,
      "learning_rate": 1.9523368367249148e-05,
      "loss": 1.6698,
      "step": 1715
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.092151641845703,
      "learning_rate": 1.952272945085136e-05,
      "loss": 1.2851,
      "step": 1716
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.141685962677002,
      "learning_rate": 1.9522090534453568e-05,
      "loss": 0.9569,
      "step": 1717
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5475616455078125,
      "learning_rate": 1.952145161805578e-05,
      "loss": 0.963,
      "step": 1718
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.831159591674805,
      "learning_rate": 1.952081270165799e-05,
      "loss": 1.2241,
      "step": 1719
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.852976083755493,
      "learning_rate": 1.9520173785260203e-05,
      "loss": 0.9657,
      "step": 1720
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9083338975906372,
      "learning_rate": 1.951953486886241e-05,
      "loss": 0.839,
      "step": 1721
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2593271732330322,
      "learning_rate": 1.9518895952464622e-05,
      "loss": 0.947,
      "step": 1722
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.291696071624756,
      "learning_rate": 1.9518257036066834e-05,
      "loss": 1.1629,
      "step": 1723
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8346199989318848,
      "learning_rate": 1.9517618119669045e-05,
      "loss": 1.0971,
      "step": 1724
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.000655174255371,
      "learning_rate": 1.9516979203271254e-05,
      "loss": 0.8467,
      "step": 1725
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.021620512008667,
      "learning_rate": 1.9516340286873465e-05,
      "loss": 0.9858,
      "step": 1726
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8305094242095947,
      "learning_rate": 1.9515701370475673e-05,
      "loss": 1.1033,
      "step": 1727
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.450038194656372,
      "learning_rate": 1.9515062454077888e-05,
      "loss": 1.0947,
      "step": 1728
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3155574798583984,
      "learning_rate": 1.9514423537680096e-05,
      "loss": 0.9471,
      "step": 1729
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.022730827331543,
      "learning_rate": 1.9513784621282308e-05,
      "loss": 1.193,
      "step": 1730
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.059061050415039,
      "learning_rate": 1.9513145704884516e-05,
      "loss": 0.9309,
      "step": 1731
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.613640308380127,
      "learning_rate": 1.9512506788486728e-05,
      "loss": 1.3424,
      "step": 1732
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.412729740142822,
      "learning_rate": 1.951186787208894e-05,
      "loss": 1.0361,
      "step": 1733
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.490983247756958,
      "learning_rate": 1.951122895569115e-05,
      "loss": 0.9276,
      "step": 1734
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3054890632629395,
      "learning_rate": 1.951059003929336e-05,
      "loss": 0.946,
      "step": 1735
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6913840770721436,
      "learning_rate": 1.950995112289557e-05,
      "loss": 1.3153,
      "step": 1736
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.60042405128479,
      "learning_rate": 1.9509312206497782e-05,
      "loss": 1.0256,
      "step": 1737
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4219324588775635,
      "learning_rate": 1.9508673290099993e-05,
      "loss": 0.9,
      "step": 1738
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.392540693283081,
      "learning_rate": 1.95080343737022e-05,
      "loss": 1.2342,
      "step": 1739
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3778059482574463,
      "learning_rate": 1.9507395457304413e-05,
      "loss": 1.1282,
      "step": 1740
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0808968544006348,
      "learning_rate": 1.9506756540906625e-05,
      "loss": 1.1525,
      "step": 1741
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.714869976043701,
      "learning_rate": 1.9506117624508833e-05,
      "loss": 1.4595,
      "step": 1742
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0478930473327637,
      "learning_rate": 1.9505478708111044e-05,
      "loss": 0.8752,
      "step": 1743
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.261387586593628,
      "learning_rate": 1.9504839791713256e-05,
      "loss": 0.8766,
      "step": 1744
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6345303058624268,
      "learning_rate": 1.9504200875315468e-05,
      "loss": 0.9982,
      "step": 1745
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9095643758773804,
      "learning_rate": 1.9503561958917676e-05,
      "loss": 0.8084,
      "step": 1746
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.465158462524414,
      "learning_rate": 1.9502923042519887e-05,
      "loss": 0.8343,
      "step": 1747
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2677001953125,
      "learning_rate": 1.95022841261221e-05,
      "loss": 0.8424,
      "step": 1748
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4949264526367188,
      "learning_rate": 1.950164520972431e-05,
      "loss": 0.9922,
      "step": 1749
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4144883155822754,
      "learning_rate": 1.950100629332652e-05,
      "loss": 0.9361,
      "step": 1750
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.791087031364441,
      "learning_rate": 1.950036737692873e-05,
      "loss": 0.7401,
      "step": 1751
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.098588466644287,
      "learning_rate": 1.949972846053094e-05,
      "loss": 0.8896,
      "step": 1752
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.651461362838745,
      "learning_rate": 1.9499089544133153e-05,
      "loss": 0.9999,
      "step": 1753
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1029787063598633,
      "learning_rate": 1.949845062773536e-05,
      "loss": 1.0184,
      "step": 1754
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.098205327987671,
      "learning_rate": 1.9497811711337573e-05,
      "loss": 1.3327,
      "step": 1755
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.147405624389648,
      "learning_rate": 1.949717279493978e-05,
      "loss": 1.454,
      "step": 1756
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3905274868011475,
      "learning_rate": 1.9496533878541996e-05,
      "loss": 1.0197,
      "step": 1757
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7408645153045654,
      "learning_rate": 1.9495894962144204e-05,
      "loss": 0.816,
      "step": 1758
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.577816009521484,
      "learning_rate": 1.9495256045746416e-05,
      "loss": 1.7688,
      "step": 1759
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.176293134689331,
      "learning_rate": 1.9494617129348624e-05,
      "loss": 0.9243,
      "step": 1760
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.399233102798462,
      "learning_rate": 1.9493978212950835e-05,
      "loss": 1.0263,
      "step": 1761
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.443876028060913,
      "learning_rate": 1.9493339296553047e-05,
      "loss": 0.2125,
      "step": 1762
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6046741008758545,
      "learning_rate": 1.949270038015526e-05,
      "loss": 1.0192,
      "step": 1763
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4124395847320557,
      "learning_rate": 1.9492061463757467e-05,
      "loss": 0.986,
      "step": 1764
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7448525428771973,
      "learning_rate": 1.9491422547359678e-05,
      "loss": 1.0943,
      "step": 1765
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7020106315612793,
      "learning_rate": 1.949078363096189e-05,
      "loss": 1.1733,
      "step": 1766
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.662539005279541,
      "learning_rate": 1.94901447145641e-05,
      "loss": 1.3439,
      "step": 1767
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.9156084060668945,
      "learning_rate": 1.948950579816631e-05,
      "loss": 1.3262,
      "step": 1768
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1270225048065186,
      "learning_rate": 1.948886688176852e-05,
      "loss": 0.8721,
      "step": 1769
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1077044010162354,
      "learning_rate": 1.9488227965370732e-05,
      "loss": 1.0508,
      "step": 1770
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.809648036956787,
      "learning_rate": 1.9487589048972944e-05,
      "loss": 1.9608,
      "step": 1771
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9374034404754639,
      "learning_rate": 1.9486950132575152e-05,
      "loss": 0.9371,
      "step": 1772
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.73801851272583,
      "learning_rate": 1.9486311216177364e-05,
      "loss": 1.704,
      "step": 1773
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1289947032928467,
      "learning_rate": 1.9485672299779575e-05,
      "loss": 0.9073,
      "step": 1774
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.471883296966553,
      "learning_rate": 1.9485033383381787e-05,
      "loss": 0.8367,
      "step": 1775
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9139869213104248,
      "learning_rate": 1.9484394466983995e-05,
      "loss": 0.9282,
      "step": 1776
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.343972682952881,
      "learning_rate": 1.9483755550586207e-05,
      "loss": 0.8064,
      "step": 1777
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0938615798950195,
      "learning_rate": 1.9483116634188418e-05,
      "loss": 0.9442,
      "step": 1778
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.293064594268799,
      "learning_rate": 1.948247771779063e-05,
      "loss": 1.1248,
      "step": 1779
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.663634300231934,
      "learning_rate": 1.9481838801392838e-05,
      "loss": 0.9269,
      "step": 1780
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3073418140411377,
      "learning_rate": 1.948119988499505e-05,
      "loss": 0.8372,
      "step": 1781
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.59123420715332,
      "learning_rate": 1.948056096859726e-05,
      "loss": 0.9091,
      "step": 1782
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8457179069519043,
      "learning_rate": 1.9479922052199472e-05,
      "loss": 0.9447,
      "step": 1783
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1111433506011963,
      "learning_rate": 1.947928313580168e-05,
      "loss": 0.9931,
      "step": 1784
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3003389835357666,
      "learning_rate": 1.9478644219403892e-05,
      "loss": 0.8401,
      "step": 1785
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9232228994369507,
      "learning_rate": 1.9478005303006104e-05,
      "loss": 0.8571,
      "step": 1786
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9559450149536133,
      "learning_rate": 1.9477366386608315e-05,
      "loss": 1.0346,
      "step": 1787
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.5294318199157715,
      "learning_rate": 1.9476727470210523e-05,
      "loss": 0.9732,
      "step": 1788
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2886199951171875,
      "learning_rate": 1.9476088553812735e-05,
      "loss": 0.903,
      "step": 1789
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3430633544921875,
      "learning_rate": 1.9475449637414946e-05,
      "loss": 1.2334,
      "step": 1790
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.97873854637146,
      "learning_rate": 1.9474810721017158e-05,
      "loss": 0.8753,
      "step": 1791
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2541258335113525,
      "learning_rate": 1.9474171804619366e-05,
      "loss": 1.1707,
      "step": 1792
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8637255430221558,
      "learning_rate": 1.9473532888221578e-05,
      "loss": 0.7786,
      "step": 1793
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.098342180252075,
      "learning_rate": 1.9472893971823786e-05,
      "loss": 0.8464,
      "step": 1794
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.219749689102173,
      "learning_rate": 1.9472255055426e-05,
      "loss": 0.9464,
      "step": 1795
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.615691900253296,
      "learning_rate": 1.947161613902821e-05,
      "loss": 1.0127,
      "step": 1796
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.804385662078857,
      "learning_rate": 1.947097722263042e-05,
      "loss": 1.0379,
      "step": 1797
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3730525970458984,
      "learning_rate": 1.947033830623263e-05,
      "loss": 0.9432,
      "step": 1798
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3779518604278564,
      "learning_rate": 1.946969938983484e-05,
      "loss": 0.9873,
      "step": 1799
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0547966957092285,
      "learning_rate": 1.9469060473437052e-05,
      "loss": 0.823,
      "step": 1800
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1423909664154053,
      "learning_rate": 1.9468421557039263e-05,
      "loss": 1.0074,
      "step": 1801
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.263395071029663,
      "learning_rate": 1.946778264064147e-05,
      "loss": 1.0448,
      "step": 1802
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.683442115783691,
      "learning_rate": 1.9467143724243683e-05,
      "loss": 1.5124,
      "step": 1803
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.157219648361206,
      "learning_rate": 1.9466504807845895e-05,
      "loss": 1.148,
      "step": 1804
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.101602792739868,
      "learning_rate": 1.9465865891448106e-05,
      "loss": 0.8211,
      "step": 1805
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.152327060699463,
      "learning_rate": 1.9465226975050314e-05,
      "loss": 1.5059,
      "step": 1806
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4516191482543945,
      "learning_rate": 1.9464588058652526e-05,
      "loss": 0.9477,
      "step": 1807
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2719571590423584,
      "learning_rate": 1.9463949142254737e-05,
      "loss": 0.9921,
      "step": 1808
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0890798568725586,
      "learning_rate": 1.946331022585695e-05,
      "loss": 1.1805,
      "step": 1809
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.627657175064087,
      "learning_rate": 1.9462671309459157e-05,
      "loss": 1.0378,
      "step": 1810
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9693790674209595,
      "learning_rate": 1.946203239306137e-05,
      "loss": 0.9496,
      "step": 1811
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.880436897277832,
      "learning_rate": 1.946139347666358e-05,
      "loss": 0.8402,
      "step": 1812
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1021761894226074,
      "learning_rate": 1.9460754560265792e-05,
      "loss": 0.9946,
      "step": 1813
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.479384899139404,
      "learning_rate": 1.9460115643868e-05,
      "loss": 1.7791,
      "step": 1814
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1282663345336914,
      "learning_rate": 1.945947672747021e-05,
      "loss": 0.8186,
      "step": 1815
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.948993444442749,
      "learning_rate": 1.9458837811072423e-05,
      "loss": 0.8209,
      "step": 1816
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0931899547576904,
      "learning_rate": 1.9458198894674635e-05,
      "loss": 1.1276,
      "step": 1817
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.258861780166626,
      "learning_rate": 1.9457559978276843e-05,
      "loss": 1.127,
      "step": 1818
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1972737312316895,
      "learning_rate": 1.9456921061879054e-05,
      "loss": 0.8723,
      "step": 1819
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.239754676818848,
      "learning_rate": 1.9456282145481266e-05,
      "loss": 1.0582,
      "step": 1820
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.355415105819702,
      "learning_rate": 1.9455643229083477e-05,
      "loss": 0.8981,
      "step": 1821
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.45019793510437,
      "learning_rate": 1.9455004312685686e-05,
      "loss": 1.0233,
      "step": 1822
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6721062660217285,
      "learning_rate": 1.9454365396287897e-05,
      "loss": 0.817,
      "step": 1823
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.486032247543335,
      "learning_rate": 1.945372647989011e-05,
      "loss": 0.9335,
      "step": 1824
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5880074501037598,
      "learning_rate": 1.945308756349232e-05,
      "loss": 1.0813,
      "step": 1825
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.292520046234131,
      "learning_rate": 1.9452448647094528e-05,
      "loss": 0.9176,
      "step": 1826
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.683072090148926,
      "learning_rate": 1.945180973069674e-05,
      "loss": 0.8305,
      "step": 1827
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7850228548049927,
      "learning_rate": 1.945117081429895e-05,
      "loss": 0.8932,
      "step": 1828
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.038041591644287,
      "learning_rate": 1.9450531897901163e-05,
      "loss": 0.8698,
      "step": 1829
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.548752784729004,
      "learning_rate": 1.944989298150337e-05,
      "loss": 1.1368,
      "step": 1830
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9375946521759033,
      "learning_rate": 1.9449254065105583e-05,
      "loss": 0.9855,
      "step": 1831
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.419451951980591,
      "learning_rate": 1.944861514870779e-05,
      "loss": 0.8629,
      "step": 1832
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.650305271148682,
      "learning_rate": 1.9447976232310006e-05,
      "loss": 1.4066,
      "step": 1833
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.142850875854492,
      "learning_rate": 1.9447337315912214e-05,
      "loss": 1.2515,
      "step": 1834
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.726780652999878,
      "learning_rate": 1.9446698399514425e-05,
      "loss": 1.1468,
      "step": 1835
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.378596782684326,
      "learning_rate": 1.9446059483116634e-05,
      "loss": 1.1957,
      "step": 1836
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3658578395843506,
      "learning_rate": 1.9445420566718845e-05,
      "loss": 0.8263,
      "step": 1837
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0097599029541016,
      "learning_rate": 1.9444781650321057e-05,
      "loss": 0.9479,
      "step": 1838
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.192838430404663,
      "learning_rate": 1.9444142733923268e-05,
      "loss": 0.7862,
      "step": 1839
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.056793689727783,
      "learning_rate": 1.9443503817525476e-05,
      "loss": 0.9369,
      "step": 1840
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.152803897857666,
      "learning_rate": 1.9442864901127688e-05,
      "loss": 0.9314,
      "step": 1841
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2747724056243896,
      "learning_rate": 1.94422259847299e-05,
      "loss": 1.0546,
      "step": 1842
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9415912628173828,
      "learning_rate": 1.944158706833211e-05,
      "loss": 0.8413,
      "step": 1843
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9687708616256714,
      "learning_rate": 1.944094815193432e-05,
      "loss": 0.9016,
      "step": 1844
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4343485832214355,
      "learning_rate": 1.944030923553653e-05,
      "loss": 0.9788,
      "step": 1845
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2844254970550537,
      "learning_rate": 1.9439670319138742e-05,
      "loss": 0.8269,
      "step": 1846
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.507843494415283,
      "learning_rate": 1.9439031402740954e-05,
      "loss": 0.8896,
      "step": 1847
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0719404220581055,
      "learning_rate": 1.9438392486343162e-05,
      "loss": 0.8449,
      "step": 1848
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.148508310317993,
      "learning_rate": 1.9437753569945374e-05,
      "loss": 1.096,
      "step": 1849
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.031774520874023,
      "learning_rate": 1.9437114653547585e-05,
      "loss": 1.2275,
      "step": 1850
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3173110485076904,
      "learning_rate": 1.9436475737149797e-05,
      "loss": 1.1899,
      "step": 1851
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8189027309417725,
      "learning_rate": 1.9435836820752005e-05,
      "loss": 0.7603,
      "step": 1852
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.0786542892456055,
      "learning_rate": 1.9435197904354216e-05,
      "loss": 1.354,
      "step": 1853
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.562329888343811,
      "learning_rate": 1.9434558987956428e-05,
      "loss": 0.306,
      "step": 1854
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.033944606781006,
      "learning_rate": 1.943392007155864e-05,
      "loss": 0.9739,
      "step": 1855
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.460275888442993,
      "learning_rate": 1.9433281155160848e-05,
      "loss": 0.9579,
      "step": 1856
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1003453731536865,
      "learning_rate": 1.943264223876306e-05,
      "loss": 0.9585,
      "step": 1857
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4913218021392822,
      "learning_rate": 1.943200332236527e-05,
      "loss": 0.9906,
      "step": 1858
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.012505531311035,
      "learning_rate": 1.9431364405967482e-05,
      "loss": 0.9229,
      "step": 1859
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.780675411224365,
      "learning_rate": 1.943072548956969e-05,
      "loss": 0.809,
      "step": 1860
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3094282150268555,
      "learning_rate": 1.9430086573171902e-05,
      "loss": 1.0091,
      "step": 1861
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.689946174621582,
      "learning_rate": 1.9429447656774114e-05,
      "loss": 1.23,
      "step": 1862
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.087407112121582,
      "learning_rate": 1.9428808740376325e-05,
      "loss": 1.4397,
      "step": 1863
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5185375213623047,
      "learning_rate": 1.9428169823978533e-05,
      "loss": 1.1892,
      "step": 1864
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.588743805885315,
      "learning_rate": 1.9427530907580745e-05,
      "loss": 0.2778,
      "step": 1865
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9197118282318115,
      "learning_rate": 1.9426891991182953e-05,
      "loss": 0.9546,
      "step": 1866
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.15446400642395,
      "learning_rate": 1.9426253074785168e-05,
      "loss": 1.0176,
      "step": 1867
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.391594409942627,
      "learning_rate": 1.9425614158387376e-05,
      "loss": 0.8435,
      "step": 1868
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.095692157745361,
      "learning_rate": 1.9424975241989588e-05,
      "loss": 1.1105,
      "step": 1869
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.404879570007324,
      "learning_rate": 1.9424336325591796e-05,
      "loss": 1.1558,
      "step": 1870
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8817425966262817,
      "learning_rate": 1.9423697409194007e-05,
      "loss": 0.913,
      "step": 1871
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.807167410850525,
      "learning_rate": 1.942305849279622e-05,
      "loss": 0.8394,
      "step": 1872
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.811161994934082,
      "learning_rate": 1.942241957639843e-05,
      "loss": 1.1542,
      "step": 1873
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4738268852233887,
      "learning_rate": 1.942178066000064e-05,
      "loss": 1.2203,
      "step": 1874
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.3764166831970215,
      "learning_rate": 1.942114174360285e-05,
      "loss": 0.7074,
      "step": 1875
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2862377166748047,
      "learning_rate": 1.942050282720506e-05,
      "loss": 1.0273,
      "step": 1876
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6179476976394653,
      "learning_rate": 1.9419863910807273e-05,
      "loss": 0.2686,
      "step": 1877
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.770372748374939,
      "learning_rate": 1.941922499440948e-05,
      "loss": 0.9322,
      "step": 1878
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6283390522003174,
      "learning_rate": 1.9418586078011693e-05,
      "loss": 1.2642,
      "step": 1879
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.812283039093018,
      "learning_rate": 1.9417947161613904e-05,
      "loss": 1.083,
      "step": 1880
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9616124629974365,
      "learning_rate": 1.9417308245216116e-05,
      "loss": 1.0046,
      "step": 1881
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.256605386734009,
      "learning_rate": 1.9416669328818324e-05,
      "loss": 0.9141,
      "step": 1882
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.178913593292236,
      "learning_rate": 1.9416030412420536e-05,
      "loss": 0.9777,
      "step": 1883
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.570955991744995,
      "learning_rate": 1.9415391496022747e-05,
      "loss": 0.8461,
      "step": 1884
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.253389358520508,
      "learning_rate": 1.941475257962496e-05,
      "loss": 0.8487,
      "step": 1885
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8577451705932617,
      "learning_rate": 1.9414113663227167e-05,
      "loss": 1.1097,
      "step": 1886
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.72947096824646,
      "learning_rate": 1.941347474682938e-05,
      "loss": 0.87,
      "step": 1887
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5815439224243164,
      "learning_rate": 1.941283583043159e-05,
      "loss": 1.0899,
      "step": 1888
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.175244331359863,
      "learning_rate": 1.94121969140338e-05,
      "loss": 1.2743,
      "step": 1889
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5369322299957275,
      "learning_rate": 1.941155799763601e-05,
      "loss": 0.9219,
      "step": 1890
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4738752841949463,
      "learning_rate": 1.941091908123822e-05,
      "loss": 0.679,
      "step": 1891
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1217093467712402,
      "learning_rate": 1.9410280164840433e-05,
      "loss": 0.8671,
      "step": 1892
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8108692169189453,
      "learning_rate": 1.9409641248442644e-05,
      "loss": 0.362,
      "step": 1893
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.8722124099731445,
      "learning_rate": 1.9409002332044853e-05,
      "loss": 1.0185,
      "step": 1894
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7940268516540527,
      "learning_rate": 1.9408363415647064e-05,
      "loss": 0.9964,
      "step": 1895
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.546579360961914,
      "learning_rate": 1.9407724499249276e-05,
      "loss": 1.4027,
      "step": 1896
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3934929370880127,
      "learning_rate": 1.9407085582851487e-05,
      "loss": 0.805,
      "step": 1897
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.399897813796997,
      "learning_rate": 1.9406446666453695e-05,
      "loss": 1.0569,
      "step": 1898
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4307444095611572,
      "learning_rate": 1.9405807750055907e-05,
      "loss": 0.9136,
      "step": 1899
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2894985675811768,
      "learning_rate": 1.940516883365812e-05,
      "loss": 0.686,
      "step": 1900
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9571607112884521,
      "learning_rate": 1.940452991726033e-05,
      "loss": 0.8387,
      "step": 1901
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3271968364715576,
      "learning_rate": 1.9403891000862538e-05,
      "loss": 0.9611,
      "step": 1902
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8311856985092163,
      "learning_rate": 1.940325208446475e-05,
      "loss": 0.8637,
      "step": 1903
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5785274505615234,
      "learning_rate": 1.9402613168066958e-05,
      "loss": 1.1641,
      "step": 1904
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7591626644134521,
      "learning_rate": 1.9401974251669173e-05,
      "loss": 0.7707,
      "step": 1905
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0068204402923584,
      "learning_rate": 1.940133533527138e-05,
      "loss": 0.8714,
      "step": 1906
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1939234733581543,
      "learning_rate": 1.9400696418873592e-05,
      "loss": 0.9783,
      "step": 1907
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1876943111419678,
      "learning_rate": 1.94000575024758e-05,
      "loss": 0.8717,
      "step": 1908
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4074289798736572,
      "learning_rate": 1.9399418586078012e-05,
      "loss": 1.0074,
      "step": 1909
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.712606191635132,
      "learning_rate": 1.9398779669680224e-05,
      "loss": 1.2526,
      "step": 1910
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.329608201980591,
      "learning_rate": 1.9398140753282435e-05,
      "loss": 0.7687,
      "step": 1911
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8395588397979736,
      "learning_rate": 1.9397501836884643e-05,
      "loss": 0.9692,
      "step": 1912
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2904694080352783,
      "learning_rate": 1.9396862920486855e-05,
      "loss": 0.9625,
      "step": 1913
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.134796142578125,
      "learning_rate": 1.9396224004089067e-05,
      "loss": 0.9545,
      "step": 1914
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.364236831665039,
      "learning_rate": 1.9395585087691278e-05,
      "loss": 0.3755,
      "step": 1915
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0879299640655518,
      "learning_rate": 1.9394946171293486e-05,
      "loss": 0.9451,
      "step": 1916
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4119045734405518,
      "learning_rate": 1.9394307254895698e-05,
      "loss": 0.9819,
      "step": 1917
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9653438329696655,
      "learning_rate": 1.939366833849791e-05,
      "loss": 0.8765,
      "step": 1918
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.089247703552246,
      "learning_rate": 1.939302942210012e-05,
      "loss": 1.0129,
      "step": 1919
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5612308979034424,
      "learning_rate": 1.939239050570233e-05,
      "loss": 1.3099,
      "step": 1920
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.732125997543335,
      "learning_rate": 1.939175158930454e-05,
      "loss": 0.7873,
      "step": 1921
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.132417917251587,
      "learning_rate": 1.9391112672906752e-05,
      "loss": 0.7668,
      "step": 1922
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.631495952606201,
      "learning_rate": 1.9390473756508964e-05,
      "loss": 0.932,
      "step": 1923
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.389344573020935,
      "learning_rate": 1.9389834840111172e-05,
      "loss": 0.3796,
      "step": 1924
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.446890830993652,
      "learning_rate": 1.9389195923713383e-05,
      "loss": 1.0595,
      "step": 1925
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.232012987136841,
      "learning_rate": 1.9388557007315595e-05,
      "loss": 0.9128,
      "step": 1926
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.696296215057373,
      "learning_rate": 1.9387918090917806e-05,
      "loss": 0.9804,
      "step": 1927
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.767951488494873,
      "learning_rate": 1.9387279174520015e-05,
      "loss": 0.9207,
      "step": 1928
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.36132550239563,
      "learning_rate": 1.9386640258122226e-05,
      "loss": 1.0925,
      "step": 1929
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7568690776824951,
      "learning_rate": 1.9386001341724438e-05,
      "loss": 0.8659,
      "step": 1930
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.665513038635254,
      "learning_rate": 1.938536242532665e-05,
      "loss": 1.3062,
      "step": 1931
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.292670726776123,
      "learning_rate": 1.9384723508928857e-05,
      "loss": 1.2645,
      "step": 1932
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9462733268737793,
      "learning_rate": 1.938408459253107e-05,
      "loss": 1.1703,
      "step": 1933
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.136542558670044,
      "learning_rate": 1.938344567613328e-05,
      "loss": 0.8463,
      "step": 1934
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.491992235183716,
      "learning_rate": 1.9382806759735492e-05,
      "loss": 0.8309,
      "step": 1935
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.560408592224121,
      "learning_rate": 1.93821678433377e-05,
      "loss": 1.1748,
      "step": 1936
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9004861116409302,
      "learning_rate": 1.9381528926939912e-05,
      "loss": 0.8585,
      "step": 1937
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5677032470703125,
      "learning_rate": 1.938089001054212e-05,
      "loss": 1.1142,
      "step": 1938
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9058512449264526,
      "learning_rate": 1.9380251094144335e-05,
      "loss": 0.8239,
      "step": 1939
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.622999429702759,
      "learning_rate": 1.9379612177746543e-05,
      "loss": 0.9895,
      "step": 1940
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.757875680923462,
      "learning_rate": 1.9378973261348755e-05,
      "loss": 0.7871,
      "step": 1941
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4343323707580566,
      "learning_rate": 1.9378334344950963e-05,
      "loss": 0.8346,
      "step": 1942
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0447216033935547,
      "learning_rate": 1.9377695428553178e-05,
      "loss": 0.8287,
      "step": 1943
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.223669052124023,
      "learning_rate": 1.9377056512155386e-05,
      "loss": 1.115,
      "step": 1944
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.092203140258789,
      "learning_rate": 1.9376417595757597e-05,
      "loss": 0.7083,
      "step": 1945
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0143229961395264,
      "learning_rate": 1.9375778679359806e-05,
      "loss": 0.8792,
      "step": 1946
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.097198724746704,
      "learning_rate": 1.9375139762962017e-05,
      "loss": 0.9642,
      "step": 1947
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.268231391906738,
      "learning_rate": 1.937450084656423e-05,
      "loss": 1.6283,
      "step": 1948
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.138715982437134,
      "learning_rate": 1.937386193016644e-05,
      "loss": 0.9132,
      "step": 1949
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3601412773132324,
      "learning_rate": 1.937322301376865e-05,
      "loss": 0.9549,
      "step": 1950
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8253885507583618,
      "learning_rate": 1.937258409737086e-05,
      "loss": 0.9104,
      "step": 1951
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.040511131286621,
      "learning_rate": 1.937194518097307e-05,
      "loss": 0.8157,
      "step": 1952
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0217974185943604,
      "learning_rate": 1.9371306264575283e-05,
      "loss": 0.9262,
      "step": 1953
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.047793388366699,
      "learning_rate": 1.937066734817749e-05,
      "loss": 0.8689,
      "step": 1954
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4443124532699585,
      "learning_rate": 1.9370028431779703e-05,
      "loss": 0.3596,
      "step": 1955
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9734774827957153,
      "learning_rate": 1.9369389515381914e-05,
      "loss": 0.9072,
      "step": 1956
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3870576620101929,
      "learning_rate": 1.9368750598984126e-05,
      "loss": 0.3689,
      "step": 1957
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4625213146209717,
      "learning_rate": 1.9368111682586334e-05,
      "loss": 0.9402,
      "step": 1958
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1092820167541504,
      "learning_rate": 1.9367472766188545e-05,
      "loss": 0.9395,
      "step": 1959
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6904534101486206,
      "learning_rate": 1.9366833849790757e-05,
      "loss": 0.8467,
      "step": 1960
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4210665225982666,
      "learning_rate": 1.936619493339297e-05,
      "loss": 0.9539,
      "step": 1961
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3469200134277344,
      "learning_rate": 1.9365556016995177e-05,
      "loss": 1.0321,
      "step": 1962
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7815744876861572,
      "learning_rate": 1.9364917100597388e-05,
      "loss": 0.9272,
      "step": 1963
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7430222034454346,
      "learning_rate": 1.93642781841996e-05,
      "loss": 1.3549,
      "step": 1964
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.6781411170959473,
      "learning_rate": 1.936363926780181e-05,
      "loss": 0.9491,
      "step": 1965
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3880860805511475,
      "learning_rate": 1.936300035140402e-05,
      "loss": 0.8101,
      "step": 1966
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1784145832061768,
      "learning_rate": 1.936236143500623e-05,
      "loss": 1.1639,
      "step": 1967
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.715250253677368,
      "learning_rate": 1.9361722518608443e-05,
      "loss": 0.9156,
      "step": 1968
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5821077823638916,
      "learning_rate": 1.9361083602210654e-05,
      "loss": 0.9905,
      "step": 1969
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3487701416015625,
      "learning_rate": 1.9360444685812862e-05,
      "loss": 0.3775,
      "step": 1970
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.829071521759033,
      "learning_rate": 1.9359805769415074e-05,
      "loss": 1.489,
      "step": 1971
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.078498363494873,
      "learning_rate": 1.9359166853017285e-05,
      "loss": 0.84,
      "step": 1972
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2729434967041016,
      "learning_rate": 1.9358527936619497e-05,
      "loss": 0.9635,
      "step": 1973
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.174825668334961,
      "learning_rate": 1.9357889020221705e-05,
      "loss": 0.8468,
      "step": 1974
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7997961044311523,
      "learning_rate": 1.9357250103823917e-05,
      "loss": 0.8873,
      "step": 1975
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4736456871032715,
      "learning_rate": 1.9356611187426125e-05,
      "loss": 1.2594,
      "step": 1976
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2327173948287964,
      "learning_rate": 1.935597227102834e-05,
      "loss": 0.282,
      "step": 1977
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.578660011291504,
      "learning_rate": 1.9355333354630548e-05,
      "loss": 0.9053,
      "step": 1978
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.749781847000122,
      "learning_rate": 1.935469443823276e-05,
      "loss": 0.9642,
      "step": 1979
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7229807376861572,
      "learning_rate": 1.9354055521834968e-05,
      "loss": 0.9607,
      "step": 1980
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9839211702346802,
      "learning_rate": 1.935341660543718e-05,
      "loss": 0.8003,
      "step": 1981
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.170877695083618,
      "learning_rate": 1.935277768903939e-05,
      "loss": 0.9108,
      "step": 1982
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.232337474822998,
      "learning_rate": 1.9352138772641602e-05,
      "loss": 0.9415,
      "step": 1983
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.607276201248169,
      "learning_rate": 1.935149985624381e-05,
      "loss": 1.1576,
      "step": 1984
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5387892723083496,
      "learning_rate": 1.9350860939846022e-05,
      "loss": 1.0792,
      "step": 1985
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0290281772613525,
      "learning_rate": 1.9350222023448234e-05,
      "loss": 0.9705,
      "step": 1986
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.7469425201416016,
      "learning_rate": 1.9349583107050445e-05,
      "loss": 0.8638,
      "step": 1987
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3750417232513428,
      "learning_rate": 1.9348944190652653e-05,
      "loss": 0.9123,
      "step": 1988
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.283529281616211,
      "learning_rate": 1.9348305274254865e-05,
      "loss": 0.9559,
      "step": 1989
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.184332847595215,
      "learning_rate": 1.9347666357857076e-05,
      "loss": 1.3238,
      "step": 1990
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.913553476333618,
      "learning_rate": 1.9347027441459288e-05,
      "loss": 0.8616,
      "step": 1991
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0316872596740723,
      "learning_rate": 1.9346388525061496e-05,
      "loss": 0.9423,
      "step": 1992
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.656050443649292,
      "learning_rate": 1.9345749608663708e-05,
      "loss": 1.3166,
      "step": 1993
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.210760593414307,
      "learning_rate": 1.934511069226592e-05,
      "loss": 1.2047,
      "step": 1994
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3379392623901367,
      "learning_rate": 1.934447177586813e-05,
      "loss": 0.8917,
      "step": 1995
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.121729373931885,
      "learning_rate": 1.934383285947034e-05,
      "loss": 1.0616,
      "step": 1996
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2869796752929688,
      "learning_rate": 1.934319394307255e-05,
      "loss": 0.9097,
      "step": 1997
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9487024545669556,
      "learning_rate": 1.9342555026674762e-05,
      "loss": 0.9191,
      "step": 1998
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.848737359046936,
      "learning_rate": 1.9341916110276973e-05,
      "loss": 0.9295,
      "step": 1999
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3528096675872803,
      "learning_rate": 1.934127719387918e-05,
      "loss": 1.0705,
      "step": 2000
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.50563645362854,
      "learning_rate": 1.9340638277481393e-05,
      "loss": 0.2239,
      "step": 2001
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9553742408752441,
      "learning_rate": 1.9339999361083605e-05,
      "loss": 0.8662,
      "step": 2002
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6789199113845825,
      "learning_rate": 1.9339360444685816e-05,
      "loss": 0.4082,
      "step": 2003
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.432516098022461,
      "learning_rate": 1.9338721528288024e-05,
      "loss": 0.6839,
      "step": 2004
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.39449405670166,
      "learning_rate": 1.9338082611890236e-05,
      "loss": 0.9185,
      "step": 2005
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.595817565917969,
      "learning_rate": 1.9337443695492448e-05,
      "loss": 1.1415,
      "step": 2006
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.439223051071167,
      "learning_rate": 1.933680477909466e-05,
      "loss": 0.9991,
      "step": 2007
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.9376649856567383,
      "learning_rate": 1.9336165862696867e-05,
      "loss": 1.1284,
      "step": 2008
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.221384525299072,
      "learning_rate": 1.933552694629908e-05,
      "loss": 1.0987,
      "step": 2009
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3356237411499023,
      "learning_rate": 1.933488802990129e-05,
      "loss": 0.9548,
      "step": 2010
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1291167736053467,
      "learning_rate": 1.9334249113503502e-05,
      "loss": 0.9113,
      "step": 2011
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3636720180511475,
      "learning_rate": 1.933361019710571e-05,
      "loss": 0.8943,
      "step": 2012
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.256298303604126,
      "learning_rate": 1.933297128070792e-05,
      "loss": 0.9887,
      "step": 2013
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.209174156188965,
      "learning_rate": 1.933233236431013e-05,
      "loss": 0.9573,
      "step": 2014
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.962010383605957,
      "learning_rate": 1.9331693447912345e-05,
      "loss": 1.0174,
      "step": 2015
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.274061441421509,
      "learning_rate": 1.9331054531514553e-05,
      "loss": 0.9432,
      "step": 2016
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.032893180847168,
      "learning_rate": 1.9330415615116764e-05,
      "loss": 1.0441,
      "step": 2017
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.88721764087677,
      "learning_rate": 1.9329776698718973e-05,
      "loss": 0.9129,
      "step": 2018
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7369104623794556,
      "learning_rate": 1.9329137782321184e-05,
      "loss": 0.8417,
      "step": 2019
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.358527421951294,
      "learning_rate": 1.9328498865923396e-05,
      "loss": 0.3769,
      "step": 2020
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7446160316467285,
      "learning_rate": 1.9327859949525607e-05,
      "loss": 1.0369,
      "step": 2021
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2002480030059814,
      "learning_rate": 1.9327221033127815e-05,
      "loss": 1.0013,
      "step": 2022
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.06001353263855,
      "learning_rate": 1.9326582116730027e-05,
      "loss": 0.7913,
      "step": 2023
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0012989044189453,
      "learning_rate": 1.932594320033224e-05,
      "loss": 0.8233,
      "step": 2024
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1846628189086914,
      "learning_rate": 1.932530428393445e-05,
      "loss": 1.115,
      "step": 2025
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5673460960388184,
      "learning_rate": 1.9324665367536658e-05,
      "loss": 0.9664,
      "step": 2026
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0314934253692627,
      "learning_rate": 1.932402645113887e-05,
      "loss": 0.7964,
      "step": 2027
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3296687602996826,
      "learning_rate": 1.932338753474108e-05,
      "loss": 0.8939,
      "step": 2028
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1601741313934326,
      "learning_rate": 1.9322748618343293e-05,
      "loss": 1.2674,
      "step": 2029
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.239692211151123,
      "learning_rate": 1.93221097019455e-05,
      "loss": 0.966,
      "step": 2030
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.492405891418457,
      "learning_rate": 1.9321470785547712e-05,
      "loss": 1.3481,
      "step": 2031
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.8703694343566895,
      "learning_rate": 1.9320831869149924e-05,
      "loss": 1.1727,
      "step": 2032
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1398215293884277,
      "learning_rate": 1.9320192952752136e-05,
      "loss": 0.9507,
      "step": 2033
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.1814751625061035,
      "learning_rate": 1.9319554036354344e-05,
      "loss": 0.8169,
      "step": 2034
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.07225227355957,
      "learning_rate": 1.9318915119956555e-05,
      "loss": 0.9269,
      "step": 2035
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.287281036376953,
      "learning_rate": 1.9318276203558767e-05,
      "loss": 0.9628,
      "step": 2036
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8678892850875854,
      "learning_rate": 1.931763728716098e-05,
      "loss": 0.9193,
      "step": 2037
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1334190368652344,
      "learning_rate": 1.9316998370763187e-05,
      "loss": 0.8255,
      "step": 2038
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.080183982849121,
      "learning_rate": 1.9316359454365398e-05,
      "loss": 0.9023,
      "step": 2039
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5493032932281494,
      "learning_rate": 1.931572053796761e-05,
      "loss": 0.3839,
      "step": 2040
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7999520301818848,
      "learning_rate": 1.931508162156982e-05,
      "loss": 1.1886,
      "step": 2041
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1936185359954834,
      "learning_rate": 1.931444270517203e-05,
      "loss": 1.3212,
      "step": 2042
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.059288740158081,
      "learning_rate": 1.931380378877424e-05,
      "loss": 0.8794,
      "step": 2043
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4383914470672607,
      "learning_rate": 1.9313164872376452e-05,
      "loss": 0.929,
      "step": 2044
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8312760591506958,
      "learning_rate": 1.9312525955978664e-05,
      "loss": 0.9121,
      "step": 2045
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.189910888671875,
      "learning_rate": 1.9311887039580872e-05,
      "loss": 0.9188,
      "step": 2046
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.1522321701049805,
      "learning_rate": 1.9311248123183084e-05,
      "loss": 1.5236,
      "step": 2047
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.775269031524658,
      "learning_rate": 1.9310609206785292e-05,
      "loss": 1.0514,
      "step": 2048
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.04127836227417,
      "learning_rate": 1.9309970290387507e-05,
      "loss": 1.1918,
      "step": 2049
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7832202911376953,
      "learning_rate": 1.9309331373989715e-05,
      "loss": 1.0275,
      "step": 2050
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5487087965011597,
      "learning_rate": 1.9308692457591926e-05,
      "loss": 0.2459,
      "step": 2051
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.6759791374206543,
      "learning_rate": 1.9308053541194135e-05,
      "loss": 1.144,
      "step": 2052
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4055862426757812,
      "learning_rate": 1.9307414624796346e-05,
      "loss": 0.9267,
      "step": 2053
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.119424343109131,
      "learning_rate": 1.9306775708398558e-05,
      "loss": 0.9206,
      "step": 2054
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.191646099090576,
      "learning_rate": 1.930613679200077e-05,
      "loss": 0.9095,
      "step": 2055
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0265023708343506,
      "learning_rate": 1.9305497875602977e-05,
      "loss": 1.0301,
      "step": 2056
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0963008403778076,
      "learning_rate": 1.930485895920519e-05,
      "loss": 1.0384,
      "step": 2057
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1640384197235107,
      "learning_rate": 1.93042200428074e-05,
      "loss": 1.0314,
      "step": 2058
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.1825666427612305,
      "learning_rate": 1.9303581126409612e-05,
      "loss": 1.3527,
      "step": 2059
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.745335102081299,
      "learning_rate": 1.930294221001182e-05,
      "loss": 1.2774,
      "step": 2060
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1543149948120117,
      "learning_rate": 1.9302303293614032e-05,
      "loss": 0.9507,
      "step": 2061
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.207505226135254,
      "learning_rate": 1.9301664377216243e-05,
      "loss": 0.9017,
      "step": 2062
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.132317066192627,
      "learning_rate": 1.9301025460818455e-05,
      "loss": 0.8626,
      "step": 2063
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.876447319984436,
      "learning_rate": 1.9300386544420663e-05,
      "loss": 0.8601,
      "step": 2064
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2949414253234863,
      "learning_rate": 1.9299747628022875e-05,
      "loss": 0.9421,
      "step": 2065
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.295488119125366,
      "learning_rate": 1.9299108711625086e-05,
      "loss": 1.0569,
      "step": 2066
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.597936630249023,
      "learning_rate": 1.9298469795227298e-05,
      "loss": 1.4066,
      "step": 2067
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.829970359802246,
      "learning_rate": 1.9297830878829506e-05,
      "loss": 0.9734,
      "step": 2068
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.748550295829773,
      "learning_rate": 1.9297191962431717e-05,
      "loss": 0.9033,
      "step": 2069
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4248723983764648,
      "learning_rate": 1.929655304603393e-05,
      "loss": 0.242,
      "step": 2070
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3028299808502197,
      "learning_rate": 1.929591412963614e-05,
      "loss": 1.0097,
      "step": 2071
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1938235759735107,
      "learning_rate": 1.929527521323835e-05,
      "loss": 1.1991,
      "step": 2072
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.41437029838562,
      "learning_rate": 1.929463629684056e-05,
      "loss": 0.9923,
      "step": 2073
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.082589626312256,
      "learning_rate": 1.9293997380442772e-05,
      "loss": 0.8425,
      "step": 2074
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1909611225128174,
      "learning_rate": 1.9293358464044983e-05,
      "loss": 0.8874,
      "step": 2075
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3550808429718018,
      "learning_rate": 1.929271954764719e-05,
      "loss": 0.8892,
      "step": 2076
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.318342924118042,
      "learning_rate": 1.9292080631249403e-05,
      "loss": 0.9116,
      "step": 2077
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0760273933410645,
      "learning_rate": 1.9291441714851615e-05,
      "loss": 0.7279,
      "step": 2078
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.550447702407837,
      "learning_rate": 1.9290802798453826e-05,
      "loss": 0.9062,
      "step": 2079
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0270543098449707,
      "learning_rate": 1.9290163882056034e-05,
      "loss": 1.0819,
      "step": 2080
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2936532497406006,
      "learning_rate": 1.9289524965658246e-05,
      "loss": 1.4524,
      "step": 2081
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.106909990310669,
      "learning_rate": 1.9288886049260457e-05,
      "loss": 0.864,
      "step": 2082
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2652812004089355,
      "learning_rate": 1.928824713286267e-05,
      "loss": 1.1434,
      "step": 2083
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1158812046051025,
      "learning_rate": 1.9287608216464877e-05,
      "loss": 0.9189,
      "step": 2084
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.6567606925964355,
      "learning_rate": 1.928696930006709e-05,
      "loss": 1.2733,
      "step": 2085
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.51979923248291,
      "learning_rate": 1.9286330383669297e-05,
      "loss": 0.7676,
      "step": 2086
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9761040210723877,
      "learning_rate": 1.928569146727151e-05,
      "loss": 0.8634,
      "step": 2087
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4880170822143555,
      "learning_rate": 1.928505255087372e-05,
      "loss": 1.0435,
      "step": 2088
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.090041637420654,
      "learning_rate": 1.928441363447593e-05,
      "loss": 1.2055,
      "step": 2089
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.349850177764893,
      "learning_rate": 1.928377471807814e-05,
      "loss": 1.8205,
      "step": 2090
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.403886556625366,
      "learning_rate": 1.928313580168035e-05,
      "loss": 0.8864,
      "step": 2091
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.694874286651611,
      "learning_rate": 1.9282496885282563e-05,
      "loss": 1.3184,
      "step": 2092
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4413907527923584,
      "learning_rate": 1.9281857968884774e-05,
      "loss": 0.9,
      "step": 2093
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2221992015838623,
      "learning_rate": 1.9281219052486982e-05,
      "loss": 0.9106,
      "step": 2094
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.438576579093933,
      "learning_rate": 1.9280580136089194e-05,
      "loss": 0.2546,
      "step": 2095
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3711813688278198,
      "learning_rate": 1.9279941219691405e-05,
      "loss": 0.2829,
      "step": 2096
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.298685312271118,
      "learning_rate": 1.9279302303293617e-05,
      "loss": 0.8794,
      "step": 2097
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.049978256225586,
      "learning_rate": 1.9278663386895825e-05,
      "loss": 1.0064,
      "step": 2098
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4543092250823975,
      "learning_rate": 1.9278024470498037e-05,
      "loss": 1.0668,
      "step": 2099
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9732669591903687,
      "learning_rate": 1.9277385554100248e-05,
      "loss": 0.8477,
      "step": 2100
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9505079984664917,
      "learning_rate": 1.927674663770246e-05,
      "loss": 0.8958,
      "step": 2101
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.806627035140991,
      "learning_rate": 1.9276107721304668e-05,
      "loss": 1.158,
      "step": 2102
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0236754417419434,
      "learning_rate": 1.927546880490688e-05,
      "loss": 0.9687,
      "step": 2103
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2223703861236572,
      "learning_rate": 1.927482988850909e-05,
      "loss": 1.1278,
      "step": 2104
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.451612710952759,
      "learning_rate": 1.9274190972111303e-05,
      "loss": 1.2975,
      "step": 2105
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.11836314201355,
      "learning_rate": 1.927355205571351e-05,
      "loss": 0.5192,
      "step": 2106
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.073347091674805,
      "learning_rate": 1.9272913139315722e-05,
      "loss": 1.3372,
      "step": 2107
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7082130908966064,
      "learning_rate": 1.9272274222917934e-05,
      "loss": 0.85,
      "step": 2108
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2358434200286865,
      "learning_rate": 1.9271635306520145e-05,
      "loss": 0.9008,
      "step": 2109
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.158740758895874,
      "learning_rate": 1.9270996390122354e-05,
      "loss": 0.7011,
      "step": 2110
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.546088933944702,
      "learning_rate": 1.9270357473724565e-05,
      "loss": 1.0706,
      "step": 2111
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.421164035797119,
      "learning_rate": 1.9269718557326777e-05,
      "loss": 1.4398,
      "step": 2112
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.545656204223633,
      "learning_rate": 1.9269079640928988e-05,
      "loss": 1.1488,
      "step": 2113
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3985755443573,
      "learning_rate": 1.9268440724531196e-05,
      "loss": 0.9673,
      "step": 2114
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3597357273101807,
      "learning_rate": 1.9267801808133408e-05,
      "loss": 0.8127,
      "step": 2115
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.0595476627349854,
      "learning_rate": 1.926716289173562e-05,
      "loss": 0.9722,
      "step": 2116
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.535168409347534,
      "learning_rate": 1.926652397533783e-05,
      "loss": 1.2028,
      "step": 2117
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.183052062988281,
      "learning_rate": 1.926588505894004e-05,
      "loss": 1.274,
      "step": 2118
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2192699909210205,
      "learning_rate": 1.926524614254225e-05,
      "loss": 0.8831,
      "step": 2119
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.651463508605957,
      "learning_rate": 1.9264607226144462e-05,
      "loss": 0.8457,
      "step": 2120
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.09071683883667,
      "learning_rate": 1.9263968309746674e-05,
      "loss": 0.9014,
      "step": 2121
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0930135250091553,
      "learning_rate": 1.9263329393348882e-05,
      "loss": 0.8461,
      "step": 2122
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3142383098602295,
      "learning_rate": 1.9262690476951093e-05,
      "loss": 0.8945,
      "step": 2123
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3771932125091553,
      "learning_rate": 1.92620515605533e-05,
      "loss": 0.9462,
      "step": 2124
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5270462036132812,
      "learning_rate": 1.9261412644155517e-05,
      "loss": 0.8704,
      "step": 2125
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4092214107513428,
      "learning_rate": 1.9260773727757725e-05,
      "loss": 1.0435,
      "step": 2126
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.055203676223755,
      "learning_rate": 1.9260134811359936e-05,
      "loss": 1.0138,
      "step": 2127
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.01180100440979,
      "learning_rate": 1.9259495894962144e-05,
      "loss": 0.8704,
      "step": 2128
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.796834945678711,
      "learning_rate": 1.9258856978564356e-05,
      "loss": 0.7082,
      "step": 2129
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8657913208007812,
      "learning_rate": 1.9258218062166568e-05,
      "loss": 0.8695,
      "step": 2130
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.495215892791748,
      "learning_rate": 1.9257579145768776e-05,
      "loss": 0.8957,
      "step": 2131
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1156489849090576,
      "learning_rate": 1.9256940229370987e-05,
      "loss": 0.939,
      "step": 2132
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3898253440856934,
      "learning_rate": 1.92563013129732e-05,
      "loss": 1.118,
      "step": 2133
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7633426189422607,
      "learning_rate": 1.925566239657541e-05,
      "loss": 0.8964,
      "step": 2134
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.149896144866943,
      "learning_rate": 1.925502348017762e-05,
      "loss": 0.8205,
      "step": 2135
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0723350048065186,
      "learning_rate": 1.925438456377983e-05,
      "loss": 0.8754,
      "step": 2136
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2348620891571045,
      "learning_rate": 1.925374564738204e-05,
      "loss": 1.3092,
      "step": 2137
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.064091920852661,
      "learning_rate": 1.9253106730984253e-05,
      "loss": 0.9425,
      "step": 2138
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.891624689102173,
      "learning_rate": 1.925246781458646e-05,
      "loss": 0.9154,
      "step": 2139
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.973506450653076,
      "learning_rate": 1.9251828898188673e-05,
      "loss": 1.648,
      "step": 2140
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1620402336120605,
      "learning_rate": 1.9251189981790884e-05,
      "loss": 0.9336,
      "step": 2141
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.421262502670288,
      "learning_rate": 1.9250551065393096e-05,
      "loss": 1.403,
      "step": 2142
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8009089231491089,
      "learning_rate": 1.9249912148995304e-05,
      "loss": 0.8702,
      "step": 2143
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9183356761932373,
      "learning_rate": 1.9249273232597516e-05,
      "loss": 0.8713,
      "step": 2144
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.026468515396118,
      "learning_rate": 1.9248634316199727e-05,
      "loss": 0.6279,
      "step": 2145
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.332148790359497,
      "learning_rate": 1.924799539980194e-05,
      "loss": 0.3849,
      "step": 2146
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2466723918914795,
      "learning_rate": 1.9247356483404147e-05,
      "loss": 1.0695,
      "step": 2147
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8553680181503296,
      "learning_rate": 1.924671756700636e-05,
      "loss": 0.8192,
      "step": 2148
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.545372486114502,
      "learning_rate": 1.924607865060857e-05,
      "loss": 0.9241,
      "step": 2149
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1921069622039795,
      "learning_rate": 1.924543973421078e-05,
      "loss": 0.7946,
      "step": 2150
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.01261305809021,
      "learning_rate": 1.924480081781299e-05,
      "loss": 0.8915,
      "step": 2151
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.982446551322937,
      "learning_rate": 1.92441619014152e-05,
      "loss": 0.8522,
      "step": 2152
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5766842365264893,
      "learning_rate": 1.924352298501741e-05,
      "loss": 0.8617,
      "step": 2153
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7308622598648071,
      "learning_rate": 1.9242884068619624e-05,
      "loss": 0.8566,
      "step": 2154
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.243965148925781,
      "learning_rate": 1.9242245152221832e-05,
      "loss": 1.6766,
      "step": 2155
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6635398864746094,
      "learning_rate": 1.9241606235824044e-05,
      "loss": 1.1175,
      "step": 2156
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5634615421295166,
      "learning_rate": 1.9240967319426252e-05,
      "loss": 0.9885,
      "step": 2157
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.6000518798828125,
      "learning_rate": 1.9240328403028464e-05,
      "loss": 1.5307,
      "step": 2158
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6005148887634277,
      "learning_rate": 1.9239689486630675e-05,
      "loss": 0.9723,
      "step": 2159
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.907137393951416,
      "learning_rate": 1.9239050570232887e-05,
      "loss": 0.967,
      "step": 2160
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.443040370941162,
      "learning_rate": 1.9238411653835095e-05,
      "loss": 0.863,
      "step": 2161
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7777438163757324,
      "learning_rate": 1.9237772737437307e-05,
      "loss": 0.7432,
      "step": 2162
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.725320816040039,
      "learning_rate": 1.9237133821039518e-05,
      "loss": 1.6248,
      "step": 2163
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8869173526763916,
      "learning_rate": 1.923649490464173e-05,
      "loss": 1.09,
      "step": 2164
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7570463418960571,
      "learning_rate": 1.9235855988243938e-05,
      "loss": 0.8261,
      "step": 2165
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3790178298950195,
      "learning_rate": 1.923521707184615e-05,
      "loss": 0.8591,
      "step": 2166
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.742076873779297,
      "learning_rate": 1.923457815544836e-05,
      "loss": 0.8307,
      "step": 2167
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.467884540557861,
      "learning_rate": 1.9233939239050572e-05,
      "loss": 0.8372,
      "step": 2168
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.617090225219727,
      "learning_rate": 1.923330032265278e-05,
      "loss": 1.4415,
      "step": 2169
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6709694862365723,
      "learning_rate": 1.9232661406254992e-05,
      "loss": 0.3532,
      "step": 2170
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1249539852142334,
      "learning_rate": 1.9232022489857204e-05,
      "loss": 0.7484,
      "step": 2171
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6806650161743164,
      "learning_rate": 1.9231383573459415e-05,
      "loss": 0.788,
      "step": 2172
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4244587421417236,
      "learning_rate": 1.9230744657061623e-05,
      "loss": 0.9951,
      "step": 2173
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4582860469818115,
      "learning_rate": 1.9230105740663835e-05,
      "loss": 1.2041,
      "step": 2174
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6898446083068848,
      "learning_rate": 1.9229466824266046e-05,
      "loss": 0.8693,
      "step": 2175
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.416229248046875,
      "learning_rate": 1.9228827907868258e-05,
      "loss": 1.3099,
      "step": 2176
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3748884201049805,
      "learning_rate": 1.9228188991470466e-05,
      "loss": 1.0475,
      "step": 2177
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.616032123565674,
      "learning_rate": 1.9227550075072678e-05,
      "loss": 0.9265,
      "step": 2178
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7139036655426025,
      "learning_rate": 1.922691115867489e-05,
      "loss": 0.8306,
      "step": 2179
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6819283962249756,
      "learning_rate": 1.92262722422771e-05,
      "loss": 1.3495,
      "step": 2180
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1807961463928223,
      "learning_rate": 1.922563332587931e-05,
      "loss": 0.9596,
      "step": 2181
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6199848651885986,
      "learning_rate": 1.922499440948152e-05,
      "loss": 0.6803,
      "step": 2182
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.702264308929443,
      "learning_rate": 1.9224355493083732e-05,
      "loss": 1.4342,
      "step": 2183
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.386053562164307,
      "learning_rate": 1.9223716576685944e-05,
      "loss": 1.5296,
      "step": 2184
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4078869819641113,
      "learning_rate": 1.9223077660288152e-05,
      "loss": 1.0409,
      "step": 2185
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0592665672302246,
      "learning_rate": 1.9222438743890363e-05,
      "loss": 0.9669,
      "step": 2186
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.648874282836914,
      "learning_rate": 1.9221799827492575e-05,
      "loss": 1.9375,
      "step": 2187
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.144535541534424,
      "learning_rate": 1.9221160911094786e-05,
      "loss": 1.2079,
      "step": 2188
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5430924892425537,
      "learning_rate": 1.9220521994696995e-05,
      "loss": 0.9627,
      "step": 2189
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.679417610168457,
      "learning_rate": 1.9219883078299206e-05,
      "loss": 0.9786,
      "step": 2190
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2906997203826904,
      "learning_rate": 1.9219244161901414e-05,
      "loss": 1.073,
      "step": 2191
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.053155422210693,
      "learning_rate": 1.921860524550363e-05,
      "loss": 1.4555,
      "step": 2192
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1325626373291016,
      "learning_rate": 1.9217966329105837e-05,
      "loss": 1.1068,
      "step": 2193
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1818575859069824,
      "learning_rate": 1.921732741270805e-05,
      "loss": 1.0776,
      "step": 2194
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4680137634277344,
      "learning_rate": 1.9216688496310257e-05,
      "loss": 0.8062,
      "step": 2195
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2430858612060547,
      "learning_rate": 1.921604957991247e-05,
      "loss": 1.2822,
      "step": 2196
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8605482578277588,
      "learning_rate": 1.921541066351468e-05,
      "loss": 0.9028,
      "step": 2197
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4372730255126953,
      "learning_rate": 1.9214771747116892e-05,
      "loss": 1.0371,
      "step": 2198
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.745333194732666,
      "learning_rate": 1.92141328307191e-05,
      "loss": 0.9307,
      "step": 2199
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.901869297027588,
      "learning_rate": 1.921349391432131e-05,
      "loss": 0.9691,
      "step": 2200
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.268097162246704,
      "learning_rate": 1.9212854997923523e-05,
      "loss": 0.9593,
      "step": 2201
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.182837963104248,
      "learning_rate": 1.9212216081525735e-05,
      "loss": 0.9393,
      "step": 2202
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8713284730911255,
      "learning_rate": 1.9211577165127943e-05,
      "loss": 0.7658,
      "step": 2203
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.097191572189331,
      "learning_rate": 1.9210938248730154e-05,
      "loss": 0.9125,
      "step": 2204
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.099985361099243,
      "learning_rate": 1.9210299332332366e-05,
      "loss": 1.0958,
      "step": 2205
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2862677574157715,
      "learning_rate": 1.9209660415934577e-05,
      "loss": 0.9775,
      "step": 2206
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6098637580871582,
      "learning_rate": 1.9209021499536786e-05,
      "loss": 0.3396,
      "step": 2207
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.950356125831604,
      "learning_rate": 1.9208382583138997e-05,
      "loss": 0.8549,
      "step": 2208
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3981704711914062,
      "learning_rate": 1.920774366674121e-05,
      "loss": 0.8071,
      "step": 2209
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.527930736541748,
      "learning_rate": 1.920710475034342e-05,
      "loss": 1.0502,
      "step": 2210
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.68705153465271,
      "learning_rate": 1.9206465833945628e-05,
      "loss": 1.0941,
      "step": 2211
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.300614833831787,
      "learning_rate": 1.920582691754784e-05,
      "loss": 0.8656,
      "step": 2212
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.152959108352661,
      "learning_rate": 1.920518800115005e-05,
      "loss": 0.88,
      "step": 2213
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.034928798675537,
      "learning_rate": 1.9204549084752263e-05,
      "loss": 0.9388,
      "step": 2214
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4442694187164307,
      "learning_rate": 1.920391016835447e-05,
      "loss": 1.0231,
      "step": 2215
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.387388229370117,
      "learning_rate": 1.9203271251956683e-05,
      "loss": 0.8398,
      "step": 2216
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.59877347946167,
      "learning_rate": 1.9202632335558894e-05,
      "loss": 0.9134,
      "step": 2217
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.103191614151001,
      "learning_rate": 1.9201993419161106e-05,
      "loss": 0.8897,
      "step": 2218
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7825950384140015,
      "learning_rate": 1.9201354502763314e-05,
      "loss": 0.9259,
      "step": 2219
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.240096092224121,
      "learning_rate": 1.9200715586365525e-05,
      "loss": 1.7813,
      "step": 2220
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3525848388671875,
      "learning_rate": 1.9200076669967737e-05,
      "loss": 0.9879,
      "step": 2221
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9277397394180298,
      "learning_rate": 1.919943775356995e-05,
      "loss": 1.0248,
      "step": 2222
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6520825624465942,
      "learning_rate": 1.9198798837172157e-05,
      "loss": 0.8716,
      "step": 2223
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2436031103134155,
      "learning_rate": 1.9198159920774368e-05,
      "loss": 0.3337,
      "step": 2224
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.710008144378662,
      "learning_rate": 1.9197521004376576e-05,
      "loss": 1.109,
      "step": 2225
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2289812564849854,
      "learning_rate": 1.919688208797879e-05,
      "loss": 0.8792,
      "step": 2226
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.970508098602295,
      "learning_rate": 1.9196243171581e-05,
      "loss": 1.2114,
      "step": 2227
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.536412239074707,
      "learning_rate": 1.919560425518321e-05,
      "loss": 0.8136,
      "step": 2228
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0330183506011963,
      "learning_rate": 1.919496533878542e-05,
      "loss": 1.028,
      "step": 2229
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7892765998840332,
      "learning_rate": 1.919432642238763e-05,
      "loss": 0.844,
      "step": 2230
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7820818424224854,
      "learning_rate": 1.9193687505989842e-05,
      "loss": 0.8456,
      "step": 2231
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6339781284332275,
      "learning_rate": 1.9193048589592054e-05,
      "loss": 0.8134,
      "step": 2232
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2593655586242676,
      "learning_rate": 1.9192409673194262e-05,
      "loss": 0.9395,
      "step": 2233
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1286020278930664,
      "learning_rate": 1.9191770756796474e-05,
      "loss": 0.9573,
      "step": 2234
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.34700608253479,
      "learning_rate": 1.9191131840398685e-05,
      "loss": 0.9507,
      "step": 2235
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.831590175628662,
      "learning_rate": 1.9190492924000897e-05,
      "loss": 0.9483,
      "step": 2236
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1042346954345703,
      "learning_rate": 1.9189854007603105e-05,
      "loss": 0.9465,
      "step": 2237
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.740793466567993,
      "learning_rate": 1.9189215091205316e-05,
      "loss": 0.8515,
      "step": 2238
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4305436611175537,
      "learning_rate": 1.9188576174807528e-05,
      "loss": 1.0717,
      "step": 2239
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2095766067504883,
      "learning_rate": 1.918793725840974e-05,
      "loss": 0.875,
      "step": 2240
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.013923168182373,
      "learning_rate": 1.9187298342011948e-05,
      "loss": 1.0538,
      "step": 2241
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8741413354873657,
      "learning_rate": 1.918665942561416e-05,
      "loss": 0.7585,
      "step": 2242
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.099614143371582,
      "learning_rate": 1.918602050921637e-05,
      "loss": 1.0037,
      "step": 2243
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1536648273468018,
      "learning_rate": 1.9185381592818582e-05,
      "loss": 1.3566,
      "step": 2244
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.846747875213623,
      "learning_rate": 1.918474267642079e-05,
      "loss": 0.9005,
      "step": 2245
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.065394401550293,
      "learning_rate": 1.9184103760023002e-05,
      "loss": 0.8945,
      "step": 2246
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.009538412094116,
      "learning_rate": 1.9183464843625214e-05,
      "loss": 0.9964,
      "step": 2247
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.812484264373779,
      "learning_rate": 1.9182825927227425e-05,
      "loss": 1.6276,
      "step": 2248
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.514882802963257,
      "learning_rate": 1.9182187010829633e-05,
      "loss": 0.944,
      "step": 2249
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.388077259063721,
      "learning_rate": 1.9181548094431845e-05,
      "loss": 1.1422,
      "step": 2250
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7451539039611816,
      "learning_rate": 1.9180909178034056e-05,
      "loss": 1.0044,
      "step": 2251
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2775018215179443,
      "learning_rate": 1.9180270261636268e-05,
      "loss": 0.9818,
      "step": 2252
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4833662509918213,
      "learning_rate": 1.9179631345238476e-05,
      "loss": 0.8968,
      "step": 2253
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.472927093505859,
      "learning_rate": 1.9178992428840688e-05,
      "loss": 1.2963,
      "step": 2254
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8554495573043823,
      "learning_rate": 1.91783535124429e-05,
      "loss": 0.824,
      "step": 2255
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.834824562072754,
      "learning_rate": 1.917771459604511e-05,
      "loss": 1.3291,
      "step": 2256
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.104792833328247,
      "learning_rate": 1.917707567964732e-05,
      "loss": 1.015,
      "step": 2257
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.390687942504883,
      "learning_rate": 1.917643676324953e-05,
      "loss": 1.5862,
      "step": 2258
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.371107816696167,
      "learning_rate": 1.9175797846851742e-05,
      "loss": 0.8533,
      "step": 2259
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1212475299835205,
      "learning_rate": 1.9175158930453953e-05,
      "loss": 0.8965,
      "step": 2260
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.755420446395874,
      "learning_rate": 1.917452001405616e-05,
      "loss": 0.9396,
      "step": 2261
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.447981357574463,
      "learning_rate": 1.9173881097658373e-05,
      "loss": 0.8856,
      "step": 2262
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.001875162124634,
      "learning_rate": 1.917324218126058e-05,
      "loss": 1.2075,
      "step": 2263
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4803390502929688,
      "learning_rate": 1.9172603264862796e-05,
      "loss": 0.8494,
      "step": 2264
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.090632677078247,
      "learning_rate": 1.9171964348465004e-05,
      "loss": 0.9914,
      "step": 2265
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4803415536880493,
      "learning_rate": 1.9171325432067216e-05,
      "loss": 0.2865,
      "step": 2266
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.0034024715423584,
      "learning_rate": 1.9170686515669424e-05,
      "loss": 1.0344,
      "step": 2267
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.910205602645874,
      "learning_rate": 1.9170047599271636e-05,
      "loss": 1.3699,
      "step": 2268
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.84064781665802,
      "learning_rate": 1.9169408682873847e-05,
      "loss": 0.8925,
      "step": 2269
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.09651780128479,
      "learning_rate": 1.916876976647606e-05,
      "loss": 0.9506,
      "step": 2270
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7361700534820557,
      "learning_rate": 1.9168130850078267e-05,
      "loss": 0.8546,
      "step": 2271
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.025287628173828,
      "learning_rate": 1.916749193368048e-05,
      "loss": 1.2461,
      "step": 2272
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.269685745239258,
      "learning_rate": 1.916685301728269e-05,
      "loss": 0.8812,
      "step": 2273
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.312485694885254,
      "learning_rate": 1.91662141008849e-05,
      "loss": 0.8469,
      "step": 2274
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9797717332839966,
      "learning_rate": 1.916557518448711e-05,
      "loss": 0.8642,
      "step": 2275
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5225982666015625,
      "learning_rate": 1.916493626808932e-05,
      "loss": 0.4448,
      "step": 2276
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.943976879119873,
      "learning_rate": 1.9164297351691533e-05,
      "loss": 1.3442,
      "step": 2277
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2356836795806885,
      "learning_rate": 1.9163658435293744e-05,
      "loss": 0.9541,
      "step": 2278
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7330102920532227,
      "learning_rate": 1.9163019518895953e-05,
      "loss": 1.3834,
      "step": 2279
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3066112995147705,
      "learning_rate": 1.9162380602498164e-05,
      "loss": 0.8455,
      "step": 2280
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.5114827156066895,
      "learning_rate": 1.9161741686100376e-05,
      "loss": 0.9822,
      "step": 2281
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.503124237060547,
      "learning_rate": 1.9161102769702587e-05,
      "loss": 1.8037,
      "step": 2282
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.920732021331787,
      "learning_rate": 1.9160463853304795e-05,
      "loss": 0.8059,
      "step": 2283
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.302706003189087,
      "learning_rate": 1.9159824936907007e-05,
      "loss": 0.9849,
      "step": 2284
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3030896186828613,
      "learning_rate": 1.915918602050922e-05,
      "loss": 0.9834,
      "step": 2285
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.901095390319824,
      "learning_rate": 1.915854710411143e-05,
      "loss": 1.9619,
      "step": 2286
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.1698431968688965,
      "learning_rate": 1.9157908187713638e-05,
      "loss": 1.2888,
      "step": 2287
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.348926067352295,
      "learning_rate": 1.915726927131585e-05,
      "loss": 1.2828,
      "step": 2288
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.697715759277344,
      "learning_rate": 1.915663035491806e-05,
      "loss": 1.236,
      "step": 2289
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1928470134735107,
      "learning_rate": 1.9155991438520273e-05,
      "loss": 0.8984,
      "step": 2290
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4051804542541504,
      "learning_rate": 1.915535252212248e-05,
      "loss": 1.0616,
      "step": 2291
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3330307006835938,
      "learning_rate": 1.9154713605724692e-05,
      "loss": 1.1744,
      "step": 2292
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.467315673828125,
      "learning_rate": 1.9154074689326904e-05,
      "loss": 0.9595,
      "step": 2293
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.740039110183716,
      "learning_rate": 1.9153435772929116e-05,
      "loss": 0.8398,
      "step": 2294
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5022623538970947,
      "learning_rate": 1.9152796856531324e-05,
      "loss": 0.8756,
      "step": 2295
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.344515562057495,
      "learning_rate": 1.9152157940133535e-05,
      "loss": 0.9325,
      "step": 2296
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8914791345596313,
      "learning_rate": 1.9151519023735747e-05,
      "loss": 0.9315,
      "step": 2297
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0024805068969727,
      "learning_rate": 1.915088010733796e-05,
      "loss": 0.9544,
      "step": 2298
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.3841657638549805,
      "learning_rate": 1.9150241190940167e-05,
      "loss": 1.2484,
      "step": 2299
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.5163702964782715,
      "learning_rate": 1.9149602274542378e-05,
      "loss": 1.8367,
      "step": 2300
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.479973554611206,
      "learning_rate": 1.9148963358144586e-05,
      "loss": 1.3683,
      "step": 2301
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6309566497802734,
      "learning_rate": 1.91483244417468e-05,
      "loss": 1.0479,
      "step": 2302
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.645031452178955,
      "learning_rate": 1.914768552534901e-05,
      "loss": 1.5209,
      "step": 2303
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.469973564147949,
      "learning_rate": 1.914704660895122e-05,
      "loss": 1.0707,
      "step": 2304
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.654088020324707,
      "learning_rate": 1.914640769255343e-05,
      "loss": 1.2696,
      "step": 2305
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.782437324523926,
      "learning_rate": 1.914576877615564e-05,
      "loss": 0.9138,
      "step": 2306
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9137320518493652,
      "learning_rate": 1.9145129859757852e-05,
      "loss": 0.8817,
      "step": 2307
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3800835609436035,
      "learning_rate": 1.9144490943360064e-05,
      "loss": 0.9533,
      "step": 2308
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.743837356567383,
      "learning_rate": 1.9143852026962272e-05,
      "loss": 1.6237,
      "step": 2309
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.156867742538452,
      "learning_rate": 1.9143213110564483e-05,
      "loss": 0.8802,
      "step": 2310
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6965816020965576,
      "learning_rate": 1.9142574194166695e-05,
      "loss": 1.0179,
      "step": 2311
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9336599111557007,
      "learning_rate": 1.9141935277768906e-05,
      "loss": 1.0102,
      "step": 2312
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.446564197540283,
      "learning_rate": 1.9141296361371115e-05,
      "loss": 1.1237,
      "step": 2313
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.361670970916748,
      "learning_rate": 1.9140657444973326e-05,
      "loss": 0.9785,
      "step": 2314
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.449694871902466,
      "learning_rate": 1.9140018528575538e-05,
      "loss": 0.925,
      "step": 2315
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8327454328536987,
      "learning_rate": 1.913937961217775e-05,
      "loss": 0.9218,
      "step": 2316
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9274104833602905,
      "learning_rate": 1.9138740695779957e-05,
      "loss": 0.9959,
      "step": 2317
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7076637744903564,
      "learning_rate": 1.913810177938217e-05,
      "loss": 0.9045,
      "step": 2318
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.11834979057312,
      "learning_rate": 1.913746286298438e-05,
      "loss": 0.894,
      "step": 2319
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.662622928619385,
      "learning_rate": 1.9136823946586592e-05,
      "loss": 1.5039,
      "step": 2320
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.215468168258667,
      "learning_rate": 1.91361850301888e-05,
      "loss": 0.9651,
      "step": 2321
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.968608021736145,
      "learning_rate": 1.9135546113791012e-05,
      "loss": 0.8414,
      "step": 2322
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.505113124847412,
      "learning_rate": 1.9134907197393223e-05,
      "loss": 0.9212,
      "step": 2323
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9231414794921875,
      "learning_rate": 1.9134268280995435e-05,
      "loss": 0.7918,
      "step": 2324
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.7432239055633545,
      "learning_rate": 1.9133629364597643e-05,
      "loss": 1.0618,
      "step": 2325
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5163028240203857,
      "learning_rate": 1.9132990448199855e-05,
      "loss": 1.003,
      "step": 2326
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8668512105941772,
      "learning_rate": 1.9132351531802066e-05,
      "loss": 0.8579,
      "step": 2327
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3558857440948486,
      "learning_rate": 1.9131712615404278e-05,
      "loss": 0.8836,
      "step": 2328
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.142598867416382,
      "learning_rate": 1.9131073699006486e-05,
      "loss": 0.9914,
      "step": 2329
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.0432825088500977,
      "learning_rate": 1.9130434782608697e-05,
      "loss": 0.983,
      "step": 2330
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3892927169799805,
      "learning_rate": 1.912979586621091e-05,
      "loss": 0.9482,
      "step": 2331
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9114775657653809,
      "learning_rate": 1.912915694981312e-05,
      "loss": 0.8311,
      "step": 2332
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3560917377471924,
      "learning_rate": 1.912851803341533e-05,
      "loss": 1.0036,
      "step": 2333
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1753907203674316,
      "learning_rate": 1.912787911701754e-05,
      "loss": 1.1171,
      "step": 2334
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9565285444259644,
      "learning_rate": 1.912724020061975e-05,
      "loss": 0.8956,
      "step": 2335
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3609619140625,
      "learning_rate": 1.9126601284221963e-05,
      "loss": 1.0805,
      "step": 2336
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0221571922302246,
      "learning_rate": 1.912596236782417e-05,
      "loss": 0.8854,
      "step": 2337
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.090263843536377,
      "learning_rate": 1.9125323451426383e-05,
      "loss": 1.2608,
      "step": 2338
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.874399662017822,
      "learning_rate": 1.912468453502859e-05,
      "loss": 1.1894,
      "step": 2339
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6615540981292725,
      "learning_rate": 1.9124045618630803e-05,
      "loss": 0.303,
      "step": 2340
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0424697399139404,
      "learning_rate": 1.9123406702233014e-05,
      "loss": 0.964,
      "step": 2341
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.976471185684204,
      "learning_rate": 1.9122767785835226e-05,
      "loss": 0.9442,
      "step": 2342
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.340858221054077,
      "learning_rate": 1.9122128869437434e-05,
      "loss": 0.9735,
      "step": 2343
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.486485242843628,
      "learning_rate": 1.9121489953039645e-05,
      "loss": 0.9841,
      "step": 2344
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.266710042953491,
      "learning_rate": 1.9120851036641857e-05,
      "loss": 0.9014,
      "step": 2345
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.578296184539795,
      "learning_rate": 1.912021212024407e-05,
      "loss": 0.8986,
      "step": 2346
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.112906455993652,
      "learning_rate": 1.9119573203846277e-05,
      "loss": 1.2899,
      "step": 2347
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.177647590637207,
      "learning_rate": 1.9118934287448488e-05,
      "loss": 0.8976,
      "step": 2348
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2579174041748047,
      "learning_rate": 1.91182953710507e-05,
      "loss": 0.5417,
      "step": 2349
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.0670011043548584,
      "learning_rate": 1.911765645465291e-05,
      "loss": 0.9647,
      "step": 2350
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.924497127532959,
      "learning_rate": 1.911701753825512e-05,
      "loss": 0.7965,
      "step": 2351
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.438554286956787,
      "learning_rate": 1.911637862185733e-05,
      "loss": 1.3901,
      "step": 2352
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2079548835754395,
      "learning_rate": 1.9115739705459543e-05,
      "loss": 0.935,
      "step": 2353
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1748976707458496,
      "learning_rate": 1.9115100789061754e-05,
      "loss": 0.9024,
      "step": 2354
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4657303094863892,
      "learning_rate": 1.9114461872663962e-05,
      "loss": 0.2812,
      "step": 2355
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.359225273132324,
      "learning_rate": 1.9113822956266174e-05,
      "loss": 0.8131,
      "step": 2356
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.586536169052124,
      "learning_rate": 1.9113184039868385e-05,
      "loss": 1.2208,
      "step": 2357
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.142293930053711,
      "learning_rate": 1.9112545123470597e-05,
      "loss": 1.1725,
      "step": 2358
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.9894661903381348,
      "learning_rate": 1.9111906207072805e-05,
      "loss": 1.0497,
      "step": 2359
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8457967042922974,
      "learning_rate": 1.9111267290675017e-05,
      "loss": 0.8518,
      "step": 2360
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.364840507507324,
      "learning_rate": 1.9110628374277228e-05,
      "loss": 0.8387,
      "step": 2361
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.427388906478882,
      "learning_rate": 1.910998945787944e-05,
      "loss": 0.9964,
      "step": 2362
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.853506326675415,
      "learning_rate": 1.9109350541481648e-05,
      "loss": 1.0096,
      "step": 2363
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3232592344284058,
      "learning_rate": 1.910871162508386e-05,
      "loss": 0.2425,
      "step": 2364
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.629136562347412,
      "learning_rate": 1.910807270868607e-05,
      "loss": 1.2793,
      "step": 2365
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.528047561645508,
      "learning_rate": 1.9107433792288283e-05,
      "loss": 1.0387,
      "step": 2366
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.464961290359497,
      "learning_rate": 1.910679487589049e-05,
      "loss": 0.8582,
      "step": 2367
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2362470626831055,
      "learning_rate": 1.9106155959492702e-05,
      "loss": 1.0374,
      "step": 2368
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.170599460601807,
      "learning_rate": 1.9105517043094914e-05,
      "loss": 1.2052,
      "step": 2369
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.015655517578125,
      "learning_rate": 1.9104878126697125e-05,
      "loss": 1.1527,
      "step": 2370
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.08414626121521,
      "learning_rate": 1.9104239210299334e-05,
      "loss": 0.948,
      "step": 2371
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.05059552192688,
      "learning_rate": 1.9103600293901545e-05,
      "loss": 0.9295,
      "step": 2372
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9093949794769287,
      "learning_rate": 1.9102961377503753e-05,
      "loss": 0.8448,
      "step": 2373
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.168484926223755,
      "learning_rate": 1.9102322461105968e-05,
      "loss": 0.9269,
      "step": 2374
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1023781299591064,
      "learning_rate": 1.9101683544708176e-05,
      "loss": 0.8854,
      "step": 2375
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.303753614425659,
      "learning_rate": 1.9101044628310388e-05,
      "loss": 1.2073,
      "step": 2376
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6519343852996826,
      "learning_rate": 1.9100405711912596e-05,
      "loss": 0.9072,
      "step": 2377
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0197396278381348,
      "learning_rate": 1.9099766795514808e-05,
      "loss": 0.7847,
      "step": 2378
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2261335849761963,
      "learning_rate": 1.909912787911702e-05,
      "loss": 0.8664,
      "step": 2379
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.047009468078613,
      "learning_rate": 1.909848896271923e-05,
      "loss": 1.4096,
      "step": 2380
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.055981159210205,
      "learning_rate": 1.909785004632144e-05,
      "loss": 0.8827,
      "step": 2381
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.755702018737793,
      "learning_rate": 1.909721112992365e-05,
      "loss": 0.7943,
      "step": 2382
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9645837545394897,
      "learning_rate": 1.9096572213525862e-05,
      "loss": 0.8117,
      "step": 2383
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.63600492477417,
      "learning_rate": 1.9095933297128073e-05,
      "loss": 1.0034,
      "step": 2384
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.144028663635254,
      "learning_rate": 1.909529438073028e-05,
      "loss": 1.9863,
      "step": 2385
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.153836727142334,
      "learning_rate": 1.9094655464332493e-05,
      "loss": 0.9992,
      "step": 2386
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5934730768203735,
      "learning_rate": 1.9094016547934705e-05,
      "loss": 0.2762,
      "step": 2387
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.01961612701416,
      "learning_rate": 1.9093377631536916e-05,
      "loss": 0.8843,
      "step": 2388
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5598862171173096,
      "learning_rate": 1.9092738715139124e-05,
      "loss": 0.8435,
      "step": 2389
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.449244260787964,
      "learning_rate": 1.9092099798741336e-05,
      "loss": 1.2378,
      "step": 2390
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8635812997817993,
      "learning_rate": 1.9091460882343548e-05,
      "loss": 0.3529,
      "step": 2391
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.644488334655762,
      "learning_rate": 1.909082196594576e-05,
      "loss": 1.2636,
      "step": 2392
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.45539927482605,
      "learning_rate": 1.9090183049547967e-05,
      "loss": 0.9232,
      "step": 2393
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.9457197189331055,
      "learning_rate": 1.908954413315018e-05,
      "loss": 0.9884,
      "step": 2394
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8513078689575195,
      "learning_rate": 1.908890521675239e-05,
      "loss": 1.002,
      "step": 2395
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.296405553817749,
      "learning_rate": 1.9088266300354602e-05,
      "loss": 1.0333,
      "step": 2396
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.483705997467041,
      "learning_rate": 1.908762738395681e-05,
      "loss": 1.0052,
      "step": 2397
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.595982789993286,
      "learning_rate": 1.908698846755902e-05,
      "loss": 1.0734,
      "step": 2398
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2427473068237305,
      "learning_rate": 1.9086349551161233e-05,
      "loss": 0.8983,
      "step": 2399
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.281480550765991,
      "learning_rate": 1.9085710634763445e-05,
      "loss": 1.1349,
      "step": 2400
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.264986038208008,
      "learning_rate": 1.9085071718365653e-05,
      "loss": 1.0123,
      "step": 2401
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.991782307624817,
      "learning_rate": 1.9084432801967864e-05,
      "loss": 0.9392,
      "step": 2402
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7206308841705322,
      "learning_rate": 1.9083793885570076e-05,
      "loss": 0.9732,
      "step": 2403
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.518826961517334,
      "learning_rate": 1.9083154969172287e-05,
      "loss": 0.8737,
      "step": 2404
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.291282653808594,
      "learning_rate": 1.9082516052774496e-05,
      "loss": 1.5232,
      "step": 2405
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.238891839981079,
      "learning_rate": 1.9081877136376707e-05,
      "loss": 1.0019,
      "step": 2406
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.391648292541504,
      "learning_rate": 1.9081238219978915e-05,
      "loss": 1.0431,
      "step": 2407
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8432891368865967,
      "learning_rate": 1.908059930358113e-05,
      "loss": 0.9192,
      "step": 2408
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6834864616394043,
      "learning_rate": 1.907996038718334e-05,
      "loss": 1.1311,
      "step": 2409
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1283836364746094,
      "learning_rate": 1.907932147078555e-05,
      "loss": 1.3034,
      "step": 2410
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.567157745361328,
      "learning_rate": 1.9078682554387758e-05,
      "loss": 1.1798,
      "step": 2411
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2557482719421387,
      "learning_rate": 1.9078043637989973e-05,
      "loss": 0.9953,
      "step": 2412
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.36181902885437,
      "learning_rate": 1.907740472159218e-05,
      "loss": 1.011,
      "step": 2413
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.845942974090576,
      "learning_rate": 1.9076765805194393e-05,
      "loss": 1.0316,
      "step": 2414
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.065859794616699,
      "learning_rate": 1.90761268887966e-05,
      "loss": 0.9635,
      "step": 2415
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3016068935394287,
      "learning_rate": 1.9075487972398812e-05,
      "loss": 1.0659,
      "step": 2416
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0615246295928955,
      "learning_rate": 1.9074849056001024e-05,
      "loss": 0.9148,
      "step": 2417
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.897129774093628,
      "learning_rate": 1.9074210139603236e-05,
      "loss": 1.151,
      "step": 2418
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5753345489501953,
      "learning_rate": 1.9073571223205444e-05,
      "loss": 0.9255,
      "step": 2419
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.881417155265808,
      "learning_rate": 1.9072932306807655e-05,
      "loss": 0.8716,
      "step": 2420
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.37461256980896,
      "learning_rate": 1.9072293390409867e-05,
      "loss": 1.1634,
      "step": 2421
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.632211208343506,
      "learning_rate": 1.907165447401208e-05,
      "loss": 0.8788,
      "step": 2422
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.684908866882324,
      "learning_rate": 1.9071015557614287e-05,
      "loss": 0.809,
      "step": 2423
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.583989143371582,
      "learning_rate": 1.9070376641216498e-05,
      "loss": 0.7394,
      "step": 2424
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4282631874084473,
      "learning_rate": 1.906973772481871e-05,
      "loss": 1.2579,
      "step": 2425
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.551638126373291,
      "learning_rate": 1.906909880842092e-05,
      "loss": 0.9438,
      "step": 2426
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.343306303024292,
      "learning_rate": 1.906845989202313e-05,
      "loss": 0.9769,
      "step": 2427
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.395994663238525,
      "learning_rate": 1.906782097562534e-05,
      "loss": 1.3764,
      "step": 2428
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.453439712524414,
      "learning_rate": 1.9067182059227552e-05,
      "loss": 1.3017,
      "step": 2429
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.156642436981201,
      "learning_rate": 1.9066543142829764e-05,
      "loss": 1.6672,
      "step": 2430
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6467803716659546,
      "learning_rate": 1.9065904226431972e-05,
      "loss": 0.8443,
      "step": 2431
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.971972942352295,
      "learning_rate": 1.9065265310034184e-05,
      "loss": 1.0364,
      "step": 2432
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.970836639404297,
      "learning_rate": 1.9064626393636395e-05,
      "loss": 1.0814,
      "step": 2433
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.404585838317871,
      "learning_rate": 1.9063987477238607e-05,
      "loss": 0.898,
      "step": 2434
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.269421339035034,
      "learning_rate": 1.9063348560840815e-05,
      "loss": 0.6708,
      "step": 2435
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7195132970809937,
      "learning_rate": 1.9062709644443026e-05,
      "loss": 0.7727,
      "step": 2436
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.486142158508301,
      "learning_rate": 1.9062070728045238e-05,
      "loss": 1.7747,
      "step": 2437
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9531011581420898,
      "learning_rate": 1.906143181164745e-05,
      "loss": 1.0164,
      "step": 2438
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0558245182037354,
      "learning_rate": 1.9060792895249658e-05,
      "loss": 0.9892,
      "step": 2439
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7208547592163086,
      "learning_rate": 1.906015397885187e-05,
      "loss": 1.0636,
      "step": 2440
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.409665584564209,
      "learning_rate": 1.905951506245408e-05,
      "loss": 0.9687,
      "step": 2441
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.949168086051941,
      "learning_rate": 1.9058876146056292e-05,
      "loss": 1.0086,
      "step": 2442
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.003089427947998,
      "learning_rate": 1.90582372296585e-05,
      "loss": 0.9763,
      "step": 2443
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.48475980758667,
      "learning_rate": 1.9057598313260712e-05,
      "loss": 0.9375,
      "step": 2444
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7646982669830322,
      "learning_rate": 1.905695939686292e-05,
      "loss": 0.9448,
      "step": 2445
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2746329307556152,
      "learning_rate": 1.9056320480465135e-05,
      "loss": 0.9352,
      "step": 2446
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9584380388259888,
      "learning_rate": 1.9055681564067343e-05,
      "loss": 0.969,
      "step": 2447
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.653814315795898,
      "learning_rate": 1.9055042647669555e-05,
      "loss": 1.2444,
      "step": 2448
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.118910312652588,
      "learning_rate": 1.9054403731271763e-05,
      "loss": 0.8088,
      "step": 2449
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3689115047454834,
      "learning_rate": 1.9053764814873975e-05,
      "loss": 0.7714,
      "step": 2450
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.588536739349365,
      "learning_rate": 1.9053125898476186e-05,
      "loss": 1.3873,
      "step": 2451
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.605415105819702,
      "learning_rate": 1.9052486982078398e-05,
      "loss": 0.9958,
      "step": 2452
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.368305206298828,
      "learning_rate": 1.9051848065680606e-05,
      "loss": 1.3311,
      "step": 2453
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9837664365768433,
      "learning_rate": 1.9051209149282817e-05,
      "loss": 0.9551,
      "step": 2454
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8965866565704346,
      "learning_rate": 1.905057023288503e-05,
      "loss": 0.9108,
      "step": 2455
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.08974552154541,
      "learning_rate": 1.904993131648724e-05,
      "loss": 0.9895,
      "step": 2456
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7880806922912598,
      "learning_rate": 1.904929240008945e-05,
      "loss": 0.2981,
      "step": 2457
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0319302082061768,
      "learning_rate": 1.904865348369166e-05,
      "loss": 0.9524,
      "step": 2458
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3887383937835693,
      "learning_rate": 1.9048014567293872e-05,
      "loss": 0.9429,
      "step": 2459
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2296969890594482,
      "learning_rate": 1.9047375650896083e-05,
      "loss": 0.8587,
      "step": 2460
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1120998859405518,
      "learning_rate": 1.904673673449829e-05,
      "loss": 0.8862,
      "step": 2461
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.897220134735107,
      "learning_rate": 1.9046097818100503e-05,
      "loss": 2.0054,
      "step": 2462
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.640252113342285,
      "learning_rate": 1.9045458901702715e-05,
      "loss": 0.9151,
      "step": 2463
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.265585422515869,
      "learning_rate": 1.9044819985304926e-05,
      "loss": 1.339,
      "step": 2464
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.241499423980713,
      "learning_rate": 1.9044181068907134e-05,
      "loss": 0.9506,
      "step": 2465
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.557940483093262,
      "learning_rate": 1.9043542152509346e-05,
      "loss": 1.7752,
      "step": 2466
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.818798303604126,
      "learning_rate": 1.9042903236111557e-05,
      "loss": 0.8711,
      "step": 2467
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3241782188415527,
      "learning_rate": 1.904226431971377e-05,
      "loss": 1.1678,
      "step": 2468
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.485485315322876,
      "learning_rate": 1.9041625403315977e-05,
      "loss": 1.0464,
      "step": 2469
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9815266132354736,
      "learning_rate": 1.904098648691819e-05,
      "loss": 0.9191,
      "step": 2470
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.196873903274536,
      "learning_rate": 1.90403475705204e-05,
      "loss": 0.8845,
      "step": 2471
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.149211883544922,
      "learning_rate": 1.903970865412261e-05,
      "loss": 1.0305,
      "step": 2472
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.222029685974121,
      "learning_rate": 1.903906973772482e-05,
      "loss": 1.2858,
      "step": 2473
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1238644123077393,
      "learning_rate": 1.903843082132703e-05,
      "loss": 0.8635,
      "step": 2474
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.2670140266418457,
      "learning_rate": 1.9037791904929243e-05,
      "loss": 0.5549,
      "step": 2475
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4176456928253174,
      "learning_rate": 1.9037152988531454e-05,
      "loss": 1.0564,
      "step": 2476
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4055349826812744,
      "learning_rate": 1.9036514072133663e-05,
      "loss": 1.2779,
      "step": 2477
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3133533000946045,
      "learning_rate": 1.9035875155735874e-05,
      "loss": 1.1553,
      "step": 2478
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.965451717376709,
      "learning_rate": 1.9035236239338086e-05,
      "loss": 0.906,
      "step": 2479
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.066922187805176,
      "learning_rate": 1.9034597322940297e-05,
      "loss": 0.8458,
      "step": 2480
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.2009336948394775,
      "learning_rate": 1.9033958406542505e-05,
      "loss": 0.9674,
      "step": 2481
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9633591175079346,
      "learning_rate": 1.9033319490144717e-05,
      "loss": 0.3249,
      "step": 2482
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.4374418258667,
      "learning_rate": 1.9032680573746925e-05,
      "loss": 1.5744,
      "step": 2483
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7866899967193604,
      "learning_rate": 1.903204165734914e-05,
      "loss": 0.94,
      "step": 2484
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.867569923400879,
      "learning_rate": 1.9031402740951348e-05,
      "loss": 0.9568,
      "step": 2485
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.267937421798706,
      "learning_rate": 1.903076382455356e-05,
      "loss": 0.8933,
      "step": 2486
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9112589359283447,
      "learning_rate": 1.9030124908155768e-05,
      "loss": 0.968,
      "step": 2487
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4169762134552,
      "learning_rate": 1.902948599175798e-05,
      "loss": 0.8244,
      "step": 2488
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1167726516723633,
      "learning_rate": 1.902884707536019e-05,
      "loss": 0.785,
      "step": 2489
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.079864740371704,
      "learning_rate": 1.9028208158962403e-05,
      "loss": 0.8778,
      "step": 2490
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.338840007781982,
      "learning_rate": 1.902756924256461e-05,
      "loss": 0.8491,
      "step": 2491
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9887371063232422,
      "learning_rate": 1.9026930326166822e-05,
      "loss": 0.8998,
      "step": 2492
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9482367038726807,
      "learning_rate": 1.9026291409769034e-05,
      "loss": 1.0597,
      "step": 2493
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.9658713340759277,
      "learning_rate": 1.9025652493371245e-05,
      "loss": 0.9688,
      "step": 2494
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4758026599884033,
      "learning_rate": 1.9025013576973454e-05,
      "loss": 1.0273,
      "step": 2495
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.199206590652466,
      "learning_rate": 1.9024374660575665e-05,
      "loss": 0.6087,
      "step": 2496
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.510717749595642,
      "learning_rate": 1.9023735744177877e-05,
      "loss": 0.3841,
      "step": 2497
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.733616828918457,
      "learning_rate": 1.9023096827780088e-05,
      "loss": 0.9351,
      "step": 2498
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.313338041305542,
      "learning_rate": 1.9022457911382296e-05,
      "loss": 0.9693,
      "step": 2499
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.09197998046875,
      "learning_rate": 1.9021818994984508e-05,
      "loss": 0.7959,
      "step": 2500
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.761542797088623,
      "learning_rate": 1.902118007858672e-05,
      "loss": 0.7632,
      "step": 2501
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.723492622375488,
      "learning_rate": 1.902054116218893e-05,
      "loss": 1.018,
      "step": 2502
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.697354793548584,
      "learning_rate": 1.901990224579114e-05,
      "loss": 1.3459,
      "step": 2503
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.215374231338501,
      "learning_rate": 1.901926332939335e-05,
      "loss": 1.0138,
      "step": 2504
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.86029314994812,
      "learning_rate": 1.9018624412995562e-05,
      "loss": 1.0332,
      "step": 2505
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.604109287261963,
      "learning_rate": 1.9017985496597774e-05,
      "loss": 1.2143,
      "step": 2506
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9775431156158447,
      "learning_rate": 1.9017346580199982e-05,
      "loss": 0.9268,
      "step": 2507
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.298816680908203,
      "learning_rate": 1.9016707663802193e-05,
      "loss": 1.2883,
      "step": 2508
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.87468683719635,
      "learning_rate": 1.9016068747404405e-05,
      "loss": 0.9155,
      "step": 2509
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4203622341156006,
      "learning_rate": 1.9015429831006617e-05,
      "loss": 0.906,
      "step": 2510
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.436764717102051,
      "learning_rate": 1.9014790914608825e-05,
      "loss": 1.2168,
      "step": 2511
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6624295711517334,
      "learning_rate": 1.9014151998211036e-05,
      "loss": 1.0169,
      "step": 2512
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.289493560791016,
      "learning_rate": 1.9013513081813248e-05,
      "loss": 0.7703,
      "step": 2513
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.48425030708313,
      "learning_rate": 1.901287416541546e-05,
      "loss": 1.2477,
      "step": 2514
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4466488361358643,
      "learning_rate": 1.9012235249017668e-05,
      "loss": 1.1408,
      "step": 2515
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3339643478393555,
      "learning_rate": 1.9011596332619876e-05,
      "loss": 1.17,
      "step": 2516
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9746286869049072,
      "learning_rate": 1.9010957416222087e-05,
      "loss": 1.0487,
      "step": 2517
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9344091415405273,
      "learning_rate": 1.90103184998243e-05,
      "loss": 1.0203,
      "step": 2518
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.378563404083252,
      "learning_rate": 1.900967958342651e-05,
      "loss": 0.9067,
      "step": 2519
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6468713283538818,
      "learning_rate": 1.900904066702872e-05,
      "loss": 0.2442,
      "step": 2520
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.177410840988159,
      "learning_rate": 1.900840175063093e-05,
      "loss": 0.8979,
      "step": 2521
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5033750534057617,
      "learning_rate": 1.900776283423314e-05,
      "loss": 0.9989,
      "step": 2522
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4648361206054688,
      "learning_rate": 1.9007123917835353e-05,
      "loss": 1.0637,
      "step": 2523
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9626662731170654,
      "learning_rate": 1.900648500143756e-05,
      "loss": 0.9762,
      "step": 2524
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.107001781463623,
      "learning_rate": 1.9005846085039773e-05,
      "loss": 0.862,
      "step": 2525
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8054981231689453,
      "learning_rate": 1.9005207168641984e-05,
      "loss": 0.9359,
      "step": 2526
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9495707750320435,
      "learning_rate": 1.9004568252244196e-05,
      "loss": 0.8911,
      "step": 2527
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.279934883117676,
      "learning_rate": 1.9003929335846404e-05,
      "loss": 0.8607,
      "step": 2528
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9633853435516357,
      "learning_rate": 1.9003290419448616e-05,
      "loss": 0.9296,
      "step": 2529
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.981887102127075,
      "learning_rate": 1.9002651503050827e-05,
      "loss": 1.0438,
      "step": 2530
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6823259592056274,
      "learning_rate": 1.900201258665304e-05,
      "loss": 0.861,
      "step": 2531
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1608800888061523,
      "learning_rate": 1.9001373670255247e-05,
      "loss": 0.8658,
      "step": 2532
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.437505006790161,
      "learning_rate": 1.900073475385746e-05,
      "loss": 0.8715,
      "step": 2533
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.124239206314087,
      "learning_rate": 1.900009583745967e-05,
      "loss": 0.9145,
      "step": 2534
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.047616958618164,
      "learning_rate": 1.899945692106188e-05,
      "loss": 0.8064,
      "step": 2535
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9491156339645386,
      "learning_rate": 1.899881800466409e-05,
      "loss": 0.8508,
      "step": 2536
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.052607536315918,
      "learning_rate": 1.89981790882663e-05,
      "loss": 1.0361,
      "step": 2537
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.174769878387451,
      "learning_rate": 1.8997540171868513e-05,
      "loss": 1.2872,
      "step": 2538
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2567076683044434,
      "learning_rate": 1.8996901255470724e-05,
      "loss": 0.8751,
      "step": 2539
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.128190279006958,
      "learning_rate": 1.8996262339072932e-05,
      "loss": 1.02,
      "step": 2540
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8735474348068237,
      "learning_rate": 1.8995623422675144e-05,
      "loss": 0.8902,
      "step": 2541
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.920393705368042,
      "learning_rate": 1.8994984506277356e-05,
      "loss": 1.021,
      "step": 2542
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.325916051864624,
      "learning_rate": 1.8994345589879567e-05,
      "loss": 0.7764,
      "step": 2543
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.090674638748169,
      "learning_rate": 1.8993706673481775e-05,
      "loss": 0.9986,
      "step": 2544
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0027122497558594,
      "learning_rate": 1.8993067757083987e-05,
      "loss": 0.9691,
      "step": 2545
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7467236518859863,
      "learning_rate": 1.89924288406862e-05,
      "loss": 0.9723,
      "step": 2546
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4074723720550537,
      "learning_rate": 1.899178992428841e-05,
      "loss": 0.9212,
      "step": 2547
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.8467867374420166,
      "learning_rate": 1.8991151007890618e-05,
      "loss": 1.2758,
      "step": 2548
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0804479122161865,
      "learning_rate": 1.899051209149283e-05,
      "loss": 0.9938,
      "step": 2549
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9083728790283203,
      "learning_rate": 1.8989873175095038e-05,
      "loss": 1.0978,
      "step": 2550
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7206904888153076,
      "learning_rate": 1.8989234258697253e-05,
      "loss": 0.9258,
      "step": 2551
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.956744194030762,
      "learning_rate": 1.898859534229946e-05,
      "loss": 1.1349,
      "step": 2552
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.2877376079559326,
      "learning_rate": 1.8987956425901672e-05,
      "loss": 0.8727,
      "step": 2553
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.193424701690674,
      "learning_rate": 1.898731750950388e-05,
      "loss": 0.9102,
      "step": 2554
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.244184970855713,
      "learning_rate": 1.8986678593106092e-05,
      "loss": 1.3438,
      "step": 2555
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9376084804534912,
      "learning_rate": 1.8986039676708304e-05,
      "loss": 0.8958,
      "step": 2556
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9475624561309814,
      "learning_rate": 1.8985400760310515e-05,
      "loss": 1.2083,
      "step": 2557
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8810234069824219,
      "learning_rate": 1.8984761843912723e-05,
      "loss": 0.9417,
      "step": 2558
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.6071627140045166,
      "learning_rate": 1.8984122927514935e-05,
      "loss": 0.9055,
      "step": 2559
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4665424823760986,
      "learning_rate": 1.8983484011117146e-05,
      "loss": 1.0528,
      "step": 2560
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7476956844329834,
      "learning_rate": 1.8982845094719358e-05,
      "loss": 0.8316,
      "step": 2561
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4672350883483887,
      "learning_rate": 1.8982206178321566e-05,
      "loss": 0.7685,
      "step": 2562
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6615450382232666,
      "learning_rate": 1.8981567261923778e-05,
      "loss": 1.0195,
      "step": 2563
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.574824094772339,
      "learning_rate": 1.898092834552599e-05,
      "loss": 1.281,
      "step": 2564
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.059417486190796,
      "learning_rate": 1.89802894291282e-05,
      "loss": 1.4217,
      "step": 2565
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.627537965774536,
      "learning_rate": 1.897965051273041e-05,
      "loss": 0.7327,
      "step": 2566
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.262484312057495,
      "learning_rate": 1.897901159633262e-05,
      "loss": 0.8993,
      "step": 2567
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.387150526046753,
      "learning_rate": 1.8978372679934832e-05,
      "loss": 1.4295,
      "step": 2568
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0725269317626953,
      "learning_rate": 1.8977733763537044e-05,
      "loss": 1.1845,
      "step": 2569
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9599958658218384,
      "learning_rate": 1.8977094847139252e-05,
      "loss": 0.8825,
      "step": 2570
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2518362998962402,
      "learning_rate": 1.8976455930741463e-05,
      "loss": 0.9188,
      "step": 2571
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.8355753421783447,
      "learning_rate": 1.8975817014343675e-05,
      "loss": 1.1073,
      "step": 2572
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.1375532150268555,
      "learning_rate": 1.8975178097945886e-05,
      "loss": 1.3649,
      "step": 2573
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3305070400238037,
      "learning_rate": 1.8974539181548095e-05,
      "loss": 1.0514,
      "step": 2574
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2960715293884277,
      "learning_rate": 1.8973900265150306e-05,
      "loss": 0.9028,
      "step": 2575
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4658665657043457,
      "learning_rate": 1.8973261348752518e-05,
      "loss": 1.08,
      "step": 2576
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.267573356628418,
      "learning_rate": 1.897262243235473e-05,
      "loss": 1.0515,
      "step": 2577
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8993782997131348,
      "learning_rate": 1.8971983515956937e-05,
      "loss": 1.1199,
      "step": 2578
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.152618885040283,
      "learning_rate": 1.897134459955915e-05,
      "loss": 0.9784,
      "step": 2579
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.970893144607544,
      "learning_rate": 1.897070568316136e-05,
      "loss": 1.098,
      "step": 2580
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7526257038116455,
      "learning_rate": 1.8970066766763572e-05,
      "loss": 0.992,
      "step": 2581
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.8067431449890137,
      "learning_rate": 1.896942785036578e-05,
      "loss": 0.8281,
      "step": 2582
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2971062660217285,
      "learning_rate": 1.8968788933967992e-05,
      "loss": 0.9752,
      "step": 2583
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.713737487792969,
      "learning_rate": 1.89681500175702e-05,
      "loss": 1.6543,
      "step": 2584
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.114595413208008,
      "learning_rate": 1.8967511101172415e-05,
      "loss": 0.9707,
      "step": 2585
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0460407733917236,
      "learning_rate": 1.8966872184774623e-05,
      "loss": 0.9003,
      "step": 2586
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8006947040557861,
      "learning_rate": 1.8966233268376835e-05,
      "loss": 0.8647,
      "step": 2587
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.952538251876831,
      "learning_rate": 1.8965594351979043e-05,
      "loss": 0.9265,
      "step": 2588
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.304136037826538,
      "learning_rate": 1.8964955435581258e-05,
      "loss": 0.9758,
      "step": 2589
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.220650672912598,
      "learning_rate": 1.8964316519183466e-05,
      "loss": 1.3633,
      "step": 2590
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.974844932556152,
      "learning_rate": 1.8963677602785677e-05,
      "loss": 1.6792,
      "step": 2591
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2293474674224854,
      "learning_rate": 1.8963038686387886e-05,
      "loss": 1.0518,
      "step": 2592
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1564571857452393,
      "learning_rate": 1.8962399769990097e-05,
      "loss": 0.9285,
      "step": 2593
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2684969902038574,
      "learning_rate": 1.896176085359231e-05,
      "loss": 1.0882,
      "step": 2594
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.09765362739563,
      "learning_rate": 1.896112193719452e-05,
      "loss": 0.978,
      "step": 2595
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4286978244781494,
      "learning_rate": 1.896048302079673e-05,
      "loss": 1.0828,
      "step": 2596
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.3981783390045166,
      "learning_rate": 1.895984410439894e-05,
      "loss": 0.8783,
      "step": 2597
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.235564708709717,
      "learning_rate": 1.895920518800115e-05,
      "loss": 0.8904,
      "step": 2598
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.638158321380615,
      "learning_rate": 1.8958566271603363e-05,
      "loss": 0.8038,
      "step": 2599
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5086207389831543,
      "learning_rate": 1.895792735520557e-05,
      "loss": 0.931,
      "step": 2600
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9809000492095947,
      "learning_rate": 1.8957288438807783e-05,
      "loss": 0.9001,
      "step": 2601
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.095992088317871,
      "learning_rate": 1.8956649522409994e-05,
      "loss": 0.924,
      "step": 2602
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7856192588806152,
      "learning_rate": 1.8956010606012206e-05,
      "loss": 0.94,
      "step": 2603
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.107616901397705,
      "learning_rate": 1.8955371689614414e-05,
      "loss": 1.0069,
      "step": 2604
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.95796012878418,
      "learning_rate": 1.8954732773216625e-05,
      "loss": 1.7064,
      "step": 2605
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6948243379592896,
      "learning_rate": 1.8954093856818837e-05,
      "loss": 0.8641,
      "step": 2606
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2166476249694824,
      "learning_rate": 1.895345494042105e-05,
      "loss": 0.5285,
      "step": 2607
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.81399393081665,
      "learning_rate": 1.8952816024023257e-05,
      "loss": 1.1213,
      "step": 2608
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7610111236572266,
      "learning_rate": 1.8952177107625468e-05,
      "loss": 0.8303,
      "step": 2609
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0618622303009033,
      "learning_rate": 1.895153819122768e-05,
      "loss": 0.9056,
      "step": 2610
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2279961109161377,
      "learning_rate": 1.895089927482989e-05,
      "loss": 0.891,
      "step": 2611
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9919012784957886,
      "learning_rate": 1.89502603584321e-05,
      "loss": 0.9561,
      "step": 2612
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3457376956939697,
      "learning_rate": 1.894962144203431e-05,
      "loss": 1.0197,
      "step": 2613
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.228304862976074,
      "learning_rate": 1.8948982525636523e-05,
      "loss": 1.0509,
      "step": 2614
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.875772476196289,
      "learning_rate": 1.8948343609238734e-05,
      "loss": 0.8511,
      "step": 2615
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.424355983734131,
      "learning_rate": 1.8947704692840942e-05,
      "loss": 0.8006,
      "step": 2616
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.201442241668701,
      "learning_rate": 1.8947065776443154e-05,
      "loss": 1.1672,
      "step": 2617
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.255181789398193,
      "learning_rate": 1.8946426860045365e-05,
      "loss": 1.2822,
      "step": 2618
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.280576467514038,
      "learning_rate": 1.8945787943647577e-05,
      "loss": 1.0943,
      "step": 2619
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.172052383422852,
      "learning_rate": 1.8945149027249785e-05,
      "loss": 1.8249,
      "step": 2620
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.233455181121826,
      "learning_rate": 1.8944510110851997e-05,
      "loss": 0.9797,
      "step": 2621
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.967585802078247,
      "learning_rate": 1.8943871194454205e-05,
      "loss": 0.8124,
      "step": 2622
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1234939098358154,
      "learning_rate": 1.894323227805642e-05,
      "loss": 0.8302,
      "step": 2623
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.332113265991211,
      "learning_rate": 1.8942593361658628e-05,
      "loss": 1.0131,
      "step": 2624
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2797017097473145,
      "learning_rate": 1.894195444526084e-05,
      "loss": 0.9693,
      "step": 2625
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1208646297454834,
      "learning_rate": 1.8941315528863048e-05,
      "loss": 0.9649,
      "step": 2626
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7713450193405151,
      "learning_rate": 1.894067661246526e-05,
      "loss": 0.9553,
      "step": 2627
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6830058097839355,
      "learning_rate": 1.894003769606747e-05,
      "loss": 0.8625,
      "step": 2628
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2712461948394775,
      "learning_rate": 1.8939398779669682e-05,
      "loss": 1.0025,
      "step": 2629
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1122024059295654,
      "learning_rate": 1.893875986327189e-05,
      "loss": 0.9055,
      "step": 2630
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.2506818771362305,
      "learning_rate": 1.8938120946874102e-05,
      "loss": 1.1073,
      "step": 2631
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.199613332748413,
      "learning_rate": 1.8937482030476314e-05,
      "loss": 1.0469,
      "step": 2632
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3091471195220947,
      "learning_rate": 1.8936843114078525e-05,
      "loss": 0.8494,
      "step": 2633
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9334261417388916,
      "learning_rate": 1.8936204197680733e-05,
      "loss": 0.8474,
      "step": 2634
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7622554302215576,
      "learning_rate": 1.8935565281282945e-05,
      "loss": 0.8438,
      "step": 2635
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8655893802642822,
      "learning_rate": 1.8934926364885156e-05,
      "loss": 0.9713,
      "step": 2636
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.034660816192627,
      "learning_rate": 1.8934287448487368e-05,
      "loss": 1.726,
      "step": 2637
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.776820182800293,
      "learning_rate": 1.8933648532089576e-05,
      "loss": 0.9491,
      "step": 2638
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6761796474456787,
      "learning_rate": 1.8933009615691788e-05,
      "loss": 0.8503,
      "step": 2639
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3749899864196777,
      "learning_rate": 1.8932370699294e-05,
      "loss": 0.9572,
      "step": 2640
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3216845989227295,
      "learning_rate": 1.893173178289621e-05,
      "loss": 0.708,
      "step": 2641
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.648617744445801,
      "learning_rate": 1.893109286649842e-05,
      "loss": 0.8398,
      "step": 2642
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.235823631286621,
      "learning_rate": 1.893045395010063e-05,
      "loss": 1.3466,
      "step": 2643
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5180721282958984,
      "learning_rate": 1.8929815033702842e-05,
      "loss": 1.0485,
      "step": 2644
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.133052349090576,
      "learning_rate": 1.8929176117305053e-05,
      "loss": 0.8981,
      "step": 2645
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.250612258911133,
      "learning_rate": 1.892853720090726e-05,
      "loss": 0.6723,
      "step": 2646
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.209716796875,
      "learning_rate": 1.8927898284509473e-05,
      "loss": 1.5091,
      "step": 2647
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.939033031463623,
      "learning_rate": 1.8927259368111685e-05,
      "loss": 1.2218,
      "step": 2648
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.321380138397217,
      "learning_rate": 1.8926620451713896e-05,
      "loss": 0.9665,
      "step": 2649
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.47782564163208,
      "learning_rate": 1.8925981535316104e-05,
      "loss": 1.227,
      "step": 2650
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7790621519088745,
      "learning_rate": 1.8925342618918316e-05,
      "loss": 0.9153,
      "step": 2651
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9322017431259155,
      "learning_rate": 1.8924703702520528e-05,
      "loss": 0.3921,
      "step": 2652
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1116816997528076,
      "learning_rate": 1.892406478612274e-05,
      "loss": 0.887,
      "step": 2653
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9837100505828857,
      "learning_rate": 1.8923425869724947e-05,
      "loss": 0.9145,
      "step": 2654
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.463181972503662,
      "learning_rate": 1.892278695332716e-05,
      "loss": 1.3254,
      "step": 2655
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.056731224060059,
      "learning_rate": 1.892214803692937e-05,
      "loss": 1.2019,
      "step": 2656
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0853421688079834,
      "learning_rate": 1.8921509120531582e-05,
      "loss": 0.6668,
      "step": 2657
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.31976318359375,
      "learning_rate": 1.892087020413379e-05,
      "loss": 1.0954,
      "step": 2658
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9170479774475098,
      "learning_rate": 1.8920231287736e-05,
      "loss": 1.0923,
      "step": 2659
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.257427215576172,
      "learning_rate": 1.891959237133821e-05,
      "loss": 1.1349,
      "step": 2660
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8718433380126953,
      "learning_rate": 1.8918953454940425e-05,
      "loss": 0.8671,
      "step": 2661
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.753464937210083,
      "learning_rate": 1.8918314538542633e-05,
      "loss": 0.9388,
      "step": 2662
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1409804821014404,
      "learning_rate": 1.8917675622144844e-05,
      "loss": 0.8742,
      "step": 2663
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3122220039367676,
      "learning_rate": 1.8917036705747053e-05,
      "loss": 0.9942,
      "step": 2664
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.110199213027954,
      "learning_rate": 1.8916397789349264e-05,
      "loss": 0.8787,
      "step": 2665
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6523783206939697,
      "learning_rate": 1.8915758872951476e-05,
      "loss": 1.0852,
      "step": 2666
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.347679615020752,
      "learning_rate": 1.8915119956553687e-05,
      "loss": 0.8998,
      "step": 2667
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.582077503204346,
      "learning_rate": 1.8914481040155895e-05,
      "loss": 1.2912,
      "step": 2668
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8002002239227295,
      "learning_rate": 1.8913842123758107e-05,
      "loss": 0.9465,
      "step": 2669
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.091912031173706,
      "learning_rate": 1.891320320736032e-05,
      "loss": 0.747,
      "step": 2670
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.219106674194336,
      "learning_rate": 1.891256429096253e-05,
      "loss": 1.0837,
      "step": 2671
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7836493253707886,
      "learning_rate": 1.8911925374564738e-05,
      "loss": 0.857,
      "step": 2672
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3639822006225586,
      "learning_rate": 1.891128645816695e-05,
      "loss": 1.0292,
      "step": 2673
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.920072317123413,
      "learning_rate": 1.891064754176916e-05,
      "loss": 0.8528,
      "step": 2674
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8455588817596436,
      "learning_rate": 1.8910008625371373e-05,
      "loss": 1.1998,
      "step": 2675
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7488999366760254,
      "learning_rate": 1.890936970897358e-05,
      "loss": 0.9139,
      "step": 2676
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.053942918777466,
      "learning_rate": 1.8908730792575792e-05,
      "loss": 1.001,
      "step": 2677
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.755919337272644,
      "learning_rate": 1.8908091876178004e-05,
      "loss": 0.9645,
      "step": 2678
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.3811731338500977,
      "learning_rate": 1.8907452959780216e-05,
      "loss": 0.806,
      "step": 2679
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0582516193389893,
      "learning_rate": 1.8906814043382424e-05,
      "loss": 0.3413,
      "step": 2680
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4186205863952637,
      "learning_rate": 1.8906175126984635e-05,
      "loss": 0.9651,
      "step": 2681
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0950520038604736,
      "learning_rate": 1.8905536210586847e-05,
      "loss": 0.9292,
      "step": 2682
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.040485382080078,
      "learning_rate": 1.890489729418906e-05,
      "loss": 1.7166,
      "step": 2683
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.158782958984375,
      "learning_rate": 1.8904258377791267e-05,
      "loss": 1.3502,
      "step": 2684
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.470160484313965,
      "learning_rate": 1.8903619461393478e-05,
      "loss": 0.9817,
      "step": 2685
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1204867362976074,
      "learning_rate": 1.890298054499569e-05,
      "loss": 0.858,
      "step": 2686
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.8508758544921875,
      "learning_rate": 1.89023416285979e-05,
      "loss": 1.4885,
      "step": 2687
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9959497451782227,
      "learning_rate": 1.890170271220011e-05,
      "loss": 0.8251,
      "step": 2688
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2598202228546143,
      "learning_rate": 1.890106379580232e-05,
      "loss": 1.0266,
      "step": 2689
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2755041122436523,
      "learning_rate": 1.8900424879404532e-05,
      "loss": 1.3317,
      "step": 2690
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2950408458709717,
      "learning_rate": 1.8899785963006744e-05,
      "loss": 0.9794,
      "step": 2691
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2768640518188477,
      "learning_rate": 1.8899147046608952e-05,
      "loss": 0.8816,
      "step": 2692
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.04232120513916,
      "learning_rate": 1.8898508130211164e-05,
      "loss": 1.0458,
      "step": 2693
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.2472352981567383,
      "learning_rate": 1.8897869213813372e-05,
      "loss": 0.953,
      "step": 2694
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.228783369064331,
      "learning_rate": 1.8897230297415587e-05,
      "loss": 0.9837,
      "step": 2695
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2822206020355225,
      "learning_rate": 1.8896591381017795e-05,
      "loss": 0.7947,
      "step": 2696
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.768847107887268,
      "learning_rate": 1.8895952464620006e-05,
      "loss": 0.9313,
      "step": 2697
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.773916482925415,
      "learning_rate": 1.8895313548222215e-05,
      "loss": 1.333,
      "step": 2698
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8424572944641113,
      "learning_rate": 1.8894674631824426e-05,
      "loss": 1.1523,
      "step": 2699
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5524590015411377,
      "learning_rate": 1.8894035715426638e-05,
      "loss": 1.3161,
      "step": 2700
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.405174493789673,
      "learning_rate": 1.889339679902885e-05,
      "loss": 1.1471,
      "step": 2701
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.444225549697876,
      "learning_rate": 1.8892757882631057e-05,
      "loss": 1.0228,
      "step": 2702
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.0400390625,
      "learning_rate": 1.889211896623327e-05,
      "loss": 1.2187,
      "step": 2703
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8563152551651,
      "learning_rate": 1.889148004983548e-05,
      "loss": 0.8579,
      "step": 2704
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0918753147125244,
      "learning_rate": 1.8890841133437692e-05,
      "loss": 0.8826,
      "step": 2705
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.283050775527954,
      "learning_rate": 1.88902022170399e-05,
      "loss": 0.9814,
      "step": 2706
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.808218002319336,
      "learning_rate": 1.8889563300642112e-05,
      "loss": 0.8929,
      "step": 2707
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1486265659332275,
      "learning_rate": 1.8888924384244323e-05,
      "loss": 0.9882,
      "step": 2708
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8944387435913086,
      "learning_rate": 1.8888285467846535e-05,
      "loss": 1.1243,
      "step": 2709
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.381124258041382,
      "learning_rate": 1.8887646551448743e-05,
      "loss": 0.975,
      "step": 2710
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.145134925842285,
      "learning_rate": 1.8887007635050955e-05,
      "loss": 1.0046,
      "step": 2711
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2717370986938477,
      "learning_rate": 1.8886368718653166e-05,
      "loss": 0.9315,
      "step": 2712
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2312240600585938,
      "learning_rate": 1.8885729802255378e-05,
      "loss": 0.9935,
      "step": 2713
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3047127723693848,
      "learning_rate": 1.8885090885857586e-05,
      "loss": 0.8837,
      "step": 2714
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6837177276611328,
      "learning_rate": 1.8884451969459797e-05,
      "loss": 0.8036,
      "step": 2715
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.061830997467041,
      "learning_rate": 1.888381305306201e-05,
      "loss": 1.0605,
      "step": 2716
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8622190952301025,
      "learning_rate": 1.888317413666422e-05,
      "loss": 0.8993,
      "step": 2717
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1580650806427,
      "learning_rate": 1.888253522026643e-05,
      "loss": 1.023,
      "step": 2718
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.622425556182861,
      "learning_rate": 1.888189630386864e-05,
      "loss": 1.3958,
      "step": 2719
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.008563041687012,
      "learning_rate": 1.8881257387470852e-05,
      "loss": 0.9499,
      "step": 2720
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.096903085708618,
      "learning_rate": 1.8880618471073063e-05,
      "loss": 0.9825,
      "step": 2721
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.451470375061035,
      "learning_rate": 1.887997955467527e-05,
      "loss": 1.2278,
      "step": 2722
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5685503482818604,
      "learning_rate": 1.8879340638277483e-05,
      "loss": 0.9834,
      "step": 2723
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1228954792022705,
      "learning_rate": 1.8878701721879695e-05,
      "loss": 0.878,
      "step": 2724
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3839802742004395,
      "learning_rate": 1.8878062805481906e-05,
      "loss": 0.8789,
      "step": 2725
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7754294872283936,
      "learning_rate": 1.8877423889084114e-05,
      "loss": 1.1136,
      "step": 2726
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.396484851837158,
      "learning_rate": 1.8876784972686326e-05,
      "loss": 1.0795,
      "step": 2727
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6561238765716553,
      "learning_rate": 1.8876146056288537e-05,
      "loss": 0.8799,
      "step": 2728
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.015659332275391,
      "learning_rate": 1.887550713989075e-05,
      "loss": 1.3934,
      "step": 2729
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0266265869140625,
      "learning_rate": 1.8874868223492957e-05,
      "loss": 0.9504,
      "step": 2730
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7897136211395264,
      "learning_rate": 1.887422930709517e-05,
      "loss": 0.8396,
      "step": 2731
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7846840620040894,
      "learning_rate": 1.8873590390697377e-05,
      "loss": 0.8561,
      "step": 2732
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4446678161621094,
      "learning_rate": 1.887295147429959e-05,
      "loss": 0.8969,
      "step": 2733
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.559652090072632,
      "learning_rate": 1.88723125579018e-05,
      "loss": 1.1586,
      "step": 2734
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.936115026473999,
      "learning_rate": 1.887167364150401e-05,
      "loss": 1.103,
      "step": 2735
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0338103771209717,
      "learning_rate": 1.887103472510622e-05,
      "loss": 0.8834,
      "step": 2736
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.075331926345825,
      "learning_rate": 1.887039580870843e-05,
      "loss": 0.879,
      "step": 2737
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.161404848098755,
      "learning_rate": 1.8869756892310643e-05,
      "loss": 0.9055,
      "step": 2738
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.153888702392578,
      "learning_rate": 1.8869117975912854e-05,
      "loss": 1.0655,
      "step": 2739
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2419464588165283,
      "learning_rate": 1.8868479059515062e-05,
      "loss": 0.8619,
      "step": 2740
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.151702404022217,
      "learning_rate": 1.8867840143117274e-05,
      "loss": 0.8851,
      "step": 2741
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.360954523086548,
      "learning_rate": 1.8867201226719485e-05,
      "loss": 1.122,
      "step": 2742
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.097952365875244,
      "learning_rate": 1.8866562310321697e-05,
      "loss": 0.9576,
      "step": 2743
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.91329288482666,
      "learning_rate": 1.8865923393923905e-05,
      "loss": 0.8415,
      "step": 2744
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7526285648345947,
      "learning_rate": 1.8865284477526117e-05,
      "loss": 0.9613,
      "step": 2745
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0521552562713623,
      "learning_rate": 1.8864645561128328e-05,
      "loss": 0.8689,
      "step": 2746
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.060915946960449,
      "learning_rate": 1.886400664473054e-05,
      "loss": 1.2918,
      "step": 2747
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3284828662872314,
      "learning_rate": 1.8863367728332748e-05,
      "loss": 0.9177,
      "step": 2748
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0388717651367188,
      "learning_rate": 1.886272881193496e-05,
      "loss": 0.9223,
      "step": 2749
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.7103211879730225,
      "learning_rate": 1.886208989553717e-05,
      "loss": 0.9113,
      "step": 2750
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.287144184112549,
      "learning_rate": 1.8861450979139383e-05,
      "loss": 0.853,
      "step": 2751
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6159112453460693,
      "learning_rate": 1.886081206274159e-05,
      "loss": 0.8669,
      "step": 2752
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1376235485076904,
      "learning_rate": 1.8860173146343802e-05,
      "loss": 0.7834,
      "step": 2753
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1141726970672607,
      "learning_rate": 1.8859534229946014e-05,
      "loss": 0.9504,
      "step": 2754
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.29422926902771,
      "learning_rate": 1.8858895313548225e-05,
      "loss": 0.9148,
      "step": 2755
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0313847064971924,
      "learning_rate": 1.8858256397150434e-05,
      "loss": 0.8985,
      "step": 2756
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.246426820755005,
      "learning_rate": 1.8857617480752645e-05,
      "loss": 0.9634,
      "step": 2757
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.004064083099365,
      "learning_rate": 1.8856978564354857e-05,
      "loss": 1.6111,
      "step": 2758
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2120883464813232,
      "learning_rate": 1.8856339647957068e-05,
      "loss": 0.9426,
      "step": 2759
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0894291400909424,
      "learning_rate": 1.8855700731559276e-05,
      "loss": 0.9283,
      "step": 2760
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1865532398223877,
      "learning_rate": 1.8855061815161488e-05,
      "loss": 0.8891,
      "step": 2761
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.380870819091797,
      "learning_rate": 1.88544228987637e-05,
      "loss": 0.9619,
      "step": 2762
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.744016170501709,
      "learning_rate": 1.885378398236591e-05,
      "loss": 0.3264,
      "step": 2763
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5414061546325684,
      "learning_rate": 1.885314506596812e-05,
      "loss": 1.1576,
      "step": 2764
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.5865654945373535,
      "learning_rate": 1.885250614957033e-05,
      "loss": 1.1998,
      "step": 2765
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0595462322235107,
      "learning_rate": 1.8851867233172542e-05,
      "loss": 0.9234,
      "step": 2766
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.048800230026245,
      "learning_rate": 1.8851228316774754e-05,
      "loss": 0.8963,
      "step": 2767
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.276423931121826,
      "learning_rate": 1.8850589400376962e-05,
      "loss": 1.2554,
      "step": 2768
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.084111452102661,
      "learning_rate": 1.8849950483979173e-05,
      "loss": 0.813,
      "step": 2769
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5791895389556885,
      "learning_rate": 1.884931156758138e-05,
      "loss": 1.0558,
      "step": 2770
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5034735202789307,
      "learning_rate": 1.8848672651183597e-05,
      "loss": 0.3959,
      "step": 2771
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9063963890075684,
      "learning_rate": 1.8848033734785805e-05,
      "loss": 1.012,
      "step": 2772
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0352942943573,
      "learning_rate": 1.8847394818388016e-05,
      "loss": 0.9199,
      "step": 2773
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.316969156265259,
      "learning_rate": 1.8846755901990224e-05,
      "loss": 1.3005,
      "step": 2774
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3118488788604736,
      "learning_rate": 1.8846116985592436e-05,
      "loss": 0.8673,
      "step": 2775
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.4501383304595947,
      "learning_rate": 1.8845478069194648e-05,
      "loss": 0.9882,
      "step": 2776
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9069578647613525,
      "learning_rate": 1.884483915279686e-05,
      "loss": 0.8988,
      "step": 2777
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.124939203262329,
      "learning_rate": 1.8844200236399067e-05,
      "loss": 0.8911,
      "step": 2778
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3501389026641846,
      "learning_rate": 1.884356132000128e-05,
      "loss": 1.0107,
      "step": 2779
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.457054853439331,
      "learning_rate": 1.884292240360349e-05,
      "loss": 1.2665,
      "step": 2780
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.158658504486084,
      "learning_rate": 1.8842283487205702e-05,
      "loss": 0.9957,
      "step": 2781
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5768210887908936,
      "learning_rate": 1.884164457080791e-05,
      "loss": 1.027,
      "step": 2782
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.015674591064453,
      "learning_rate": 1.884100565441012e-05,
      "loss": 0.8511,
      "step": 2783
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8904629945755005,
      "learning_rate": 1.8840366738012333e-05,
      "loss": 0.8086,
      "step": 2784
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.218064308166504,
      "learning_rate": 1.8839727821614545e-05,
      "loss": 0.9407,
      "step": 2785
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.285562753677368,
      "learning_rate": 1.8839088905216753e-05,
      "loss": 0.9119,
      "step": 2786
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.518388032913208,
      "learning_rate": 1.8838449988818964e-05,
      "loss": 0.8693,
      "step": 2787
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.408343553543091,
      "learning_rate": 1.8837811072421176e-05,
      "loss": 0.8998,
      "step": 2788
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.005943298339844,
      "learning_rate": 1.8837172156023387e-05,
      "loss": 1.4756,
      "step": 2789
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.997965693473816,
      "learning_rate": 1.8836533239625596e-05,
      "loss": 0.8797,
      "step": 2790
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1642017364501953,
      "learning_rate": 1.8835894323227807e-05,
      "loss": 1.0587,
      "step": 2791
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2025094032287598,
      "learning_rate": 1.883525540683002e-05,
      "loss": 0.862,
      "step": 2792
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0840015411376953,
      "learning_rate": 1.883461649043223e-05,
      "loss": 0.9549,
      "step": 2793
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.418726921081543,
      "learning_rate": 1.883397757403444e-05,
      "loss": 0.7406,
      "step": 2794
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.8275980949401855,
      "learning_rate": 1.883333865763665e-05,
      "loss": 1.6093,
      "step": 2795
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.334591865539551,
      "learning_rate": 1.883269974123886e-05,
      "loss": 0.9906,
      "step": 2796
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.638467788696289,
      "learning_rate": 1.8832060824841073e-05,
      "loss": 1.0983,
      "step": 2797
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.255857229232788,
      "learning_rate": 1.883142190844328e-05,
      "loss": 0.8795,
      "step": 2798
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.433406352996826,
      "learning_rate": 1.8830782992045493e-05,
      "loss": 0.9963,
      "step": 2799
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9299521446228027,
      "learning_rate": 1.8830144075647704e-05,
      "loss": 1.0073,
      "step": 2800
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9663938283920288,
      "learning_rate": 1.8829505159249916e-05,
      "loss": 0.936,
      "step": 2801
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.005422830581665,
      "learning_rate": 1.8828866242852124e-05,
      "loss": 0.9169,
      "step": 2802
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8538905382156372,
      "learning_rate": 1.8828227326454336e-05,
      "loss": 0.9067,
      "step": 2803
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.936367392539978,
      "learning_rate": 1.8827588410056544e-05,
      "loss": 0.9543,
      "step": 2804
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.093832492828369,
      "learning_rate": 1.882694949365876e-05,
      "loss": 1.2374,
      "step": 2805
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.949831485748291,
      "learning_rate": 1.8826310577260967e-05,
      "loss": 1.0861,
      "step": 2806
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9768568277359009,
      "learning_rate": 1.882567166086318e-05,
      "loss": 0.8026,
      "step": 2807
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.184228420257568,
      "learning_rate": 1.8825032744465387e-05,
      "loss": 1.1418,
      "step": 2808
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7546330690383911,
      "learning_rate": 1.8824393828067598e-05,
      "loss": 0.8094,
      "step": 2809
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2259621620178223,
      "learning_rate": 1.882375491166981e-05,
      "loss": 0.9794,
      "step": 2810
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7580024003982544,
      "learning_rate": 1.882311599527202e-05,
      "loss": 0.8391,
      "step": 2811
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0364859104156494,
      "learning_rate": 1.882247707887423e-05,
      "loss": 0.8826,
      "step": 2812
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9403581619262695,
      "learning_rate": 1.882183816247644e-05,
      "loss": 0.962,
      "step": 2813
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.604898452758789,
      "learning_rate": 1.8821199246078652e-05,
      "loss": 0.959,
      "step": 2814
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.120598077774048,
      "learning_rate": 1.8820560329680864e-05,
      "loss": 1.0284,
      "step": 2815
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9152966737747192,
      "learning_rate": 1.8819921413283072e-05,
      "loss": 0.847,
      "step": 2816
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.177194118499756,
      "learning_rate": 1.8819282496885284e-05,
      "loss": 0.8708,
      "step": 2817
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.545839548110962,
      "learning_rate": 1.8818643580487495e-05,
      "loss": 0.8881,
      "step": 2818
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5500833988189697,
      "learning_rate": 1.8818004664089707e-05,
      "loss": 0.5598,
      "step": 2819
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.213244676589966,
      "learning_rate": 1.8817365747691915e-05,
      "loss": 0.8956,
      "step": 2820
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6796982288360596,
      "learning_rate": 1.8816726831294126e-05,
      "loss": 1.1264,
      "step": 2821
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9817050695419312,
      "learning_rate": 1.8816087914896338e-05,
      "loss": 0.7862,
      "step": 2822
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.2720181941986084,
      "learning_rate": 1.881544899849855e-05,
      "loss": 1.068,
      "step": 2823
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.253742218017578,
      "learning_rate": 1.8814810082100758e-05,
      "loss": 1.0954,
      "step": 2824
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0996339321136475,
      "learning_rate": 1.881417116570297e-05,
      "loss": 1.0895,
      "step": 2825
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3848659992218018,
      "learning_rate": 1.881353224930518e-05,
      "loss": 0.9334,
      "step": 2826
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7857601642608643,
      "learning_rate": 1.8812893332907392e-05,
      "loss": 0.8706,
      "step": 2827
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.793375015258789,
      "learning_rate": 1.88122544165096e-05,
      "loss": 0.9049,
      "step": 2828
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7959331274032593,
      "learning_rate": 1.8811615500111812e-05,
      "loss": 0.8988,
      "step": 2829
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1240651607513428,
      "learning_rate": 1.8810976583714024e-05,
      "loss": 0.8319,
      "step": 2830
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1149373054504395,
      "learning_rate": 1.8810337667316235e-05,
      "loss": 0.8823,
      "step": 2831
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.866724729537964,
      "learning_rate": 1.8809698750918443e-05,
      "loss": 0.67,
      "step": 2832
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.621433258056641,
      "learning_rate": 1.8809059834520655e-05,
      "loss": 1.045,
      "step": 2833
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.5124430656433105,
      "learning_rate": 1.8808420918122866e-05,
      "loss": 1.1958,
      "step": 2834
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.477445363998413,
      "learning_rate": 1.8807782001725078e-05,
      "loss": 0.8916,
      "step": 2835
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.548694133758545,
      "learning_rate": 1.8807143085327286e-05,
      "loss": 1.0862,
      "step": 2836
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.093477487564087,
      "learning_rate": 1.8806504168929498e-05,
      "loss": 0.893,
      "step": 2837
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0418195724487305,
      "learning_rate": 1.880586525253171e-05,
      "loss": 0.9444,
      "step": 2838
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0572457313537598,
      "learning_rate": 1.880522633613392e-05,
      "loss": 0.9663,
      "step": 2839
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.0833001136779785,
      "learning_rate": 1.880458741973613e-05,
      "loss": 0.8787,
      "step": 2840
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.572627544403076,
      "learning_rate": 1.880394850333834e-05,
      "loss": 0.9971,
      "step": 2841
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8530596494674683,
      "learning_rate": 1.880330958694055e-05,
      "loss": 0.8139,
      "step": 2842
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.03692626953125,
      "learning_rate": 1.8802670670542764e-05,
      "loss": 0.9403,
      "step": 2843
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4412269592285156,
      "learning_rate": 1.8802031754144972e-05,
      "loss": 1.2332,
      "step": 2844
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2776072025299072,
      "learning_rate": 1.8801392837747183e-05,
      "loss": 1.2641,
      "step": 2845
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0558876991271973,
      "learning_rate": 1.880075392134939e-05,
      "loss": 0.9365,
      "step": 2846
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7133002281188965,
      "learning_rate": 1.8800115004951603e-05,
      "loss": 0.8382,
      "step": 2847
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.574601411819458,
      "learning_rate": 1.8799476088553815e-05,
      "loss": 0.9457,
      "step": 2848
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6767518520355225,
      "learning_rate": 1.8798837172156026e-05,
      "loss": 1.0567,
      "step": 2849
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.2563421726226807,
      "learning_rate": 1.8798198255758234e-05,
      "loss": 1.1807,
      "step": 2850
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1694788932800293,
      "learning_rate": 1.8797559339360446e-05,
      "loss": 0.918,
      "step": 2851
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2603466510772705,
      "learning_rate": 1.8796920422962657e-05,
      "loss": 0.9601,
      "step": 2852
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.746283769607544,
      "learning_rate": 1.879628150656487e-05,
      "loss": 0.7092,
      "step": 2853
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.50864839553833,
      "learning_rate": 1.8795642590167077e-05,
      "loss": 0.8519,
      "step": 2854
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7251827716827393,
      "learning_rate": 1.879500367376929e-05,
      "loss": 0.884,
      "step": 2855
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.365022659301758,
      "learning_rate": 1.87943647573715e-05,
      "loss": 0.881,
      "step": 2856
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0631864070892334,
      "learning_rate": 1.879372584097371e-05,
      "loss": 0.941,
      "step": 2857
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2722482681274414,
      "learning_rate": 1.879308692457592e-05,
      "loss": 0.9945,
      "step": 2858
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8659651279449463,
      "learning_rate": 1.879244800817813e-05,
      "loss": 0.9841,
      "step": 2859
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.67450749874115,
      "learning_rate": 1.8791809091780343e-05,
      "loss": 0.9191,
      "step": 2860
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1108946800231934,
      "learning_rate": 1.8791170175382554e-05,
      "loss": 1.1524,
      "step": 2861
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.028041124343872,
      "learning_rate": 1.8790531258984763e-05,
      "loss": 1.0437,
      "step": 2862
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9737144708633423,
      "learning_rate": 1.8789892342586974e-05,
      "loss": 0.829,
      "step": 2863
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.293100118637085,
      "learning_rate": 1.8789253426189186e-05,
      "loss": 0.8881,
      "step": 2864
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.259281873703003,
      "learning_rate": 1.8788614509791397e-05,
      "loss": 0.9322,
      "step": 2865
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0667810440063477,
      "learning_rate": 1.8787975593393605e-05,
      "loss": 1.0786,
      "step": 2866
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8204309940338135,
      "learning_rate": 1.8787336676995817e-05,
      "loss": 0.9458,
      "step": 2867
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7682209014892578,
      "learning_rate": 1.878669776059803e-05,
      "loss": 0.9276,
      "step": 2868
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.452179431915283,
      "learning_rate": 1.878605884420024e-05,
      "loss": 0.7784,
      "step": 2869
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7115042209625244,
      "learning_rate": 1.8785419927802448e-05,
      "loss": 0.603,
      "step": 2870
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.017474889755249,
      "learning_rate": 1.878478101140466e-05,
      "loss": 0.849,
      "step": 2871
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4536519050598145,
      "learning_rate": 1.878414209500687e-05,
      "loss": 1.1199,
      "step": 2872
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.4918596744537354,
      "learning_rate": 1.8783503178609083e-05,
      "loss": 0.8707,
      "step": 2873
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4131460189819336,
      "learning_rate": 1.878286426221129e-05,
      "loss": 0.2914,
      "step": 2874
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4841504096984863,
      "learning_rate": 1.8782225345813503e-05,
      "loss": 0.8654,
      "step": 2875
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.710389256477356,
      "learning_rate": 1.878158642941571e-05,
      "loss": 0.8859,
      "step": 2876
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.181408643722534,
      "learning_rate": 1.8780947513017926e-05,
      "loss": 0.9937,
      "step": 2877
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9606831073760986,
      "learning_rate": 1.8780308596620134e-05,
      "loss": 0.3421,
      "step": 2878
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5826565027236938,
      "learning_rate": 1.8779669680222345e-05,
      "loss": 0.8782,
      "step": 2879
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.905769109725952,
      "learning_rate": 1.8779030763824554e-05,
      "loss": 0.8969,
      "step": 2880
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8625385761260986,
      "learning_rate": 1.877839184742677e-05,
      "loss": 0.8259,
      "step": 2881
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.491992473602295,
      "learning_rate": 1.8777752931028977e-05,
      "loss": 0.8403,
      "step": 2882
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.322661876678467,
      "learning_rate": 1.8777114014631188e-05,
      "loss": 0.8778,
      "step": 2883
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.944758415222168,
      "learning_rate": 1.8776475098233396e-05,
      "loss": 1.2944,
      "step": 2884
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.145782709121704,
      "learning_rate": 1.8775836181835608e-05,
      "loss": 1.0208,
      "step": 2885
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6926441192626953,
      "learning_rate": 1.877519726543782e-05,
      "loss": 0.9413,
      "step": 2886
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8813846111297607,
      "learning_rate": 1.877455834904003e-05,
      "loss": 0.9418,
      "step": 2887
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.019930362701416,
      "learning_rate": 1.877391943264224e-05,
      "loss": 0.8806,
      "step": 2888
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9532098770141602,
      "learning_rate": 1.877328051624445e-05,
      "loss": 0.968,
      "step": 2889
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5419223308563232,
      "learning_rate": 1.8772641599846662e-05,
      "loss": 1.2268,
      "step": 2890
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.931226134300232,
      "learning_rate": 1.8772002683448874e-05,
      "loss": 0.3339,
      "step": 2891
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.084148645401001,
      "learning_rate": 1.8771363767051082e-05,
      "loss": 0.9594,
      "step": 2892
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7558002471923828,
      "learning_rate": 1.8770724850653293e-05,
      "loss": 0.9287,
      "step": 2893
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2493793964385986,
      "learning_rate": 1.8770085934255505e-05,
      "loss": 0.8433,
      "step": 2894
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0255846977233887,
      "learning_rate": 1.8769447017857717e-05,
      "loss": 0.7127,
      "step": 2895
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0321524143218994,
      "learning_rate": 1.8768808101459925e-05,
      "loss": 0.8162,
      "step": 2896
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5539567470550537,
      "learning_rate": 1.8768169185062136e-05,
      "loss": 0.8633,
      "step": 2897
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.008767127990723,
      "learning_rate": 1.8767530268664348e-05,
      "loss": 1.2858,
      "step": 2898
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8343074321746826,
      "learning_rate": 1.876689135226656e-05,
      "loss": 1.0846,
      "step": 2899
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.80177903175354,
      "learning_rate": 1.8766252435868768e-05,
      "loss": 1.184,
      "step": 2900
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1066489219665527,
      "learning_rate": 1.876561351947098e-05,
      "loss": 0.9698,
      "step": 2901
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.808041572570801,
      "learning_rate": 1.876497460307319e-05,
      "loss": 1.2469,
      "step": 2902
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.461367130279541,
      "learning_rate": 1.87643356866754e-05,
      "loss": 1.0652,
      "step": 2903
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9037742614746094,
      "learning_rate": 1.876369677027761e-05,
      "loss": 1.1034,
      "step": 2904
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6451401710510254,
      "learning_rate": 1.8763057853879822e-05,
      "loss": 1.3185,
      "step": 2905
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1359312534332275,
      "learning_rate": 1.8762418937482033e-05,
      "loss": 0.9629,
      "step": 2906
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9175865650177002,
      "learning_rate": 1.876178002108424e-05,
      "loss": 0.9649,
      "step": 2907
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7072584629058838,
      "learning_rate": 1.8761141104686453e-05,
      "loss": 0.9111,
      "step": 2908
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.202583312988281,
      "learning_rate": 1.876050218828866e-05,
      "loss": 0.9747,
      "step": 2909
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2417819499969482,
      "learning_rate": 1.8759863271890876e-05,
      "loss": 0.9235,
      "step": 2910
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3948127031326294,
      "learning_rate": 1.8759224355493084e-05,
      "loss": 0.3281,
      "step": 2911
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0649797916412354,
      "learning_rate": 1.8758585439095296e-05,
      "loss": 0.9215,
      "step": 2912
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.021291255950928,
      "learning_rate": 1.8757946522697504e-05,
      "loss": 0.6169,
      "step": 2913
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2464675903320312,
      "learning_rate": 1.8757307606299716e-05,
      "loss": 0.9771,
      "step": 2914
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.326329469680786,
      "learning_rate": 1.8756668689901927e-05,
      "loss": 1.1668,
      "step": 2915
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0382657051086426,
      "learning_rate": 1.875602977350414e-05,
      "loss": 0.8901,
      "step": 2916
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.332636594772339,
      "learning_rate": 1.8755390857106347e-05,
      "loss": 0.939,
      "step": 2917
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8923842906951904,
      "learning_rate": 1.875475194070856e-05,
      "loss": 0.9048,
      "step": 2918
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0064032077789307,
      "learning_rate": 1.875411302431077e-05,
      "loss": 1.0352,
      "step": 2919
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3566219806671143,
      "learning_rate": 1.875347410791298e-05,
      "loss": 0.8933,
      "step": 2920
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.496364116668701,
      "learning_rate": 1.875283519151519e-05,
      "loss": 0.9207,
      "step": 2921
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.845947265625,
      "learning_rate": 1.87521962751174e-05,
      "loss": 1.78,
      "step": 2922
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5055192708969116,
      "learning_rate": 1.8751557358719613e-05,
      "loss": 0.2824,
      "step": 2923
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0870087146759033,
      "learning_rate": 1.8750918442321824e-05,
      "loss": 0.9166,
      "step": 2924
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.727475643157959,
      "learning_rate": 1.8750279525924032e-05,
      "loss": 0.7466,
      "step": 2925
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.0190277099609375,
      "learning_rate": 1.8749640609526244e-05,
      "loss": 1.2259,
      "step": 2926
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.319664716720581,
      "learning_rate": 1.8749001693128456e-05,
      "loss": 0.9185,
      "step": 2927
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2731547355651855,
      "learning_rate": 1.8748362776730667e-05,
      "loss": 0.8856,
      "step": 2928
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.382082223892212,
      "learning_rate": 1.8747723860332875e-05,
      "loss": 0.9813,
      "step": 2929
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7538238763809204,
      "learning_rate": 1.8747084943935087e-05,
      "loss": 0.864,
      "step": 2930
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.073563575744629,
      "learning_rate": 1.87464460275373e-05,
      "loss": 1.4634,
      "step": 2931
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0134053230285645,
      "learning_rate": 1.874580711113951e-05,
      "loss": 0.8787,
      "step": 2932
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8425822257995605,
      "learning_rate": 1.8745168194741718e-05,
      "loss": 0.9055,
      "step": 2933
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0290515422821045,
      "learning_rate": 1.874452927834393e-05,
      "loss": 0.9529,
      "step": 2934
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.111288070678711,
      "learning_rate": 1.874389036194614e-05,
      "loss": 1.6077,
      "step": 2935
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.0011420249938965,
      "learning_rate": 1.8743251445548353e-05,
      "loss": 1.973,
      "step": 2936
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6542694568634033,
      "learning_rate": 1.874261252915056e-05,
      "loss": 0.7877,
      "step": 2937
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.446066379547119,
      "learning_rate": 1.8741973612752772e-05,
      "loss": 0.7892,
      "step": 2938
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.826918601989746,
      "learning_rate": 1.8741334696354984e-05,
      "loss": 1.1255,
      "step": 2939
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3661069869995117,
      "learning_rate": 1.8740695779957196e-05,
      "loss": 0.9789,
      "step": 2940
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8668655157089233,
      "learning_rate": 1.8740056863559404e-05,
      "loss": 0.8897,
      "step": 2941
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.152114152908325,
      "learning_rate": 1.8739417947161615e-05,
      "loss": 0.9382,
      "step": 2942
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4817962646484375,
      "learning_rate": 1.8738779030763823e-05,
      "loss": 1.027,
      "step": 2943
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9082074165344238,
      "learning_rate": 1.873814011436604e-05,
      "loss": 0.3209,
      "step": 2944
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8614568710327148,
      "learning_rate": 1.8737501197968246e-05,
      "loss": 0.8648,
      "step": 2945
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.601143836975098,
      "learning_rate": 1.8736862281570458e-05,
      "loss": 1.3009,
      "step": 2946
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8887637853622437,
      "learning_rate": 1.8736223365172666e-05,
      "loss": 0.8554,
      "step": 2947
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.004011392593384,
      "learning_rate": 1.873558444877488e-05,
      "loss": 0.8935,
      "step": 2948
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.326249361038208,
      "learning_rate": 1.873494553237709e-05,
      "loss": 0.9966,
      "step": 2949
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7648309469223022,
      "learning_rate": 1.87343066159793e-05,
      "loss": 0.9087,
      "step": 2950
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6309237480163574,
      "learning_rate": 1.873366769958151e-05,
      "loss": 1.0493,
      "step": 2951
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.711310863494873,
      "learning_rate": 1.873302878318372e-05,
      "loss": 0.7266,
      "step": 2952
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.0472564697265625,
      "learning_rate": 1.8732389866785932e-05,
      "loss": 0.8376,
      "step": 2953
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.511296033859253,
      "learning_rate": 1.8731750950388144e-05,
      "loss": 0.9971,
      "step": 2954
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.556180477142334,
      "learning_rate": 1.8731112033990352e-05,
      "loss": 1.0188,
      "step": 2955
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0603787899017334,
      "learning_rate": 1.8730473117592563e-05,
      "loss": 0.8818,
      "step": 2956
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.801767349243164,
      "learning_rate": 1.8729834201194775e-05,
      "loss": 0.9684,
      "step": 2957
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7307515144348145,
      "learning_rate": 1.8729195284796986e-05,
      "loss": 1.0113,
      "step": 2958
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.008878231048584,
      "learning_rate": 1.8728556368399195e-05,
      "loss": 0.9195,
      "step": 2959
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.301246404647827,
      "learning_rate": 1.8727917452001406e-05,
      "loss": 1.0697,
      "step": 2960
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.389826536178589,
      "learning_rate": 1.8727278535603618e-05,
      "loss": 0.9954,
      "step": 2961
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4544153213500977,
      "learning_rate": 1.872663961920583e-05,
      "loss": 1.0039,
      "step": 2962
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2160205841064453,
      "learning_rate": 1.8726000702808037e-05,
      "loss": 1.0756,
      "step": 2963
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9066624641418457,
      "learning_rate": 1.872536178641025e-05,
      "loss": 0.8385,
      "step": 2964
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3901119232177734,
      "learning_rate": 1.872472287001246e-05,
      "loss": 0.8771,
      "step": 2965
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7653504610061646,
      "learning_rate": 1.8724083953614672e-05,
      "loss": 0.4708,
      "step": 2966
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3065762519836426,
      "learning_rate": 1.872344503721688e-05,
      "loss": 1.0182,
      "step": 2967
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.939436912536621,
      "learning_rate": 1.8722806120819092e-05,
      "loss": 0.8431,
      "step": 2968
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0090465545654297,
      "learning_rate": 1.8722167204421303e-05,
      "loss": 0.9048,
      "step": 2969
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0382704734802246,
      "learning_rate": 1.8721528288023515e-05,
      "loss": 1.044,
      "step": 2970
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1702725887298584,
      "learning_rate": 1.8720889371625723e-05,
      "loss": 1.0051,
      "step": 2971
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9373871088027954,
      "learning_rate": 1.8720250455227935e-05,
      "loss": 0.9676,
      "step": 2972
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9616351127624512,
      "learning_rate": 1.8719611538830146e-05,
      "loss": 0.9036,
      "step": 2973
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.2845346927642822,
      "learning_rate": 1.8718972622432358e-05,
      "loss": 0.9149,
      "step": 2974
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0224742889404297,
      "learning_rate": 1.8718333706034566e-05,
      "loss": 0.8893,
      "step": 2975
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.617356061935425,
      "learning_rate": 1.8717694789636777e-05,
      "loss": 0.8831,
      "step": 2976
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8563077449798584,
      "learning_rate": 1.871705587323899e-05,
      "loss": 0.9233,
      "step": 2977
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7301949262619019,
      "learning_rate": 1.87164169568412e-05,
      "loss": 0.381,
      "step": 2978
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.109931707382202,
      "learning_rate": 1.871577804044341e-05,
      "loss": 0.8762,
      "step": 2979
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.274794340133667,
      "learning_rate": 1.871513912404562e-05,
      "loss": 1.156,
      "step": 2980
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.763292908668518,
      "learning_rate": 1.871450020764783e-05,
      "loss": 1.0102,
      "step": 2981
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.4824976921081543,
      "learning_rate": 1.8713861291250043e-05,
      "loss": 1.1666,
      "step": 2982
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.593067169189453,
      "learning_rate": 1.871322237485225e-05,
      "loss": 0.8133,
      "step": 2983
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8049535751342773,
      "learning_rate": 1.8712583458454463e-05,
      "loss": 0.9508,
      "step": 2984
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.196632385253906,
      "learning_rate": 1.871194454205667e-05,
      "loss": 0.8896,
      "step": 2985
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.322352886199951,
      "learning_rate": 1.8711305625658883e-05,
      "loss": 1.1277,
      "step": 2986
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.106292486190796,
      "learning_rate": 1.8710666709261094e-05,
      "loss": 0.9358,
      "step": 2987
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.021156072616577,
      "learning_rate": 1.8710027792863306e-05,
      "loss": 0.8877,
      "step": 2988
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9068231582641602,
      "learning_rate": 1.8709388876465514e-05,
      "loss": 0.8754,
      "step": 2989
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.445395588874817,
      "learning_rate": 1.8708749960067725e-05,
      "loss": 0.4389,
      "step": 2990
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.463834285736084,
      "learning_rate": 1.8708111043669937e-05,
      "loss": 1.0205,
      "step": 2991
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3570868968963623,
      "learning_rate": 1.870747212727215e-05,
      "loss": 1.252,
      "step": 2992
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.954904079437256,
      "learning_rate": 1.8706833210874357e-05,
      "loss": 1.8085,
      "step": 2993
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.141510009765625,
      "learning_rate": 1.8706194294476568e-05,
      "loss": 0.9138,
      "step": 2994
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1960272789001465,
      "learning_rate": 1.870555537807878e-05,
      "loss": 0.995,
      "step": 2995
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.095510482788086,
      "learning_rate": 1.870491646168099e-05,
      "loss": 0.9539,
      "step": 2996
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7500660419464111,
      "learning_rate": 1.87042775452832e-05,
      "loss": 0.9015,
      "step": 2997
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1376795768737793,
      "learning_rate": 1.870363862888541e-05,
      "loss": 1.0106,
      "step": 2998
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7387574911117554,
      "learning_rate": 1.8702999712487623e-05,
      "loss": 0.9108,
      "step": 2999
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.630407691001892,
      "learning_rate": 1.8702360796089834e-05,
      "loss": 0.893,
      "step": 3000
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9377089738845825,
      "learning_rate": 1.8701721879692042e-05,
      "loss": 0.95,
      "step": 3001
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.96919584274292,
      "learning_rate": 1.8701082963294254e-05,
      "loss": 1.1066,
      "step": 3002
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1347429752349854,
      "learning_rate": 1.8700444046896465e-05,
      "loss": 1.0346,
      "step": 3003
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.272820472717285,
      "learning_rate": 1.8699805130498677e-05,
      "loss": 1.2596,
      "step": 3004
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4921504259109497,
      "learning_rate": 1.8699166214100885e-05,
      "loss": 0.8663,
      "step": 3005
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9489586353302002,
      "learning_rate": 1.8698527297703097e-05,
      "loss": 1.205,
      "step": 3006
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2654147148132324,
      "learning_rate": 1.8697888381305308e-05,
      "loss": 0.9182,
      "step": 3007
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8487422466278076,
      "learning_rate": 1.869724946490752e-05,
      "loss": 0.8327,
      "step": 3008
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4044525623321533,
      "learning_rate": 1.8696610548509728e-05,
      "loss": 0.8978,
      "step": 3009
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.045391798019409,
      "learning_rate": 1.869597163211194e-05,
      "loss": 0.9162,
      "step": 3010
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6472735404968262,
      "learning_rate": 1.869533271571415e-05,
      "loss": 0.8213,
      "step": 3011
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.167555093765259,
      "learning_rate": 1.8694693799316363e-05,
      "loss": 0.811,
      "step": 3012
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.187978267669678,
      "learning_rate": 1.869405488291857e-05,
      "loss": 1.145,
      "step": 3013
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7088561058044434,
      "learning_rate": 1.8693415966520782e-05,
      "loss": 0.928,
      "step": 3014
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3820221424102783,
      "learning_rate": 1.8692777050122994e-05,
      "loss": 0.8246,
      "step": 3015
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.011990070343018,
      "learning_rate": 1.8692138133725205e-05,
      "loss": 1.4577,
      "step": 3016
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.497660517692566,
      "learning_rate": 1.8691499217327414e-05,
      "loss": 0.4404,
      "step": 3017
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.403208017349243,
      "learning_rate": 1.8690860300929625e-05,
      "loss": 0.9942,
      "step": 3018
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.7056527137756348,
      "learning_rate": 1.8690221384531833e-05,
      "loss": 0.9826,
      "step": 3019
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.356358289718628,
      "learning_rate": 1.8689582468134048e-05,
      "loss": 1.1145,
      "step": 3020
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.329005718231201,
      "learning_rate": 1.8688943551736256e-05,
      "loss": 0.9526,
      "step": 3021
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0706710815429688,
      "learning_rate": 1.8688304635338468e-05,
      "loss": 0.966,
      "step": 3022
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1377859115600586,
      "learning_rate": 1.8687665718940676e-05,
      "loss": 0.8775,
      "step": 3023
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8863459825515747,
      "learning_rate": 1.8687026802542888e-05,
      "loss": 0.8659,
      "step": 3024
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.453444004058838,
      "learning_rate": 1.86863878861451e-05,
      "loss": 1.0436,
      "step": 3025
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.242579936981201,
      "learning_rate": 1.868574896974731e-05,
      "loss": 0.9273,
      "step": 3026
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8735891580581665,
      "learning_rate": 1.868511005334952e-05,
      "loss": 0.9979,
      "step": 3027
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.360692024230957,
      "learning_rate": 1.868447113695173e-05,
      "loss": 1.4693,
      "step": 3028
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.43572998046875,
      "learning_rate": 1.8683832220553942e-05,
      "loss": 0.942,
      "step": 3029
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0774967670440674,
      "learning_rate": 1.8683193304156153e-05,
      "loss": 0.9115,
      "step": 3030
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.182703733444214,
      "learning_rate": 1.868255438775836e-05,
      "loss": 0.8723,
      "step": 3031
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6443099975585938,
      "learning_rate": 1.8681915471360573e-05,
      "loss": 0.9041,
      "step": 3032
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.012002468109131,
      "learning_rate": 1.8681276554962785e-05,
      "loss": 0.9364,
      "step": 3033
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.9121813774108887,
      "learning_rate": 1.8680637638564996e-05,
      "loss": 1.231,
      "step": 3034
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.6395530700683594,
      "learning_rate": 1.8679998722167204e-05,
      "loss": 1.2587,
      "step": 3035
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.408542633056641,
      "learning_rate": 1.8679359805769416e-05,
      "loss": 1.0914,
      "step": 3036
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9067542552948,
      "learning_rate": 1.8678720889371628e-05,
      "loss": 0.9341,
      "step": 3037
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1743485927581787,
      "learning_rate": 1.867808197297384e-05,
      "loss": 0.8991,
      "step": 3038
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.735149621963501,
      "learning_rate": 1.8677443056576047e-05,
      "loss": 0.8618,
      "step": 3039
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.437373638153076,
      "learning_rate": 1.867680414017826e-05,
      "loss": 1.0368,
      "step": 3040
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8740919828414917,
      "learning_rate": 1.867616522378047e-05,
      "loss": 0.7963,
      "step": 3041
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8535125255584717,
      "learning_rate": 1.8675526307382682e-05,
      "loss": 0.9043,
      "step": 3042
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0332603454589844,
      "learning_rate": 1.867488739098489e-05,
      "loss": 0.957,
      "step": 3043
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.324193477630615,
      "learning_rate": 1.86742484745871e-05,
      "loss": 1.4983,
      "step": 3044
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9998538494110107,
      "learning_rate": 1.8673609558189313e-05,
      "loss": 0.8779,
      "step": 3045
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9518113136291504,
      "learning_rate": 1.8672970641791525e-05,
      "loss": 0.832,
      "step": 3046
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8231637477874756,
      "learning_rate": 1.8672331725393733e-05,
      "loss": 0.7949,
      "step": 3047
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0453298091888428,
      "learning_rate": 1.8671692808995944e-05,
      "loss": 1.1542,
      "step": 3048
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.441867351531982,
      "learning_rate": 1.8671053892598156e-05,
      "loss": 1.6595,
      "step": 3049
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.376884698867798,
      "learning_rate": 1.8670414976200367e-05,
      "loss": 1.0054,
      "step": 3050
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.962162971496582,
      "learning_rate": 1.8669776059802576e-05,
      "loss": 0.9682,
      "step": 3051
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.034261703491211,
      "learning_rate": 1.8669137143404787e-05,
      "loss": 0.9374,
      "step": 3052
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.273355007171631,
      "learning_rate": 1.8668498227006995e-05,
      "loss": 1.0231,
      "step": 3053
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1587371826171875,
      "learning_rate": 1.866785931060921e-05,
      "loss": 1.1354,
      "step": 3054
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4199063777923584,
      "learning_rate": 1.866722039421142e-05,
      "loss": 1.0095,
      "step": 3055
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6739380359649658,
      "learning_rate": 1.866658147781363e-05,
      "loss": 0.897,
      "step": 3056
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.179447650909424,
      "learning_rate": 1.8665942561415838e-05,
      "loss": 1.1434,
      "step": 3057
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.192523002624512,
      "learning_rate": 1.8665303645018053e-05,
      "loss": 1.2943,
      "step": 3058
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.184136867523193,
      "learning_rate": 1.866466472862026e-05,
      "loss": 1.1655,
      "step": 3059
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.863399505615234,
      "learning_rate": 1.8664025812222473e-05,
      "loss": 0.912,
      "step": 3060
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6351780891418457,
      "learning_rate": 1.866338689582468e-05,
      "loss": 0.9406,
      "step": 3061
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0204660892486572,
      "learning_rate": 1.8662747979426892e-05,
      "loss": 1.1124,
      "step": 3062
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.187896728515625,
      "learning_rate": 1.8662109063029104e-05,
      "loss": 0.9616,
      "step": 3063
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.099271774291992,
      "learning_rate": 1.8661470146631316e-05,
      "loss": 0.9978,
      "step": 3064
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.295907735824585,
      "learning_rate": 1.8660831230233524e-05,
      "loss": 0.9245,
      "step": 3065
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.7904698848724365,
      "learning_rate": 1.8660192313835735e-05,
      "loss": 0.8619,
      "step": 3066
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9069229364395142,
      "learning_rate": 1.8659553397437947e-05,
      "loss": 1.0221,
      "step": 3067
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.2566399574279785,
      "learning_rate": 1.865891448104016e-05,
      "loss": 0.8709,
      "step": 3068
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1371304988861084,
      "learning_rate": 1.8658275564642367e-05,
      "loss": 0.7853,
      "step": 3069
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.453718900680542,
      "learning_rate": 1.8657636648244578e-05,
      "loss": 1.1336,
      "step": 3070
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.28164005279541,
      "learning_rate": 1.865699773184679e-05,
      "loss": 0.9317,
      "step": 3071
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.375873327255249,
      "learning_rate": 1.8656358815449e-05,
      "loss": 0.9659,
      "step": 3072
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.566956639289856,
      "learning_rate": 1.865571989905121e-05,
      "loss": 0.7735,
      "step": 3073
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.106168746948242,
      "learning_rate": 1.865508098265342e-05,
      "loss": 0.9245,
      "step": 3074
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.705724835395813,
      "learning_rate": 1.8654442066255632e-05,
      "loss": 0.919,
      "step": 3075
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.959343910217285,
      "learning_rate": 1.8653803149857844e-05,
      "loss": 1.0424,
      "step": 3076
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1336915493011475,
      "learning_rate": 1.8653164233460052e-05,
      "loss": 0.8961,
      "step": 3077
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2338240146636963,
      "learning_rate": 1.8652525317062264e-05,
      "loss": 0.913,
      "step": 3078
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.506389617919922,
      "learning_rate": 1.8651886400664475e-05,
      "loss": 1.5557,
      "step": 3079
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1127188205718994,
      "learning_rate": 1.8651247484266687e-05,
      "loss": 0.8504,
      "step": 3080
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5032591819763184,
      "learning_rate": 1.8650608567868895e-05,
      "loss": 0.8465,
      "step": 3081
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6809558868408203,
      "learning_rate": 1.8649969651471106e-05,
      "loss": 0.95,
      "step": 3082
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.484898567199707,
      "learning_rate": 1.8649330735073318e-05,
      "loss": 0.9594,
      "step": 3083
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.288557767868042,
      "learning_rate": 1.864869181867553e-05,
      "loss": 0.8978,
      "step": 3084
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9446245431900024,
      "learning_rate": 1.8648052902277738e-05,
      "loss": 0.9415,
      "step": 3085
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5043389797210693,
      "learning_rate": 1.864741398587995e-05,
      "loss": 0.845,
      "step": 3086
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.184736728668213,
      "learning_rate": 1.864677506948216e-05,
      "loss": 1.0339,
      "step": 3087
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9964790344238281,
      "learning_rate": 1.8646136153084372e-05,
      "loss": 0.9978,
      "step": 3088
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2790534496307373,
      "learning_rate": 1.864549723668658e-05,
      "loss": 0.8737,
      "step": 3089
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1929030418395996,
      "learning_rate": 1.8644858320288792e-05,
      "loss": 0.9117,
      "step": 3090
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3427183628082275,
      "learning_rate": 1.8644219403891e-05,
      "loss": 1.0067,
      "step": 3091
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.400193929672241,
      "learning_rate": 1.8643580487493215e-05,
      "loss": 1.0334,
      "step": 3092
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1830387115478516,
      "learning_rate": 1.8642941571095423e-05,
      "loss": 0.8692,
      "step": 3093
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.507704496383667,
      "learning_rate": 1.8642302654697635e-05,
      "loss": 1.2874,
      "step": 3094
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.38785719871521,
      "learning_rate": 1.8641663738299843e-05,
      "loss": 1.0041,
      "step": 3095
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.624049186706543,
      "learning_rate": 1.8641024821902055e-05,
      "loss": 0.8837,
      "step": 3096
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8388357162475586,
      "learning_rate": 1.8640385905504266e-05,
      "loss": 1.0931,
      "step": 3097
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.391326665878296,
      "learning_rate": 1.8639746989106478e-05,
      "loss": 0.5611,
      "step": 3098
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.970073938369751,
      "learning_rate": 1.8639108072708686e-05,
      "loss": 0.8459,
      "step": 3099
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3702118396759033,
      "learning_rate": 1.8638469156310897e-05,
      "loss": 1.0645,
      "step": 3100
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1362435817718506,
      "learning_rate": 1.863783023991311e-05,
      "loss": 1.0324,
      "step": 3101
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2198100090026855,
      "learning_rate": 1.863719132351532e-05,
      "loss": 1.0368,
      "step": 3102
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.416517734527588,
      "learning_rate": 1.863655240711753e-05,
      "loss": 1.9327,
      "step": 3103
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0539722442626953,
      "learning_rate": 1.863591349071974e-05,
      "loss": 0.9266,
      "step": 3104
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1343836784362793,
      "learning_rate": 1.8635274574321952e-05,
      "loss": 0.97,
      "step": 3105
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.299767255783081,
      "learning_rate": 1.8634635657924163e-05,
      "loss": 0.9379,
      "step": 3106
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.087568759918213,
      "learning_rate": 1.863399674152637e-05,
      "loss": 1.2305,
      "step": 3107
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9390041828155518,
      "learning_rate": 1.8633357825128583e-05,
      "loss": 0.8938,
      "step": 3108
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8712090253829956,
      "learning_rate": 1.8632718908730795e-05,
      "loss": 0.8503,
      "step": 3109
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.851959705352783,
      "learning_rate": 1.8632079992333006e-05,
      "loss": 1.0903,
      "step": 3110
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.482043743133545,
      "learning_rate": 1.8631441075935214e-05,
      "loss": 0.7259,
      "step": 3111
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.182940721511841,
      "learning_rate": 1.8630802159537426e-05,
      "loss": 0.8654,
      "step": 3112
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.2951736450195312,
      "learning_rate": 1.8630163243139637e-05,
      "loss": 1.1334,
      "step": 3113
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5765280723571777,
      "learning_rate": 1.862952432674185e-05,
      "loss": 1.224,
      "step": 3114
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3705430030822754,
      "learning_rate": 1.8628885410344057e-05,
      "loss": 0.2839,
      "step": 3115
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.788770914077759,
      "learning_rate": 1.862824649394627e-05,
      "loss": 1.0676,
      "step": 3116
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.724809169769287,
      "learning_rate": 1.862760757754848e-05,
      "loss": 1.2659,
      "step": 3117
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.238180637359619,
      "learning_rate": 1.862696866115069e-05,
      "loss": 1.1716,
      "step": 3118
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.742247104644775,
      "learning_rate": 1.86263297447529e-05,
      "loss": 1.5412,
      "step": 3119
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0706145763397217,
      "learning_rate": 1.862569082835511e-05,
      "loss": 0.8664,
      "step": 3120
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6073217391967773,
      "learning_rate": 1.8625051911957323e-05,
      "loss": 0.9482,
      "step": 3121
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.071000814437866,
      "learning_rate": 1.8624412995559534e-05,
      "loss": 1.0157,
      "step": 3122
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8934367895126343,
      "learning_rate": 1.8623774079161743e-05,
      "loss": 0.9548,
      "step": 3123
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.832393169403076,
      "learning_rate": 1.8623135162763954e-05,
      "loss": 1.062,
      "step": 3124
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.135838270187378,
      "learning_rate": 1.8622496246366166e-05,
      "loss": 0.8984,
      "step": 3125
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.11308217048645,
      "learning_rate": 1.8621857329968377e-05,
      "loss": 0.9877,
      "step": 3126
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.805570363998413,
      "learning_rate": 1.8621218413570585e-05,
      "loss": 0.9878,
      "step": 3127
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.620581030845642,
      "learning_rate": 1.8620579497172797e-05,
      "loss": 0.8966,
      "step": 3128
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.180757999420166,
      "learning_rate": 1.8619940580775005e-05,
      "loss": 0.9643,
      "step": 3129
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.104332447052002,
      "learning_rate": 1.861930166437722e-05,
      "loss": 1.1211,
      "step": 3130
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.374673843383789,
      "learning_rate": 1.8618662747979428e-05,
      "loss": 1.1424,
      "step": 3131
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3793365955352783,
      "learning_rate": 1.861802383158164e-05,
      "loss": 1.0487,
      "step": 3132
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9304274320602417,
      "learning_rate": 1.8617384915183848e-05,
      "loss": 0.9841,
      "step": 3133
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5441527366638184,
      "learning_rate": 1.861674599878606e-05,
      "loss": 0.7701,
      "step": 3134
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6984566450119019,
      "learning_rate": 1.861610708238827e-05,
      "loss": 0.9919,
      "step": 3135
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.513954162597656,
      "learning_rate": 1.8615468165990483e-05,
      "loss": 1.3954,
      "step": 3136
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.2579991817474365,
      "learning_rate": 1.861482924959269e-05,
      "loss": 0.8229,
      "step": 3137
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.68748140335083,
      "learning_rate": 1.8614190333194902e-05,
      "loss": 1.9703,
      "step": 3138
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.133279323577881,
      "learning_rate": 1.8613551416797114e-05,
      "loss": 1.0314,
      "step": 3139
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.284088134765625,
      "learning_rate": 1.8612912500399325e-05,
      "loss": 0.9283,
      "step": 3140
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2432003021240234,
      "learning_rate": 1.8612273584001534e-05,
      "loss": 1.0789,
      "step": 3141
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.110929012298584,
      "learning_rate": 1.8611634667603745e-05,
      "loss": 0.9336,
      "step": 3142
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6532723903656006,
      "learning_rate": 1.8610995751205957e-05,
      "loss": 0.7585,
      "step": 3143
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.573731541633606,
      "learning_rate": 1.8610356834808168e-05,
      "loss": 0.8468,
      "step": 3144
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.089641571044922,
      "learning_rate": 1.8609717918410376e-05,
      "loss": 0.8577,
      "step": 3145
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.639690637588501,
      "learning_rate": 1.8609079002012588e-05,
      "loss": 0.9849,
      "step": 3146
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1685848236083984,
      "learning_rate": 1.86084400856148e-05,
      "loss": 1.1577,
      "step": 3147
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4284722805023193,
      "learning_rate": 1.860780116921701e-05,
      "loss": 0.8476,
      "step": 3148
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.005877733230591,
      "learning_rate": 1.860716225281922e-05,
      "loss": 0.8825,
      "step": 3149
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4055681228637695,
      "learning_rate": 1.860652333642143e-05,
      "loss": 1.1052,
      "step": 3150
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.249534845352173,
      "learning_rate": 1.8605884420023642e-05,
      "loss": 0.8975,
      "step": 3151
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3901050090789795,
      "learning_rate": 1.8605245503625854e-05,
      "loss": 1.3311,
      "step": 3152
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0972609519958496,
      "learning_rate": 1.8604606587228062e-05,
      "loss": 1.3455,
      "step": 3153
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3668456077575684,
      "learning_rate": 1.8603967670830273e-05,
      "loss": 0.9999,
      "step": 3154
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0836212635040283,
      "learning_rate": 1.8603328754432485e-05,
      "loss": 0.8071,
      "step": 3155
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.837707757949829,
      "learning_rate": 1.8602689838034697e-05,
      "loss": 0.8495,
      "step": 3156
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.525103807449341,
      "learning_rate": 1.8602050921636905e-05,
      "loss": 0.7567,
      "step": 3157
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.223778486251831,
      "learning_rate": 1.8601412005239116e-05,
      "loss": 0.9875,
      "step": 3158
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7244677543640137,
      "learning_rate": 1.8600773088841328e-05,
      "loss": 0.8873,
      "step": 3159
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.100513458251953,
      "learning_rate": 1.860013417244354e-05,
      "loss": 1.0162,
      "step": 3160
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.173680305480957,
      "learning_rate": 1.8599495256045748e-05,
      "loss": 1.0243,
      "step": 3161
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.652312278747559,
      "learning_rate": 1.859885633964796e-05,
      "loss": 1.1099,
      "step": 3162
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6298564672470093,
      "learning_rate": 1.8598217423250167e-05,
      "loss": 0.9072,
      "step": 3163
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.893800973892212,
      "learning_rate": 1.8597578506852382e-05,
      "loss": 0.8152,
      "step": 3164
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4401211738586426,
      "learning_rate": 1.859693959045459e-05,
      "loss": 0.2992,
      "step": 3165
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0596354007720947,
      "learning_rate": 1.8596300674056802e-05,
      "loss": 0.8815,
      "step": 3166
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.552586555480957,
      "learning_rate": 1.859566175765901e-05,
      "loss": 0.8925,
      "step": 3167
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.051283359527588,
      "learning_rate": 1.859502284126122e-05,
      "loss": 0.8629,
      "step": 3168
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.480764865875244,
      "learning_rate": 1.8594383924863433e-05,
      "loss": 1.2107,
      "step": 3169
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1964807510375977,
      "learning_rate": 1.8593745008465645e-05,
      "loss": 0.9765,
      "step": 3170
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.212545394897461,
      "learning_rate": 1.8593106092067853e-05,
      "loss": 0.823,
      "step": 3171
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0071210861206055,
      "learning_rate": 1.8592467175670064e-05,
      "loss": 0.8744,
      "step": 3172
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8129382133483887,
      "learning_rate": 1.8591828259272276e-05,
      "loss": 0.7447,
      "step": 3173
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.783261775970459,
      "learning_rate": 1.8591189342874487e-05,
      "loss": 1.51,
      "step": 3174
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.388198137283325,
      "learning_rate": 1.8590550426476696e-05,
      "loss": 1.0179,
      "step": 3175
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6075607538223267,
      "learning_rate": 1.8589911510078907e-05,
      "loss": 0.3414,
      "step": 3176
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6156249046325684,
      "learning_rate": 1.858927259368112e-05,
      "loss": 0.8764,
      "step": 3177
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3177428245544434,
      "learning_rate": 1.858863367728333e-05,
      "loss": 1.0118,
      "step": 3178
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4427645206451416,
      "learning_rate": 1.858799476088554e-05,
      "loss": 0.2874,
      "step": 3179
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9901624917984009,
      "learning_rate": 1.858735584448775e-05,
      "loss": 0.816,
      "step": 3180
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.079345703125,
      "learning_rate": 1.858671692808996e-05,
      "loss": 0.7011,
      "step": 3181
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.696869134902954,
      "learning_rate": 1.8586078011692173e-05,
      "loss": 0.5626,
      "step": 3182
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.266602039337158,
      "learning_rate": 1.858543909529438e-05,
      "loss": 1.2122,
      "step": 3183
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5273122787475586,
      "learning_rate": 1.8584800178896593e-05,
      "loss": 0.8922,
      "step": 3184
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0798723697662354,
      "learning_rate": 1.8584161262498804e-05,
      "loss": 0.9474,
      "step": 3185
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2509195804595947,
      "learning_rate": 1.8583522346101016e-05,
      "loss": 0.9713,
      "step": 3186
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.5884428024292,
      "learning_rate": 1.8582883429703224e-05,
      "loss": 1.1005,
      "step": 3187
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.430293083190918,
      "learning_rate": 1.8582244513305436e-05,
      "loss": 0.8965,
      "step": 3188
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5453176498413086,
      "learning_rate": 1.8581605596907647e-05,
      "loss": 0.9375,
      "step": 3189
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1603176593780518,
      "learning_rate": 1.858096668050986e-05,
      "loss": 0.9745,
      "step": 3190
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.091418981552124,
      "learning_rate": 1.8580327764112067e-05,
      "loss": 0.8557,
      "step": 3191
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9988703727722168,
      "learning_rate": 1.857968884771428e-05,
      "loss": 0.9744,
      "step": 3192
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5392465591430664,
      "learning_rate": 1.857904993131649e-05,
      "loss": 0.9298,
      "step": 3193
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.707406044006348,
      "learning_rate": 1.85784110149187e-05,
      "loss": 1.0879,
      "step": 3194
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.160834312438965,
      "learning_rate": 1.857777209852091e-05,
      "loss": 0.8223,
      "step": 3195
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9429653882980347,
      "learning_rate": 1.857713318212312e-05,
      "loss": 0.879,
      "step": 3196
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0911076068878174,
      "learning_rate": 1.8576494265725333e-05,
      "loss": 1.0758,
      "step": 3197
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5581870079040527,
      "learning_rate": 1.8575855349327544e-05,
      "loss": 0.9113,
      "step": 3198
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.934255599975586,
      "learning_rate": 1.8575216432929752e-05,
      "loss": 1.0529,
      "step": 3199
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.598219394683838,
      "learning_rate": 1.8574577516531964e-05,
      "loss": 0.8563,
      "step": 3200
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2319111824035645,
      "learning_rate": 1.8573938600134172e-05,
      "loss": 0.9925,
      "step": 3201
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.7600204944610596,
      "learning_rate": 1.8573299683736387e-05,
      "loss": 0.7986,
      "step": 3202
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1781890392303467,
      "learning_rate": 1.8572660767338595e-05,
      "loss": 1.0246,
      "step": 3203
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5670363903045654,
      "learning_rate": 1.8572021850940807e-05,
      "loss": 1.0732,
      "step": 3204
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8886507749557495,
      "learning_rate": 1.8571382934543015e-05,
      "loss": 0.8667,
      "step": 3205
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.704421043395996,
      "learning_rate": 1.8570744018145226e-05,
      "loss": 1.2252,
      "step": 3206
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3817875385284424,
      "learning_rate": 1.8570105101747438e-05,
      "loss": 0.9146,
      "step": 3207
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.400550603866577,
      "learning_rate": 1.856946618534965e-05,
      "loss": 0.9528,
      "step": 3208
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1680755615234375,
      "learning_rate": 1.8568827268951858e-05,
      "loss": 0.8813,
      "step": 3209
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.59212327003479,
      "learning_rate": 1.856818835255407e-05,
      "loss": 0.8412,
      "step": 3210
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.814990758895874,
      "learning_rate": 1.856754943615628e-05,
      "loss": 1.0683,
      "step": 3211
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.31132173538208,
      "learning_rate": 1.8566910519758492e-05,
      "loss": 0.8103,
      "step": 3212
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.638480186462402,
      "learning_rate": 1.85662716033607e-05,
      "loss": 1.4653,
      "step": 3213
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9772264957427979,
      "learning_rate": 1.8565632686962912e-05,
      "loss": 0.9163,
      "step": 3214
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4663524627685547,
      "learning_rate": 1.8564993770565124e-05,
      "loss": 1.034,
      "step": 3215
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2477877140045166,
      "learning_rate": 1.8564354854167335e-05,
      "loss": 0.9469,
      "step": 3216
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8438465595245361,
      "learning_rate": 1.8563715937769543e-05,
      "loss": 0.9288,
      "step": 3217
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.3037333488464355,
      "learning_rate": 1.8563077021371755e-05,
      "loss": 1.4662,
      "step": 3218
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.8748562335968018,
      "learning_rate": 1.8562438104973966e-05,
      "loss": 0.9787,
      "step": 3219
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.642373561859131,
      "learning_rate": 1.8561799188576178e-05,
      "loss": 0.955,
      "step": 3220
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.594945907592773,
      "learning_rate": 1.8561160272178386e-05,
      "loss": 1.3622,
      "step": 3221
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.786722183227539,
      "learning_rate": 1.8560521355780598e-05,
      "loss": 1.0519,
      "step": 3222
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8846462965011597,
      "learning_rate": 1.855988243938281e-05,
      "loss": 1.0245,
      "step": 3223
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5722355842590332,
      "learning_rate": 1.855924352298502e-05,
      "loss": 0.3577,
      "step": 3224
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.940067291259766,
      "learning_rate": 1.855860460658723e-05,
      "loss": 1.8827,
      "step": 3225
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.10619854927063,
      "learning_rate": 1.855796569018944e-05,
      "loss": 0.8413,
      "step": 3226
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2279021739959717,
      "learning_rate": 1.8557326773791652e-05,
      "loss": 0.8443,
      "step": 3227
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8112789392471313,
      "learning_rate": 1.8556687857393864e-05,
      "loss": 0.9474,
      "step": 3228
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.266406297683716,
      "learning_rate": 1.8556048940996072e-05,
      "loss": 1.0941,
      "step": 3229
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3033456802368164,
      "learning_rate": 1.8555410024598283e-05,
      "loss": 0.9656,
      "step": 3230
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.3815789222717285,
      "learning_rate": 1.8554771108200495e-05,
      "loss": 1.2941,
      "step": 3231
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4782423973083496,
      "learning_rate": 1.8554132191802706e-05,
      "loss": 0.9595,
      "step": 3232
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8456140756607056,
      "learning_rate": 1.8553493275404915e-05,
      "loss": 0.9318,
      "step": 3233
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1906142234802246,
      "learning_rate": 1.8552854359007126e-05,
      "loss": 1.0885,
      "step": 3234
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3447282314300537,
      "learning_rate": 1.8552215442609338e-05,
      "loss": 0.8688,
      "step": 3235
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0819218158721924,
      "learning_rate": 1.855157652621155e-05,
      "loss": 1.1543,
      "step": 3236
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0301785469055176,
      "learning_rate": 1.8550937609813757e-05,
      "loss": 0.913,
      "step": 3237
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.130214214324951,
      "learning_rate": 1.855029869341597e-05,
      "loss": 1.3357,
      "step": 3238
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1387453079223633,
      "learning_rate": 1.8549659777018177e-05,
      "loss": 0.9301,
      "step": 3239
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.436265707015991,
      "learning_rate": 1.8549020860620392e-05,
      "loss": 1.2249,
      "step": 3240
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3619511127471924,
      "learning_rate": 1.85483819442226e-05,
      "loss": 0.9954,
      "step": 3241
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1979780197143555,
      "learning_rate": 1.854774302782481e-05,
      "loss": 0.8387,
      "step": 3242
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.708841800689697,
      "learning_rate": 1.854710411142702e-05,
      "loss": 1.2215,
      "step": 3243
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9158413410186768,
      "learning_rate": 1.854646519502923e-05,
      "loss": 0.4658,
      "step": 3244
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.631286144256592,
      "learning_rate": 1.8545826278631443e-05,
      "loss": 1.211,
      "step": 3245
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2130424976348877,
      "learning_rate": 1.8545187362233654e-05,
      "loss": 0.9752,
      "step": 3246
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.103140354156494,
      "learning_rate": 1.8544548445835863e-05,
      "loss": 0.9359,
      "step": 3247
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8069639205932617,
      "learning_rate": 1.8543909529438074e-05,
      "loss": 0.8472,
      "step": 3248
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0687363147735596,
      "learning_rate": 1.8543270613040286e-05,
      "loss": 0.8865,
      "step": 3249
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.82816219329834,
      "learning_rate": 1.8542631696642497e-05,
      "loss": 0.9174,
      "step": 3250
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.54250431060791,
      "learning_rate": 1.8541992780244705e-05,
      "loss": 1.0059,
      "step": 3251
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.068869113922119,
      "learning_rate": 1.8541353863846917e-05,
      "loss": 0.8952,
      "step": 3252
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7470040321350098,
      "learning_rate": 1.854071494744913e-05,
      "loss": 0.8667,
      "step": 3253
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.348299980163574,
      "learning_rate": 1.854007603105134e-05,
      "loss": 0.8583,
      "step": 3254
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2069671154022217,
      "learning_rate": 1.8539437114653548e-05,
      "loss": 1.0148,
      "step": 3255
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9411250352859497,
      "learning_rate": 1.853879819825576e-05,
      "loss": 0.854,
      "step": 3256
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8537451028823853,
      "learning_rate": 1.853815928185797e-05,
      "loss": 0.7925,
      "step": 3257
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.751725196838379,
      "learning_rate": 1.8537520365460183e-05,
      "loss": 1.4128,
      "step": 3258
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.705793857574463,
      "learning_rate": 1.853688144906239e-05,
      "loss": 0.8731,
      "step": 3259
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.321829319000244,
      "learning_rate": 1.8536242532664603e-05,
      "loss": 0.9909,
      "step": 3260
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.4796082973480225,
      "learning_rate": 1.8535603616266814e-05,
      "loss": 0.8104,
      "step": 3261
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.110236883163452,
      "learning_rate": 1.8534964699869026e-05,
      "loss": 0.8392,
      "step": 3262
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0320699214935303,
      "learning_rate": 1.8534325783471234e-05,
      "loss": 0.8705,
      "step": 3263
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.080838203430176,
      "learning_rate": 1.8533686867073445e-05,
      "loss": 0.9435,
      "step": 3264
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.437544345855713,
      "learning_rate": 1.8533047950675657e-05,
      "loss": 1.0317,
      "step": 3265
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.92244291305542,
      "learning_rate": 1.853240903427787e-05,
      "loss": 1.3521,
      "step": 3266
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.864655494689941,
      "learning_rate": 1.8531770117880077e-05,
      "loss": 1.3438,
      "step": 3267
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.220135688781738,
      "learning_rate": 1.8531131201482288e-05,
      "loss": 1.0876,
      "step": 3268
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.6775732040405273,
      "learning_rate": 1.85304922850845e-05,
      "loss": 1.0342,
      "step": 3269
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.180143356323242,
      "learning_rate": 1.852985336868671e-05,
      "loss": 0.835,
      "step": 3270
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.329256057739258,
      "learning_rate": 1.852921445228892e-05,
      "loss": 0.7889,
      "step": 3271
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1507256031036377,
      "learning_rate": 1.852857553589113e-05,
      "loss": 0.8565,
      "step": 3272
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4472315311431885,
      "learning_rate": 1.852793661949334e-05,
      "loss": 1.0975,
      "step": 3273
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8676406145095825,
      "learning_rate": 1.8527297703095554e-05,
      "loss": 0.7317,
      "step": 3274
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.16942024230957,
      "learning_rate": 1.8526658786697762e-05,
      "loss": 1.1978,
      "step": 3275
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.626316070556641,
      "learning_rate": 1.8526019870299974e-05,
      "loss": 1.2065,
      "step": 3276
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.076629400253296,
      "learning_rate": 1.8525380953902182e-05,
      "loss": 1.3066,
      "step": 3277
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8132879734039307,
      "learning_rate": 1.8524742037504393e-05,
      "loss": 0.9531,
      "step": 3278
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.29512882232666,
      "learning_rate": 1.8524103121106605e-05,
      "loss": 0.9624,
      "step": 3279
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.132913589477539,
      "learning_rate": 1.8523464204708817e-05,
      "loss": 0.9236,
      "step": 3280
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5456807613372803,
      "learning_rate": 1.8522825288311025e-05,
      "loss": 0.9704,
      "step": 3281
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9905903339385986,
      "learning_rate": 1.8522186371913236e-05,
      "loss": 0.8552,
      "step": 3282
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1628246307373047,
      "learning_rate": 1.8521547455515448e-05,
      "loss": 0.9723,
      "step": 3283
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.311760902404785,
      "learning_rate": 1.852090853911766e-05,
      "loss": 1.0935,
      "step": 3284
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8462491035461426,
      "learning_rate": 1.8520269622719868e-05,
      "loss": 1.082,
      "step": 3285
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.948768138885498,
      "learning_rate": 1.851963070632208e-05,
      "loss": 0.8231,
      "step": 3286
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.116199254989624,
      "learning_rate": 1.851899178992429e-05,
      "loss": 0.9864,
      "step": 3287
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4916210174560547,
      "learning_rate": 1.85183528735265e-05,
      "loss": 0.9117,
      "step": 3288
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2343173027038574,
      "learning_rate": 1.851771395712871e-05,
      "loss": 0.9199,
      "step": 3289
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9881515502929688,
      "learning_rate": 1.8517075040730922e-05,
      "loss": 0.9989,
      "step": 3290
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.212899923324585,
      "learning_rate": 1.8516436124333133e-05,
      "loss": 0.9875,
      "step": 3291
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.249922513961792,
      "learning_rate": 1.851579720793534e-05,
      "loss": 0.8906,
      "step": 3292
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.81699275970459,
      "learning_rate": 1.8515158291537553e-05,
      "loss": 1.4881,
      "step": 3293
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1889498233795166,
      "learning_rate": 1.8514519375139765e-05,
      "loss": 1.0193,
      "step": 3294
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3906521797180176,
      "learning_rate": 1.8513880458741976e-05,
      "loss": 0.9847,
      "step": 3295
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.3189568519592285,
      "learning_rate": 1.8513241542344184e-05,
      "loss": 0.9761,
      "step": 3296
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.934360027313232,
      "learning_rate": 1.8512602625946396e-05,
      "loss": 1.1037,
      "step": 3297
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2700889110565186,
      "learning_rate": 1.8511963709548607e-05,
      "loss": 0.8841,
      "step": 3298
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.736238956451416,
      "learning_rate": 1.851132479315082e-05,
      "loss": 1.098,
      "step": 3299
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9376546144485474,
      "learning_rate": 1.8510685876753027e-05,
      "loss": 0.8748,
      "step": 3300
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.224464416503906,
      "learning_rate": 1.851004696035524e-05,
      "loss": 1.4478,
      "step": 3301
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9275422096252441,
      "learning_rate": 1.850940804395745e-05,
      "loss": 0.887,
      "step": 3302
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3263096809387207,
      "learning_rate": 1.8508769127559662e-05,
      "loss": 1.0328,
      "step": 3303
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8701093196868896,
      "learning_rate": 1.850813021116187e-05,
      "loss": 0.828,
      "step": 3304
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.100017547607422,
      "learning_rate": 1.850749129476408e-05,
      "loss": 1.4552,
      "step": 3305
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8080090284347534,
      "learning_rate": 1.850685237836629e-05,
      "loss": 0.3483,
      "step": 3306
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.679697036743164,
      "learning_rate": 1.8506213461968505e-05,
      "loss": 0.9648,
      "step": 3307
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.030341386795044,
      "learning_rate": 1.8505574545570713e-05,
      "loss": 0.8253,
      "step": 3308
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.280055046081543,
      "learning_rate": 1.8504935629172924e-05,
      "loss": 0.9548,
      "step": 3309
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5982842445373535,
      "learning_rate": 1.8504296712775133e-05,
      "loss": 1.6369,
      "step": 3310
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.001849174499512,
      "learning_rate": 1.8503657796377344e-05,
      "loss": 1.0642,
      "step": 3311
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.523384094238281,
      "learning_rate": 1.8503018879979556e-05,
      "loss": 0.9236,
      "step": 3312
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1761205196380615,
      "learning_rate": 1.8502379963581767e-05,
      "loss": 0.9647,
      "step": 3313
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8347917795181274,
      "learning_rate": 1.8501741047183975e-05,
      "loss": 1.0218,
      "step": 3314
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1177117824554443,
      "learning_rate": 1.8501102130786187e-05,
      "loss": 0.9788,
      "step": 3315
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.998793601989746,
      "learning_rate": 1.85004632143884e-05,
      "loss": 1.4077,
      "step": 3316
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.315750122070312,
      "learning_rate": 1.849982429799061e-05,
      "loss": 1.8543,
      "step": 3317
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5079197883605957,
      "learning_rate": 1.8499185381592818e-05,
      "loss": 0.9271,
      "step": 3318
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.548269271850586,
      "learning_rate": 1.849854646519503e-05,
      "loss": 0.8241,
      "step": 3319
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.015165090560913,
      "learning_rate": 1.849790754879724e-05,
      "loss": 0.8843,
      "step": 3320
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.8197338581085205,
      "learning_rate": 1.8497268632399453e-05,
      "loss": 1.1753,
      "step": 3321
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1538753509521484,
      "learning_rate": 1.849662971600166e-05,
      "loss": 0.9762,
      "step": 3322
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.750715732574463,
      "learning_rate": 1.8495990799603872e-05,
      "loss": 1.1854,
      "step": 3323
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.248043060302734,
      "learning_rate": 1.8495351883206084e-05,
      "loss": 1.2646,
      "step": 3324
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1289639472961426,
      "learning_rate": 1.8494712966808296e-05,
      "loss": 0.8736,
      "step": 3325
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.900242805480957,
      "learning_rate": 1.8494074050410504e-05,
      "loss": 0.9741,
      "step": 3326
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1552188396453857,
      "learning_rate": 1.8493435134012715e-05,
      "loss": 0.9138,
      "step": 3327
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6044784784317017,
      "learning_rate": 1.8492796217614927e-05,
      "loss": 0.4166,
      "step": 3328
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.764051079750061,
      "learning_rate": 1.849215730121714e-05,
      "loss": 0.8366,
      "step": 3329
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.2212653160095215,
      "learning_rate": 1.8491518384819347e-05,
      "loss": 1.2648,
      "step": 3330
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8201427459716797,
      "learning_rate": 1.8490879468421558e-05,
      "loss": 0.9113,
      "step": 3331
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.2868452072143555,
      "learning_rate": 1.849024055202377e-05,
      "loss": 1.1614,
      "step": 3332
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.519316554069519,
      "learning_rate": 1.848960163562598e-05,
      "loss": 0.3728,
      "step": 3333
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.035576581954956,
      "learning_rate": 1.848896271922819e-05,
      "loss": 0.8667,
      "step": 3334
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9304152727127075,
      "learning_rate": 1.84883238028304e-05,
      "loss": 0.9127,
      "step": 3335
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.224907398223877,
      "learning_rate": 1.8487684886432612e-05,
      "loss": 1.4,
      "step": 3336
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8288577795028687,
      "learning_rate": 1.8487045970034824e-05,
      "loss": 0.8038,
      "step": 3337
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2178735733032227,
      "learning_rate": 1.8486407053637032e-05,
      "loss": 0.342,
      "step": 3338
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.921918511390686,
      "learning_rate": 1.8485768137239244e-05,
      "loss": 0.9251,
      "step": 3339
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1784796714782715,
      "learning_rate": 1.8485129220841452e-05,
      "loss": 0.885,
      "step": 3340
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.894394874572754,
      "learning_rate": 1.8484490304443667e-05,
      "loss": 1.0878,
      "step": 3341
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.881700038909912,
      "learning_rate": 1.8483851388045875e-05,
      "loss": 1.5514,
      "step": 3342
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.101863145828247,
      "learning_rate": 1.8483212471648086e-05,
      "loss": 0.9902,
      "step": 3343
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9573118686676025,
      "learning_rate": 1.8482573555250295e-05,
      "loss": 0.951,
      "step": 3344
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9375922679901123,
      "learning_rate": 1.8481934638852506e-05,
      "loss": 0.6605,
      "step": 3345
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5091183185577393,
      "learning_rate": 1.8481295722454718e-05,
      "loss": 0.8556,
      "step": 3346
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5176312923431396,
      "learning_rate": 1.848065680605693e-05,
      "loss": 0.7754,
      "step": 3347
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7423229217529297,
      "learning_rate": 1.8480017889659137e-05,
      "loss": 0.2857,
      "step": 3348
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0459580421447754,
      "learning_rate": 1.847937897326135e-05,
      "loss": 0.961,
      "step": 3349
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3869450092315674,
      "learning_rate": 1.847874005686356e-05,
      "loss": 0.9529,
      "step": 3350
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9129812717437744,
      "learning_rate": 1.8478101140465772e-05,
      "loss": 0.8748,
      "step": 3351
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.7368860244750977,
      "learning_rate": 1.847746222406798e-05,
      "loss": 1.1865,
      "step": 3352
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5554683208465576,
      "learning_rate": 1.8476823307670192e-05,
      "loss": 1.0505,
      "step": 3353
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6268880367279053,
      "learning_rate": 1.8476184391272403e-05,
      "loss": 1.0676,
      "step": 3354
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0342276096343994,
      "learning_rate": 1.8475545474874615e-05,
      "loss": 1.1996,
      "step": 3355
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2893948554992676,
      "learning_rate": 1.8474906558476823e-05,
      "loss": 1.4625,
      "step": 3356
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.1249213218688965,
      "learning_rate": 1.8474267642079035e-05,
      "loss": 1.43,
      "step": 3357
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.06441068649292,
      "learning_rate": 1.8473628725681246e-05,
      "loss": 1.0502,
      "step": 3358
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9524078369140625,
      "learning_rate": 1.8472989809283458e-05,
      "loss": 0.9276,
      "step": 3359
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.40604829788208,
      "learning_rate": 1.8472350892885666e-05,
      "loss": 1.118,
      "step": 3360
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3217012882232666,
      "learning_rate": 1.8471711976487877e-05,
      "loss": 0.9996,
      "step": 3361
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.827420949935913,
      "learning_rate": 1.847107306009009e-05,
      "loss": 0.8421,
      "step": 3362
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.611901044845581,
      "learning_rate": 1.84704341436923e-05,
      "loss": 1.1176,
      "step": 3363
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.691182017326355,
      "learning_rate": 1.846979522729451e-05,
      "loss": 0.8154,
      "step": 3364
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.695875406265259,
      "learning_rate": 1.846915631089672e-05,
      "loss": 1.1336,
      "step": 3365
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.368209362030029,
      "learning_rate": 1.8468517394498932e-05,
      "loss": 1.38,
      "step": 3366
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8830536603927612,
      "learning_rate": 1.8467878478101143e-05,
      "loss": 0.9211,
      "step": 3367
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9078408479690552,
      "learning_rate": 1.846723956170335e-05,
      "loss": 0.875,
      "step": 3368
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.121924877166748,
      "learning_rate": 1.8466600645305563e-05,
      "loss": 1.0205,
      "step": 3369
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.88864004611969,
      "learning_rate": 1.8465961728907774e-05,
      "loss": 0.9715,
      "step": 3370
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.031933307647705,
      "learning_rate": 1.8465322812509986e-05,
      "loss": 0.9277,
      "step": 3371
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.968535900115967,
      "learning_rate": 1.8464683896112194e-05,
      "loss": 1.256,
      "step": 3372
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5618743896484375,
      "learning_rate": 1.8464044979714406e-05,
      "loss": 1.0872,
      "step": 3373
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.215756416320801,
      "learning_rate": 1.8463406063316617e-05,
      "loss": 1.3682,
      "step": 3374
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9432299137115479,
      "learning_rate": 1.846276714691883e-05,
      "loss": 0.9311,
      "step": 3375
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5111536979675293,
      "learning_rate": 1.8462128230521037e-05,
      "loss": 0.8833,
      "step": 3376
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.651467800140381,
      "learning_rate": 1.846148931412325e-05,
      "loss": 1.3088,
      "step": 3377
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.062910079956055,
      "learning_rate": 1.8460850397725457e-05,
      "loss": 1.4695,
      "step": 3378
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0637052059173584,
      "learning_rate": 1.846021148132767e-05,
      "loss": 0.8392,
      "step": 3379
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.356740951538086,
      "learning_rate": 1.845957256492988e-05,
      "loss": 0.8684,
      "step": 3380
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.781744003295898,
      "learning_rate": 1.845893364853209e-05,
      "loss": 1.5776,
      "step": 3381
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.309470176696777,
      "learning_rate": 1.84582947321343e-05,
      "loss": 1.104,
      "step": 3382
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.852245569229126,
      "learning_rate": 1.845765581573651e-05,
      "loss": 0.9945,
      "step": 3383
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7425780296325684,
      "learning_rate": 1.8457016899338723e-05,
      "loss": 0.6999,
      "step": 3384
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.805240273475647,
      "learning_rate": 1.8456377982940934e-05,
      "loss": 0.8202,
      "step": 3385
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2124617099761963,
      "learning_rate": 1.8455739066543142e-05,
      "loss": 1.0807,
      "step": 3386
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.6514031887054443,
      "learning_rate": 1.8455100150145354e-05,
      "loss": 1.2752,
      "step": 3387
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2243125438690186,
      "learning_rate": 1.8454461233747565e-05,
      "loss": 0.7873,
      "step": 3388
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.891832947731018,
      "learning_rate": 1.8453822317349777e-05,
      "loss": 0.9509,
      "step": 3389
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6039023399353027,
      "learning_rate": 1.8453183400951985e-05,
      "loss": 0.9451,
      "step": 3390
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8962162733078003,
      "learning_rate": 1.8452544484554197e-05,
      "loss": 0.9516,
      "step": 3391
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.3222689628601074,
      "learning_rate": 1.8451905568156408e-05,
      "loss": 0.9777,
      "step": 3392
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4216761589050293,
      "learning_rate": 1.845126665175862e-05,
      "loss": 0.8593,
      "step": 3393
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.686803936958313,
      "learning_rate": 1.8450627735360828e-05,
      "loss": 0.8916,
      "step": 3394
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7702454328536987,
      "learning_rate": 1.844998881896304e-05,
      "loss": 1.0738,
      "step": 3395
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2925124168395996,
      "learning_rate": 1.844934990256525e-05,
      "loss": 1.0603,
      "step": 3396
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.534514427185059,
      "learning_rate": 1.8448710986167463e-05,
      "loss": 1.4324,
      "step": 3397
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4967145919799805,
      "learning_rate": 1.844807206976967e-05,
      "loss": 0.8402,
      "step": 3398
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.177647352218628,
      "learning_rate": 1.8447433153371882e-05,
      "loss": 0.8114,
      "step": 3399
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.089885950088501,
      "learning_rate": 1.8446794236974094e-05,
      "loss": 1.1711,
      "step": 3400
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8034385442733765,
      "learning_rate": 1.8446155320576305e-05,
      "loss": 0.9174,
      "step": 3401
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.1282880306243896,
      "learning_rate": 1.8445516404178514e-05,
      "loss": 1.2196,
      "step": 3402
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8924643993377686,
      "learning_rate": 1.8444877487780725e-05,
      "loss": 0.8226,
      "step": 3403
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8727502822875977,
      "learning_rate": 1.8444238571382937e-05,
      "loss": 0.8498,
      "step": 3404
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9913746118545532,
      "learning_rate": 1.8443599654985148e-05,
      "loss": 0.9918,
      "step": 3405
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2317490577697754,
      "learning_rate": 1.8442960738587356e-05,
      "loss": 1.3214,
      "step": 3406
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.707105040550232,
      "learning_rate": 1.8442321822189568e-05,
      "loss": 0.803,
      "step": 3407
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3865771293640137,
      "learning_rate": 1.844168290579178e-05,
      "loss": 1.355,
      "step": 3408
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.937232494354248,
      "learning_rate": 1.844104398939399e-05,
      "loss": 1.6301,
      "step": 3409
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3275845050811768,
      "learning_rate": 1.84404050729962e-05,
      "loss": 0.9219,
      "step": 3410
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1704261302948,
      "learning_rate": 1.843976615659841e-05,
      "loss": 0.8725,
      "step": 3411
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6753838062286377,
      "learning_rate": 1.843912724020062e-05,
      "loss": 1.0314,
      "step": 3412
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0940239429473877,
      "learning_rate": 1.8438488323802834e-05,
      "loss": 1.0573,
      "step": 3413
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.97603440284729,
      "learning_rate": 1.8437849407405042e-05,
      "loss": 0.8316,
      "step": 3414
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9003336429595947,
      "learning_rate": 1.8437210491007253e-05,
      "loss": 1.09,
      "step": 3415
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0742342472076416,
      "learning_rate": 1.843657157460946e-05,
      "loss": 0.8726,
      "step": 3416
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5962826013565063,
      "learning_rate": 1.8435932658211677e-05,
      "loss": 0.8735,
      "step": 3417
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5710283517837524,
      "learning_rate": 1.8435293741813885e-05,
      "loss": 0.8757,
      "step": 3418
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.957421898841858,
      "learning_rate": 1.8434654825416096e-05,
      "loss": 1.1364,
      "step": 3419
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4339263439178467,
      "learning_rate": 1.8434015909018304e-05,
      "loss": 0.9298,
      "step": 3420
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4885642528533936,
      "learning_rate": 1.8433376992620516e-05,
      "loss": 0.7827,
      "step": 3421
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9493321180343628,
      "learning_rate": 1.8432738076222728e-05,
      "loss": 1.0401,
      "step": 3422
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.190752983093262,
      "learning_rate": 1.843209915982494e-05,
      "loss": 1.2045,
      "step": 3423
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.779163122177124,
      "learning_rate": 1.8431460243427147e-05,
      "loss": 0.9232,
      "step": 3424
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6020455360412598,
      "learning_rate": 1.843082132702936e-05,
      "loss": 1.1902,
      "step": 3425
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2591426372528076,
      "learning_rate": 1.843018241063157e-05,
      "loss": 1.4429,
      "step": 3426
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2991995811462402,
      "learning_rate": 1.8429543494233782e-05,
      "loss": 1.0253,
      "step": 3427
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8056632280349731,
      "learning_rate": 1.842890457783599e-05,
      "loss": 0.8843,
      "step": 3428
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.016899347305298,
      "learning_rate": 1.84282656614382e-05,
      "loss": 0.9552,
      "step": 3429
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7108796834945679,
      "learning_rate": 1.8427626745040413e-05,
      "loss": 0.8428,
      "step": 3430
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.297821521759033,
      "learning_rate": 1.8426987828642625e-05,
      "loss": 0.8272,
      "step": 3431
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.231411933898926,
      "learning_rate": 1.8426348912244833e-05,
      "loss": 0.9753,
      "step": 3432
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.6155195236206055,
      "learning_rate": 1.8425709995847044e-05,
      "loss": 1.0576,
      "step": 3433
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.100853443145752,
      "learning_rate": 1.8425071079449256e-05,
      "loss": 0.9313,
      "step": 3434
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0208911895751953,
      "learning_rate": 1.8424432163051467e-05,
      "loss": 0.8654,
      "step": 3435
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5678610801696777,
      "learning_rate": 1.8423793246653676e-05,
      "loss": 1.0499,
      "step": 3436
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0081124305725098,
      "learning_rate": 1.8423154330255887e-05,
      "loss": 0.9484,
      "step": 3437
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.2919046878814697,
      "learning_rate": 1.84225154138581e-05,
      "loss": 1.0565,
      "step": 3438
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8518404960632324,
      "learning_rate": 1.842187649746031e-05,
      "loss": 0.777,
      "step": 3439
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8306782245635986,
      "learning_rate": 1.842123758106252e-05,
      "loss": 0.9361,
      "step": 3440
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1578361988067627,
      "learning_rate": 1.842059866466473e-05,
      "loss": 1.0156,
      "step": 3441
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.077042579650879,
      "learning_rate": 1.841995974826694e-05,
      "loss": 0.9581,
      "step": 3442
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.995229959487915,
      "learning_rate": 1.8419320831869153e-05,
      "loss": 1.1062,
      "step": 3443
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3047425746917725,
      "learning_rate": 1.841868191547136e-05,
      "loss": 1.3138,
      "step": 3444
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0550193786621094,
      "learning_rate": 1.8418042999073573e-05,
      "loss": 1.0305,
      "step": 3445
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9990415573120117,
      "learning_rate": 1.8417404082675784e-05,
      "loss": 0.9597,
      "step": 3446
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.105858087539673,
      "learning_rate": 1.8416765166277996e-05,
      "loss": 0.9967,
      "step": 3447
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.419959306716919,
      "learning_rate": 1.8416126249880204e-05,
      "loss": 0.9925,
      "step": 3448
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3384950160980225,
      "learning_rate": 1.8415487333482416e-05,
      "loss": 1.0583,
      "step": 3449
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.191084623336792,
      "learning_rate": 1.8414848417084624e-05,
      "loss": 1.0812,
      "step": 3450
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.977779507637024,
      "learning_rate": 1.841420950068684e-05,
      "loss": 0.9515,
      "step": 3451
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.694352149963379,
      "learning_rate": 1.8413570584289047e-05,
      "loss": 0.9276,
      "step": 3452
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9321823120117188,
      "learning_rate": 1.841293166789126e-05,
      "loss": 0.8788,
      "step": 3453
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.933696746826172,
      "learning_rate": 1.8412292751493467e-05,
      "loss": 1.1078,
      "step": 3454
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6223608255386353,
      "learning_rate": 1.8411653835095678e-05,
      "loss": 0.3283,
      "step": 3455
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0490002632141113,
      "learning_rate": 1.841101491869789e-05,
      "loss": 0.8238,
      "step": 3456
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.621617555618286,
      "learning_rate": 1.84103760023001e-05,
      "loss": 0.9694,
      "step": 3457
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4430056810379028,
      "learning_rate": 1.840973708590231e-05,
      "loss": 0.2932,
      "step": 3458
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.927320122718811,
      "learning_rate": 1.840909816950452e-05,
      "loss": 0.9099,
      "step": 3459
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6705586910247803,
      "learning_rate": 1.8408459253106732e-05,
      "loss": 1.002,
      "step": 3460
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.112653970718384,
      "learning_rate": 1.8407820336708944e-05,
      "loss": 1.197,
      "step": 3461
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1362650394439697,
      "learning_rate": 1.8407181420311152e-05,
      "loss": 0.8583,
      "step": 3462
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.099853992462158,
      "learning_rate": 1.8406542503913364e-05,
      "loss": 0.8648,
      "step": 3463
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.281020641326904,
      "learning_rate": 1.8405903587515575e-05,
      "loss": 1.3331,
      "step": 3464
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8395253419876099,
      "learning_rate": 1.8405264671117787e-05,
      "loss": 0.8363,
      "step": 3465
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1463284492492676,
      "learning_rate": 1.8404625754719995e-05,
      "loss": 0.8954,
      "step": 3466
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.641561985015869,
      "learning_rate": 1.8403986838322206e-05,
      "loss": 1.0542,
      "step": 3467
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8138986825942993,
      "learning_rate": 1.8403347921924418e-05,
      "loss": 0.9681,
      "step": 3468
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.093481540679932,
      "learning_rate": 1.840270900552663e-05,
      "loss": 1.2839,
      "step": 3469
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.465627908706665,
      "learning_rate": 1.8402070089128838e-05,
      "loss": 0.5074,
      "step": 3470
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.248992919921875,
      "learning_rate": 1.840143117273105e-05,
      "loss": 0.92,
      "step": 3471
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.094337224960327,
      "learning_rate": 1.840079225633326e-05,
      "loss": 0.9335,
      "step": 3472
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.183683395385742,
      "learning_rate": 1.8400153339935472e-05,
      "loss": 0.8723,
      "step": 3473
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6183757781982422,
      "learning_rate": 1.839951442353768e-05,
      "loss": 0.3583,
      "step": 3474
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.55511212348938,
      "learning_rate": 1.8398875507139892e-05,
      "loss": 1.0706,
      "step": 3475
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.700712203979492,
      "learning_rate": 1.8398236590742104e-05,
      "loss": 1.3964,
      "step": 3476
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.149217367172241,
      "learning_rate": 1.8397597674344315e-05,
      "loss": 1.4564,
      "step": 3477
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5515408515930176,
      "learning_rate": 1.8396958757946523e-05,
      "loss": 1.1653,
      "step": 3478
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.395256757736206,
      "learning_rate": 1.8396319841548735e-05,
      "loss": 0.9396,
      "step": 3479
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9208619594573975,
      "learning_rate": 1.8395680925150946e-05,
      "loss": 0.8266,
      "step": 3480
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4241132736206055,
      "learning_rate": 1.8395042008753158e-05,
      "loss": 1.226,
      "step": 3481
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.886986255645752,
      "learning_rate": 1.8394403092355366e-05,
      "loss": 1.1316,
      "step": 3482
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7213106155395508,
      "learning_rate": 1.8393764175957578e-05,
      "loss": 0.9116,
      "step": 3483
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4737484455108643,
      "learning_rate": 1.839312525955979e-05,
      "loss": 0.9233,
      "step": 3484
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9096381664276123,
      "learning_rate": 1.8392486343162e-05,
      "loss": 0.8505,
      "step": 3485
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8973511457443237,
      "learning_rate": 1.839184742676421e-05,
      "loss": 0.9045,
      "step": 3486
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.10264253616333,
      "learning_rate": 1.839120851036642e-05,
      "loss": 0.8497,
      "step": 3487
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9363906383514404,
      "learning_rate": 1.839056959396863e-05,
      "loss": 0.9535,
      "step": 3488
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8982760906219482,
      "learning_rate": 1.8389930677570844e-05,
      "loss": 0.9042,
      "step": 3489
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.126300573348999,
      "learning_rate": 1.8389291761173052e-05,
      "loss": 1.0359,
      "step": 3490
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4322550296783447,
      "learning_rate": 1.8388652844775263e-05,
      "loss": 0.8439,
      "step": 3491
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8453917503356934,
      "learning_rate": 1.838801392837747e-05,
      "loss": 0.9578,
      "step": 3492
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0243947505950928,
      "learning_rate": 1.8387375011979683e-05,
      "loss": 0.9846,
      "step": 3493
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8579394817352295,
      "learning_rate": 1.8386736095581895e-05,
      "loss": 0.9524,
      "step": 3494
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.8435378074646,
      "learning_rate": 1.8386097179184106e-05,
      "loss": 1.103,
      "step": 3495
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1278469562530518,
      "learning_rate": 1.8385458262786314e-05,
      "loss": 0.9897,
      "step": 3496
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.103015422821045,
      "learning_rate": 1.8384819346388526e-05,
      "loss": 0.9274,
      "step": 3497
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4464433193206787,
      "learning_rate": 1.8384180429990737e-05,
      "loss": 1.014,
      "step": 3498
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6187071800231934,
      "learning_rate": 1.838354151359295e-05,
      "loss": 0.623,
      "step": 3499
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1996755599975586,
      "learning_rate": 1.8382902597195157e-05,
      "loss": 0.9152,
      "step": 3500
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5687540769577026,
      "learning_rate": 1.838226368079737e-05,
      "loss": 0.3735,
      "step": 3501
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.627020835876465,
      "learning_rate": 1.838162476439958e-05,
      "loss": 1.7978,
      "step": 3502
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.0821852684021,
      "learning_rate": 1.838098584800179e-05,
      "loss": 1.4033,
      "step": 3503
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.206882476806641,
      "learning_rate": 1.8380346931604e-05,
      "loss": 1.1773,
      "step": 3504
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9988985061645508,
      "learning_rate": 1.837970801520621e-05,
      "loss": 1.0174,
      "step": 3505
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.921305775642395,
      "learning_rate": 1.8379069098808423e-05,
      "loss": 1.077,
      "step": 3506
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.297555446624756,
      "learning_rate": 1.8378430182410634e-05,
      "loss": 1.0018,
      "step": 3507
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.34051775932312,
      "learning_rate": 1.8377791266012843e-05,
      "loss": 1.0692,
      "step": 3508
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.722496509552002,
      "learning_rate": 1.8377152349615054e-05,
      "loss": 1.1871,
      "step": 3509
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8332107067108154,
      "learning_rate": 1.8376513433217266e-05,
      "loss": 0.9235,
      "step": 3510
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2172274589538574,
      "learning_rate": 1.8375874516819477e-05,
      "loss": 1.0622,
      "step": 3511
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.097430467605591,
      "learning_rate": 1.8375235600421685e-05,
      "loss": 0.8468,
      "step": 3512
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3853867053985596,
      "learning_rate": 1.8374596684023897e-05,
      "loss": 1.0423,
      "step": 3513
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0912985801696777,
      "learning_rate": 1.837395776762611e-05,
      "loss": 0.9523,
      "step": 3514
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.373770236968994,
      "learning_rate": 1.837331885122832e-05,
      "loss": 1.4002,
      "step": 3515
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3534090518951416,
      "learning_rate": 1.8372679934830528e-05,
      "loss": 1.0636,
      "step": 3516
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8898303508758545,
      "learning_rate": 1.837204101843274e-05,
      "loss": 0.9745,
      "step": 3517
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.648470640182495,
      "learning_rate": 1.837140210203495e-05,
      "loss": 0.6437,
      "step": 3518
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.1297607421875,
      "learning_rate": 1.8370763185637163e-05,
      "loss": 1.0738,
      "step": 3519
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.027113199234009,
      "learning_rate": 1.837012426923937e-05,
      "loss": 0.8328,
      "step": 3520
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.9335927963256836,
      "learning_rate": 1.8369485352841583e-05,
      "loss": 1.0377,
      "step": 3521
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.136914014816284,
      "learning_rate": 1.836884643644379e-05,
      "loss": 0.8703,
      "step": 3522
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5146331787109375,
      "learning_rate": 1.8368207520046006e-05,
      "loss": 0.9977,
      "step": 3523
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8302972316741943,
      "learning_rate": 1.8367568603648214e-05,
      "loss": 0.8482,
      "step": 3524
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5285298824310303,
      "learning_rate": 1.8366929687250425e-05,
      "loss": 0.386,
      "step": 3525
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8690478801727295,
      "learning_rate": 1.8366290770852634e-05,
      "loss": 0.8926,
      "step": 3526
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1736719608306885,
      "learning_rate": 1.836565185445485e-05,
      "loss": 0.9876,
      "step": 3527
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.352221727371216,
      "learning_rate": 1.8365012938057057e-05,
      "loss": 0.9681,
      "step": 3528
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.54349684715271,
      "learning_rate": 1.8364374021659268e-05,
      "loss": 1.1967,
      "step": 3529
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5877618789672852,
      "learning_rate": 1.8363735105261476e-05,
      "loss": 0.7665,
      "step": 3530
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3537628650665283,
      "learning_rate": 1.8363096188863688e-05,
      "loss": 1.0367,
      "step": 3531
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6751692295074463,
      "learning_rate": 1.83624572724659e-05,
      "loss": 0.9265,
      "step": 3532
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.357048988342285,
      "learning_rate": 1.836181835606811e-05,
      "loss": 0.9778,
      "step": 3533
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.157527446746826,
      "learning_rate": 1.836117943967032e-05,
      "loss": 0.7777,
      "step": 3534
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8600562810897827,
      "learning_rate": 1.836054052327253e-05,
      "loss": 0.842,
      "step": 3535
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.740166425704956,
      "learning_rate": 1.8359901606874742e-05,
      "loss": 1.143,
      "step": 3536
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.615959405899048,
      "learning_rate": 1.8359262690476954e-05,
      "loss": 0.8068,
      "step": 3537
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3866262435913086,
      "learning_rate": 1.8358623774079162e-05,
      "loss": 0.8938,
      "step": 3538
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.003356695175171,
      "learning_rate": 1.8357984857681373e-05,
      "loss": 1.3131,
      "step": 3539
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9623382091522217,
      "learning_rate": 1.8357345941283585e-05,
      "loss": 0.9054,
      "step": 3540
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.449147939682007,
      "learning_rate": 1.8356707024885797e-05,
      "loss": 1.1049,
      "step": 3541
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.853694438934326,
      "learning_rate": 1.8356068108488005e-05,
      "loss": 1.2855,
      "step": 3542
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.7140049934387207,
      "learning_rate": 1.8355429192090216e-05,
      "loss": 1.1541,
      "step": 3543
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7799630165100098,
      "learning_rate": 1.8354790275692428e-05,
      "loss": 0.874,
      "step": 3544
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.345017671585083,
      "learning_rate": 1.835415135929464e-05,
      "loss": 0.8352,
      "step": 3545
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.03035569190979,
      "learning_rate": 1.8353512442896848e-05,
      "loss": 0.9521,
      "step": 3546
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9007377624511719,
      "learning_rate": 1.835287352649906e-05,
      "loss": 0.9376,
      "step": 3547
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2282183170318604,
      "learning_rate": 1.835223461010127e-05,
      "loss": 0.9307,
      "step": 3548
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.906407356262207,
      "learning_rate": 1.8351595693703482e-05,
      "loss": 1.0762,
      "step": 3549
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.289536952972412,
      "learning_rate": 1.835095677730569e-05,
      "loss": 1.118,
      "step": 3550
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.026733875274658,
      "learning_rate": 1.8350317860907902e-05,
      "loss": 0.8611,
      "step": 3551
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.553523063659668,
      "learning_rate": 1.8349678944510113e-05,
      "loss": 1.5002,
      "step": 3552
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6641390323638916,
      "learning_rate": 1.8349040028112325e-05,
      "loss": 0.9302,
      "step": 3553
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.030266046524048,
      "learning_rate": 1.8348401111714533e-05,
      "loss": 0.9149,
      "step": 3554
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.695951461791992,
      "learning_rate": 1.8347762195316745e-05,
      "loss": 0.9715,
      "step": 3555
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0453031063079834,
      "learning_rate": 1.8347123278918956e-05,
      "loss": 0.9712,
      "step": 3556
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.739091396331787,
      "learning_rate": 1.8346484362521168e-05,
      "loss": 1.521,
      "step": 3557
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7918360233306885,
      "learning_rate": 1.8345845446123376e-05,
      "loss": 0.4227,
      "step": 3558
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.62179696559906,
      "learning_rate": 1.8345206529725587e-05,
      "loss": 0.417,
      "step": 3559
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4761462211608887,
      "learning_rate": 1.8344567613327796e-05,
      "loss": 1.1962,
      "step": 3560
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4986753463745117,
      "learning_rate": 1.834392869693001e-05,
      "loss": 0.8979,
      "step": 3561
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.792422890663147,
      "learning_rate": 1.834328978053222e-05,
      "loss": 0.9028,
      "step": 3562
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1771912574768066,
      "learning_rate": 1.834265086413443e-05,
      "loss": 0.9077,
      "step": 3563
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7669597864151,
      "learning_rate": 1.834201194773664e-05,
      "loss": 0.8664,
      "step": 3564
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.649499416351318,
      "learning_rate": 1.834137303133885e-05,
      "loss": 1.2752,
      "step": 3565
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.584031343460083,
      "learning_rate": 1.834073411494106e-05,
      "loss": 0.9716,
      "step": 3566
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6870243549346924,
      "learning_rate": 1.8340095198543273e-05,
      "loss": 0.8509,
      "step": 3567
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7526248693466187,
      "learning_rate": 1.833945628214548e-05,
      "loss": 0.8712,
      "step": 3568
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7132184505462646,
      "learning_rate": 1.8338817365747693e-05,
      "loss": 0.9861,
      "step": 3569
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7291144132614136,
      "learning_rate": 1.8338178449349904e-05,
      "loss": 0.8117,
      "step": 3570
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2132530212402344,
      "learning_rate": 1.8337539532952116e-05,
      "loss": 1.1613,
      "step": 3571
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.877360224723816,
      "learning_rate": 1.8336900616554324e-05,
      "loss": 0.9039,
      "step": 3572
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.040097951889038,
      "learning_rate": 1.8336261700156536e-05,
      "loss": 0.888,
      "step": 3573
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1781797409057617,
      "learning_rate": 1.8335622783758747e-05,
      "loss": 0.9164,
      "step": 3574
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.723676085472107,
      "learning_rate": 1.833498386736096e-05,
      "loss": 0.8865,
      "step": 3575
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.505903720855713,
      "learning_rate": 1.8334344950963167e-05,
      "loss": 1.6342,
      "step": 3576
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.164165735244751,
      "learning_rate": 1.833370603456538e-05,
      "loss": 1.1744,
      "step": 3577
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2620158195495605,
      "learning_rate": 1.833306711816759e-05,
      "loss": 0.3368,
      "step": 3578
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.922288417816162,
      "learning_rate": 1.83324282017698e-05,
      "loss": 1.3013,
      "step": 3579
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.933642387390137,
      "learning_rate": 1.833178928537201e-05,
      "loss": 1.047,
      "step": 3580
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3002867698669434,
      "learning_rate": 1.833115036897422e-05,
      "loss": 0.842,
      "step": 3581
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.693537712097168,
      "learning_rate": 1.8330511452576433e-05,
      "loss": 1.3344,
      "step": 3582
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9410136938095093,
      "learning_rate": 1.8329872536178644e-05,
      "loss": 0.883,
      "step": 3583
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.159869909286499,
      "learning_rate": 1.8329233619780852e-05,
      "loss": 1.0739,
      "step": 3584
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5011744499206543,
      "learning_rate": 1.8328594703383064e-05,
      "loss": 1.4615,
      "step": 3585
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0836188793182373,
      "learning_rate": 1.8327955786985276e-05,
      "loss": 0.8663,
      "step": 3586
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.3072521686553955,
      "learning_rate": 1.8327316870587487e-05,
      "loss": 0.7342,
      "step": 3587
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9801386594772339,
      "learning_rate": 1.8326677954189695e-05,
      "loss": 0.9908,
      "step": 3588
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.992935061454773,
      "learning_rate": 1.8326039037791907e-05,
      "loss": 0.9678,
      "step": 3589
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0740199089050293,
      "learning_rate": 1.832540012139412e-05,
      "loss": 0.9599,
      "step": 3590
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9171816110610962,
      "learning_rate": 1.832476120499633e-05,
      "loss": 0.8517,
      "step": 3591
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.996054768562317,
      "learning_rate": 1.8324122288598538e-05,
      "loss": 0.8263,
      "step": 3592
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6497591733932495,
      "learning_rate": 1.832348337220075e-05,
      "loss": 0.8414,
      "step": 3593
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.991119861602783,
      "learning_rate": 1.832284445580296e-05,
      "loss": 0.7474,
      "step": 3594
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.464759111404419,
      "learning_rate": 1.8322205539405173e-05,
      "loss": 0.8734,
      "step": 3595
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.022757053375244,
      "learning_rate": 1.832156662300738e-05,
      "loss": 1.0039,
      "step": 3596
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6674762964248657,
      "learning_rate": 1.8320927706609592e-05,
      "loss": 0.8277,
      "step": 3597
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9377366304397583,
      "learning_rate": 1.83202887902118e-05,
      "loss": 0.9058,
      "step": 3598
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0754456520080566,
      "learning_rate": 1.8319649873814015e-05,
      "loss": 0.9901,
      "step": 3599
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.220576286315918,
      "learning_rate": 1.8319010957416224e-05,
      "loss": 0.7351,
      "step": 3600
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3717291355133057,
      "learning_rate": 1.8318372041018435e-05,
      "loss": 1.2977,
      "step": 3601
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.58380925655365,
      "learning_rate": 1.8317733124620643e-05,
      "loss": 0.4346,
      "step": 3602
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.236187219619751,
      "learning_rate": 1.8317094208222855e-05,
      "loss": 0.933,
      "step": 3603
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.078392028808594,
      "learning_rate": 1.8316455291825066e-05,
      "loss": 1.6752,
      "step": 3604
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7781902551651,
      "learning_rate": 1.8315816375427278e-05,
      "loss": 0.9439,
      "step": 3605
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8612048625946045,
      "learning_rate": 1.8315177459029486e-05,
      "loss": 0.9163,
      "step": 3606
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8423606157302856,
      "learning_rate": 1.8314538542631698e-05,
      "loss": 0.9817,
      "step": 3607
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9961296319961548,
      "learning_rate": 1.831389962623391e-05,
      "loss": 0.9015,
      "step": 3608
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3267595767974854,
      "learning_rate": 1.831326070983612e-05,
      "loss": 1.2515,
      "step": 3609
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.3895158767700195,
      "learning_rate": 1.831262179343833e-05,
      "loss": 0.9629,
      "step": 3610
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9114291667938232,
      "learning_rate": 1.831198287704054e-05,
      "loss": 0.8915,
      "step": 3611
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8964341878890991,
      "learning_rate": 1.8311343960642752e-05,
      "loss": 0.9109,
      "step": 3612
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8512979745864868,
      "learning_rate": 1.8310705044244964e-05,
      "loss": 0.8454,
      "step": 3613
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.178316593170166,
      "learning_rate": 1.8310066127847172e-05,
      "loss": 0.9216,
      "step": 3614
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0579185485839844,
      "learning_rate": 1.8309427211449383e-05,
      "loss": 0.8967,
      "step": 3615
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9234484434127808,
      "learning_rate": 1.8308788295051595e-05,
      "loss": 0.826,
      "step": 3616
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.34159517288208,
      "learning_rate": 1.8308149378653806e-05,
      "loss": 1.2602,
      "step": 3617
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8640704154968262,
      "learning_rate": 1.8307510462256015e-05,
      "loss": 0.8444,
      "step": 3618
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8713507652282715,
      "learning_rate": 1.8306871545858226e-05,
      "loss": 1.1364,
      "step": 3619
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3288819789886475,
      "learning_rate": 1.8306232629460438e-05,
      "loss": 1.0484,
      "step": 3620
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1810076236724854,
      "learning_rate": 1.830559371306265e-05,
      "loss": 0.8821,
      "step": 3621
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.14316987991333,
      "learning_rate": 1.8304954796664857e-05,
      "loss": 0.9569,
      "step": 3622
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.500370502471924,
      "learning_rate": 1.830431588026707e-05,
      "loss": 1.5979,
      "step": 3623
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.967734932899475,
      "learning_rate": 1.830367696386928e-05,
      "loss": 1.3369,
      "step": 3624
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.5835490226745605,
      "learning_rate": 1.8303038047471492e-05,
      "loss": 0.8121,
      "step": 3625
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.092219829559326,
      "learning_rate": 1.83023991310737e-05,
      "loss": 0.8292,
      "step": 3626
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.666396379470825,
      "learning_rate": 1.830176021467591e-05,
      "loss": 1.1571,
      "step": 3627
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.625775098800659,
      "learning_rate": 1.8301121298278123e-05,
      "loss": 0.8194,
      "step": 3628
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2728958129882812,
      "learning_rate": 1.8300482381880335e-05,
      "loss": 0.9687,
      "step": 3629
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.130690574645996,
      "learning_rate": 1.8299843465482543e-05,
      "loss": 1.4691,
      "step": 3630
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.391550540924072,
      "learning_rate": 1.8299204549084754e-05,
      "loss": 1.5884,
      "step": 3631
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5520691871643066,
      "learning_rate": 1.8298565632686963e-05,
      "loss": 0.8,
      "step": 3632
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.30204439163208,
      "learning_rate": 1.8297926716289178e-05,
      "loss": 0.9746,
      "step": 3633
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8221454620361328,
      "learning_rate": 1.8297287799891386e-05,
      "loss": 0.9127,
      "step": 3634
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.32832407951355,
      "learning_rate": 1.8296648883493597e-05,
      "loss": 0.9735,
      "step": 3635
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.216700553894043,
      "learning_rate": 1.8296009967095805e-05,
      "loss": 1.0356,
      "step": 3636
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.048532962799072,
      "learning_rate": 1.8295371050698017e-05,
      "loss": 1.2397,
      "step": 3637
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4511988162994385,
      "learning_rate": 1.829473213430023e-05,
      "loss": 0.9018,
      "step": 3638
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7423954010009766,
      "learning_rate": 1.829409321790244e-05,
      "loss": 0.9716,
      "step": 3639
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9092298746109009,
      "learning_rate": 1.8293454301504648e-05,
      "loss": 0.868,
      "step": 3640
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.912536144256592,
      "learning_rate": 1.829281538510686e-05,
      "loss": 1.0149,
      "step": 3641
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9086291790008545,
      "learning_rate": 1.829217646870907e-05,
      "loss": 0.6853,
      "step": 3642
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7480428218841553,
      "learning_rate": 1.8291537552311283e-05,
      "loss": 1.0223,
      "step": 3643
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3560030460357666,
      "learning_rate": 1.829089863591349e-05,
      "loss": 1.2822,
      "step": 3644
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.470912933349609,
      "learning_rate": 1.8290259719515703e-05,
      "loss": 1.1497,
      "step": 3645
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8648349046707153,
      "learning_rate": 1.8289620803117914e-05,
      "loss": 0.9439,
      "step": 3646
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.506410598754883,
      "learning_rate": 1.8288981886720126e-05,
      "loss": 0.8892,
      "step": 3647
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.120635986328125,
      "learning_rate": 1.8288342970322334e-05,
      "loss": 0.8877,
      "step": 3648
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3037679195404053,
      "learning_rate": 1.8287704053924545e-05,
      "loss": 0.9948,
      "step": 3649
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0261118412017822,
      "learning_rate": 1.8287065137526757e-05,
      "loss": 0.8926,
      "step": 3650
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.148906707763672,
      "learning_rate": 1.828642622112897e-05,
      "loss": 1.7071,
      "step": 3651
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.108664035797119,
      "learning_rate": 1.8285787304731177e-05,
      "loss": 1.2003,
      "step": 3652
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8674241304397583,
      "learning_rate": 1.8285148388333388e-05,
      "loss": 0.9621,
      "step": 3653
    },
    {
      "epoch": 0.11,
      "grad_norm": 8.504881858825684,
      "learning_rate": 1.82845094719356e-05,
      "loss": 1.1203,
      "step": 3654
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.464015483856201,
      "learning_rate": 1.828387055553781e-05,
      "loss": 1.0255,
      "step": 3655
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9010728597640991,
      "learning_rate": 1.828323163914002e-05,
      "loss": 0.8756,
      "step": 3656
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7959238290786743,
      "learning_rate": 1.828259272274223e-05,
      "loss": 0.9524,
      "step": 3657
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.0158233642578125,
      "learning_rate": 1.8281953806344443e-05,
      "loss": 0.9307,
      "step": 3658
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.304175615310669,
      "learning_rate": 1.8281314889946654e-05,
      "loss": 1.0117,
      "step": 3659
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.249037265777588,
      "learning_rate": 1.8280675973548862e-05,
      "loss": 1.3296,
      "step": 3660
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1946911811828613,
      "learning_rate": 1.8280037057151074e-05,
      "loss": 0.9498,
      "step": 3661
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4530119895935059,
      "learning_rate": 1.8279398140753285e-05,
      "loss": 0.4455,
      "step": 3662
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8540911674499512,
      "learning_rate": 1.8278759224355497e-05,
      "loss": 0.9448,
      "step": 3663
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7482119798660278,
      "learning_rate": 1.8278120307957705e-05,
      "loss": 0.9885,
      "step": 3664
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.17557430267334,
      "learning_rate": 1.8277481391559917e-05,
      "loss": 1.2144,
      "step": 3665
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3930563926696777,
      "learning_rate": 1.8276842475162128e-05,
      "loss": 1.0604,
      "step": 3666
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9444066286087036,
      "learning_rate": 1.827620355876434e-05,
      "loss": 0.9188,
      "step": 3667
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.316973924636841,
      "learning_rate": 1.8275564642366548e-05,
      "loss": 0.9891,
      "step": 3668
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.012366533279419,
      "learning_rate": 1.827492572596876e-05,
      "loss": 0.9606,
      "step": 3669
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.913607120513916,
      "learning_rate": 1.8274286809570968e-05,
      "loss": 0.8479,
      "step": 3670
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.799675464630127,
      "learning_rate": 1.8273647893173182e-05,
      "loss": 1.2064,
      "step": 3671
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7570559978485107,
      "learning_rate": 1.827300897677539e-05,
      "loss": 0.9296,
      "step": 3672
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2676947116851807,
      "learning_rate": 1.8272370060377602e-05,
      "loss": 0.9446,
      "step": 3673
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.602163314819336,
      "learning_rate": 1.827173114397981e-05,
      "loss": 1.0968,
      "step": 3674
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.86849844455719,
      "learning_rate": 1.8271092227582022e-05,
      "loss": 0.9117,
      "step": 3675
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.2077066898345947,
      "learning_rate": 1.8270453311184233e-05,
      "loss": 0.7709,
      "step": 3676
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.337382435798645,
      "learning_rate": 1.826981439478644e-05,
      "loss": 0.3536,
      "step": 3677
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.845011234283447,
      "learning_rate": 1.8269175478388653e-05,
      "loss": 1.048,
      "step": 3678
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.219052314758301,
      "learning_rate": 1.8268536561990865e-05,
      "loss": 0.9433,
      "step": 3679
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4609949588775635,
      "learning_rate": 1.8267897645593076e-05,
      "loss": 1.0253,
      "step": 3680
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.478889465332031,
      "learning_rate": 1.8267258729195284e-05,
      "loss": 1.7811,
      "step": 3681
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1263487339019775,
      "learning_rate": 1.8266619812797496e-05,
      "loss": 1.3159,
      "step": 3682
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0618929862976074,
      "learning_rate": 1.8265980896399707e-05,
      "loss": 0.9105,
      "step": 3683
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8150360584259033,
      "learning_rate": 1.826534198000192e-05,
      "loss": 0.939,
      "step": 3684
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.129368782043457,
      "learning_rate": 1.8264703063604127e-05,
      "loss": 1.1009,
      "step": 3685
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2795515060424805,
      "learning_rate": 1.826406414720634e-05,
      "loss": 0.8959,
      "step": 3686
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.552712082862854,
      "learning_rate": 1.826342523080855e-05,
      "loss": 0.8541,
      "step": 3687
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9625134468078613,
      "learning_rate": 1.8262786314410762e-05,
      "loss": 1.0846,
      "step": 3688
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.319282054901123,
      "learning_rate": 1.826214739801297e-05,
      "loss": 0.8772,
      "step": 3689
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7210018634796143,
      "learning_rate": 1.826150848161518e-05,
      "loss": 0.9717,
      "step": 3690
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.214320182800293,
      "learning_rate": 1.8260869565217393e-05,
      "loss": 0.8849,
      "step": 3691
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8333253860473633,
      "learning_rate": 1.8260230648819605e-05,
      "loss": 0.9085,
      "step": 3692
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9974682331085205,
      "learning_rate": 1.8259591732421813e-05,
      "loss": 0.9747,
      "step": 3693
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.953477621078491,
      "learning_rate": 1.8258952816024024e-05,
      "loss": 1.0501,
      "step": 3694
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5370099544525146,
      "learning_rate": 1.8258313899626236e-05,
      "loss": 1.0406,
      "step": 3695
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6374871730804443,
      "learning_rate": 1.8257674983228447e-05,
      "loss": 1.0238,
      "step": 3696
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9641112089157104,
      "learning_rate": 1.8257036066830656e-05,
      "loss": 0.9003,
      "step": 3697
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.659392356872559,
      "learning_rate": 1.8256397150432867e-05,
      "loss": 1.172,
      "step": 3698
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.166433334350586,
      "learning_rate": 1.8255758234035075e-05,
      "loss": 0.9998,
      "step": 3699
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.1889500617980957,
      "learning_rate": 1.825511931763729e-05,
      "loss": 0.9606,
      "step": 3700
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.819247007369995,
      "learning_rate": 1.82544804012395e-05,
      "loss": 1.0093,
      "step": 3701
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.261760234832764,
      "learning_rate": 1.825384148484171e-05,
      "loss": 1.4359,
      "step": 3702
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.789781093597412,
      "learning_rate": 1.8253202568443918e-05,
      "loss": 1.1295,
      "step": 3703
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1675684452056885,
      "learning_rate": 1.825256365204613e-05,
      "loss": 0.8775,
      "step": 3704
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2890677452087402,
      "learning_rate": 1.825192473564834e-05,
      "loss": 0.9497,
      "step": 3705
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.929953932762146,
      "learning_rate": 1.8251285819250553e-05,
      "loss": 0.8898,
      "step": 3706
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4394848346710205,
      "learning_rate": 1.825064690285276e-05,
      "loss": 1.0483,
      "step": 3707
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.085194110870361,
      "learning_rate": 1.8250007986454972e-05,
      "loss": 1.0043,
      "step": 3708
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7543511390686035,
      "learning_rate": 1.8249369070057184e-05,
      "loss": 1.1192,
      "step": 3709
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6407012939453125,
      "learning_rate": 1.8248730153659396e-05,
      "loss": 1.2506,
      "step": 3710
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.39667010307312,
      "learning_rate": 1.8248091237261604e-05,
      "loss": 1.0891,
      "step": 3711
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.080909013748169,
      "learning_rate": 1.8247452320863815e-05,
      "loss": 0.9891,
      "step": 3712
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5161561965942383,
      "learning_rate": 1.8246813404466027e-05,
      "loss": 0.9521,
      "step": 3713
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3131563663482666,
      "learning_rate": 1.824617448806824e-05,
      "loss": 0.892,
      "step": 3714
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9231481552124023,
      "learning_rate": 1.8245535571670447e-05,
      "loss": 0.9098,
      "step": 3715
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3453521728515625,
      "learning_rate": 1.8244896655272658e-05,
      "loss": 0.9414,
      "step": 3716
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0123493671417236,
      "learning_rate": 1.824425773887487e-05,
      "loss": 0.8145,
      "step": 3717
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0622599124908447,
      "learning_rate": 1.824361882247708e-05,
      "loss": 0.9474,
      "step": 3718
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7235782146453857,
      "learning_rate": 1.824297990607929e-05,
      "loss": 0.9572,
      "step": 3719
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.8977904319763184,
      "learning_rate": 1.82423409896815e-05,
      "loss": 0.9155,
      "step": 3720
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3991352319717407,
      "learning_rate": 1.8241702073283712e-05,
      "loss": 0.3352,
      "step": 3721
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1101791858673096,
      "learning_rate": 1.8241063156885924e-05,
      "loss": 0.8407,
      "step": 3722
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7982875108718872,
      "learning_rate": 1.8240424240488132e-05,
      "loss": 1.0178,
      "step": 3723
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.328704595565796,
      "learning_rate": 1.8239785324090344e-05,
      "loss": 0.7946,
      "step": 3724
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.8988568782806396,
      "learning_rate": 1.8239146407692555e-05,
      "loss": 0.8823,
      "step": 3725
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.225759506225586,
      "learning_rate": 1.8238507491294767e-05,
      "loss": 0.8992,
      "step": 3726
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9118925333023071,
      "learning_rate": 1.8237868574896975e-05,
      "loss": 0.9306,
      "step": 3727
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.12876558303833,
      "learning_rate": 1.8237229658499186e-05,
      "loss": 1.0222,
      "step": 3728
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.908208847045898,
      "learning_rate": 1.8236590742101398e-05,
      "loss": 1.2366,
      "step": 3729
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0212502479553223,
      "learning_rate": 1.823595182570361e-05,
      "loss": 0.8439,
      "step": 3730
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.321563959121704,
      "learning_rate": 1.8235312909305818e-05,
      "loss": 1.0874,
      "step": 3731
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0448920726776123,
      "learning_rate": 1.823467399290803e-05,
      "loss": 0.9006,
      "step": 3732
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3287298679351807,
      "learning_rate": 1.823403507651024e-05,
      "loss": 0.3998,
      "step": 3733
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.986404299736023,
      "learning_rate": 1.8233396160112452e-05,
      "loss": 0.9801,
      "step": 3734
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.381943702697754,
      "learning_rate": 1.823275724371466e-05,
      "loss": 1.0409,
      "step": 3735
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.170820474624634,
      "learning_rate": 1.8232118327316872e-05,
      "loss": 0.9881,
      "step": 3736
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.173966884613037,
      "learning_rate": 1.823147941091908e-05,
      "loss": 1.0218,
      "step": 3737
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.380496025085449,
      "learning_rate": 1.8230840494521295e-05,
      "loss": 1.1819,
      "step": 3738
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.796816349029541,
      "learning_rate": 1.8230201578123503e-05,
      "loss": 0.9884,
      "step": 3739
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7294968366622925,
      "learning_rate": 1.8229562661725715e-05,
      "loss": 0.7885,
      "step": 3740
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8139493465423584,
      "learning_rate": 1.8228923745327923e-05,
      "loss": 0.9005,
      "step": 3741
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8933539390563965,
      "learning_rate": 1.8228284828930135e-05,
      "loss": 0.8607,
      "step": 3742
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.35506534576416,
      "learning_rate": 1.8227645912532346e-05,
      "loss": 1.5561,
      "step": 3743
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0469577312469482,
      "learning_rate": 1.8227006996134558e-05,
      "loss": 0.8191,
      "step": 3744
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1240108013153076,
      "learning_rate": 1.8226368079736766e-05,
      "loss": 1.0226,
      "step": 3745
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.132436752319336,
      "learning_rate": 1.8225729163338977e-05,
      "loss": 0.8973,
      "step": 3746
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.915339946746826,
      "learning_rate": 1.822509024694119e-05,
      "loss": 1.4087,
      "step": 3747
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6749893426895142,
      "learning_rate": 1.82244513305434e-05,
      "loss": 0.8121,
      "step": 3748
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9794450998306274,
      "learning_rate": 1.822381241414561e-05,
      "loss": 0.8234,
      "step": 3749
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.107466459274292,
      "learning_rate": 1.822317349774782e-05,
      "loss": 0.8253,
      "step": 3750
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.514404296875,
      "learning_rate": 1.8222534581350032e-05,
      "loss": 1.502,
      "step": 3751
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2313549518585205,
      "learning_rate": 1.8221895664952243e-05,
      "loss": 1.2532,
      "step": 3752
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0209391117095947,
      "learning_rate": 1.822125674855445e-05,
      "loss": 1.0799,
      "step": 3753
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2357265949249268,
      "learning_rate": 1.8220617832156663e-05,
      "loss": 0.5916,
      "step": 3754
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1881637573242188,
      "learning_rate": 1.8219978915758875e-05,
      "loss": 1.1357,
      "step": 3755
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9364700317382812,
      "learning_rate": 1.8219339999361086e-05,
      "loss": 0.9703,
      "step": 3756
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0229761600494385,
      "learning_rate": 1.8218701082963294e-05,
      "loss": 1.0307,
      "step": 3757
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.935824394226074,
      "learning_rate": 1.8218062166565506e-05,
      "loss": 1.4121,
      "step": 3758
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7045392990112305,
      "learning_rate": 1.8217423250167717e-05,
      "loss": 0.838,
      "step": 3759
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7242071628570557,
      "learning_rate": 1.821678433376993e-05,
      "loss": 0.9724,
      "step": 3760
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.622766971588135,
      "learning_rate": 1.8216145417372137e-05,
      "loss": 1.281,
      "step": 3761
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.880216360092163,
      "learning_rate": 1.821550650097435e-05,
      "loss": 0.9068,
      "step": 3762
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.131597518920898,
      "learning_rate": 1.821486758457656e-05,
      "loss": 1.5246,
      "step": 3763
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.211106538772583,
      "learning_rate": 1.821422866817877e-05,
      "loss": 0.8898,
      "step": 3764
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.001763820648193,
      "learning_rate": 1.821358975178098e-05,
      "loss": 0.9962,
      "step": 3765
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.028096914291382,
      "learning_rate": 1.821295083538319e-05,
      "loss": 1.0003,
      "step": 3766
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8682085275650024,
      "learning_rate": 1.8212311918985403e-05,
      "loss": 0.8637,
      "step": 3767
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.074222087860107,
      "learning_rate": 1.8211673002587614e-05,
      "loss": 1.5356,
      "step": 3768
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.965671420097351,
      "learning_rate": 1.8211034086189823e-05,
      "loss": 0.9262,
      "step": 3769
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9850107431411743,
      "learning_rate": 1.8210395169792034e-05,
      "loss": 0.9734,
      "step": 3770
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5077133178710938,
      "learning_rate": 1.8209756253394246e-05,
      "loss": 0.954,
      "step": 3771
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4909589290618896,
      "learning_rate": 1.8209117336996457e-05,
      "loss": 0.9007,
      "step": 3772
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6394448280334473,
      "learning_rate": 1.8208478420598665e-05,
      "loss": 0.8821,
      "step": 3773
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4913793802261353,
      "learning_rate": 1.8207839504200877e-05,
      "loss": 0.4461,
      "step": 3774
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4430623054504395,
      "learning_rate": 1.8207200587803085e-05,
      "loss": 1.3283,
      "step": 3775
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7672449350357056,
      "learning_rate": 1.82065616714053e-05,
      "loss": 0.3853,
      "step": 3776
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6685643196105957,
      "learning_rate": 1.8205922755007508e-05,
      "loss": 0.9134,
      "step": 3777
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2251930236816406,
      "learning_rate": 1.820528383860972e-05,
      "loss": 1.0878,
      "step": 3778
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6232918500900269,
      "learning_rate": 1.8204644922211928e-05,
      "loss": 1.2597,
      "step": 3779
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.644141674041748,
      "learning_rate": 1.820400600581414e-05,
      "loss": 1.3004,
      "step": 3780
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8515630960464478,
      "learning_rate": 1.820336708941635e-05,
      "loss": 0.9363,
      "step": 3781
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.2441234588623047,
      "learning_rate": 1.8202728173018563e-05,
      "loss": 1.1118,
      "step": 3782
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.475237846374512,
      "learning_rate": 1.820208925662077e-05,
      "loss": 1.1813,
      "step": 3783
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6093802452087402,
      "learning_rate": 1.8201450340222982e-05,
      "loss": 0.8451,
      "step": 3784
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.020789623260498,
      "learning_rate": 1.8200811423825194e-05,
      "loss": 0.8788,
      "step": 3785
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9366170167922974,
      "learning_rate": 1.8200172507427405e-05,
      "loss": 0.825,
      "step": 3786
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.653478741645813,
      "learning_rate": 1.8199533591029614e-05,
      "loss": 0.954,
      "step": 3787
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.052370071411133,
      "learning_rate": 1.8198894674631825e-05,
      "loss": 0.9199,
      "step": 3788
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.970066070556641,
      "learning_rate": 1.8198255758234037e-05,
      "loss": 1.5017,
      "step": 3789
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2112975120544434,
      "learning_rate": 1.8197616841836248e-05,
      "loss": 0.894,
      "step": 3790
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2919628620147705,
      "learning_rate": 1.8196977925438456e-05,
      "loss": 0.9559,
      "step": 3791
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.823486804962158,
      "learning_rate": 1.8196339009040668e-05,
      "loss": 1.3079,
      "step": 3792
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.370690107345581,
      "learning_rate": 1.819570009264288e-05,
      "loss": 1.4813,
      "step": 3793
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0596516132354736,
      "learning_rate": 1.819506117624509e-05,
      "loss": 0.7336,
      "step": 3794
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.288677930831909,
      "learning_rate": 1.81944222598473e-05,
      "loss": 0.7485,
      "step": 3795
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6299359798431396,
      "learning_rate": 1.819378334344951e-05,
      "loss": 0.6836,
      "step": 3796
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.108745813369751,
      "learning_rate": 1.8193144427051722e-05,
      "loss": 0.8757,
      "step": 3797
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1866209506988525,
      "learning_rate": 1.8192505510653934e-05,
      "loss": 0.9202,
      "step": 3798
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.33923077583313,
      "learning_rate": 1.8191866594256142e-05,
      "loss": 1.1388,
      "step": 3799
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.8047797679901123,
      "learning_rate": 1.8191227677858353e-05,
      "loss": 0.7119,
      "step": 3800
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.974924087524414,
      "learning_rate": 1.8190588761460565e-05,
      "loss": 1.0686,
      "step": 3801
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6303906440734863,
      "learning_rate": 1.8189949845062777e-05,
      "loss": 0.7552,
      "step": 3802
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.256242036819458,
      "learning_rate": 1.8189310928664985e-05,
      "loss": 0.8937,
      "step": 3803
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9336761236190796,
      "learning_rate": 1.8188672012267196e-05,
      "loss": 0.9021,
      "step": 3804
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.354475975036621,
      "learning_rate": 1.8188033095869408e-05,
      "loss": 1.3949,
      "step": 3805
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.19531512260437,
      "learning_rate": 1.818739417947162e-05,
      "loss": 0.9878,
      "step": 3806
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3331971168518066,
      "learning_rate": 1.8186755263073828e-05,
      "loss": 0.6523,
      "step": 3807
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.338529109954834,
      "learning_rate": 1.818611634667604e-05,
      "loss": 1.5869,
      "step": 3808
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.39510440826416,
      "learning_rate": 1.8185477430278247e-05,
      "loss": 1.311,
      "step": 3809
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.823892593383789,
      "learning_rate": 1.8184838513880462e-05,
      "loss": 0.8527,
      "step": 3810
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.154358386993408,
      "learning_rate": 1.818419959748267e-05,
      "loss": 1.1565,
      "step": 3811
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.8321290016174316,
      "learning_rate": 1.8183560681084882e-05,
      "loss": 0.7618,
      "step": 3812
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.682908773422241,
      "learning_rate": 1.818292176468709e-05,
      "loss": 1.0016,
      "step": 3813
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7490029335021973,
      "learning_rate": 1.81822828482893e-05,
      "loss": 0.8265,
      "step": 3814
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0757758617401123,
      "learning_rate": 1.8181643931891513e-05,
      "loss": 0.9999,
      "step": 3815
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7519776821136475,
      "learning_rate": 1.8181005015493725e-05,
      "loss": 0.9837,
      "step": 3816
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9760771989822388,
      "learning_rate": 1.8180366099095933e-05,
      "loss": 0.8785,
      "step": 3817
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.930633544921875,
      "learning_rate": 1.8179727182698144e-05,
      "loss": 0.7725,
      "step": 3818
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3006186485290527,
      "learning_rate": 1.8179088266300356e-05,
      "loss": 0.9543,
      "step": 3819
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.804962396621704,
      "learning_rate": 1.8178449349902567e-05,
      "loss": 1.0123,
      "step": 3820
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.645322799682617,
      "learning_rate": 1.8177810433504776e-05,
      "loss": 0.831,
      "step": 3821
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1787374019622803,
      "learning_rate": 1.8177171517106987e-05,
      "loss": 0.9316,
      "step": 3822
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9657788276672363,
      "learning_rate": 1.81765326007092e-05,
      "loss": 0.891,
      "step": 3823
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.986390471458435,
      "learning_rate": 1.817589368431141e-05,
      "loss": 0.9524,
      "step": 3824
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8409579992294312,
      "learning_rate": 1.817525476791362e-05,
      "loss": 0.9414,
      "step": 3825
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.262730121612549,
      "learning_rate": 1.817461585151583e-05,
      "loss": 1.0835,
      "step": 3826
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7098915576934814,
      "learning_rate": 1.817397693511804e-05,
      "loss": 1.146,
      "step": 3827
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9560977220535278,
      "learning_rate": 1.8173338018720253e-05,
      "loss": 0.9202,
      "step": 3828
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.445359230041504,
      "learning_rate": 1.817269910232246e-05,
      "loss": 1.3319,
      "step": 3829
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7896249294281006,
      "learning_rate": 1.8172060185924673e-05,
      "loss": 0.8906,
      "step": 3830
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.12229585647583,
      "learning_rate": 1.8171421269526884e-05,
      "loss": 1.2267,
      "step": 3831
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.109236717224121,
      "learning_rate": 1.8170782353129096e-05,
      "loss": 0.9047,
      "step": 3832
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.640958786010742,
      "learning_rate": 1.8170143436731304e-05,
      "loss": 1.655,
      "step": 3833
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0127222537994385,
      "learning_rate": 1.8169504520333516e-05,
      "loss": 0.8718,
      "step": 3834
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1517913341522217,
      "learning_rate": 1.8168865603935727e-05,
      "loss": 1.113,
      "step": 3835
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.743886947631836,
      "learning_rate": 1.816822668753794e-05,
      "loss": 0.8222,
      "step": 3836
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.917025566101074,
      "learning_rate": 1.8167587771140147e-05,
      "loss": 1.3271,
      "step": 3837
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9101344347000122,
      "learning_rate": 1.816694885474236e-05,
      "loss": 0.9884,
      "step": 3838
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.885966181755066,
      "learning_rate": 1.816630993834457e-05,
      "loss": 0.9334,
      "step": 3839
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.367081880569458,
      "learning_rate": 1.816567102194678e-05,
      "loss": 0.8545,
      "step": 3840
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8773900270462036,
      "learning_rate": 1.816503210554899e-05,
      "loss": 0.8806,
      "step": 3841
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.169243097305298,
      "learning_rate": 1.81643931891512e-05,
      "loss": 0.9333,
      "step": 3842
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.140977382659912,
      "learning_rate": 1.8163754272753413e-05,
      "loss": 0.8003,
      "step": 3843
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.7847771644592285,
      "learning_rate": 1.8163115356355624e-05,
      "loss": 1.2266,
      "step": 3844
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.838381290435791,
      "learning_rate": 1.8162476439957832e-05,
      "loss": 0.8536,
      "step": 3845
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.02542781829834,
      "learning_rate": 1.8161837523560044e-05,
      "loss": 0.7995,
      "step": 3846
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7804510593414307,
      "learning_rate": 1.8161198607162252e-05,
      "loss": 0.5088,
      "step": 3847
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.599433183670044,
      "learning_rate": 1.8160559690764467e-05,
      "loss": 0.3632,
      "step": 3848
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9521472454071045,
      "learning_rate": 1.8159920774366675e-05,
      "loss": 0.9679,
      "step": 3849
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.815164089202881,
      "learning_rate": 1.8159281857968887e-05,
      "loss": 1.1551,
      "step": 3850
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8657336235046387,
      "learning_rate": 1.8158642941571095e-05,
      "loss": 0.9736,
      "step": 3851
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.011740207672119,
      "learning_rate": 1.8158004025173306e-05,
      "loss": 0.9542,
      "step": 3852
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.388166904449463,
      "learning_rate": 1.8157365108775518e-05,
      "loss": 0.374,
      "step": 3853
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9685654640197754,
      "learning_rate": 1.815672619237773e-05,
      "loss": 0.9698,
      "step": 3854
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.4100141525268555,
      "learning_rate": 1.8156087275979938e-05,
      "loss": 1.0635,
      "step": 3855
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.520895004272461,
      "learning_rate": 1.815544835958215e-05,
      "loss": 0.9229,
      "step": 3856
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8895084857940674,
      "learning_rate": 1.815480944318436e-05,
      "loss": 0.8082,
      "step": 3857
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.414257049560547,
      "learning_rate": 1.8154170526786572e-05,
      "loss": 0.9221,
      "step": 3858
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5528712272644043,
      "learning_rate": 1.815353161038878e-05,
      "loss": 0.9426,
      "step": 3859
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.10109806060791,
      "learning_rate": 1.8152892693990992e-05,
      "loss": 0.9741,
      "step": 3860
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9961785078048706,
      "learning_rate": 1.8152253777593204e-05,
      "loss": 1.0059,
      "step": 3861
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.018216848373413,
      "learning_rate": 1.8151614861195415e-05,
      "loss": 0.9253,
      "step": 3862
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.156235456466675,
      "learning_rate": 1.8150975944797623e-05,
      "loss": 0.972,
      "step": 3863
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.9118266105651855,
      "learning_rate": 1.8150337028399835e-05,
      "loss": 1.1526,
      "step": 3864
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.165517807006836,
      "learning_rate": 1.8149698112002046e-05,
      "loss": 0.8767,
      "step": 3865
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2570254802703857,
      "learning_rate": 1.8149059195604258e-05,
      "loss": 0.8272,
      "step": 3866
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1865956783294678,
      "learning_rate": 1.8148420279206466e-05,
      "loss": 0.8872,
      "step": 3867
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6024675369262695,
      "learning_rate": 1.8147781362808678e-05,
      "loss": 0.8107,
      "step": 3868
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5023908615112305,
      "learning_rate": 1.814714244641089e-05,
      "loss": 1.1012,
      "step": 3869
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3170106410980225,
      "learning_rate": 1.81465035300131e-05,
      "loss": 1.0586,
      "step": 3870
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.500729560852051,
      "learning_rate": 1.814586461361531e-05,
      "loss": 0.9362,
      "step": 3871
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0832555294036865,
      "learning_rate": 1.814522569721752e-05,
      "loss": 0.8817,
      "step": 3872
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1871378421783447,
      "learning_rate": 1.8144586780819732e-05,
      "loss": 0.9779,
      "step": 3873
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7529553174972534,
      "learning_rate": 1.8143947864421944e-05,
      "loss": 0.914,
      "step": 3874
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7780191898345947,
      "learning_rate": 1.8143308948024152e-05,
      "loss": 0.9523,
      "step": 3875
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4257984161376953,
      "learning_rate": 1.8142670031626363e-05,
      "loss": 0.915,
      "step": 3876
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2243196964263916,
      "learning_rate": 1.8142031115228575e-05,
      "loss": 0.954,
      "step": 3877
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8495850563049316,
      "learning_rate": 1.8141392198830786e-05,
      "loss": 0.9871,
      "step": 3878
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0038247108459473,
      "learning_rate": 1.8140753282432995e-05,
      "loss": 0.4178,
      "step": 3879
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.748234748840332,
      "learning_rate": 1.8140114366035206e-05,
      "loss": 0.9287,
      "step": 3880
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0547854900360107,
      "learning_rate": 1.8139475449637414e-05,
      "loss": 0.9982,
      "step": 3881
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6005218029022217,
      "learning_rate": 1.813883653323963e-05,
      "loss": 0.9475,
      "step": 3882
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3566412925720215,
      "learning_rate": 1.8138197616841837e-05,
      "loss": 1.0263,
      "step": 3883
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.61686897277832,
      "learning_rate": 1.813755870044405e-05,
      "loss": 0.9618,
      "step": 3884
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7511411905288696,
      "learning_rate": 1.8136919784046257e-05,
      "loss": 0.9561,
      "step": 3885
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7336158752441406,
      "learning_rate": 1.8136280867648472e-05,
      "loss": 1.0461,
      "step": 3886
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.230351209640503,
      "learning_rate": 1.813564195125068e-05,
      "loss": 0.7571,
      "step": 3887
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.374104022979736,
      "learning_rate": 1.813500303485289e-05,
      "loss": 1.1749,
      "step": 3888
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.983067750930786,
      "learning_rate": 1.81343641184551e-05,
      "loss": 1.0409,
      "step": 3889
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9515206813812256,
      "learning_rate": 1.813372520205731e-05,
      "loss": 0.5263,
      "step": 3890
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.438436508178711,
      "learning_rate": 1.8133086285659523e-05,
      "loss": 1.2999,
      "step": 3891
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.691555380821228,
      "learning_rate": 1.8132447369261734e-05,
      "loss": 0.3577,
      "step": 3892
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2553744316101074,
      "learning_rate": 1.8131808452863943e-05,
      "loss": 1.0107,
      "step": 3893
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1066205501556396,
      "learning_rate": 1.8131169536466154e-05,
      "loss": 1.1246,
      "step": 3894
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6222779750823975,
      "learning_rate": 1.8130530620068366e-05,
      "loss": 1.0171,
      "step": 3895
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7253541946411133,
      "learning_rate": 1.8129891703670577e-05,
      "loss": 0.9959,
      "step": 3896
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8921765089035034,
      "learning_rate": 1.8129252787272785e-05,
      "loss": 0.9121,
      "step": 3897
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.8732330799102783,
      "learning_rate": 1.8128613870874997e-05,
      "loss": 1.2142,
      "step": 3898
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2804248332977295,
      "learning_rate": 1.812797495447721e-05,
      "loss": 0.6821,
      "step": 3899
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.919019341468811,
      "learning_rate": 1.812733603807942e-05,
      "loss": 0.9341,
      "step": 3900
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.953647136688232,
      "learning_rate": 1.8126697121681628e-05,
      "loss": 1.2863,
      "step": 3901
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.648763656616211,
      "learning_rate": 1.812605820528384e-05,
      "loss": 1.3107,
      "step": 3902
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.686291694641113,
      "learning_rate": 1.812541928888605e-05,
      "loss": 1.0057,
      "step": 3903
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0623278617858887,
      "learning_rate": 1.8124780372488263e-05,
      "loss": 0.9515,
      "step": 3904
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3190810680389404,
      "learning_rate": 1.812414145609047e-05,
      "loss": 0.964,
      "step": 3905
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7811434268951416,
      "learning_rate": 1.8123502539692683e-05,
      "loss": 1.4303,
      "step": 3906
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.242266893386841,
      "learning_rate": 1.8122863623294894e-05,
      "loss": 0.8314,
      "step": 3907
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8379933834075928,
      "learning_rate": 1.8122224706897106e-05,
      "loss": 0.8981,
      "step": 3908
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4353013038635254,
      "learning_rate": 1.8121585790499314e-05,
      "loss": 1.0459,
      "step": 3909
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1387033462524414,
      "learning_rate": 1.8120946874101525e-05,
      "loss": 0.9285,
      "step": 3910
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9841313362121582,
      "learning_rate": 1.8120307957703737e-05,
      "loss": 0.9138,
      "step": 3911
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.323944091796875,
      "learning_rate": 1.811966904130595e-05,
      "loss": 0.725,
      "step": 3912
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.022588014602661,
      "learning_rate": 1.8119030124908157e-05,
      "loss": 0.8107,
      "step": 3913
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.446131706237793,
      "learning_rate": 1.8118391208510368e-05,
      "loss": 0.8163,
      "step": 3914
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.260047197341919,
      "learning_rate": 1.811775229211258e-05,
      "loss": 0.8812,
      "step": 3915
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.619595527648926,
      "learning_rate": 1.811711337571479e-05,
      "loss": 2.2539,
      "step": 3916
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1753690242767334,
      "learning_rate": 1.8116474459317e-05,
      "loss": 0.9557,
      "step": 3917
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3104093074798584,
      "learning_rate": 1.811583554291921e-05,
      "loss": 1.0329,
      "step": 3918
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.180870056152344,
      "learning_rate": 1.811519662652142e-05,
      "loss": 1.0502,
      "step": 3919
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.033047676086426,
      "learning_rate": 1.8114557710123634e-05,
      "loss": 1.4449,
      "step": 3920
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.505342960357666,
      "learning_rate": 1.8113918793725842e-05,
      "loss": 1.3245,
      "step": 3921
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.136878490447998,
      "learning_rate": 1.8113279877328054e-05,
      "loss": 0.917,
      "step": 3922
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1096901893615723,
      "learning_rate": 1.8112640960930262e-05,
      "loss": 1.4274,
      "step": 3923
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.43171501159668,
      "learning_rate": 1.8112002044532473e-05,
      "loss": 1.4873,
      "step": 3924
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.2323384284973145,
      "learning_rate": 1.8111363128134685e-05,
      "loss": 0.602,
      "step": 3925
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6790406703948975,
      "learning_rate": 1.8110724211736897e-05,
      "loss": 1.0967,
      "step": 3926
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.8594720363616943,
      "learning_rate": 1.8110085295339105e-05,
      "loss": 1.3109,
      "step": 3927
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2485787868499756,
      "learning_rate": 1.8109446378941316e-05,
      "loss": 0.9524,
      "step": 3928
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.135132312774658,
      "learning_rate": 1.8108807462543528e-05,
      "loss": 1.0252,
      "step": 3929
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.292874813079834,
      "learning_rate": 1.810816854614574e-05,
      "loss": 1.0459,
      "step": 3930
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.785853147506714,
      "learning_rate": 1.8107529629747948e-05,
      "loss": 0.6912,
      "step": 3931
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0338385105133057,
      "learning_rate": 1.810689071335016e-05,
      "loss": 0.8562,
      "step": 3932
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.707195997238159,
      "learning_rate": 1.810625179695237e-05,
      "loss": 1.2221,
      "step": 3933
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4820613861083984,
      "learning_rate": 1.8105612880554582e-05,
      "loss": 1.0662,
      "step": 3934
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.670917272567749,
      "learning_rate": 1.810497396415679e-05,
      "loss": 0.8135,
      "step": 3935
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8292789459228516,
      "learning_rate": 1.8104335047759002e-05,
      "loss": 0.9995,
      "step": 3936
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4385526180267334,
      "learning_rate": 1.8103696131361213e-05,
      "loss": 0.9351,
      "step": 3937
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.143730401992798,
      "learning_rate": 1.8103057214963425e-05,
      "loss": 0.869,
      "step": 3938
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.126939058303833,
      "learning_rate": 1.8102418298565633e-05,
      "loss": 0.814,
      "step": 3939
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0987706184387207,
      "learning_rate": 1.8101779382167845e-05,
      "loss": 0.8778,
      "step": 3940
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5575830936431885,
      "learning_rate": 1.8101140465770056e-05,
      "loss": 0.9849,
      "step": 3941
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.409205913543701,
      "learning_rate": 1.8100501549372268e-05,
      "loss": 0.9331,
      "step": 3942
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.652177333831787,
      "learning_rate": 1.8099862632974476e-05,
      "loss": 0.3097,
      "step": 3943
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0504674911499023,
      "learning_rate": 1.8099223716576687e-05,
      "loss": 0.8845,
      "step": 3944
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.333702325820923,
      "learning_rate": 1.80985848001789e-05,
      "loss": 0.9887,
      "step": 3945
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.220580577850342,
      "learning_rate": 1.809794588378111e-05,
      "loss": 1.0055,
      "step": 3946
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.996968150138855,
      "learning_rate": 1.809730696738332e-05,
      "loss": 0.8835,
      "step": 3947
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9002777338027954,
      "learning_rate": 1.809666805098553e-05,
      "loss": 0.9794,
      "step": 3948
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.011012077331543,
      "learning_rate": 1.8096029134587742e-05,
      "loss": 0.8405,
      "step": 3949
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.858822345733643,
      "learning_rate": 1.8095390218189953e-05,
      "loss": 1.5634,
      "step": 3950
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.149097919464111,
      "learning_rate": 1.809475130179216e-05,
      "loss": 1.0913,
      "step": 3951
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.200770378112793,
      "learning_rate": 1.8094112385394373e-05,
      "loss": 0.9544,
      "step": 3952
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7200677394866943,
      "learning_rate": 1.8093473468996585e-05,
      "loss": 1.1009,
      "step": 3953
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8333461284637451,
      "learning_rate": 1.8092834552598796e-05,
      "loss": 0.9048,
      "step": 3954
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.422584056854248,
      "learning_rate": 1.8092195636201004e-05,
      "loss": 1.0337,
      "step": 3955
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.673441171646118,
      "learning_rate": 1.8091556719803216e-05,
      "loss": 1.0131,
      "step": 3956
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4477286338806152,
      "learning_rate": 1.8090917803405424e-05,
      "loss": 0.9052,
      "step": 3957
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.695598125457764,
      "learning_rate": 1.809027888700764e-05,
      "loss": 1.0421,
      "step": 3958
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5290441513061523,
      "learning_rate": 1.8089639970609847e-05,
      "loss": 1.2849,
      "step": 3959
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.888442873954773,
      "learning_rate": 1.808900105421206e-05,
      "loss": 1.0241,
      "step": 3960
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.805205821990967,
      "learning_rate": 1.8088362137814267e-05,
      "loss": 1.3103,
      "step": 3961
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3430209159851074,
      "learning_rate": 1.808772322141648e-05,
      "loss": 0.6369,
      "step": 3962
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5438077449798584,
      "learning_rate": 1.808708430501869e-05,
      "loss": 0.6949,
      "step": 3963
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6283173561096191,
      "learning_rate": 1.80864453886209e-05,
      "loss": 0.8713,
      "step": 3964
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9979029893875122,
      "learning_rate": 1.808580647222311e-05,
      "loss": 0.9194,
      "step": 3965
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1865930557250977,
      "learning_rate": 1.808516755582532e-05,
      "loss": 1.0976,
      "step": 3966
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.409299612045288,
      "learning_rate": 1.8084528639427533e-05,
      "loss": 0.884,
      "step": 3967
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.553718090057373,
      "learning_rate": 1.8083889723029744e-05,
      "loss": 1.3257,
      "step": 3968
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1667163372039795,
      "learning_rate": 1.8083250806631952e-05,
      "loss": 0.7651,
      "step": 3969
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5974512100219727,
      "learning_rate": 1.8082611890234164e-05,
      "loss": 1.0299,
      "step": 3970
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.949954628944397,
      "learning_rate": 1.8081972973836376e-05,
      "loss": 0.97,
      "step": 3971
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7353224754333496,
      "learning_rate": 1.8081334057438587e-05,
      "loss": 1.1347,
      "step": 3972
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.581190347671509,
      "learning_rate": 1.8080695141040795e-05,
      "loss": 1.0397,
      "step": 3973
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5613105297088623,
      "learning_rate": 1.8080056224643007e-05,
      "loss": 0.8931,
      "step": 3974
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0638327598571777,
      "learning_rate": 1.807941730824522e-05,
      "loss": 0.9554,
      "step": 3975
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7023065090179443,
      "learning_rate": 1.807877839184743e-05,
      "loss": 0.8135,
      "step": 3976
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9877309799194336,
      "learning_rate": 1.8078139475449638e-05,
      "loss": 0.9462,
      "step": 3977
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.339700698852539,
      "learning_rate": 1.807750055905185e-05,
      "loss": 0.9943,
      "step": 3978
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9201174974441528,
      "learning_rate": 1.807686164265406e-05,
      "loss": 0.8944,
      "step": 3979
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.443821668624878,
      "learning_rate": 1.8076222726256273e-05,
      "loss": 1.2494,
      "step": 3980
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.239234447479248,
      "learning_rate": 1.807558380985848e-05,
      "loss": 1.0226,
      "step": 3981
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0033490657806396,
      "learning_rate": 1.8074944893460692e-05,
      "loss": 0.9807,
      "step": 3982
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7107553482055664,
      "learning_rate": 1.8074305977062904e-05,
      "loss": 0.9876,
      "step": 3983
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9475719928741455,
      "learning_rate": 1.8073667060665115e-05,
      "loss": 0.9777,
      "step": 3984
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8886027336120605,
      "learning_rate": 1.8073028144267324e-05,
      "loss": 0.9268,
      "step": 3985
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8931413888931274,
      "learning_rate": 1.8072389227869535e-05,
      "loss": 0.8779,
      "step": 3986
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.110715866088867,
      "learning_rate": 1.8071750311471747e-05,
      "loss": 0.7976,
      "step": 3987
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3246655464172363,
      "learning_rate": 1.8071111395073958e-05,
      "loss": 0.8034,
      "step": 3988
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6972765922546387,
      "learning_rate": 1.8070472478676166e-05,
      "loss": 0.8915,
      "step": 3989
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4095351696014404,
      "learning_rate": 1.8069833562278378e-05,
      "loss": 1.1096,
      "step": 3990
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1002492904663086,
      "learning_rate": 1.8069194645880586e-05,
      "loss": 0.9686,
      "step": 3991
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1438722610473633,
      "learning_rate": 1.80685557294828e-05,
      "loss": 0.843,
      "step": 3992
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8078579902648926,
      "learning_rate": 1.806791681308501e-05,
      "loss": 0.4285,
      "step": 3993
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3472342491149902,
      "learning_rate": 1.806727789668722e-05,
      "loss": 0.8084,
      "step": 3994
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5301060676574707,
      "learning_rate": 1.806663898028943e-05,
      "loss": 0.9669,
      "step": 3995
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8250240087509155,
      "learning_rate": 1.8066000063891644e-05,
      "loss": 0.9778,
      "step": 3996
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8446630239486694,
      "learning_rate": 1.8065361147493852e-05,
      "loss": 0.9227,
      "step": 3997
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.304030179977417,
      "learning_rate": 1.8064722231096064e-05,
      "loss": 1.0065,
      "step": 3998
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1207447052001953,
      "learning_rate": 1.8064083314698272e-05,
      "loss": 0.9544,
      "step": 3999
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1600255966186523,
      "learning_rate": 1.8063444398300483e-05,
      "loss": 1.0723,
      "step": 4000
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.0507652759552,
      "learning_rate": 1.8062805481902695e-05,
      "loss": 0.9295,
      "step": 4001
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.246624946594238,
      "learning_rate": 1.8062166565504906e-05,
      "loss": 1.6954,
      "step": 4002
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.904010772705078,
      "learning_rate": 1.8061527649107115e-05,
      "loss": 1.5568,
      "step": 4003
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4492292404174805,
      "learning_rate": 1.8060888732709326e-05,
      "loss": 1.1202,
      "step": 4004
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4127042293548584,
      "learning_rate": 1.8060249816311538e-05,
      "loss": 1.0835,
      "step": 4005
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8113576173782349,
      "learning_rate": 1.805961089991375e-05,
      "loss": 0.9381,
      "step": 4006
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9060251712799072,
      "learning_rate": 1.8058971983515957e-05,
      "loss": 0.928,
      "step": 4007
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.902251958847046,
      "learning_rate": 1.805833306711817e-05,
      "loss": 0.9283,
      "step": 4008
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6601027250289917,
      "learning_rate": 1.805769415072038e-05,
      "loss": 0.3668,
      "step": 4009
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1823134422302246,
      "learning_rate": 1.8057055234322592e-05,
      "loss": 0.8707,
      "step": 4010
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5912290811538696,
      "learning_rate": 1.80564163179248e-05,
      "loss": 0.3307,
      "step": 4011
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.200188636779785,
      "learning_rate": 1.805577740152701e-05,
      "loss": 1.0599,
      "step": 4012
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8056244850158691,
      "learning_rate": 1.8055138485129223e-05,
      "loss": 0.8646,
      "step": 4013
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3199093341827393,
      "learning_rate": 1.8054499568731435e-05,
      "loss": 0.8577,
      "step": 4014
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4094550609588623,
      "learning_rate": 1.8053860652333643e-05,
      "loss": 0.878,
      "step": 4015
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.808178424835205,
      "learning_rate": 1.8053221735935854e-05,
      "loss": 1.1862,
      "step": 4016
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8087109327316284,
      "learning_rate": 1.8052582819538066e-05,
      "loss": 0.9059,
      "step": 4017
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4042491912841797,
      "learning_rate": 1.8051943903140278e-05,
      "loss": 1.0095,
      "step": 4018
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.032186985015869,
      "learning_rate": 1.8051304986742486e-05,
      "loss": 0.9832,
      "step": 4019
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8647853136062622,
      "learning_rate": 1.8050666070344697e-05,
      "loss": 0.9463,
      "step": 4020
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.049546003341675,
      "learning_rate": 1.805002715394691e-05,
      "loss": 1.285,
      "step": 4021
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2409005165100098,
      "learning_rate": 1.804938823754912e-05,
      "loss": 0.8989,
      "step": 4022
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7474989891052246,
      "learning_rate": 1.804874932115133e-05,
      "loss": 1.4541,
      "step": 4023
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9915217161178589,
      "learning_rate": 1.804811040475354e-05,
      "loss": 0.8995,
      "step": 4024
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.606420040130615,
      "learning_rate": 1.804747148835575e-05,
      "loss": 1.6721,
      "step": 4025
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7281570434570312,
      "learning_rate": 1.8046832571957963e-05,
      "loss": 0.8599,
      "step": 4026
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0002150535583496,
      "learning_rate": 1.804619365556017e-05,
      "loss": 0.8437,
      "step": 4027
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.885533332824707,
      "learning_rate": 1.8045554739162383e-05,
      "loss": 0.9107,
      "step": 4028
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.078749418258667,
      "learning_rate": 1.804491582276459e-05,
      "loss": 0.9003,
      "step": 4029
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.382111072540283,
      "learning_rate": 1.8044276906366806e-05,
      "loss": 1.3743,
      "step": 4030
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8922033309936523,
      "learning_rate": 1.8043637989969014e-05,
      "loss": 0.8461,
      "step": 4031
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.4006049633026123,
      "learning_rate": 1.8042999073571226e-05,
      "loss": 0.7247,
      "step": 4032
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.241305351257324,
      "learning_rate": 1.8042360157173434e-05,
      "loss": 0.8055,
      "step": 4033
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.772927284240723,
      "learning_rate": 1.8041721240775645e-05,
      "loss": 1.7247,
      "step": 4034
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5165178775787354,
      "learning_rate": 1.8041082324377857e-05,
      "loss": 0.8467,
      "step": 4035
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.487222671508789,
      "learning_rate": 1.804044340798007e-05,
      "loss": 1.2438,
      "step": 4036
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5475428104400635,
      "learning_rate": 1.8039804491582277e-05,
      "loss": 0.9313,
      "step": 4037
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2852978706359863,
      "learning_rate": 1.8039165575184488e-05,
      "loss": 1.0315,
      "step": 4038
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.002103805541992,
      "learning_rate": 1.80385266587867e-05,
      "loss": 0.9673,
      "step": 4039
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.326790809631348,
      "learning_rate": 1.803788774238891e-05,
      "loss": 1.3891,
      "step": 4040
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.937024712562561,
      "learning_rate": 1.803724882599112e-05,
      "loss": 0.8726,
      "step": 4041
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3273684978485107,
      "learning_rate": 1.803660990959333e-05,
      "loss": 1.033,
      "step": 4042
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.551341533660889,
      "learning_rate": 1.8035970993195543e-05,
      "loss": 1.2612,
      "step": 4043
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.203831434249878,
      "learning_rate": 1.8035332076797754e-05,
      "loss": 0.943,
      "step": 4044
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1175198554992676,
      "learning_rate": 1.8034693160399962e-05,
      "loss": 0.3196,
      "step": 4045
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0740344524383545,
      "learning_rate": 1.8034054244002174e-05,
      "loss": 1.1167,
      "step": 4046
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6995055675506592,
      "learning_rate": 1.8033415327604385e-05,
      "loss": 0.3778,
      "step": 4047
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.003896713256836,
      "learning_rate": 1.8032776411206597e-05,
      "loss": 1.9869,
      "step": 4048
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5116400718688965,
      "learning_rate": 1.8032137494808805e-05,
      "loss": 0.4059,
      "step": 4049
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.074239492416382,
      "learning_rate": 1.8031498578411017e-05,
      "loss": 0.9925,
      "step": 4050
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.562626361846924,
      "learning_rate": 1.8030859662013228e-05,
      "loss": 0.9077,
      "step": 4051
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7271490097045898,
      "learning_rate": 1.803022074561544e-05,
      "loss": 0.9281,
      "step": 4052
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9061174392700195,
      "learning_rate": 1.8029581829217648e-05,
      "loss": 0.9688,
      "step": 4053
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.490224599838257,
      "learning_rate": 1.802894291281986e-05,
      "loss": 1.2001,
      "step": 4054
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7766103744506836,
      "learning_rate": 1.802830399642207e-05,
      "loss": 0.9352,
      "step": 4055
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3202455043792725,
      "learning_rate": 1.8027665080024282e-05,
      "loss": 1.0346,
      "step": 4056
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5205305814743042,
      "learning_rate": 1.802702616362649e-05,
      "loss": 0.3425,
      "step": 4057
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.04549503326416,
      "learning_rate": 1.8026387247228702e-05,
      "loss": 1.0507,
      "step": 4058
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8093043565750122,
      "learning_rate": 1.8025748330830914e-05,
      "loss": 0.9451,
      "step": 4059
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2055721282958984,
      "learning_rate": 1.8025109414433125e-05,
      "loss": 1.1482,
      "step": 4060
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.097034215927124,
      "learning_rate": 1.8024470498035333e-05,
      "loss": 1.0321,
      "step": 4061
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.260301351547241,
      "learning_rate": 1.802383158163754e-05,
      "loss": 0.8718,
      "step": 4062
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0195846557617188,
      "learning_rate": 1.8023192665239757e-05,
      "loss": 0.9007,
      "step": 4063
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.871911883354187,
      "learning_rate": 1.8022553748841965e-05,
      "loss": 0.916,
      "step": 4064
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9611542224884033,
      "learning_rate": 1.8021914832444176e-05,
      "loss": 0.9309,
      "step": 4065
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7391608953475952,
      "learning_rate": 1.8021275916046384e-05,
      "loss": 0.8593,
      "step": 4066
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.143954277038574,
      "learning_rate": 1.8020636999648596e-05,
      "loss": 0.9631,
      "step": 4067
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5959510803222656,
      "learning_rate": 1.8019998083250807e-05,
      "loss": 0.9473,
      "step": 4068
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8459750413894653,
      "learning_rate": 1.801935916685302e-05,
      "loss": 0.9511,
      "step": 4069
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.932417869567871,
      "learning_rate": 1.8018720250455227e-05,
      "loss": 1.467,
      "step": 4070
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.4788830280303955,
      "learning_rate": 1.801808133405744e-05,
      "loss": 1.385,
      "step": 4071
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8090784549713135,
      "learning_rate": 1.801744241765965e-05,
      "loss": 0.9981,
      "step": 4072
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.2288432121276855,
      "learning_rate": 1.8016803501261862e-05,
      "loss": 0.6784,
      "step": 4073
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.401160955429077,
      "learning_rate": 1.801616458486407e-05,
      "loss": 1.2036,
      "step": 4074
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1127445697784424,
      "learning_rate": 1.801552566846628e-05,
      "loss": 0.9416,
      "step": 4075
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4239654541015625,
      "learning_rate": 1.8014886752068493e-05,
      "loss": 1.1865,
      "step": 4076
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.075138568878174,
      "learning_rate": 1.8014247835670705e-05,
      "loss": 0.9838,
      "step": 4077
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.645785331726074,
      "learning_rate": 1.8013608919272913e-05,
      "loss": 0.9477,
      "step": 4078
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.619982123374939,
      "learning_rate": 1.8012970002875124e-05,
      "loss": 0.934,
      "step": 4079
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9237653017044067,
      "learning_rate": 1.8012331086477336e-05,
      "loss": 0.9262,
      "step": 4080
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7327260971069336,
      "learning_rate": 1.8011692170079547e-05,
      "loss": 0.8757,
      "step": 4081
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3174068927764893,
      "learning_rate": 1.8011053253681756e-05,
      "loss": 0.3582,
      "step": 4082
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.233417510986328,
      "learning_rate": 1.8010414337283967e-05,
      "loss": 1.5127,
      "step": 4083
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9677395820617676,
      "learning_rate": 1.800977542088618e-05,
      "loss": 1.093,
      "step": 4084
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8324116468429565,
      "learning_rate": 1.800913650448839e-05,
      "loss": 0.9532,
      "step": 4085
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.262821674346924,
      "learning_rate": 1.80084975880906e-05,
      "loss": 1.0338,
      "step": 4086
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.519679069519043,
      "learning_rate": 1.800785867169281e-05,
      "loss": 1.0012,
      "step": 4087
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.956332206726074,
      "learning_rate": 1.800721975529502e-05,
      "loss": 0.9442,
      "step": 4088
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.338953971862793,
      "learning_rate": 1.8006580838897233e-05,
      "loss": 0.9569,
      "step": 4089
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.975981593132019,
      "learning_rate": 1.800594192249944e-05,
      "loss": 0.9338,
      "step": 4090
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5573023557662964,
      "learning_rate": 1.8005303006101653e-05,
      "loss": 0.9061,
      "step": 4091
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.850947380065918,
      "learning_rate": 1.8004664089703864e-05,
      "loss": 0.9264,
      "step": 4092
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.448526620864868,
      "learning_rate": 1.8004025173306076e-05,
      "loss": 0.544,
      "step": 4093
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.4328787326812744,
      "learning_rate": 1.8003386256908284e-05,
      "loss": 0.9374,
      "step": 4094
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4438815116882324,
      "learning_rate": 1.8002747340510496e-05,
      "loss": 0.8566,
      "step": 4095
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9748791456222534,
      "learning_rate": 1.8002108424112704e-05,
      "loss": 0.993,
      "step": 4096
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.386690139770508,
      "learning_rate": 1.800146950771492e-05,
      "loss": 1.0838,
      "step": 4097
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1402969360351562,
      "learning_rate": 1.8000830591317127e-05,
      "loss": 0.9817,
      "step": 4098
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.123481273651123,
      "learning_rate": 1.800019167491934e-05,
      "loss": 1.2141,
      "step": 4099
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.440894603729248,
      "learning_rate": 1.7999552758521547e-05,
      "loss": 1.0367,
      "step": 4100
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7558212280273438,
      "learning_rate": 1.7998913842123758e-05,
      "loss": 1.1376,
      "step": 4101
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4553247690200806,
      "learning_rate": 1.799827492572597e-05,
      "loss": 0.3123,
      "step": 4102
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.581752061843872,
      "learning_rate": 1.799763600932818e-05,
      "loss": 1.314,
      "step": 4103
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.988076090812683,
      "learning_rate": 1.799699709293039e-05,
      "loss": 0.9837,
      "step": 4104
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.447181224822998,
      "learning_rate": 1.79963581765326e-05,
      "loss": 1.1883,
      "step": 4105
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9364702701568604,
      "learning_rate": 1.7995719260134812e-05,
      "loss": 0.9036,
      "step": 4106
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.1210668087005615,
      "learning_rate": 1.7995080343737024e-05,
      "loss": 0.9547,
      "step": 4107
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.657075047492981,
      "learning_rate": 1.7994441427339232e-05,
      "loss": 0.8921,
      "step": 4108
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.213796854019165,
      "learning_rate": 1.7993802510941444e-05,
      "loss": 1.113,
      "step": 4109
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.115016460418701,
      "learning_rate": 1.7993163594543655e-05,
      "loss": 1.2018,
      "step": 4110
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1971373558044434,
      "learning_rate": 1.7992524678145867e-05,
      "loss": 0.9944,
      "step": 4111
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.873905181884766,
      "learning_rate": 1.7991885761748075e-05,
      "loss": 1.3348,
      "step": 4112
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9219127893447876,
      "learning_rate": 1.7991246845350286e-05,
      "loss": 0.8954,
      "step": 4113
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.2953834533691406,
      "learning_rate": 1.7990607928952498e-05,
      "loss": 0.6522,
      "step": 4114
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9276072978973389,
      "learning_rate": 1.798996901255471e-05,
      "loss": 0.9921,
      "step": 4115
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7529795169830322,
      "learning_rate": 1.7989330096156918e-05,
      "loss": 1.0363,
      "step": 4116
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.046188831329346,
      "learning_rate": 1.798869117975913e-05,
      "loss": 1.5324,
      "step": 4117
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6762568950653076,
      "learning_rate": 1.798805226336134e-05,
      "loss": 0.9254,
      "step": 4118
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.248411655426025,
      "learning_rate": 1.7987413346963552e-05,
      "loss": 1.4576,
      "step": 4119
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.800467014312744,
      "learning_rate": 1.798677443056576e-05,
      "loss": 1.0616,
      "step": 4120
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.777981758117676,
      "learning_rate": 1.7986135514167972e-05,
      "loss": 1.0653,
      "step": 4121
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1534581184387207,
      "learning_rate": 1.7985496597770184e-05,
      "loss": 1.1299,
      "step": 4122
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0429391860961914,
      "learning_rate": 1.7984857681372395e-05,
      "loss": 0.9337,
      "step": 4123
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.223536729812622,
      "learning_rate": 1.7984218764974603e-05,
      "loss": 0.8457,
      "step": 4124
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0693676471710205,
      "learning_rate": 1.7983579848576815e-05,
      "loss": 1.0105,
      "step": 4125
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6185191869735718,
      "learning_rate": 1.7982940932179026e-05,
      "loss": 0.9456,
      "step": 4126
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.381260395050049,
      "learning_rate": 1.7982302015781238e-05,
      "loss": 1.104,
      "step": 4127
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.924172043800354,
      "learning_rate": 1.7981663099383446e-05,
      "loss": 0.8809,
      "step": 4128
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9462918043136597,
      "learning_rate": 1.7981024182985658e-05,
      "loss": 0.7706,
      "step": 4129
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.94804048538208,
      "learning_rate": 1.798038526658787e-05,
      "loss": 1.348,
      "step": 4130
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4207005500793457,
      "learning_rate": 1.797974635019008e-05,
      "loss": 1.0892,
      "step": 4131
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4180870056152344,
      "learning_rate": 1.797910743379229e-05,
      "loss": 1.0442,
      "step": 4132
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.146380424499512,
      "learning_rate": 1.79784685173945e-05,
      "loss": 0.8653,
      "step": 4133
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6743444204330444,
      "learning_rate": 1.797782960099671e-05,
      "loss": 0.4661,
      "step": 4134
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9342609643936157,
      "learning_rate": 1.7977190684598924e-05,
      "loss": 0.9089,
      "step": 4135
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7215924263000488,
      "learning_rate": 1.7976551768201132e-05,
      "loss": 0.4395,
      "step": 4136
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.180903911590576,
      "learning_rate": 1.7975912851803343e-05,
      "loss": 0.6711,
      "step": 4137
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.116053819656372,
      "learning_rate": 1.797527393540555e-05,
      "loss": 0.8163,
      "step": 4138
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6475263833999634,
      "learning_rate": 1.7974635019007763e-05,
      "loss": 0.9282,
      "step": 4139
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.835670232772827,
      "learning_rate": 1.7973996102609975e-05,
      "loss": 1.3121,
      "step": 4140
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0185437202453613,
      "learning_rate": 1.7973357186212186e-05,
      "loss": 0.9298,
      "step": 4141
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.411710262298584,
      "learning_rate": 1.7972718269814394e-05,
      "loss": 0.9206,
      "step": 4142
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1976871490478516,
      "learning_rate": 1.7972079353416606e-05,
      "loss": 0.9251,
      "step": 4143
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4002419710159302,
      "learning_rate": 1.7971440437018817e-05,
      "loss": 0.373,
      "step": 4144
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9541218280792236,
      "learning_rate": 1.797080152062103e-05,
      "loss": 0.9039,
      "step": 4145
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4141831398010254,
      "learning_rate": 1.7970162604223237e-05,
      "loss": 1.3515,
      "step": 4146
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.058356523513794,
      "learning_rate": 1.796952368782545e-05,
      "loss": 0.9459,
      "step": 4147
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.779131293296814,
      "learning_rate": 1.796888477142766e-05,
      "loss": 0.8979,
      "step": 4148
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9468214511871338,
      "learning_rate": 1.796824585502987e-05,
      "loss": 1.3865,
      "step": 4149
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.902732253074646,
      "learning_rate": 1.796760693863208e-05,
      "loss": 0.971,
      "step": 4150
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.496809720993042,
      "learning_rate": 1.796696802223429e-05,
      "loss": 0.9326,
      "step": 4151
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.09078049659729,
      "learning_rate": 1.7966329105836503e-05,
      "loss": 0.8682,
      "step": 4152
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6979634761810303,
      "learning_rate": 1.7965690189438714e-05,
      "loss": 0.959,
      "step": 4153
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6701600551605225,
      "learning_rate": 1.7965051273040923e-05,
      "loss": 0.9381,
      "step": 4154
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.026172399520874,
      "learning_rate": 1.7964412356643134e-05,
      "loss": 0.8111,
      "step": 4155
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.586867094039917,
      "learning_rate": 1.7963773440245346e-05,
      "loss": 0.9865,
      "step": 4156
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.641577959060669,
      "learning_rate": 1.7963134523847557e-05,
      "loss": 0.4787,
      "step": 4157
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9518615007400513,
      "learning_rate": 1.7962495607449765e-05,
      "loss": 0.9736,
      "step": 4158
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.870121955871582,
      "learning_rate": 1.7961856691051977e-05,
      "loss": 1.63,
      "step": 4159
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.432897567749023,
      "learning_rate": 1.796121777465419e-05,
      "loss": 1.4653,
      "step": 4160
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.376126527786255,
      "learning_rate": 1.79605788582564e-05,
      "loss": 1.0355,
      "step": 4161
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.550896406173706,
      "learning_rate": 1.7959939941858608e-05,
      "loss": 0.8844,
      "step": 4162
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.966223120689392,
      "learning_rate": 1.795930102546082e-05,
      "loss": 1.0777,
      "step": 4163
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8977115154266357,
      "learning_rate": 1.795866210906303e-05,
      "loss": 0.8396,
      "step": 4164
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9135215282440186,
      "learning_rate": 1.7958023192665243e-05,
      "loss": 0.9251,
      "step": 4165
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2660722732543945,
      "learning_rate": 1.795738427626745e-05,
      "loss": 1.0755,
      "step": 4166
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8845256567001343,
      "learning_rate": 1.7956745359869663e-05,
      "loss": 0.86,
      "step": 4167
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.110185146331787,
      "learning_rate": 1.795610644347187e-05,
      "loss": 1.2757,
      "step": 4168
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3536486625671387,
      "learning_rate": 1.7955467527074086e-05,
      "loss": 1.1902,
      "step": 4169
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.012890100479126,
      "learning_rate": 1.7954828610676294e-05,
      "loss": 0.928,
      "step": 4170
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.037006855010986,
      "learning_rate": 1.7954189694278505e-05,
      "loss": 1.2208,
      "step": 4171
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.723720073699951,
      "learning_rate": 1.7953550777880714e-05,
      "loss": 1.4535,
      "step": 4172
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.129293918609619,
      "learning_rate": 1.7952911861482925e-05,
      "loss": 0.957,
      "step": 4173
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.019098281860352,
      "learning_rate": 1.7952272945085137e-05,
      "loss": 1.66,
      "step": 4174
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.9592764377593994,
      "learning_rate": 1.7951634028687348e-05,
      "loss": 1.5203,
      "step": 4175
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1292154788970947,
      "learning_rate": 1.7950995112289556e-05,
      "loss": 0.8598,
      "step": 4176
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7052276134490967,
      "learning_rate": 1.7950356195891768e-05,
      "loss": 0.9395,
      "step": 4177
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.349065065383911,
      "learning_rate": 1.794971727949398e-05,
      "loss": 0.8127,
      "step": 4178
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5512733459472656,
      "learning_rate": 1.794907836309619e-05,
      "loss": 0.8269,
      "step": 4179
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.435318946838379,
      "learning_rate": 1.79484394466984e-05,
      "loss": 1.5739,
      "step": 4180
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9411084651947021,
      "learning_rate": 1.794780053030061e-05,
      "loss": 0.8891,
      "step": 4181
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.158576011657715,
      "learning_rate": 1.7947161613902822e-05,
      "loss": 1.001,
      "step": 4182
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8861156702041626,
      "learning_rate": 1.7946522697505034e-05,
      "loss": 0.9331,
      "step": 4183
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4742562770843506,
      "learning_rate": 1.7945883781107242e-05,
      "loss": 0.8567,
      "step": 4184
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.214125156402588,
      "learning_rate": 1.7945244864709453e-05,
      "loss": 0.9536,
      "step": 4185
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.735701084136963,
      "learning_rate": 1.7944605948311665e-05,
      "loss": 1.4111,
      "step": 4186
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.279323101043701,
      "learning_rate": 1.7943967031913877e-05,
      "loss": 0.9233,
      "step": 4187
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.332491636276245,
      "learning_rate": 1.7943328115516085e-05,
      "loss": 1.068,
      "step": 4188
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3110201358795166,
      "learning_rate": 1.7942689199118296e-05,
      "loss": 1.2437,
      "step": 4189
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.9805707931518555,
      "learning_rate": 1.7942050282720508e-05,
      "loss": 1.2826,
      "step": 4190
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.00240421295166,
      "learning_rate": 1.794141136632272e-05,
      "loss": 1.2646,
      "step": 4191
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.533231735229492,
      "learning_rate": 1.7940772449924928e-05,
      "loss": 1.2318,
      "step": 4192
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.906728982925415,
      "learning_rate": 1.794013353352714e-05,
      "loss": 1.0072,
      "step": 4193
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8357362747192383,
      "learning_rate": 1.793949461712935e-05,
      "loss": 0.8652,
      "step": 4194
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.098702907562256,
      "learning_rate": 1.7938855700731562e-05,
      "loss": 1.4207,
      "step": 4195
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.061864137649536,
      "learning_rate": 1.793821678433377e-05,
      "loss": 1.1258,
      "step": 4196
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7882450819015503,
      "learning_rate": 1.7937577867935982e-05,
      "loss": 0.9205,
      "step": 4197
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.929834246635437,
      "learning_rate": 1.7936938951538193e-05,
      "loss": 0.9178,
      "step": 4198
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2076666355133057,
      "learning_rate": 1.7936300035140405e-05,
      "loss": 1.0085,
      "step": 4199
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.374936580657959,
      "learning_rate": 1.7935661118742613e-05,
      "loss": 1.5108,
      "step": 4200
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9035791158676147,
      "learning_rate": 1.7935022202344825e-05,
      "loss": 0.8794,
      "step": 4201
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0296525955200195,
      "learning_rate": 1.7934383285947036e-05,
      "loss": 1.0169,
      "step": 4202
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5378077030181885,
      "learning_rate": 1.7933744369549248e-05,
      "loss": 1.0165,
      "step": 4203
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5869007110595703,
      "learning_rate": 1.7933105453151456e-05,
      "loss": 0.9363,
      "step": 4204
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8092795610427856,
      "learning_rate": 1.7932466536753667e-05,
      "loss": 0.975,
      "step": 4205
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.191985607147217,
      "learning_rate": 1.7931827620355876e-05,
      "loss": 1.3519,
      "step": 4206
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0612106323242188,
      "learning_rate": 1.793118870395809e-05,
      "loss": 0.9514,
      "step": 4207
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.107701301574707,
      "learning_rate": 1.79305497875603e-05,
      "loss": 0.9068,
      "step": 4208
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4630156755447388,
      "learning_rate": 1.792991087116251e-05,
      "loss": 0.4363,
      "step": 4209
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.625314950942993,
      "learning_rate": 1.792927195476472e-05,
      "loss": 0.932,
      "step": 4210
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.825882911682129,
      "learning_rate": 1.792863303836693e-05,
      "loss": 1.733,
      "step": 4211
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1845126152038574,
      "learning_rate": 1.792799412196914e-05,
      "loss": 0.9734,
      "step": 4212
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4869096279144287,
      "learning_rate": 1.7927355205571353e-05,
      "loss": 0.8919,
      "step": 4213
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.213869333267212,
      "learning_rate": 1.792671628917356e-05,
      "loss": 0.8494,
      "step": 4214
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.258122682571411,
      "learning_rate": 1.7926077372775773e-05,
      "loss": 0.9689,
      "step": 4215
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.311787128448486,
      "learning_rate": 1.7925438456377984e-05,
      "loss": 1.3826,
      "step": 4216
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.273566484451294,
      "learning_rate": 1.7924799539980196e-05,
      "loss": 0.9454,
      "step": 4217
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9249502420425415,
      "learning_rate": 1.7924160623582404e-05,
      "loss": 1.0488,
      "step": 4218
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.469590187072754,
      "learning_rate": 1.7923521707184616e-05,
      "loss": 1.0756,
      "step": 4219
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.043673276901245,
      "learning_rate": 1.7922882790786827e-05,
      "loss": 0.96,
      "step": 4220
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2366108894348145,
      "learning_rate": 1.792224387438904e-05,
      "loss": 1.0422,
      "step": 4221
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.878956913948059,
      "learning_rate": 1.7921604957991247e-05,
      "loss": 0.9631,
      "step": 4222
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.220128297805786,
      "learning_rate": 1.792096604159346e-05,
      "loss": 0.8235,
      "step": 4223
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0560567378997803,
      "learning_rate": 1.792032712519567e-05,
      "loss": 0.9787,
      "step": 4224
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5446151494979858,
      "learning_rate": 1.791968820879788e-05,
      "loss": 0.4477,
      "step": 4225
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.3422858715057373,
      "learning_rate": 1.791904929240009e-05,
      "loss": 0.9709,
      "step": 4226
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.806326150894165,
      "learning_rate": 1.79184103760023e-05,
      "loss": 0.8875,
      "step": 4227
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.44787859916687,
      "learning_rate": 1.7917771459604513e-05,
      "loss": 1.1734,
      "step": 4228
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3415122032165527,
      "learning_rate": 1.7917132543206724e-05,
      "loss": 1.062,
      "step": 4229
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9148871898651123,
      "learning_rate": 1.7916493626808932e-05,
      "loss": 0.9287,
      "step": 4230
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9634788036346436,
      "learning_rate": 1.7915854710411144e-05,
      "loss": 0.9298,
      "step": 4231
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.714482545852661,
      "learning_rate": 1.7915215794013356e-05,
      "loss": 0.9092,
      "step": 4232
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.068376541137695,
      "learning_rate": 1.7914576877615567e-05,
      "loss": 1.4587,
      "step": 4233
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.817884922027588,
      "learning_rate": 1.7913937961217775e-05,
      "loss": 0.6259,
      "step": 4234
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6530871391296387,
      "learning_rate": 1.7913299044819987e-05,
      "loss": 0.9227,
      "step": 4235
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.067897796630859,
      "learning_rate": 1.79126601284222e-05,
      "loss": 1.1895,
      "step": 4236
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0728585720062256,
      "learning_rate": 1.791202121202441e-05,
      "loss": 0.7745,
      "step": 4237
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9459929466247559,
      "learning_rate": 1.7911382295626618e-05,
      "loss": 0.8812,
      "step": 4238
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7679952383041382,
      "learning_rate": 1.791074337922883e-05,
      "loss": 0.9056,
      "step": 4239
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0740952491760254,
      "learning_rate": 1.791010446283104e-05,
      "loss": 0.9982,
      "step": 4240
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.275023937225342,
      "learning_rate": 1.7909465546433253e-05,
      "loss": 1.0662,
      "step": 4241
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.298269033432007,
      "learning_rate": 1.790882663003546e-05,
      "loss": 0.8857,
      "step": 4242
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.154430866241455,
      "learning_rate": 1.7908187713637672e-05,
      "loss": 0.9531,
      "step": 4243
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.455099105834961,
      "learning_rate": 1.790754879723988e-05,
      "loss": 1.3056,
      "step": 4244
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7873785495758057,
      "learning_rate": 1.7906909880842095e-05,
      "loss": 1.0252,
      "step": 4245
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7751927375793457,
      "learning_rate": 1.7906270964444304e-05,
      "loss": 0.984,
      "step": 4246
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.9888672828674316,
      "learning_rate": 1.7905632048046515e-05,
      "loss": 0.985,
      "step": 4247
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.887552261352539,
      "learning_rate": 1.7904993131648723e-05,
      "loss": 1.1162,
      "step": 4248
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.333267688751221,
      "learning_rate": 1.7904354215250935e-05,
      "loss": 1.0572,
      "step": 4249
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.393611431121826,
      "learning_rate": 1.7903715298853146e-05,
      "loss": 1.134,
      "step": 4250
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.24013352394104,
      "learning_rate": 1.7903076382455358e-05,
      "loss": 1.0066,
      "step": 4251
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2991771697998047,
      "learning_rate": 1.7902437466057566e-05,
      "loss": 0.9966,
      "step": 4252
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4314987659454346,
      "learning_rate": 1.7901798549659778e-05,
      "loss": 1.023,
      "step": 4253
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1238625049591064,
      "learning_rate": 1.790115963326199e-05,
      "loss": 1.0417,
      "step": 4254
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.157000780105591,
      "learning_rate": 1.79005207168642e-05,
      "loss": 0.8927,
      "step": 4255
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6716666221618652,
      "learning_rate": 1.789988180046641e-05,
      "loss": 0.3688,
      "step": 4256
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3635783195495605,
      "learning_rate": 1.789924288406862e-05,
      "loss": 0.9477,
      "step": 4257
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5455877780914307,
      "learning_rate": 1.7898603967670832e-05,
      "loss": 0.9606,
      "step": 4258
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.970682144165039,
      "learning_rate": 1.7897965051273044e-05,
      "loss": 1.3947,
      "step": 4259
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9520083665847778,
      "learning_rate": 1.7897326134875252e-05,
      "loss": 0.9543,
      "step": 4260
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.288649320602417,
      "learning_rate": 1.7896687218477463e-05,
      "loss": 1.0213,
      "step": 4261
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.515903949737549,
      "learning_rate": 1.7896048302079675e-05,
      "loss": 0.9905,
      "step": 4262
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.205826759338379,
      "learning_rate": 1.7895409385681886e-05,
      "loss": 1.0115,
      "step": 4263
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5543644428253174,
      "learning_rate": 1.7894770469284095e-05,
      "loss": 0.3473,
      "step": 4264
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.09032940864563,
      "learning_rate": 1.7894131552886306e-05,
      "loss": 0.9643,
      "step": 4265
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.0960798263549805,
      "learning_rate": 1.7893492636488518e-05,
      "loss": 1.563,
      "step": 4266
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7643933296203613,
      "learning_rate": 1.789285372009073e-05,
      "loss": 0.9893,
      "step": 4267
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8066495656967163,
      "learning_rate": 1.7892214803692937e-05,
      "loss": 0.9163,
      "step": 4268
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.087265729904175,
      "learning_rate": 1.789157588729515e-05,
      "loss": 0.942,
      "step": 4269
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7369937896728516,
      "learning_rate": 1.789093697089736e-05,
      "loss": 0.897,
      "step": 4270
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8031495809555054,
      "learning_rate": 1.7890298054499572e-05,
      "loss": 0.9894,
      "step": 4271
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8662800788879395,
      "learning_rate": 1.788965913810178e-05,
      "loss": 0.9967,
      "step": 4272
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0006284713745117,
      "learning_rate": 1.788902022170399e-05,
      "loss": 0.9635,
      "step": 4273
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.610555171966553,
      "learning_rate": 1.7888381305306203e-05,
      "loss": 1.5591,
      "step": 4274
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.880951881408691,
      "learning_rate": 1.7887742388908415e-05,
      "loss": 1.2024,
      "step": 4275
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3111772537231445,
      "learning_rate": 1.7887103472510623e-05,
      "loss": 0.9198,
      "step": 4276
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9356889724731445,
      "learning_rate": 1.7886464556112834e-05,
      "loss": 0.9425,
      "step": 4277
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.907867908477783,
      "learning_rate": 1.7885825639715043e-05,
      "loss": 1.3175,
      "step": 4278
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2147562503814697,
      "learning_rate": 1.7885186723317258e-05,
      "loss": 0.8923,
      "step": 4279
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8298046588897705,
      "learning_rate": 1.7884547806919466e-05,
      "loss": 1.0116,
      "step": 4280
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1265993118286133,
      "learning_rate": 1.7883908890521677e-05,
      "loss": 0.8669,
      "step": 4281
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2321462631225586,
      "learning_rate": 1.7883269974123885e-05,
      "loss": 1.1636,
      "step": 4282
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.747370481491089,
      "learning_rate": 1.7882631057726097e-05,
      "loss": 0.8375,
      "step": 4283
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.722195625305176,
      "learning_rate": 1.788199214132831e-05,
      "loss": 0.8896,
      "step": 4284
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.882800579071045,
      "learning_rate": 1.788135322493052e-05,
      "loss": 1.0995,
      "step": 4285
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.243198394775391,
      "learning_rate": 1.7880714308532728e-05,
      "loss": 1.0853,
      "step": 4286
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0302462577819824,
      "learning_rate": 1.788007539213494e-05,
      "loss": 0.9413,
      "step": 4287
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.597681760787964,
      "learning_rate": 1.787943647573715e-05,
      "loss": 0.9955,
      "step": 4288
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7666974067687988,
      "learning_rate": 1.7878797559339363e-05,
      "loss": 0.9219,
      "step": 4289
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.4218802452087402,
      "learning_rate": 1.787815864294157e-05,
      "loss": 0.9776,
      "step": 4290
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3203110694885254,
      "learning_rate": 1.7877519726543783e-05,
      "loss": 0.9299,
      "step": 4291
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9743642807006836,
      "learning_rate": 1.7876880810145994e-05,
      "loss": 0.848,
      "step": 4292
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.323456764221191,
      "learning_rate": 1.7876241893748206e-05,
      "loss": 1.9153,
      "step": 4293
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6479820013046265,
      "learning_rate": 1.7875602977350414e-05,
      "loss": 0.9043,
      "step": 4294
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5271177291870117,
      "learning_rate": 1.7874964060952625e-05,
      "loss": 0.9788,
      "step": 4295
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.274207830429077,
      "learning_rate": 1.7874325144554837e-05,
      "loss": 1.0281,
      "step": 4296
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.107072353363037,
      "learning_rate": 1.787368622815705e-05,
      "loss": 0.9815,
      "step": 4297
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6030800342559814,
      "learning_rate": 1.7873047311759257e-05,
      "loss": 0.9192,
      "step": 4298
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9408729076385498,
      "learning_rate": 1.7872408395361468e-05,
      "loss": 0.9114,
      "step": 4299
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.379605293273926,
      "learning_rate": 1.787176947896368e-05,
      "loss": 1.2171,
      "step": 4300
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.101680040359497,
      "learning_rate": 1.787113056256589e-05,
      "loss": 0.8703,
      "step": 4301
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3198699951171875,
      "learning_rate": 1.78704916461681e-05,
      "loss": 0.9609,
      "step": 4302
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.358860969543457,
      "learning_rate": 1.786985272977031e-05,
      "loss": 0.9799,
      "step": 4303
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.177762031555176,
      "learning_rate": 1.7869213813372523e-05,
      "loss": 0.9694,
      "step": 4304
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.248699426651001,
      "learning_rate": 1.7868574896974734e-05,
      "loss": 0.8955,
      "step": 4305
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.830725908279419,
      "learning_rate": 1.7867935980576942e-05,
      "loss": 1.2402,
      "step": 4306
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5855374336242676,
      "learning_rate": 1.7867297064179154e-05,
      "loss": 0.8918,
      "step": 4307
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1579554080963135,
      "learning_rate": 1.7866658147781365e-05,
      "loss": 1.0574,
      "step": 4308
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.875331401824951,
      "learning_rate": 1.7866019231383577e-05,
      "loss": 0.9771,
      "step": 4309
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9068386554718018,
      "learning_rate": 1.7865380314985785e-05,
      "loss": 0.8567,
      "step": 4310
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.831751346588135,
      "learning_rate": 1.7864741398587997e-05,
      "loss": 1.2015,
      "step": 4311
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.414736270904541,
      "learning_rate": 1.7864102482190208e-05,
      "loss": 0.9477,
      "step": 4312
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.754981517791748,
      "learning_rate": 1.786346356579242e-05,
      "loss": 1.052,
      "step": 4313
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.577895402908325,
      "learning_rate": 1.7862824649394628e-05,
      "loss": 1.093,
      "step": 4314
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.70220947265625,
      "learning_rate": 1.786218573299684e-05,
      "loss": 0.8243,
      "step": 4315
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.26369571685791,
      "learning_rate": 1.7861546816599048e-05,
      "loss": 1.5401,
      "step": 4316
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.851898193359375,
      "learning_rate": 1.7860907900201262e-05,
      "loss": 0.8694,
      "step": 4317
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.381801128387451,
      "learning_rate": 1.786026898380347e-05,
      "loss": 0.8612,
      "step": 4318
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5604026317596436,
      "learning_rate": 1.7859630067405682e-05,
      "loss": 1.0339,
      "step": 4319
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.769834041595459,
      "learning_rate": 1.785899115100789e-05,
      "loss": 0.8345,
      "step": 4320
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9741207361221313,
      "learning_rate": 1.7858352234610102e-05,
      "loss": 0.9715,
      "step": 4321
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6946663856506348,
      "learning_rate": 1.7857713318212313e-05,
      "loss": 0.8381,
      "step": 4322
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.136300802230835,
      "learning_rate": 1.7857074401814525e-05,
      "loss": 0.8387,
      "step": 4323
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.732109308242798,
      "learning_rate": 1.7856435485416733e-05,
      "loss": 1.0277,
      "step": 4324
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1295018196105957,
      "learning_rate": 1.7855796569018945e-05,
      "loss": 0.4829,
      "step": 4325
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1192758083343506,
      "learning_rate": 1.7855157652621156e-05,
      "loss": 1.0419,
      "step": 4326
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.867835521697998,
      "learning_rate": 1.7854518736223368e-05,
      "loss": 0.9147,
      "step": 4327
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.341405868530273,
      "learning_rate": 1.7853879819825576e-05,
      "loss": 1.2285,
      "step": 4328
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1241531372070312,
      "learning_rate": 1.7853240903427787e-05,
      "loss": 0.9344,
      "step": 4329
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.290278911590576,
      "learning_rate": 1.785260198703e-05,
      "loss": 0.9793,
      "step": 4330
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.548276662826538,
      "learning_rate": 1.785196307063221e-05,
      "loss": 0.8807,
      "step": 4331
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.3777565956115723,
      "learning_rate": 1.785132415423442e-05,
      "loss": 0.9338,
      "step": 4332
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0339558124542236,
      "learning_rate": 1.785068523783663e-05,
      "loss": 1.046,
      "step": 4333
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1148600578308105,
      "learning_rate": 1.7850046321438842e-05,
      "loss": 1.2921,
      "step": 4334
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.237025260925293,
      "learning_rate": 1.7849407405041053e-05,
      "loss": 1.1589,
      "step": 4335
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2071666717529297,
      "learning_rate": 1.784876848864326e-05,
      "loss": 1.0667,
      "step": 4336
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8825851678848267,
      "learning_rate": 1.7848129572245473e-05,
      "loss": 0.9629,
      "step": 4337
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1021361351013184,
      "learning_rate": 1.7847490655847685e-05,
      "loss": 1.0047,
      "step": 4338
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.136439323425293,
      "learning_rate": 1.7846851739449896e-05,
      "loss": 1.0236,
      "step": 4339
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.348240852355957,
      "learning_rate": 1.7846212823052104e-05,
      "loss": 0.8617,
      "step": 4340
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0025522708892822,
      "learning_rate": 1.7845573906654316e-05,
      "loss": 1.3188,
      "step": 4341
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8893263339996338,
      "learning_rate": 1.7844934990256527e-05,
      "loss": 0.9203,
      "step": 4342
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.424417495727539,
      "learning_rate": 1.784429607385874e-05,
      "loss": 0.9949,
      "step": 4343
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0800387859344482,
      "learning_rate": 1.7843657157460947e-05,
      "loss": 0.9994,
      "step": 4344
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5629842281341553,
      "learning_rate": 1.784301824106316e-05,
      "loss": 0.9384,
      "step": 4345
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9580544233322144,
      "learning_rate": 1.784237932466537e-05,
      "loss": 0.9128,
      "step": 4346
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7775523662567139,
      "learning_rate": 1.7841740408267582e-05,
      "loss": 0.9077,
      "step": 4347
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.862536907196045,
      "learning_rate": 1.784110149186979e-05,
      "loss": 0.7619,
      "step": 4348
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.872759819030762,
      "learning_rate": 1.7840462575472e-05,
      "loss": 1.4146,
      "step": 4349
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3372530937194824,
      "learning_rate": 1.783982365907421e-05,
      "loss": 0.9861,
      "step": 4350
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4039348363876343,
      "learning_rate": 1.7839184742676425e-05,
      "loss": 0.4751,
      "step": 4351
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.558065414428711,
      "learning_rate": 1.7838545826278633e-05,
      "loss": 1.1622,
      "step": 4352
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9194196462631226,
      "learning_rate": 1.7837906909880844e-05,
      "loss": 0.8147,
      "step": 4353
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.934797286987305,
      "learning_rate": 1.7837267993483052e-05,
      "loss": 1.7473,
      "step": 4354
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.6793806552886963,
      "learning_rate": 1.7836629077085267e-05,
      "loss": 0.684,
      "step": 4355
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3511221408843994,
      "learning_rate": 1.7835990160687476e-05,
      "loss": 0.9192,
      "step": 4356
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.740138053894043,
      "learning_rate": 1.7835351244289687e-05,
      "loss": 0.9417,
      "step": 4357
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1019320487976074,
      "learning_rate": 1.7834712327891895e-05,
      "loss": 0.9345,
      "step": 4358
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.4249508380889893,
      "learning_rate": 1.7834073411494107e-05,
      "loss": 0.8282,
      "step": 4359
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9051991701126099,
      "learning_rate": 1.783343449509632e-05,
      "loss": 0.2993,
      "step": 4360
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3191604614257812,
      "learning_rate": 1.783279557869853e-05,
      "loss": 0.98,
      "step": 4361
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8328728675842285,
      "learning_rate": 1.7832156662300738e-05,
      "loss": 0.9085,
      "step": 4362
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0742247104644775,
      "learning_rate": 1.783151774590295e-05,
      "loss": 0.8734,
      "step": 4363
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.042548179626465,
      "learning_rate": 1.783087882950516e-05,
      "loss": 0.9455,
      "step": 4364
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9348313808441162,
      "learning_rate": 1.7830239913107373e-05,
      "loss": 0.9306,
      "step": 4365
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.013936758041382,
      "learning_rate": 1.782960099670958e-05,
      "loss": 0.8947,
      "step": 4366
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4484670162200928,
      "learning_rate": 1.7828962080311792e-05,
      "loss": 1.3963,
      "step": 4367
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.1508073806762695,
      "learning_rate": 1.7828323163914004e-05,
      "loss": 1.3335,
      "step": 4368
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8597722053527832,
      "learning_rate": 1.7827684247516215e-05,
      "loss": 1.0085,
      "step": 4369
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9024075269699097,
      "learning_rate": 1.7827045331118424e-05,
      "loss": 0.9834,
      "step": 4370
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8998676538467407,
      "learning_rate": 1.7826406414720635e-05,
      "loss": 0.9106,
      "step": 4371
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.085021018981934,
      "learning_rate": 1.7825767498322847e-05,
      "loss": 1.1264,
      "step": 4372
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6678234338760376,
      "learning_rate": 1.7825128581925058e-05,
      "loss": 0.8522,
      "step": 4373
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9153908491134644,
      "learning_rate": 1.7824489665527266e-05,
      "loss": 0.9595,
      "step": 4374
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.794923782348633,
      "learning_rate": 1.7823850749129478e-05,
      "loss": 1.1226,
      "step": 4375
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3862709999084473,
      "learning_rate": 1.782321183273169e-05,
      "loss": 0.8082,
      "step": 4376
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.119553804397583,
      "learning_rate": 1.78225729163339e-05,
      "loss": 1.0383,
      "step": 4377
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.276399612426758,
      "learning_rate": 1.782193399993611e-05,
      "loss": 1.3277,
      "step": 4378
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3148365020751953,
      "learning_rate": 1.782129508353832e-05,
      "loss": 1.0872,
      "step": 4379
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.576531171798706,
      "learning_rate": 1.7820656167140532e-05,
      "loss": 0.4121,
      "step": 4380
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0807652473449707,
      "learning_rate": 1.7820017250742744e-05,
      "loss": 0.8854,
      "step": 4381
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.833152174949646,
      "learning_rate": 1.7819378334344952e-05,
      "loss": 0.866,
      "step": 4382
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7222753763198853,
      "learning_rate": 1.7818739417947164e-05,
      "loss": 0.3828,
      "step": 4383
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.768255710601807,
      "learning_rate": 1.7818100501549375e-05,
      "loss": 0.8814,
      "step": 4384
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.098853826522827,
      "learning_rate": 1.7817461585151587e-05,
      "loss": 0.8816,
      "step": 4385
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8478713035583496,
      "learning_rate": 1.7816822668753795e-05,
      "loss": 0.8431,
      "step": 4386
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4702014923095703,
      "learning_rate": 1.7816183752356006e-05,
      "loss": 0.9159,
      "step": 4387
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.081132173538208,
      "learning_rate": 1.7815544835958215e-05,
      "loss": 1.0148,
      "step": 4388
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9655040502548218,
      "learning_rate": 1.781490591956043e-05,
      "loss": 0.8602,
      "step": 4389
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.316915273666382,
      "learning_rate": 1.7814267003162638e-05,
      "loss": 0.9616,
      "step": 4390
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3422930240631104,
      "learning_rate": 1.781362808676485e-05,
      "loss": 0.9621,
      "step": 4391
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.141425132751465,
      "learning_rate": 1.7812989170367057e-05,
      "loss": 0.9052,
      "step": 4392
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9300113916397095,
      "learning_rate": 1.781235025396927e-05,
      "loss": 0.8863,
      "step": 4393
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8830870389938354,
      "learning_rate": 1.781171133757148e-05,
      "loss": 0.9288,
      "step": 4394
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.018293619155884,
      "learning_rate": 1.7811072421173692e-05,
      "loss": 1.1325,
      "step": 4395
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.273038625717163,
      "learning_rate": 1.78104335047759e-05,
      "loss": 0.8651,
      "step": 4396
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1731302738189697,
      "learning_rate": 1.780979458837811e-05,
      "loss": 1.0261,
      "step": 4397
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.8822691440582275,
      "learning_rate": 1.7809155671980323e-05,
      "loss": 0.8941,
      "step": 4398
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8299777507781982,
      "learning_rate": 1.7808516755582535e-05,
      "loss": 0.9074,
      "step": 4399
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8240554332733154,
      "learning_rate": 1.7807877839184743e-05,
      "loss": 0.4008,
      "step": 4400
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.974835753440857,
      "learning_rate": 1.7807238922786954e-05,
      "loss": 0.9779,
      "step": 4401
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.818440318107605,
      "learning_rate": 1.7806600006389166e-05,
      "loss": 0.8462,
      "step": 4402
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0531084537506104,
      "learning_rate": 1.7805961089991378e-05,
      "loss": 0.76,
      "step": 4403
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6783227920532227,
      "learning_rate": 1.7805322173593586e-05,
      "loss": 1.435,
      "step": 4404
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0537736415863037,
      "learning_rate": 1.7804683257195797e-05,
      "loss": 0.9113,
      "step": 4405
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1990184783935547,
      "learning_rate": 1.780404434079801e-05,
      "loss": 1.0573,
      "step": 4406
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.938688278198242,
      "learning_rate": 1.780340542440022e-05,
      "loss": 0.8789,
      "step": 4407
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.508878469467163,
      "learning_rate": 1.780276650800243e-05,
      "loss": 1.0237,
      "step": 4408
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.33207106590271,
      "learning_rate": 1.780212759160464e-05,
      "loss": 0.8188,
      "step": 4409
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.868154764175415,
      "learning_rate": 1.780148867520685e-05,
      "loss": 0.8853,
      "step": 4410
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0154800415039062,
      "learning_rate": 1.7800849758809063e-05,
      "loss": 1.3469,
      "step": 4411
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.2704482078552246,
      "learning_rate": 1.780021084241127e-05,
      "loss": 0.7516,
      "step": 4412
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.863616704940796,
      "learning_rate": 1.7799571926013483e-05,
      "loss": 0.8405,
      "step": 4413
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6540724039077759,
      "learning_rate": 1.7798933009615694e-05,
      "loss": 0.8926,
      "step": 4414
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.110196113586426,
      "learning_rate": 1.7798294093217906e-05,
      "loss": 1.1055,
      "step": 4415
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9533023834228516,
      "learning_rate": 1.7797655176820114e-05,
      "loss": 0.9149,
      "step": 4416
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.406471014022827,
      "learning_rate": 1.7797016260422326e-05,
      "loss": 1.1244,
      "step": 4417
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0123555660247803,
      "learning_rate": 1.7796377344024537e-05,
      "loss": 0.909,
      "step": 4418
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8632334470748901,
      "learning_rate": 1.779573842762675e-05,
      "loss": 0.8793,
      "step": 4419
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8541682958602905,
      "learning_rate": 1.7795099511228957e-05,
      "loss": 0.8948,
      "step": 4420
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7829413414001465,
      "learning_rate": 1.779446059483117e-05,
      "loss": 0.8999,
      "step": 4421
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.2693400382995605,
      "learning_rate": 1.779382167843338e-05,
      "loss": 0.9866,
      "step": 4422
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.714534044265747,
      "learning_rate": 1.779318276203559e-05,
      "loss": 1.0457,
      "step": 4423
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6281263828277588,
      "learning_rate": 1.77925438456378e-05,
      "loss": 0.945,
      "step": 4424
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.278595447540283,
      "learning_rate": 1.779190492924001e-05,
      "loss": 0.9148,
      "step": 4425
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9624313116073608,
      "learning_rate": 1.779126601284222e-05,
      "loss": 1.0484,
      "step": 4426
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.2392842769622803,
      "learning_rate": 1.7790627096444434e-05,
      "loss": 0.6793,
      "step": 4427
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2537295818328857,
      "learning_rate": 1.7789988180046643e-05,
      "loss": 1.0079,
      "step": 4428
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.1879119873046875,
      "learning_rate": 1.7789349263648854e-05,
      "loss": 1.1165,
      "step": 4429
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1004817485809326,
      "learning_rate": 1.7788710347251062e-05,
      "loss": 0.7873,
      "step": 4430
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.442281723022461,
      "learning_rate": 1.7788071430853274e-05,
      "loss": 0.8145,
      "step": 4431
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.565521001815796,
      "learning_rate": 1.7787432514455485e-05,
      "loss": 0.79,
      "step": 4432
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9910304546356201,
      "learning_rate": 1.7786793598057697e-05,
      "loss": 0.9371,
      "step": 4433
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9692721366882324,
      "learning_rate": 1.7786154681659905e-05,
      "loss": 0.9252,
      "step": 4434
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.099560022354126,
      "learning_rate": 1.7785515765262117e-05,
      "loss": 0.9019,
      "step": 4435
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5446224212646484,
      "learning_rate": 1.7784876848864328e-05,
      "loss": 0.8074,
      "step": 4436
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.0044569969177246,
      "learning_rate": 1.778423793246654e-05,
      "loss": 1.1051,
      "step": 4437
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6431065797805786,
      "learning_rate": 1.7783599016068748e-05,
      "loss": 0.7615,
      "step": 4438
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.163442850112915,
      "learning_rate": 1.778296009967096e-05,
      "loss": 0.9269,
      "step": 4439
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.824376344680786,
      "learning_rate": 1.778232118327317e-05,
      "loss": 0.8909,
      "step": 4440
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.080833911895752,
      "learning_rate": 1.7781682266875382e-05,
      "loss": 1.2672,
      "step": 4441
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.400814056396484,
      "learning_rate": 1.778104335047759e-05,
      "loss": 1.745,
      "step": 4442
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9473495483398438,
      "learning_rate": 1.7780404434079802e-05,
      "loss": 0.9367,
      "step": 4443
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.440542221069336,
      "learning_rate": 1.7779765517682014e-05,
      "loss": 1.0944,
      "step": 4444
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.479330062866211,
      "learning_rate": 1.7779126601284225e-05,
      "loss": 1.4875,
      "step": 4445
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.237617254257202,
      "learning_rate": 1.7778487684886433e-05,
      "loss": 0.87,
      "step": 4446
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6424953937530518,
      "learning_rate": 1.7777848768488645e-05,
      "loss": 0.9322,
      "step": 4447
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.155110836029053,
      "learning_rate": 1.7777209852090857e-05,
      "loss": 2.0717,
      "step": 4448
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.561845064163208,
      "learning_rate": 1.7776570935693065e-05,
      "loss": 1.0781,
      "step": 4449
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.867989420890808,
      "learning_rate": 1.7775932019295276e-05,
      "loss": 1.0535,
      "step": 4450
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5994377136230469,
      "learning_rate": 1.7775293102897488e-05,
      "loss": 0.3801,
      "step": 4451
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1336755752563477,
      "learning_rate": 1.77746541864997e-05,
      "loss": 1.3002,
      "step": 4452
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.140235185623169,
      "learning_rate": 1.7774015270101907e-05,
      "loss": 0.9986,
      "step": 4453
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7597689628601074,
      "learning_rate": 1.777337635370412e-05,
      "loss": 0.9363,
      "step": 4454
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9912965297698975,
      "learning_rate": 1.7772737437306327e-05,
      "loss": 0.9995,
      "step": 4455
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.544973373413086,
      "learning_rate": 1.7772098520908542e-05,
      "loss": 0.8733,
      "step": 4456
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.950016975402832,
      "learning_rate": 1.777145960451075e-05,
      "loss": 0.9785,
      "step": 4457
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.8639237880706787,
      "learning_rate": 1.7770820688112962e-05,
      "loss": 0.8763,
      "step": 4458
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3474016189575195,
      "learning_rate": 1.777018177171517e-05,
      "loss": 1.1085,
      "step": 4459
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3282580375671387,
      "learning_rate": 1.776954285531738e-05,
      "loss": 0.9361,
      "step": 4460
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6015304327011108,
      "learning_rate": 1.7768903938919593e-05,
      "loss": 0.9178,
      "step": 4461
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0005993843078613,
      "learning_rate": 1.7768265022521805e-05,
      "loss": 1.0378,
      "step": 4462
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9464658498764038,
      "learning_rate": 1.7767626106124013e-05,
      "loss": 0.8901,
      "step": 4463
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.372964382171631,
      "learning_rate": 1.7766987189726224e-05,
      "loss": 1.2084,
      "step": 4464
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4661762714385986,
      "learning_rate": 1.7766348273328436e-05,
      "loss": 0.9265,
      "step": 4465
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.590742826461792,
      "learning_rate": 1.7765709356930647e-05,
      "loss": 0.6316,
      "step": 4466
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.724029302597046,
      "learning_rate": 1.7765070440532856e-05,
      "loss": 0.9313,
      "step": 4467
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1193904876708984,
      "learning_rate": 1.7764431524135067e-05,
      "loss": 0.9172,
      "step": 4468
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7675566673278809,
      "learning_rate": 1.776379260773728e-05,
      "loss": 0.8205,
      "step": 4469
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.87398099899292,
      "learning_rate": 1.776315369133949e-05,
      "loss": 0.8956,
      "step": 4470
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9618096351623535,
      "learning_rate": 1.77625147749417e-05,
      "loss": 1.2314,
      "step": 4471
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.891627550125122,
      "learning_rate": 1.776187585854391e-05,
      "loss": 0.9578,
      "step": 4472
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.752664804458618,
      "learning_rate": 1.776123694214612e-05,
      "loss": 1.1966,
      "step": 4473
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7889996767044067,
      "learning_rate": 1.7760598025748333e-05,
      "loss": 0.9518,
      "step": 4474
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.846672534942627,
      "learning_rate": 1.775995910935054e-05,
      "loss": 0.8653,
      "step": 4475
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.286106824874878,
      "learning_rate": 1.7759320192952753e-05,
      "loss": 1.0212,
      "step": 4476
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.296513319015503,
      "learning_rate": 1.7758681276554964e-05,
      "loss": 1.0927,
      "step": 4477
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5212924480438232,
      "learning_rate": 1.7758042360157176e-05,
      "loss": 1.1182,
      "step": 4478
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.4490694999694824,
      "learning_rate": 1.7757403443759384e-05,
      "loss": 0.6901,
      "step": 4479
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1814932823181152,
      "learning_rate": 1.7756764527361596e-05,
      "loss": 1.1164,
      "step": 4480
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7741421461105347,
      "learning_rate": 1.7756125610963807e-05,
      "loss": 0.9038,
      "step": 4481
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8531889915466309,
      "learning_rate": 1.775548669456602e-05,
      "loss": 1.0703,
      "step": 4482
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.319608449935913,
      "learning_rate": 1.7754847778168227e-05,
      "loss": 0.7613,
      "step": 4483
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7553757429122925,
      "learning_rate": 1.775420886177044e-05,
      "loss": 0.9421,
      "step": 4484
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9146697521209717,
      "learning_rate": 1.775356994537265e-05,
      "loss": 1.0163,
      "step": 4485
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.366197347640991,
      "learning_rate": 1.775293102897486e-05,
      "loss": 0.944,
      "step": 4486
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5234029293060303,
      "learning_rate": 1.775229211257707e-05,
      "loss": 0.9593,
      "step": 4487
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1294898986816406,
      "learning_rate": 1.775165319617928e-05,
      "loss": 1.0513,
      "step": 4488
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2698652744293213,
      "learning_rate": 1.7751014279781493e-05,
      "loss": 0.8944,
      "step": 4489
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.036929130554199,
      "learning_rate": 1.7750375363383704e-05,
      "loss": 0.8133,
      "step": 4490
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4327435493469238,
      "learning_rate": 1.7749736446985912e-05,
      "loss": 0.848,
      "step": 4491
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7707664966583252,
      "learning_rate": 1.7749097530588124e-05,
      "loss": 0.8824,
      "step": 4492
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0720882415771484,
      "learning_rate": 1.7748458614190332e-05,
      "loss": 1.0349,
      "step": 4493
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6045646667480469,
      "learning_rate": 1.7747819697792547e-05,
      "loss": 0.8613,
      "step": 4494
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.191189289093018,
      "learning_rate": 1.7747180781394755e-05,
      "loss": 1.1494,
      "step": 4495
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.259594440460205,
      "learning_rate": 1.7746541864996967e-05,
      "loss": 0.9748,
      "step": 4496
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.410604476928711,
      "learning_rate": 1.7745902948599175e-05,
      "loss": 1.3791,
      "step": 4497
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.731772422790527,
      "learning_rate": 1.7745264032201386e-05,
      "loss": 1.2484,
      "step": 4498
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.512901544570923,
      "learning_rate": 1.7744625115803598e-05,
      "loss": 1.0809,
      "step": 4499
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.827806234359741,
      "learning_rate": 1.774398619940581e-05,
      "loss": 1.0435,
      "step": 4500
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.3279356956481934,
      "learning_rate": 1.7743347283008018e-05,
      "loss": 1.1007,
      "step": 4501
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6147868633270264,
      "learning_rate": 1.774270836661023e-05,
      "loss": 1.0517,
      "step": 4502
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0507025718688965,
      "learning_rate": 1.774206945021244e-05,
      "loss": 0.882,
      "step": 4503
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8164530992507935,
      "learning_rate": 1.7741430533814652e-05,
      "loss": 0.831,
      "step": 4504
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.5010955333709717,
      "learning_rate": 1.774079161741686e-05,
      "loss": 0.8692,
      "step": 4505
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.825929641723633,
      "learning_rate": 1.7740152701019072e-05,
      "loss": 0.9322,
      "step": 4506
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0715999603271484,
      "learning_rate": 1.7739513784621284e-05,
      "loss": 0.9529,
      "step": 4507
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.219081401824951,
      "learning_rate": 1.7738874868223495e-05,
      "loss": 1.2413,
      "step": 4508
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.427678346633911,
      "learning_rate": 1.7738235951825703e-05,
      "loss": 0.9864,
      "step": 4509
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.033996820449829,
      "learning_rate": 1.7737597035427915e-05,
      "loss": 0.9027,
      "step": 4510
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2615602016448975,
      "learning_rate": 1.7736958119030126e-05,
      "loss": 1.0038,
      "step": 4511
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4698235988616943,
      "learning_rate": 1.7736319202632338e-05,
      "loss": 0.9963,
      "step": 4512
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.304547071456909,
      "learning_rate": 1.7735680286234546e-05,
      "loss": 0.7893,
      "step": 4513
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.329659461975098,
      "learning_rate": 1.7735041369836758e-05,
      "loss": 1.2725,
      "step": 4514
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.733420968055725,
      "learning_rate": 1.773440245343897e-05,
      "loss": 0.9204,
      "step": 4515
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.032382011413574,
      "learning_rate": 1.773376353704118e-05,
      "loss": 1.2849,
      "step": 4516
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.883339762687683,
      "learning_rate": 1.773312462064339e-05,
      "loss": 0.8747,
      "step": 4517
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6018428802490234,
      "learning_rate": 1.77324857042456e-05,
      "loss": 1.1068,
      "step": 4518
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9071428775787354,
      "learning_rate": 1.7731846787847812e-05,
      "loss": 0.8887,
      "step": 4519
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.756415843963623,
      "learning_rate": 1.7731207871450024e-05,
      "loss": 0.9317,
      "step": 4520
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6151835918426514,
      "learning_rate": 1.7730568955052232e-05,
      "loss": 0.9562,
      "step": 4521
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6883604526519775,
      "learning_rate": 1.7729930038654443e-05,
      "loss": 0.8218,
      "step": 4522
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7518872022628784,
      "learning_rate": 1.7729291122256655e-05,
      "loss": 0.8719,
      "step": 4523
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9101289510726929,
      "learning_rate": 1.7728652205858866e-05,
      "loss": 1.0827,
      "step": 4524
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.504746437072754,
      "learning_rate": 1.7728013289461075e-05,
      "loss": 0.5455,
      "step": 4525
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1432011127471924,
      "learning_rate": 1.7727374373063286e-05,
      "loss": 1.3517,
      "step": 4526
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8909008502960205,
      "learning_rate": 1.7726735456665494e-05,
      "loss": 0.9508,
      "step": 4527
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6375603675842285,
      "learning_rate": 1.772609654026771e-05,
      "loss": 1.0511,
      "step": 4528
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9482266902923584,
      "learning_rate": 1.7725457623869917e-05,
      "loss": 0.9174,
      "step": 4529
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.984627366065979,
      "learning_rate": 1.772481870747213e-05,
      "loss": 0.8039,
      "step": 4530
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2311315536499023,
      "learning_rate": 1.7724179791074337e-05,
      "loss": 0.965,
      "step": 4531
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.697864532470703,
      "learning_rate": 1.7723540874676552e-05,
      "loss": 1.04,
      "step": 4532
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.195472240447998,
      "learning_rate": 1.772290195827876e-05,
      "loss": 1.1262,
      "step": 4533
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0769524574279785,
      "learning_rate": 1.772226304188097e-05,
      "loss": 0.973,
      "step": 4534
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4798290729522705,
      "learning_rate": 1.772162412548318e-05,
      "loss": 0.9247,
      "step": 4535
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7801203727722168,
      "learning_rate": 1.772098520908539e-05,
      "loss": 0.9119,
      "step": 4536
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1646575927734375,
      "learning_rate": 1.7720346292687603e-05,
      "loss": 1.0258,
      "step": 4537
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9026318788528442,
      "learning_rate": 1.7719707376289814e-05,
      "loss": 1.0379,
      "step": 4538
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.684508800506592,
      "learning_rate": 1.7719068459892023e-05,
      "loss": 1.1755,
      "step": 4539
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1770331859588623,
      "learning_rate": 1.7718429543494234e-05,
      "loss": 1.3047,
      "step": 4540
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.0629923343658447,
      "learning_rate": 1.7717790627096446e-05,
      "loss": 0.9142,
      "step": 4541
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1237878799438477,
      "learning_rate": 1.7717151710698657e-05,
      "loss": 1.2903,
      "step": 4542
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.136113166809082,
      "learning_rate": 1.7716512794300865e-05,
      "loss": 1.2783,
      "step": 4543
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.834190011024475,
      "learning_rate": 1.7715873877903077e-05,
      "loss": 0.8137,
      "step": 4544
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.316063165664673,
      "learning_rate": 1.771523496150529e-05,
      "loss": 0.9698,
      "step": 4545
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.859490156173706,
      "learning_rate": 1.77145960451075e-05,
      "loss": 0.9141,
      "step": 4546
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9829634428024292,
      "learning_rate": 1.7713957128709708e-05,
      "loss": 0.8924,
      "step": 4547
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.593411684036255,
      "learning_rate": 1.771331821231192e-05,
      "loss": 0.9758,
      "step": 4548
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.097710609436035,
      "learning_rate": 1.771267929591413e-05,
      "loss": 0.9625,
      "step": 4549
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0470941066741943,
      "learning_rate": 1.7712040379516343e-05,
      "loss": 0.9255,
      "step": 4550
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8827388286590576,
      "learning_rate": 1.771140146311855e-05,
      "loss": 1.0103,
      "step": 4551
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3095476627349854,
      "learning_rate": 1.7710762546720763e-05,
      "loss": 1.1249,
      "step": 4552
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4762656688690186,
      "learning_rate": 1.7710123630322974e-05,
      "loss": 0.4275,
      "step": 4553
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.7165071964263916,
      "learning_rate": 1.7709484713925186e-05,
      "loss": 0.983,
      "step": 4554
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.425276279449463,
      "learning_rate": 1.7708845797527394e-05,
      "loss": 0.4013,
      "step": 4555
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.893636703491211,
      "learning_rate": 1.7708206881129605e-05,
      "loss": 1.1584,
      "step": 4556
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6477636098861694,
      "learning_rate": 1.7707567964731817e-05,
      "loss": 0.9268,
      "step": 4557
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.0804786682128906,
      "learning_rate": 1.770692904833403e-05,
      "loss": 0.933,
      "step": 4558
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1041982173919678,
      "learning_rate": 1.7706290131936237e-05,
      "loss": 1.116,
      "step": 4559
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6561825275421143,
      "learning_rate": 1.7705651215538448e-05,
      "loss": 0.8563,
      "step": 4560
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3740313053131104,
      "learning_rate": 1.770501229914066e-05,
      "loss": 0.5571,
      "step": 4561
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8903796672821045,
      "learning_rate": 1.770437338274287e-05,
      "loss": 0.8603,
      "step": 4562
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2293410301208496,
      "learning_rate": 1.770373446634508e-05,
      "loss": 0.9585,
      "step": 4563
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6946752071380615,
      "learning_rate": 1.770309554994729e-05,
      "loss": 1.0181,
      "step": 4564
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6774985790252686,
      "learning_rate": 1.77024566335495e-05,
      "loss": 0.8526,
      "step": 4565
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1741580963134766,
      "learning_rate": 1.7701817717151714e-05,
      "loss": 1.022,
      "step": 4566
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1715192794799805,
      "learning_rate": 1.7701178800753922e-05,
      "loss": 0.8047,
      "step": 4567
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.2873494625091553,
      "learning_rate": 1.7700539884356134e-05,
      "loss": 0.9164,
      "step": 4568
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6686017513275146,
      "learning_rate": 1.7699900967958342e-05,
      "loss": 0.9039,
      "step": 4569
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.5627951622009277,
      "learning_rate": 1.7699262051560553e-05,
      "loss": 0.8699,
      "step": 4570
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5586286783218384,
      "learning_rate": 1.7698623135162765e-05,
      "loss": 0.4525,
      "step": 4571
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5134146213531494,
      "learning_rate": 1.7697984218764977e-05,
      "loss": 1.0044,
      "step": 4572
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.098644971847534,
      "learning_rate": 1.7697345302367185e-05,
      "loss": 0.8537,
      "step": 4573
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2773547172546387,
      "learning_rate": 1.7696706385969396e-05,
      "loss": 0.9478,
      "step": 4574
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0235326290130615,
      "learning_rate": 1.7696067469571608e-05,
      "loss": 0.9878,
      "step": 4575
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0811691284179688,
      "learning_rate": 1.769542855317382e-05,
      "loss": 1.3304,
      "step": 4576
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.9800004959106445,
      "learning_rate": 1.7694789636776028e-05,
      "loss": 1.0227,
      "step": 4577
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1575381755828857,
      "learning_rate": 1.769415072037824e-05,
      "loss": 1.0331,
      "step": 4578
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.133237600326538,
      "learning_rate": 1.769351180398045e-05,
      "loss": 0.9317,
      "step": 4579
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4535467624664307,
      "learning_rate": 1.7692872887582662e-05,
      "loss": 1.0285,
      "step": 4580
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.404927253723145,
      "learning_rate": 1.769223397118487e-05,
      "loss": 1.0295,
      "step": 4581
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.468865156173706,
      "learning_rate": 1.7691595054787082e-05,
      "loss": 0.8201,
      "step": 4582
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.8159754276275635,
      "learning_rate": 1.7690956138389293e-05,
      "loss": 0.7701,
      "step": 4583
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.597788095474243,
      "learning_rate": 1.7690317221991505e-05,
      "loss": 1.0416,
      "step": 4584
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4865467548370361,
      "learning_rate": 1.7689678305593713e-05,
      "loss": 0.8241,
      "step": 4585
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7415378093719482,
      "learning_rate": 1.7689039389195925e-05,
      "loss": 0.8676,
      "step": 4586
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.441254615783691,
      "learning_rate": 1.7688400472798136e-05,
      "loss": 1.6383,
      "step": 4587
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6853431463241577,
      "learning_rate": 1.7687761556400348e-05,
      "loss": 0.9683,
      "step": 4588
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8282370567321777,
      "learning_rate": 1.7687122640002556e-05,
      "loss": 0.9103,
      "step": 4589
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.829575538635254,
      "learning_rate": 1.7686483723604767e-05,
      "loss": 0.8868,
      "step": 4590
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.827430248260498,
      "learning_rate": 1.768584480720698e-05,
      "loss": 1.5202,
      "step": 4591
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.396112442016602,
      "learning_rate": 1.768520589080919e-05,
      "loss": 1.1072,
      "step": 4592
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.266626358032227,
      "learning_rate": 1.76845669744114e-05,
      "loss": 0.3894,
      "step": 4593
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1947126388549805,
      "learning_rate": 1.768392805801361e-05,
      "loss": 0.8913,
      "step": 4594
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7331889867782593,
      "learning_rate": 1.7683289141615822e-05,
      "loss": 0.8534,
      "step": 4595
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2281429767608643,
      "learning_rate": 1.7682650225218033e-05,
      "loss": 0.9999,
      "step": 4596
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.115945816040039,
      "learning_rate": 1.768201130882024e-05,
      "loss": 0.965,
      "step": 4597
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.096811294555664,
      "learning_rate": 1.7681372392422453e-05,
      "loss": 1.3199,
      "step": 4598
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5774362087249756,
      "learning_rate": 1.7680733476024665e-05,
      "loss": 0.9304,
      "step": 4599
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8844828605651855,
      "learning_rate": 1.7680094559626876e-05,
      "loss": 0.9065,
      "step": 4600
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8369863033294678,
      "learning_rate": 1.7679455643229084e-05,
      "loss": 1.008,
      "step": 4601
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.576084613800049,
      "learning_rate": 1.7678816726831296e-05,
      "loss": 0.7447,
      "step": 4602
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2645483016967773,
      "learning_rate": 1.7678177810433504e-05,
      "loss": 1.387,
      "step": 4603
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8262805938720703,
      "learning_rate": 1.767753889403572e-05,
      "loss": 0.9472,
      "step": 4604
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.933044672012329,
      "learning_rate": 1.7676899977637927e-05,
      "loss": 0.8904,
      "step": 4605
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.020717144012451,
      "learning_rate": 1.767626106124014e-05,
      "loss": 1.0016,
      "step": 4606
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.272510051727295,
      "learning_rate": 1.7675622144842347e-05,
      "loss": 0.8234,
      "step": 4607
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.465714931488037,
      "learning_rate": 1.767498322844456e-05,
      "loss": 1.2386,
      "step": 4608
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.483593702316284,
      "learning_rate": 1.767434431204677e-05,
      "loss": 1.1209,
      "step": 4609
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.128514766693115,
      "learning_rate": 1.767370539564898e-05,
      "loss": 0.8526,
      "step": 4610
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9268606901168823,
      "learning_rate": 1.767306647925119e-05,
      "loss": 0.8671,
      "step": 4611
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.702999472618103,
      "learning_rate": 1.76724275628534e-05,
      "loss": 0.8447,
      "step": 4612
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.2264530658721924,
      "learning_rate": 1.7671788646455613e-05,
      "loss": 1.0274,
      "step": 4613
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.840912342071533,
      "learning_rate": 1.7671149730057824e-05,
      "loss": 1.2006,
      "step": 4614
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1168155670166016,
      "learning_rate": 1.7670510813660032e-05,
      "loss": 0.903,
      "step": 4615
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8140798807144165,
      "learning_rate": 1.7669871897262244e-05,
      "loss": 1.0156,
      "step": 4616
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.534400463104248,
      "learning_rate": 1.7669232980864456e-05,
      "loss": 0.8643,
      "step": 4617
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.69134783744812,
      "learning_rate": 1.7668594064466667e-05,
      "loss": 1.0172,
      "step": 4618
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6451550722122192,
      "learning_rate": 1.7667955148068875e-05,
      "loss": 0.8937,
      "step": 4619
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.027078151702881,
      "learning_rate": 1.7667316231671087e-05,
      "loss": 0.8753,
      "step": 4620
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.03029727935791,
      "learning_rate": 1.76666773152733e-05,
      "loss": 0.8966,
      "step": 4621
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.491285562515259,
      "learning_rate": 1.766603839887551e-05,
      "loss": 0.9226,
      "step": 4622
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5468039512634277,
      "learning_rate": 1.7665399482477718e-05,
      "loss": 1.0547,
      "step": 4623
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.050536632537842,
      "learning_rate": 1.766476056607993e-05,
      "loss": 1.1389,
      "step": 4624
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.98658287525177,
      "learning_rate": 1.766412164968214e-05,
      "loss": 0.8983,
      "step": 4625
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.161168098449707,
      "learning_rate": 1.7663482733284353e-05,
      "loss": 1.4976,
      "step": 4626
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6348371505737305,
      "learning_rate": 1.766284381688656e-05,
      "loss": 0.3577,
      "step": 4627
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8582462072372437,
      "learning_rate": 1.7662204900488772e-05,
      "loss": 0.3725,
      "step": 4628
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.485228538513184,
      "learning_rate": 1.7661565984090984e-05,
      "loss": 1.3549,
      "step": 4629
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.9641191959381104,
      "learning_rate": 1.7660927067693195e-05,
      "loss": 1.0884,
      "step": 4630
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.470064163208008,
      "learning_rate": 1.7660288151295404e-05,
      "loss": 1.137,
      "step": 4631
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8076339960098267,
      "learning_rate": 1.7659649234897615e-05,
      "loss": 0.928,
      "step": 4632
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.330509662628174,
      "learning_rate": 1.7659010318499827e-05,
      "loss": 1.5166,
      "step": 4633
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8534202575683594,
      "learning_rate": 1.7658371402102038e-05,
      "loss": 0.9381,
      "step": 4634
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1722259521484375,
      "learning_rate": 1.7657732485704246e-05,
      "loss": 1.02,
      "step": 4635
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7098243236541748,
      "learning_rate": 1.7657093569306458e-05,
      "loss": 0.89,
      "step": 4636
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.310946464538574,
      "learning_rate": 1.7656454652908666e-05,
      "loss": 1.3619,
      "step": 4637
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7974027395248413,
      "learning_rate": 1.765581573651088e-05,
      "loss": 0.8717,
      "step": 4638
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8946831226348877,
      "learning_rate": 1.765517682011309e-05,
      "loss": 0.9276,
      "step": 4639
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1064937114715576,
      "learning_rate": 1.76545379037153e-05,
      "loss": 0.8654,
      "step": 4640
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.49876594543457,
      "learning_rate": 1.765389898731751e-05,
      "loss": 1.4682,
      "step": 4641
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7045085430145264,
      "learning_rate": 1.765326007091972e-05,
      "loss": 0.86,
      "step": 4642
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9900882244110107,
      "learning_rate": 1.7652621154521932e-05,
      "loss": 0.8679,
      "step": 4643
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.7155797481536865,
      "learning_rate": 1.7651982238124144e-05,
      "loss": 0.9784,
      "step": 4644
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.281928062438965,
      "learning_rate": 1.7651343321726352e-05,
      "loss": 0.7953,
      "step": 4645
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.6752872467041016,
      "learning_rate": 1.7650704405328563e-05,
      "loss": 1.1154,
      "step": 4646
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.63063907623291,
      "learning_rate": 1.7650065488930775e-05,
      "loss": 1.32,
      "step": 4647
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.428651809692383,
      "learning_rate": 1.7649426572532986e-05,
      "loss": 1.1946,
      "step": 4648
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9720826148986816,
      "learning_rate": 1.7648787656135195e-05,
      "loss": 0.9799,
      "step": 4649
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.873772382736206,
      "learning_rate": 1.7648148739737406e-05,
      "loss": 0.9163,
      "step": 4650
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.9200925827026367,
      "learning_rate": 1.7647509823339618e-05,
      "loss": 0.9479,
      "step": 4651
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.85508394241333,
      "learning_rate": 1.764687090694183e-05,
      "loss": 1.1156,
      "step": 4652
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.430647850036621,
      "learning_rate": 1.7646231990544037e-05,
      "loss": 1.2934,
      "step": 4653
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0284745693206787,
      "learning_rate": 1.764559307414625e-05,
      "loss": 1.3142,
      "step": 4654
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1399025917053223,
      "learning_rate": 1.764495415774846e-05,
      "loss": 0.9562,
      "step": 4655
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.089315414428711,
      "learning_rate": 1.7644315241350672e-05,
      "loss": 0.8926,
      "step": 4656
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4795539379119873,
      "learning_rate": 1.764367632495288e-05,
      "loss": 0.8412,
      "step": 4657
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.467172622680664,
      "learning_rate": 1.764303740855509e-05,
      "loss": 1.0674,
      "step": 4658
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9849005937576294,
      "learning_rate": 1.7642398492157303e-05,
      "loss": 0.9628,
      "step": 4659
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.084327220916748,
      "learning_rate": 1.7641759575759515e-05,
      "loss": 1.0535,
      "step": 4660
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6554714441299438,
      "learning_rate": 1.7641120659361723e-05,
      "loss": 1.2054,
      "step": 4661
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9105170965194702,
      "learning_rate": 1.7640481742963934e-05,
      "loss": 0.9467,
      "step": 4662
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.310356855392456,
      "learning_rate": 1.7639842826566146e-05,
      "loss": 0.6912,
      "step": 4663
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.835978627204895,
      "learning_rate": 1.7639203910168358e-05,
      "loss": 0.9009,
      "step": 4664
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0063910484313965,
      "learning_rate": 1.7638564993770566e-05,
      "loss": 0.9704,
      "step": 4665
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.15901517868042,
      "learning_rate": 1.7637926077372777e-05,
      "loss": 0.9719,
      "step": 4666
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.365661144256592,
      "learning_rate": 1.763728716097499e-05,
      "loss": 1.0833,
      "step": 4667
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.460709571838379,
      "learning_rate": 1.76366482445772e-05,
      "loss": 0.9568,
      "step": 4668
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.010331630706787,
      "learning_rate": 1.763600932817941e-05,
      "loss": 0.8777,
      "step": 4669
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.817603588104248,
      "learning_rate": 1.763537041178162e-05,
      "loss": 1.1753,
      "step": 4670
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.692947506904602,
      "learning_rate": 1.763473149538383e-05,
      "loss": 0.9514,
      "step": 4671
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.19804048538208,
      "learning_rate": 1.7634092578986043e-05,
      "loss": 0.9814,
      "step": 4672
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9070147275924683,
      "learning_rate": 1.763345366258825e-05,
      "loss": 0.8436,
      "step": 4673
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0590221881866455,
      "learning_rate": 1.7632814746190463e-05,
      "loss": 0.8056,
      "step": 4674
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2739691734313965,
      "learning_rate": 1.763217582979267e-05,
      "loss": 1.0229,
      "step": 4675
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.8231873512268066,
      "learning_rate": 1.7631536913394886e-05,
      "loss": 1.3693,
      "step": 4676
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.985683798789978,
      "learning_rate": 1.7630897996997094e-05,
      "loss": 1.0967,
      "step": 4677
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.293426752090454,
      "learning_rate": 1.7630259080599306e-05,
      "loss": 1.1307,
      "step": 4678
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6200988292694092,
      "learning_rate": 1.7629620164201514e-05,
      "loss": 0.8768,
      "step": 4679
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6738436222076416,
      "learning_rate": 1.7628981247803725e-05,
      "loss": 1.0303,
      "step": 4680
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.13329815864563,
      "learning_rate": 1.7628342331405937e-05,
      "loss": 0.9586,
      "step": 4681
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1760752201080322,
      "learning_rate": 1.762770341500815e-05,
      "loss": 0.9285,
      "step": 4682
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.286029815673828,
      "learning_rate": 1.7627064498610357e-05,
      "loss": 1.6154,
      "step": 4683
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0155694484710693,
      "learning_rate": 1.7626425582212568e-05,
      "loss": 0.8851,
      "step": 4684
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.858789086341858,
      "learning_rate": 1.762578666581478e-05,
      "loss": 0.8701,
      "step": 4685
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.031229019165039,
      "learning_rate": 1.762514774941699e-05,
      "loss": 0.9046,
      "step": 4686
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1101136207580566,
      "learning_rate": 1.76245088330192e-05,
      "loss": 0.9629,
      "step": 4687
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3333451747894287,
      "learning_rate": 1.762386991662141e-05,
      "loss": 0.9051,
      "step": 4688
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.465881109237671,
      "learning_rate": 1.7623231000223623e-05,
      "loss": 0.8917,
      "step": 4689
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.311201333999634,
      "learning_rate": 1.7622592083825834e-05,
      "loss": 1.0635,
      "step": 4690
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.2908613681793213,
      "learning_rate": 1.7621953167428042e-05,
      "loss": 1.2117,
      "step": 4691
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.1093900203704834,
      "learning_rate": 1.7621314251030254e-05,
      "loss": 0.9205,
      "step": 4692
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4282567501068115,
      "learning_rate": 1.7620675334632465e-05,
      "loss": 0.3679,
      "step": 4693
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6754989624023438,
      "learning_rate": 1.7620036418234677e-05,
      "loss": 0.8975,
      "step": 4694
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3149948120117188,
      "learning_rate": 1.7619397501836885e-05,
      "loss": 0.8057,
      "step": 4695
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4553592205047607,
      "learning_rate": 1.7618758585439097e-05,
      "loss": 0.8909,
      "step": 4696
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8699179887771606,
      "learning_rate": 1.7618119669041308e-05,
      "loss": 0.8697,
      "step": 4697
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.649656772613525,
      "learning_rate": 1.761748075264352e-05,
      "loss": 1.4874,
      "step": 4698
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9149703979492188,
      "learning_rate": 1.7616841836245728e-05,
      "loss": 0.8796,
      "step": 4699
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0679168701171875,
      "learning_rate": 1.761620291984794e-05,
      "loss": 0.9101,
      "step": 4700
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.154226064682007,
      "learning_rate": 1.761556400345015e-05,
      "loss": 0.9469,
      "step": 4701
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2358157634735107,
      "learning_rate": 1.7614925087052362e-05,
      "loss": 1.0418,
      "step": 4702
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.996971845626831,
      "learning_rate": 1.761428617065457e-05,
      "loss": 1.036,
      "step": 4703
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.613013505935669,
      "learning_rate": 1.7613647254256782e-05,
      "loss": 0.8407,
      "step": 4704
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.421922206878662,
      "learning_rate": 1.7613008337858994e-05,
      "loss": 1.1702,
      "step": 4705
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.621382236480713,
      "learning_rate": 1.7612369421461205e-05,
      "loss": 0.8565,
      "step": 4706
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1573524475097656,
      "learning_rate": 1.7611730505063413e-05,
      "loss": 1.0592,
      "step": 4707
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.79254150390625,
      "learning_rate": 1.7611091588665625e-05,
      "loss": 1.2491,
      "step": 4708
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.8005669116973877,
      "learning_rate": 1.7610452672267837e-05,
      "loss": 1.0713,
      "step": 4709
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.9853925704956055,
      "learning_rate": 1.7609813755870048e-05,
      "loss": 1.3517,
      "step": 4710
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9435243606567383,
      "learning_rate": 1.7609174839472256e-05,
      "loss": 0.8726,
      "step": 4711
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.159966230392456,
      "learning_rate": 1.7608535923074468e-05,
      "loss": 0.9605,
      "step": 4712
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1377100944519043,
      "learning_rate": 1.7607897006676676e-05,
      "loss": 1.4177,
      "step": 4713
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2274889945983887,
      "learning_rate": 1.760725809027889e-05,
      "loss": 1.2525,
      "step": 4714
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4900999069213867,
      "learning_rate": 1.76066191738811e-05,
      "loss": 0.9767,
      "step": 4715
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1159653663635254,
      "learning_rate": 1.760598025748331e-05,
      "loss": 0.9778,
      "step": 4716
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8197453022003174,
      "learning_rate": 1.760534134108552e-05,
      "loss": 0.9287,
      "step": 4717
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.310375690460205,
      "learning_rate": 1.760470242468773e-05,
      "loss": 0.8901,
      "step": 4718
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.498169422149658,
      "learning_rate": 1.7604063508289942e-05,
      "loss": 1.2231,
      "step": 4719
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.63772714138031,
      "learning_rate": 1.7603424591892153e-05,
      "loss": 0.9555,
      "step": 4720
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.218438625335693,
      "learning_rate": 1.760278567549436e-05,
      "loss": 1.0269,
      "step": 4721
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0331075191497803,
      "learning_rate": 1.7602146759096573e-05,
      "loss": 0.8898,
      "step": 4722
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.072925090789795,
      "learning_rate": 1.7601507842698785e-05,
      "loss": 1.148,
      "step": 4723
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.319355010986328,
      "learning_rate": 1.7600868926300996e-05,
      "loss": 1.0062,
      "step": 4724
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.220139980316162,
      "learning_rate": 1.7600230009903204e-05,
      "loss": 1.0897,
      "step": 4725
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.857783317565918,
      "learning_rate": 1.7599591093505416e-05,
      "loss": 0.9029,
      "step": 4726
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0361452102661133,
      "learning_rate": 1.7598952177107627e-05,
      "loss": 1.0382,
      "step": 4727
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2900586128234863,
      "learning_rate": 1.759831326070984e-05,
      "loss": 1.0434,
      "step": 4728
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1201226711273193,
      "learning_rate": 1.7597674344312047e-05,
      "loss": 1.0788,
      "step": 4729
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.493403911590576,
      "learning_rate": 1.759703542791426e-05,
      "loss": 1.0487,
      "step": 4730
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.816826343536377,
      "learning_rate": 1.759639651151647e-05,
      "loss": 0.9309,
      "step": 4731
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2874374389648438,
      "learning_rate": 1.7595757595118682e-05,
      "loss": 1.3563,
      "step": 4732
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1191718578338623,
      "learning_rate": 1.759511867872089e-05,
      "loss": 1.3487,
      "step": 4733
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.441059589385986,
      "learning_rate": 1.75944797623231e-05,
      "loss": 1.318,
      "step": 4734
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0774693489074707,
      "learning_rate": 1.7593840845925313e-05,
      "loss": 0.9615,
      "step": 4735
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9904859066009521,
      "learning_rate": 1.7593201929527525e-05,
      "loss": 0.9089,
      "step": 4736
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.7930095195770264,
      "learning_rate": 1.7592563013129733e-05,
      "loss": 0.7822,
      "step": 4737
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.435720920562744,
      "learning_rate": 1.7591924096731944e-05,
      "loss": 0.9826,
      "step": 4738
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1914737224578857,
      "learning_rate": 1.7591285180334156e-05,
      "loss": 0.7795,
      "step": 4739
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.604332685470581,
      "learning_rate": 1.7590646263936367e-05,
      "loss": 0.9792,
      "step": 4740
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.691585063934326,
      "learning_rate": 1.7590007347538576e-05,
      "loss": 0.8703,
      "step": 4741
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4972310066223145,
      "learning_rate": 1.7589368431140787e-05,
      "loss": 0.9073,
      "step": 4742
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.257338523864746,
      "learning_rate": 1.7588729514743e-05,
      "loss": 1.4325,
      "step": 4743
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.8768980503082275,
      "learning_rate": 1.758809059834521e-05,
      "loss": 1.1059,
      "step": 4744
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.582646131515503,
      "learning_rate": 1.758745168194742e-05,
      "loss": 0.3678,
      "step": 4745
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.974448323249817,
      "learning_rate": 1.758681276554963e-05,
      "loss": 0.838,
      "step": 4746
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8266104459762573,
      "learning_rate": 1.7586173849151838e-05,
      "loss": 0.8482,
      "step": 4747
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.997909426689148,
      "learning_rate": 1.7585534932754053e-05,
      "loss": 0.9737,
      "step": 4748
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.2830700874328613,
      "learning_rate": 1.758489601635626e-05,
      "loss": 1.026,
      "step": 4749
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.855865240097046,
      "learning_rate": 1.7584257099958473e-05,
      "loss": 0.9992,
      "step": 4750
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.939487099647522,
      "learning_rate": 1.758361818356068e-05,
      "loss": 0.9732,
      "step": 4751
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6492418050765991,
      "learning_rate": 1.7582979267162892e-05,
      "loss": 0.7468,
      "step": 4752
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5928595066070557,
      "learning_rate": 1.7582340350765104e-05,
      "loss": 1.1251,
      "step": 4753
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8741132020950317,
      "learning_rate": 1.7581701434367315e-05,
      "loss": 0.906,
      "step": 4754
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8908817768096924,
      "learning_rate": 1.7581062517969524e-05,
      "loss": 0.9461,
      "step": 4755
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4865825176239014,
      "learning_rate": 1.7580423601571735e-05,
      "loss": 0.9186,
      "step": 4756
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9250357151031494,
      "learning_rate": 1.7579784685173947e-05,
      "loss": 0.971,
      "step": 4757
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.029017925262451,
      "learning_rate": 1.7579145768776158e-05,
      "loss": 0.8975,
      "step": 4758
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.0470218658447266,
      "learning_rate": 1.7578506852378366e-05,
      "loss": 1.1294,
      "step": 4759
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.026160478591919,
      "learning_rate": 1.7577867935980578e-05,
      "loss": 0.9471,
      "step": 4760
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.647752285003662,
      "learning_rate": 1.757722901958279e-05,
      "loss": 0.9255,
      "step": 4761
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.993873119354248,
      "learning_rate": 1.7576590103185e-05,
      "loss": 0.9389,
      "step": 4762
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.8034019470214844,
      "learning_rate": 1.757595118678721e-05,
      "loss": 1.2351,
      "step": 4763
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.309661865234375,
      "learning_rate": 1.757531227038942e-05,
      "loss": 1.0665,
      "step": 4764
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.315774440765381,
      "learning_rate": 1.7574673353991632e-05,
      "loss": 1.0127,
      "step": 4765
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1194067001342773,
      "learning_rate": 1.7574034437593844e-05,
      "loss": 0.884,
      "step": 4766
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.353572845458984,
      "learning_rate": 1.7573395521196052e-05,
      "loss": 1.0187,
      "step": 4767
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.638954758644104,
      "learning_rate": 1.7572756604798264e-05,
      "loss": 0.8796,
      "step": 4768
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6094895601272583,
      "learning_rate": 1.7572117688400475e-05,
      "loss": 0.3034,
      "step": 4769
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.724806308746338,
      "learning_rate": 1.7571478772002687e-05,
      "loss": 0.9894,
      "step": 4770
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.443824052810669,
      "learning_rate": 1.7570839855604895e-05,
      "loss": 1.0319,
      "step": 4771
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6415421962738037,
      "learning_rate": 1.7570200939207106e-05,
      "loss": 0.9197,
      "step": 4772
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2252538204193115,
      "learning_rate": 1.7569562022809318e-05,
      "loss": 1.2101,
      "step": 4773
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6557459831237793,
      "learning_rate": 1.756892310641153e-05,
      "loss": 0.9233,
      "step": 4774
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.222379684448242,
      "learning_rate": 1.7568284190013738e-05,
      "loss": 0.8431,
      "step": 4775
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4796245098114014,
      "learning_rate": 1.756764527361595e-05,
      "loss": 1.0256,
      "step": 4776
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.8007428646087646,
      "learning_rate": 1.756700635721816e-05,
      "loss": 1.12,
      "step": 4777
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.282320737838745,
      "learning_rate": 1.7566367440820372e-05,
      "loss": 0.9584,
      "step": 4778
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0504610538482666,
      "learning_rate": 1.756572852442258e-05,
      "loss": 0.8384,
      "step": 4779
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.743913173675537,
      "learning_rate": 1.7565089608024792e-05,
      "loss": 1.0715,
      "step": 4780
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8154677152633667,
      "learning_rate": 1.7564450691627004e-05,
      "loss": 0.9105,
      "step": 4781
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.149724245071411,
      "learning_rate": 1.7563811775229215e-05,
      "loss": 1.0043,
      "step": 4782
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7703857421875,
      "learning_rate": 1.7563172858831423e-05,
      "loss": 0.7487,
      "step": 4783
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9197832345962524,
      "learning_rate": 1.7562533942433635e-05,
      "loss": 0.4407,
      "step": 4784
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4629135131835938,
      "learning_rate": 1.7561895026035843e-05,
      "loss": 1.008,
      "step": 4785
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.354393720626831,
      "learning_rate": 1.7561256109638058e-05,
      "loss": 0.8087,
      "step": 4786
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0161924362182617,
      "learning_rate": 1.7560617193240266e-05,
      "loss": 0.9238,
      "step": 4787
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4971706867218018,
      "learning_rate": 1.7559978276842478e-05,
      "loss": 1.0091,
      "step": 4788
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7070188522338867,
      "learning_rate": 1.7559339360444686e-05,
      "loss": 0.8118,
      "step": 4789
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2299561500549316,
      "learning_rate": 1.7558700444046897e-05,
      "loss": 0.8996,
      "step": 4790
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8613675832748413,
      "learning_rate": 1.755806152764911e-05,
      "loss": 0.8759,
      "step": 4791
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5846071243286133,
      "learning_rate": 1.755742261125132e-05,
      "loss": 0.9087,
      "step": 4792
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7785260677337646,
      "learning_rate": 1.755678369485353e-05,
      "loss": 0.856,
      "step": 4793
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.5130984783172607,
      "learning_rate": 1.755614477845574e-05,
      "loss": 0.9902,
      "step": 4794
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3927289247512817,
      "learning_rate": 1.755550586205795e-05,
      "loss": 0.413,
      "step": 4795
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6517536640167236,
      "learning_rate": 1.7554866945660163e-05,
      "loss": 0.9254,
      "step": 4796
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6691263914108276,
      "learning_rate": 1.755422802926237e-05,
      "loss": 0.8748,
      "step": 4797
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5391979217529297,
      "learning_rate": 1.7553589112864583e-05,
      "loss": 0.9044,
      "step": 4798
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6427468061447144,
      "learning_rate": 1.7552950196466794e-05,
      "loss": 0.794,
      "step": 4799
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.979172706604004,
      "learning_rate": 1.7552311280069006e-05,
      "loss": 1.1866,
      "step": 4800
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.023275375366211,
      "learning_rate": 1.7551672363671214e-05,
      "loss": 1.3379,
      "step": 4801
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.638887643814087,
      "learning_rate": 1.7551033447273426e-05,
      "loss": 0.8827,
      "step": 4802
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5527970790863037,
      "learning_rate": 1.7550394530875637e-05,
      "loss": 1.0013,
      "step": 4803
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.871993064880371,
      "learning_rate": 1.754975561447785e-05,
      "loss": 0.5475,
      "step": 4804
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3372154235839844,
      "learning_rate": 1.7549116698080057e-05,
      "loss": 0.9945,
      "step": 4805
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.219648599624634,
      "learning_rate": 1.754847778168227e-05,
      "loss": 0.9058,
      "step": 4806
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.46689510345459,
      "learning_rate": 1.754783886528448e-05,
      "loss": 1.189,
      "step": 4807
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.46579909324646,
      "learning_rate": 1.754719994888669e-05,
      "loss": 1.2723,
      "step": 4808
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.2636969089508057,
      "learning_rate": 1.75465610324889e-05,
      "loss": 0.8821,
      "step": 4809
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5240041017532349,
      "learning_rate": 1.754592211609111e-05,
      "loss": 0.44,
      "step": 4810
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0268306732177734,
      "learning_rate": 1.7545283199693323e-05,
      "loss": 0.834,
      "step": 4811
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9596127271652222,
      "learning_rate": 1.7544644283295534e-05,
      "loss": 0.9791,
      "step": 4812
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.564974308013916,
      "learning_rate": 1.7544005366897743e-05,
      "loss": 0.9954,
      "step": 4813
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.6025311946868896,
      "learning_rate": 1.7543366450499954e-05,
      "loss": 1.0393,
      "step": 4814
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.162019968032837,
      "learning_rate": 1.7542727534102166e-05,
      "loss": 1.0703,
      "step": 4815
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.264507293701172,
      "learning_rate": 1.7542088617704377e-05,
      "loss": 0.926,
      "step": 4816
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.38930344581604,
      "learning_rate": 1.7541449701306585e-05,
      "loss": 0.8248,
      "step": 4817
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.285026788711548,
      "learning_rate": 1.7540810784908797e-05,
      "loss": 1.0699,
      "step": 4818
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.9544451236724854,
      "learning_rate": 1.7540171868511005e-05,
      "loss": 1.0419,
      "step": 4819
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.094400405883789,
      "learning_rate": 1.753953295211322e-05,
      "loss": 1.1592,
      "step": 4820
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9873923063278198,
      "learning_rate": 1.7538894035715428e-05,
      "loss": 0.988,
      "step": 4821
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.738722562789917,
      "learning_rate": 1.753825511931764e-05,
      "loss": 0.8985,
      "step": 4822
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.8822989463806152,
      "learning_rate": 1.7537616202919848e-05,
      "loss": 1.0045,
      "step": 4823
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7955269813537598,
      "learning_rate": 1.7536977286522063e-05,
      "loss": 0.881,
      "step": 4824
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.766654014587402,
      "learning_rate": 1.753633837012427e-05,
      "loss": 1.258,
      "step": 4825
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8753598928451538,
      "learning_rate": 1.7535699453726482e-05,
      "loss": 0.9161,
      "step": 4826
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.76548171043396,
      "learning_rate": 1.753506053732869e-05,
      "loss": 0.9567,
      "step": 4827
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.088514566421509,
      "learning_rate": 1.7534421620930902e-05,
      "loss": 0.9356,
      "step": 4828
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.83580207824707,
      "learning_rate": 1.7533782704533114e-05,
      "loss": 1.0207,
      "step": 4829
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.12911319732666,
      "learning_rate": 1.7533143788135325e-05,
      "loss": 0.9385,
      "step": 4830
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1336686611175537,
      "learning_rate": 1.7532504871737533e-05,
      "loss": 0.8385,
      "step": 4831
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5664749145507812,
      "learning_rate": 1.7531865955339745e-05,
      "loss": 0.9047,
      "step": 4832
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7880779504776,
      "learning_rate": 1.7531227038941957e-05,
      "loss": 0.8945,
      "step": 4833
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9805268049240112,
      "learning_rate": 1.7530588122544165e-05,
      "loss": 1.0026,
      "step": 4834
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2982990741729736,
      "learning_rate": 1.7529949206146376e-05,
      "loss": 0.9963,
      "step": 4835
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6841154098510742,
      "learning_rate": 1.7529310289748588e-05,
      "loss": 0.8776,
      "step": 4836
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1705260276794434,
      "learning_rate": 1.75286713733508e-05,
      "loss": 0.9861,
      "step": 4837
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.190906047821045,
      "learning_rate": 1.7528032456953008e-05,
      "loss": 1.0848,
      "step": 4838
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5692228078842163,
      "learning_rate": 1.752739354055522e-05,
      "loss": 0.8336,
      "step": 4839
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.544295310974121,
      "learning_rate": 1.752675462415743e-05,
      "loss": 1.157,
      "step": 4840
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.692509412765503,
      "learning_rate": 1.7526115707759642e-05,
      "loss": 1.0907,
      "step": 4841
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0840418338775635,
      "learning_rate": 1.752547679136185e-05,
      "loss": 1.0007,
      "step": 4842
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.835312843322754,
      "learning_rate": 1.7524837874964062e-05,
      "loss": 1.1081,
      "step": 4843
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.709242582321167,
      "learning_rate": 1.7524198958566273e-05,
      "loss": 1.0796,
      "step": 4844
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.214101314544678,
      "learning_rate": 1.7523560042168485e-05,
      "loss": 1.4918,
      "step": 4845
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.65407657623291,
      "learning_rate": 1.7522921125770693e-05,
      "loss": 0.9104,
      "step": 4846
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3708109855651855,
      "learning_rate": 1.7522282209372905e-05,
      "loss": 1.0453,
      "step": 4847
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2830049991607666,
      "learning_rate": 1.7521643292975116e-05,
      "loss": 1.0376,
      "step": 4848
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8780766725540161,
      "learning_rate": 1.7521004376577328e-05,
      "loss": 0.866,
      "step": 4849
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6642146110534668,
      "learning_rate": 1.7520365460179536e-05,
      "loss": 0.9725,
      "step": 4850
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.546872138977051,
      "learning_rate": 1.7519726543781747e-05,
      "loss": 1.0761,
      "step": 4851
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8871259689331055,
      "learning_rate": 1.7519087627383956e-05,
      "loss": 0.9485,
      "step": 4852
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.9988179206848145,
      "learning_rate": 1.751844871098617e-05,
      "loss": 1.185,
      "step": 4853
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.968108892440796,
      "learning_rate": 1.751780979458838e-05,
      "loss": 0.8292,
      "step": 4854
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0947673320770264,
      "learning_rate": 1.751717087819059e-05,
      "loss": 1.0043,
      "step": 4855
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.214740753173828,
      "learning_rate": 1.75165319617928e-05,
      "loss": 1.0037,
      "step": 4856
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.606235980987549,
      "learning_rate": 1.751589304539501e-05,
      "loss": 0.883,
      "step": 4857
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6597094535827637,
      "learning_rate": 1.751525412899722e-05,
      "loss": 0.8609,
      "step": 4858
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.12598991394043,
      "learning_rate": 1.7514615212599433e-05,
      "loss": 0.9331,
      "step": 4859
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5487849712371826,
      "learning_rate": 1.751397629620164e-05,
      "loss": 1.0163,
      "step": 4860
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.778835296630859,
      "learning_rate": 1.7513337379803853e-05,
      "loss": 1.3222,
      "step": 4861
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9948982000350952,
      "learning_rate": 1.7512698463406064e-05,
      "loss": 1.0827,
      "step": 4862
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4651520252227783,
      "learning_rate": 1.7512059547008276e-05,
      "loss": 0.4668,
      "step": 4863
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8402003049850464,
      "learning_rate": 1.7511420630610484e-05,
      "loss": 1.0187,
      "step": 4864
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.145799160003662,
      "learning_rate": 1.7510781714212696e-05,
      "loss": 0.9457,
      "step": 4865
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.613007068634033,
      "learning_rate": 1.7510142797814907e-05,
      "loss": 1.0853,
      "step": 4866
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.195394515991211,
      "learning_rate": 1.750950388141712e-05,
      "loss": 1.105,
      "step": 4867
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.94186270236969,
      "learning_rate": 1.7508864965019327e-05,
      "loss": 0.9461,
      "step": 4868
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6933910846710205,
      "learning_rate": 1.750822604862154e-05,
      "loss": 1.0015,
      "step": 4869
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.212369203567505,
      "learning_rate": 1.750758713222375e-05,
      "loss": 0.4654,
      "step": 4870
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.470683574676514,
      "learning_rate": 1.750694821582596e-05,
      "loss": 1.4527,
      "step": 4871
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.634633779525757,
      "learning_rate": 1.750630929942817e-05,
      "loss": 0.913,
      "step": 4872
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.5483438968658447,
      "learning_rate": 1.750567038303038e-05,
      "loss": 0.9765,
      "step": 4873
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.5306599140167236,
      "learning_rate": 1.7505031466632593e-05,
      "loss": 0.8684,
      "step": 4874
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5964436531066895,
      "learning_rate": 1.7504392550234804e-05,
      "loss": 1.0372,
      "step": 4875
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6877355575561523,
      "learning_rate": 1.7503753633837012e-05,
      "loss": 1.0526,
      "step": 4876
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8899189233779907,
      "learning_rate": 1.7503114717439224e-05,
      "loss": 0.9108,
      "step": 4877
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8561917543411255,
      "learning_rate": 1.7502475801041435e-05,
      "loss": 0.9432,
      "step": 4878
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.342052936553955,
      "learning_rate": 1.7501836884643647e-05,
      "loss": 1.0954,
      "step": 4879
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7358922958374023,
      "learning_rate": 1.7501197968245855e-05,
      "loss": 0.8055,
      "step": 4880
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.685636520385742,
      "learning_rate": 1.7500559051848067e-05,
      "loss": 1.6025,
      "step": 4881
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.8425276279449463,
      "learning_rate": 1.7499920135450278e-05,
      "loss": 1.1444,
      "step": 4882
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.371603965759277,
      "learning_rate": 1.749928121905249e-05,
      "loss": 0.8494,
      "step": 4883
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5709502696990967,
      "learning_rate": 1.7498642302654698e-05,
      "loss": 1.0187,
      "step": 4884
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.2511329650878906,
      "learning_rate": 1.749800338625691e-05,
      "loss": 1.3449,
      "step": 4885
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8795654773712158,
      "learning_rate": 1.749736446985912e-05,
      "loss": 0.9171,
      "step": 4886
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6392433643341064,
      "learning_rate": 1.7496725553461333e-05,
      "loss": 0.9503,
      "step": 4887
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.081264972686768,
      "learning_rate": 1.749608663706354e-05,
      "loss": 1.3253,
      "step": 4888
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.828505992889404,
      "learning_rate": 1.7495447720665752e-05,
      "loss": 1.1076,
      "step": 4889
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2710154056549072,
      "learning_rate": 1.749480880426796e-05,
      "loss": 0.92,
      "step": 4890
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0416905879974365,
      "learning_rate": 1.7494169887870175e-05,
      "loss": 0.8912,
      "step": 4891
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7072935104370117,
      "learning_rate": 1.7493530971472384e-05,
      "loss": 0.523,
      "step": 4892
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9661329984664917,
      "learning_rate": 1.7492892055074595e-05,
      "loss": 0.894,
      "step": 4893
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.732728362083435,
      "learning_rate": 1.7492253138676803e-05,
      "loss": 0.8902,
      "step": 4894
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.147726535797119,
      "learning_rate": 1.7491614222279015e-05,
      "loss": 0.995,
      "step": 4895
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.728285789489746,
      "learning_rate": 1.7490975305881226e-05,
      "loss": 1.0238,
      "step": 4896
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6400837898254395,
      "learning_rate": 1.7490336389483438e-05,
      "loss": 0.9378,
      "step": 4897
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1136865615844727,
      "learning_rate": 1.7489697473085646e-05,
      "loss": 0.8852,
      "step": 4898
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.572317361831665,
      "learning_rate": 1.7489058556687858e-05,
      "loss": 1.0738,
      "step": 4899
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5267670154571533,
      "learning_rate": 1.748841964029007e-05,
      "loss": 0.3231,
      "step": 4900
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.465942859649658,
      "learning_rate": 1.748778072389228e-05,
      "loss": 1.079,
      "step": 4901
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.690451145172119,
      "learning_rate": 1.748714180749449e-05,
      "loss": 1.3269,
      "step": 4902
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8511935472488403,
      "learning_rate": 1.74865028910967e-05,
      "loss": 0.9622,
      "step": 4903
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.786144733428955,
      "learning_rate": 1.7485863974698912e-05,
      "loss": 0.8861,
      "step": 4904
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.5831615924835205,
      "learning_rate": 1.7485225058301124e-05,
      "loss": 0.9944,
      "step": 4905
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.491556167602539,
      "learning_rate": 1.7484586141903332e-05,
      "loss": 1.4894,
      "step": 4906
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8505008220672607,
      "learning_rate": 1.7483947225505543e-05,
      "loss": 1.0744,
      "step": 4907
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5525668859481812,
      "learning_rate": 1.7483308309107755e-05,
      "loss": 0.4088,
      "step": 4908
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4580090045928955,
      "learning_rate": 1.7482669392709966e-05,
      "loss": 0.3009,
      "step": 4909
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2795281410217285,
      "learning_rate": 1.7482030476312175e-05,
      "loss": 0.9955,
      "step": 4910
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7276368141174316,
      "learning_rate": 1.7481391559914386e-05,
      "loss": 0.8801,
      "step": 4911
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6701416969299316,
      "learning_rate": 1.7480752643516598e-05,
      "loss": 0.8893,
      "step": 4912
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.670125126838684,
      "learning_rate": 1.748011372711881e-05,
      "loss": 0.882,
      "step": 4913
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5313758850097656,
      "learning_rate": 1.7479474810721017e-05,
      "loss": 1.0672,
      "step": 4914
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6384315490722656,
      "learning_rate": 1.747883589432323e-05,
      "loss": 0.3727,
      "step": 4915
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2343242168426514,
      "learning_rate": 1.747819697792544e-05,
      "loss": 1.0217,
      "step": 4916
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8678337335586548,
      "learning_rate": 1.7477558061527652e-05,
      "loss": 0.8952,
      "step": 4917
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.161545991897583,
      "learning_rate": 1.747691914512986e-05,
      "loss": 0.9132,
      "step": 4918
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1621627807617188,
      "learning_rate": 1.747628022873207e-05,
      "loss": 0.9178,
      "step": 4919
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.083719491958618,
      "learning_rate": 1.7475641312334283e-05,
      "loss": 0.9662,
      "step": 4920
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4870922565460205,
      "learning_rate": 1.7475002395936495e-05,
      "loss": 0.9837,
      "step": 4921
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.711057424545288,
      "learning_rate": 1.7474363479538703e-05,
      "loss": 0.9114,
      "step": 4922
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.96218204498291,
      "learning_rate": 1.7473724563140914e-05,
      "loss": 1.2401,
      "step": 4923
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.382215738296509,
      "learning_rate": 1.7473085646743123e-05,
      "loss": 0.9289,
      "step": 4924
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.640612840652466,
      "learning_rate": 1.7472446730345338e-05,
      "loss": 1.2097,
      "step": 4925
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.532224416732788,
      "learning_rate": 1.7471807813947546e-05,
      "loss": 1.0706,
      "step": 4926
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.253603458404541,
      "learning_rate": 1.7471168897549757e-05,
      "loss": 1.4562,
      "step": 4927
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.199172019958496,
      "learning_rate": 1.7470529981151965e-05,
      "loss": 0.929,
      "step": 4928
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9850777387619019,
      "learning_rate": 1.7469891064754177e-05,
      "loss": 1.0151,
      "step": 4929
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5211445093154907,
      "learning_rate": 1.746925214835639e-05,
      "loss": 0.9033,
      "step": 4930
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.980809688568115,
      "learning_rate": 1.74686132319586e-05,
      "loss": 0.8919,
      "step": 4931
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.353780746459961,
      "learning_rate": 1.7467974315560808e-05,
      "loss": 0.8461,
      "step": 4932
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.78960919380188,
      "learning_rate": 1.746733539916302e-05,
      "loss": 0.8455,
      "step": 4933
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.380212306976318,
      "learning_rate": 1.746669648276523e-05,
      "loss": 1.4449,
      "step": 4934
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.8763997554779053,
      "learning_rate": 1.7466057566367443e-05,
      "loss": 0.877,
      "step": 4935
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.228381872177124,
      "learning_rate": 1.746541864996965e-05,
      "loss": 0.9777,
      "step": 4936
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.364335536956787,
      "learning_rate": 1.7464779733571863e-05,
      "loss": 1.0457,
      "step": 4937
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.864428162574768,
      "learning_rate": 1.7464140817174074e-05,
      "loss": 0.4858,
      "step": 4938
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9443470239639282,
      "learning_rate": 1.7463501900776286e-05,
      "loss": 0.9036,
      "step": 4939
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.5064263343811035,
      "learning_rate": 1.7462862984378494e-05,
      "loss": 1.5689,
      "step": 4940
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.964199066162109,
      "learning_rate": 1.7462224067980705e-05,
      "loss": 1.3447,
      "step": 4941
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.486341714859009,
      "learning_rate": 1.7461585151582917e-05,
      "loss": 0.9835,
      "step": 4942
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.697279691696167,
      "learning_rate": 1.746094623518513e-05,
      "loss": 0.994,
      "step": 4943
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.581631064414978,
      "learning_rate": 1.7460307318787337e-05,
      "loss": 0.8213,
      "step": 4944
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0603392124176025,
      "learning_rate": 1.7459668402389548e-05,
      "loss": 1.0345,
      "step": 4945
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.477430820465088,
      "learning_rate": 1.745902948599176e-05,
      "loss": 1.7787,
      "step": 4946
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.181870698928833,
      "learning_rate": 1.745839056959397e-05,
      "loss": 0.9313,
      "step": 4947
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.4685893058776855,
      "learning_rate": 1.745775165319618e-05,
      "loss": 1.0,
      "step": 4948
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7958623170852661,
      "learning_rate": 1.745711273679839e-05,
      "loss": 0.9472,
      "step": 4949
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.195356369018555,
      "learning_rate": 1.7456473820400603e-05,
      "loss": 1.3789,
      "step": 4950
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7957704067230225,
      "learning_rate": 1.7455834904002814e-05,
      "loss": 0.9946,
      "step": 4951
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.8687214851379395,
      "learning_rate": 1.7455195987605022e-05,
      "loss": 0.9327,
      "step": 4952
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.270661354064941,
      "learning_rate": 1.7454557071207234e-05,
      "loss": 1.4156,
      "step": 4953
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0608768463134766,
      "learning_rate": 1.7453918154809445e-05,
      "loss": 0.9561,
      "step": 4954
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.50341534614563,
      "learning_rate": 1.7453279238411657e-05,
      "loss": 0.8284,
      "step": 4955
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.2812607288360596,
      "learning_rate": 1.7452640322013865e-05,
      "loss": 0.7511,
      "step": 4956
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5734987258911133,
      "learning_rate": 1.7452001405616077e-05,
      "loss": 0.9458,
      "step": 4957
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1610946655273438,
      "learning_rate": 1.7451362489218288e-05,
      "loss": 0.9637,
      "step": 4958
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6661269664764404,
      "learning_rate": 1.74507235728205e-05,
      "loss": 0.9103,
      "step": 4959
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1436944007873535,
      "learning_rate": 1.7450084656422708e-05,
      "loss": 0.9622,
      "step": 4960
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5345172882080078,
      "learning_rate": 1.744944574002492e-05,
      "loss": 0.8275,
      "step": 4961
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3479437828063965,
      "learning_rate": 1.7448806823627128e-05,
      "loss": 0.8166,
      "step": 4962
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.043835401535034,
      "learning_rate": 1.7448167907229342e-05,
      "loss": 1.0293,
      "step": 4963
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.694709539413452,
      "learning_rate": 1.744752899083155e-05,
      "loss": 0.7937,
      "step": 4964
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2429401874542236,
      "learning_rate": 1.7446890074433762e-05,
      "loss": 0.9936,
      "step": 4965
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.417351007461548,
      "learning_rate": 1.744625115803597e-05,
      "loss": 1.0421,
      "step": 4966
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.199958562850952,
      "learning_rate": 1.7445612241638182e-05,
      "loss": 0.9145,
      "step": 4967
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.525362968444824,
      "learning_rate": 1.7444973325240393e-05,
      "loss": 0.923,
      "step": 4968
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9696946144104004,
      "learning_rate": 1.7444334408842605e-05,
      "loss": 0.9974,
      "step": 4969
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4260382652282715,
      "learning_rate": 1.7443695492444813e-05,
      "loss": 0.9182,
      "step": 4970
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.112048149108887,
      "learning_rate": 1.7443056576047025e-05,
      "loss": 1.1891,
      "step": 4971
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.97138249874115,
      "learning_rate": 1.7442417659649236e-05,
      "loss": 0.9825,
      "step": 4972
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.066227674484253,
      "learning_rate": 1.7441778743251448e-05,
      "loss": 0.9808,
      "step": 4973
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.324774742126465,
      "learning_rate": 1.7441139826853656e-05,
      "loss": 1.0219,
      "step": 4974
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.062037944793701,
      "learning_rate": 1.7440500910455867e-05,
      "loss": 0.9867,
      "step": 4975
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7184562683105469,
      "learning_rate": 1.743986199405808e-05,
      "loss": 0.8874,
      "step": 4976
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.203914165496826,
      "learning_rate": 1.743922307766029e-05,
      "loss": 0.9084,
      "step": 4977
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0539422035217285,
      "learning_rate": 1.74385841612625e-05,
      "loss": 0.8416,
      "step": 4978
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.244455814361572,
      "learning_rate": 1.743794524486471e-05,
      "loss": 1.0823,
      "step": 4979
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.767984628677368,
      "learning_rate": 1.7437306328466922e-05,
      "loss": 0.8703,
      "step": 4980
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.916008949279785,
      "learning_rate": 1.7436667412069133e-05,
      "loss": 1.6687,
      "step": 4981
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.427173376083374,
      "learning_rate": 1.743602849567134e-05,
      "loss": 0.896,
      "step": 4982
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.239779472351074,
      "learning_rate": 1.7435389579273553e-05,
      "loss": 0.9387,
      "step": 4983
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.792668104171753,
      "learning_rate": 1.7434750662875765e-05,
      "loss": 0.8187,
      "step": 4984
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.601879119873047,
      "learning_rate": 1.7434111746477976e-05,
      "loss": 1.192,
      "step": 4985
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9338027238845825,
      "learning_rate": 1.7433472830080184e-05,
      "loss": 0.8853,
      "step": 4986
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0626909732818604,
      "learning_rate": 1.7432833913682396e-05,
      "loss": 0.9016,
      "step": 4987
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.055762529373169,
      "learning_rate": 1.7432194997284607e-05,
      "loss": 0.9416,
      "step": 4988
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4075088500976562,
      "learning_rate": 1.743155608088682e-05,
      "loss": 0.9953,
      "step": 4989
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5652668476104736,
      "learning_rate": 1.7430917164489027e-05,
      "loss": 1.0281,
      "step": 4990
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.341114044189453,
      "learning_rate": 1.743027824809124e-05,
      "loss": 0.8834,
      "step": 4991
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.435060977935791,
      "learning_rate": 1.742963933169345e-05,
      "loss": 1.0741,
      "step": 4992
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.364902973175049,
      "learning_rate": 1.7429000415295662e-05,
      "loss": 0.8961,
      "step": 4993
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0307018756866455,
      "learning_rate": 1.742836149889787e-05,
      "loss": 1.049,
      "step": 4994
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.144913911819458,
      "learning_rate": 1.742772258250008e-05,
      "loss": 1.0395,
      "step": 4995
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9812097549438477,
      "learning_rate": 1.742708366610229e-05,
      "loss": 0.8273,
      "step": 4996
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.291689395904541,
      "learning_rate": 1.7426444749704505e-05,
      "loss": 0.8985,
      "step": 4997
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.16793966293335,
      "learning_rate": 1.7425805833306713e-05,
      "loss": 1.0921,
      "step": 4998
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4566580057144165,
      "learning_rate": 1.7425166916908924e-05,
      "loss": 0.8519,
      "step": 4999
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.826148748397827,
      "learning_rate": 1.7424528000511132e-05,
      "loss": 1.3962,
      "step": 5000
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9566959142684937,
      "learning_rate": 1.7423889084113347e-05,
      "loss": 1.0442,
      "step": 5001
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8435745239257812,
      "learning_rate": 1.7423250167715556e-05,
      "loss": 0.8773,
      "step": 5002
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.092982053756714,
      "learning_rate": 1.7422611251317767e-05,
      "loss": 0.8606,
      "step": 5003
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0403387546539307,
      "learning_rate": 1.7421972334919975e-05,
      "loss": 0.8582,
      "step": 5004
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8559694290161133,
      "learning_rate": 1.7421333418522187e-05,
      "loss": 0.8998,
      "step": 5005
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8355212211608887,
      "learning_rate": 1.74206945021244e-05,
      "loss": 0.8725,
      "step": 5006
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.818413496017456,
      "learning_rate": 1.742005558572661e-05,
      "loss": 0.8656,
      "step": 5007
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7438321113586426,
      "learning_rate": 1.7419416669328818e-05,
      "loss": 0.9507,
      "step": 5008
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.774025321006775,
      "learning_rate": 1.741877775293103e-05,
      "loss": 0.8577,
      "step": 5009
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.791631817817688,
      "learning_rate": 1.741813883653324e-05,
      "loss": 0.8174,
      "step": 5010
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4521126747131348,
      "learning_rate": 1.7417499920135453e-05,
      "loss": 0.9904,
      "step": 5011
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.224705934524536,
      "learning_rate": 1.741686100373766e-05,
      "loss": 0.8768,
      "step": 5012
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.645033597946167,
      "learning_rate": 1.7416222087339872e-05,
      "loss": 0.4685,
      "step": 5013
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.464358329772949,
      "learning_rate": 1.7415583170942084e-05,
      "loss": 1.0129,
      "step": 5014
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8936271667480469,
      "learning_rate": 1.7414944254544295e-05,
      "loss": 1.0067,
      "step": 5015
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.146420478820801,
      "learning_rate": 1.7414305338146504e-05,
      "loss": 1.3324,
      "step": 5016
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.991919994354248,
      "learning_rate": 1.7413666421748715e-05,
      "loss": 0.9439,
      "step": 5017
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.226841449737549,
      "learning_rate": 1.7413027505350927e-05,
      "loss": 1.0111,
      "step": 5018
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1164913177490234,
      "learning_rate": 1.7412388588953138e-05,
      "loss": 1.01,
      "step": 5019
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.463000774383545,
      "learning_rate": 1.7411749672555346e-05,
      "loss": 1.0818,
      "step": 5020
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1336169242858887,
      "learning_rate": 1.7411110756157558e-05,
      "loss": 0.6307,
      "step": 5021
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8935972452163696,
      "learning_rate": 1.741047183975977e-05,
      "loss": 1.1997,
      "step": 5022
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9806492328643799,
      "learning_rate": 1.740983292336198e-05,
      "loss": 0.9211,
      "step": 5023
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.7020466327667236,
      "learning_rate": 1.740919400696419e-05,
      "loss": 1.1858,
      "step": 5024
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8082445859909058,
      "learning_rate": 1.74085550905664e-05,
      "loss": 0.988,
      "step": 5025
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.827541351318359,
      "learning_rate": 1.7407916174168612e-05,
      "loss": 1.5765,
      "step": 5026
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.584164142608643,
      "learning_rate": 1.7407277257770824e-05,
      "loss": 0.9042,
      "step": 5027
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5915584564208984,
      "learning_rate": 1.7406638341373032e-05,
      "loss": 0.9569,
      "step": 5028
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.586694240570068,
      "learning_rate": 1.7405999424975244e-05,
      "loss": 1.6288,
      "step": 5029
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.44710373878479,
      "learning_rate": 1.7405360508577455e-05,
      "loss": 0.8515,
      "step": 5030
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.125570774078369,
      "learning_rate": 1.7404721592179667e-05,
      "loss": 1.0387,
      "step": 5031
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.685359477996826,
      "learning_rate": 1.7404082675781875e-05,
      "loss": 1.5676,
      "step": 5032
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6610512733459473,
      "learning_rate": 1.7403443759384086e-05,
      "loss": 1.0003,
      "step": 5033
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3354135751724243,
      "learning_rate": 1.7402804842986295e-05,
      "loss": 0.8282,
      "step": 5034
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6975317001342773,
      "learning_rate": 1.740216592658851e-05,
      "loss": 0.9739,
      "step": 5035
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.437267541885376,
      "learning_rate": 1.7401527010190718e-05,
      "loss": 1.0582,
      "step": 5036
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.0897367000579834,
      "learning_rate": 1.740088809379293e-05,
      "loss": 1.1284,
      "step": 5037
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.904475450515747,
      "learning_rate": 1.7400249177395137e-05,
      "loss": 1.0167,
      "step": 5038
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.91447114944458,
      "learning_rate": 1.739961026099735e-05,
      "loss": 1.0081,
      "step": 5039
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9721195697784424,
      "learning_rate": 1.739897134459956e-05,
      "loss": 0.867,
      "step": 5040
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.034385919570923,
      "learning_rate": 1.7398332428201772e-05,
      "loss": 1.0319,
      "step": 5041
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.715052127838135,
      "learning_rate": 1.739769351180398e-05,
      "loss": 1.8634,
      "step": 5042
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0327484607696533,
      "learning_rate": 1.739705459540619e-05,
      "loss": 0.8248,
      "step": 5043
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3228297233581543,
      "learning_rate": 1.7396415679008403e-05,
      "loss": 1.233,
      "step": 5044
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.6435747146606445,
      "learning_rate": 1.7395776762610615e-05,
      "loss": 0.8162,
      "step": 5045
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1480791568756104,
      "learning_rate": 1.7395137846212823e-05,
      "loss": 0.9626,
      "step": 5046
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.19697904586792,
      "learning_rate": 1.7394498929815034e-05,
      "loss": 1.2557,
      "step": 5047
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8153650760650635,
      "learning_rate": 1.7393860013417246e-05,
      "loss": 0.6947,
      "step": 5048
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9267829656600952,
      "learning_rate": 1.7393221097019458e-05,
      "loss": 1.0318,
      "step": 5049
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7558375597000122,
      "learning_rate": 1.7392582180621666e-05,
      "loss": 0.8943,
      "step": 5050
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.84099280834198,
      "learning_rate": 1.7391943264223877e-05,
      "loss": 0.9461,
      "step": 5051
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.121131420135498,
      "learning_rate": 1.739130434782609e-05,
      "loss": 0.8573,
      "step": 5052
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0184788703918457,
      "learning_rate": 1.73906654314283e-05,
      "loss": 0.9598,
      "step": 5053
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.7357563972473145,
      "learning_rate": 1.739002651503051e-05,
      "loss": 1.1429,
      "step": 5054
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.541393518447876,
      "learning_rate": 1.738938759863272e-05,
      "loss": 0.8757,
      "step": 5055
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.039945602416992,
      "learning_rate": 1.738874868223493e-05,
      "loss": 1.318,
      "step": 5056
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.15501070022583,
      "learning_rate": 1.7388109765837143e-05,
      "loss": 0.7377,
      "step": 5057
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.203123092651367,
      "learning_rate": 1.738747084943935e-05,
      "loss": 1.1468,
      "step": 5058
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7273534536361694,
      "learning_rate": 1.7386831933041563e-05,
      "loss": 0.9531,
      "step": 5059
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.110435962677002,
      "learning_rate": 1.7386193016643774e-05,
      "loss": 1.1676,
      "step": 5060
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2380282878875732,
      "learning_rate": 1.7385554100245986e-05,
      "loss": 0.9743,
      "step": 5061
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.536386489868164,
      "learning_rate": 1.7384915183848194e-05,
      "loss": 0.7882,
      "step": 5062
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.846861362457275,
      "learning_rate": 1.7384276267450406e-05,
      "loss": 1.3,
      "step": 5063
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1410131454467773,
      "learning_rate": 1.7383637351052617e-05,
      "loss": 0.8595,
      "step": 5064
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6849370002746582,
      "learning_rate": 1.738299843465483e-05,
      "loss": 0.9588,
      "step": 5065
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3475158214569092,
      "learning_rate": 1.7382359518257037e-05,
      "loss": 0.5206,
      "step": 5066
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0540761947631836,
      "learning_rate": 1.738172060185925e-05,
      "loss": 0.9654,
      "step": 5067
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.967966914176941,
      "learning_rate": 1.738108168546146e-05,
      "loss": 0.9678,
      "step": 5068
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4802157878875732,
      "learning_rate": 1.738044276906367e-05,
      "loss": 0.8759,
      "step": 5069
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.7687151432037354,
      "learning_rate": 1.737980385266588e-05,
      "loss": 1.0767,
      "step": 5070
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.971879482269287,
      "learning_rate": 1.737916493626809e-05,
      "loss": 0.8701,
      "step": 5071
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.826783537864685,
      "learning_rate": 1.73785260198703e-05,
      "loss": 0.9057,
      "step": 5072
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8903393745422363,
      "learning_rate": 1.7377887103472514e-05,
      "loss": 0.9723,
      "step": 5073
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.558688759803772,
      "learning_rate": 1.7377248187074723e-05,
      "loss": 0.493,
      "step": 5074
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.970222473144531,
      "learning_rate": 1.7376609270676934e-05,
      "loss": 1.263,
      "step": 5075
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4856367111206055,
      "learning_rate": 1.7375970354279142e-05,
      "loss": 0.6549,
      "step": 5076
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0094778537750244,
      "learning_rate": 1.7375331437881354e-05,
      "loss": 1.0946,
      "step": 5077
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.717148780822754,
      "learning_rate": 1.7374692521483565e-05,
      "loss": 1.1204,
      "step": 5078
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6396979093551636,
      "learning_rate": 1.7374053605085777e-05,
      "loss": 0.4139,
      "step": 5079
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9901546239852905,
      "learning_rate": 1.7373414688687985e-05,
      "loss": 0.9336,
      "step": 5080
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.095060110092163,
      "learning_rate": 1.7372775772290197e-05,
      "loss": 0.9489,
      "step": 5081
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5582726001739502,
      "learning_rate": 1.7372136855892408e-05,
      "loss": 0.8536,
      "step": 5082
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0266075134277344,
      "learning_rate": 1.737149793949462e-05,
      "loss": 1.0162,
      "step": 5083
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.868818759918213,
      "learning_rate": 1.7370859023096828e-05,
      "loss": 0.8054,
      "step": 5084
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1079282760620117,
      "learning_rate": 1.737022010669904e-05,
      "loss": 0.8634,
      "step": 5085
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2042510509490967,
      "learning_rate": 1.736958119030125e-05,
      "loss": 0.9908,
      "step": 5086
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1975796222686768,
      "learning_rate": 1.7368942273903462e-05,
      "loss": 0.9684,
      "step": 5087
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8903279304504395,
      "learning_rate": 1.736830335750567e-05,
      "loss": 1.1759,
      "step": 5088
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0842626094818115,
      "learning_rate": 1.7367664441107882e-05,
      "loss": 0.8767,
      "step": 5089
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9388713836669922,
      "learning_rate": 1.7367025524710094e-05,
      "loss": 1.2029,
      "step": 5090
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1528873443603516,
      "learning_rate": 1.7366386608312305e-05,
      "loss": 1.012,
      "step": 5091
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.05446195602417,
      "learning_rate": 1.7365747691914513e-05,
      "loss": 0.9016,
      "step": 5092
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.263932704925537,
      "learning_rate": 1.7365108775516725e-05,
      "loss": 1.0683,
      "step": 5093
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8899286985397339,
      "learning_rate": 1.7364469859118937e-05,
      "loss": 0.8388,
      "step": 5094
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.498696327209473,
      "learning_rate": 1.7363830942721148e-05,
      "loss": 1.8993,
      "step": 5095
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.234971761703491,
      "learning_rate": 1.7363192026323356e-05,
      "loss": 1.0477,
      "step": 5096
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.9326491355896,
      "learning_rate": 1.7362553109925568e-05,
      "loss": 1.1532,
      "step": 5097
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.854304552078247,
      "learning_rate": 1.736191419352778e-05,
      "loss": 0.8675,
      "step": 5098
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.6992857456207275,
      "learning_rate": 1.736127527712999e-05,
      "loss": 1.2261,
      "step": 5099
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.793910264968872,
      "learning_rate": 1.73606363607322e-05,
      "loss": 0.9733,
      "step": 5100
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.6123757362365723,
      "learning_rate": 1.735999744433441e-05,
      "loss": 1.052,
      "step": 5101
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8923184871673584,
      "learning_rate": 1.7359358527936622e-05,
      "loss": 1.1918,
      "step": 5102
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.268232583999634,
      "learning_rate": 1.7358719611538834e-05,
      "loss": 0.9849,
      "step": 5103
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1302058696746826,
      "learning_rate": 1.7358080695141042e-05,
      "loss": 0.952,
      "step": 5104
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.283534526824951,
      "learning_rate": 1.7357441778743253e-05,
      "loss": 0.9637,
      "step": 5105
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7640925645828247,
      "learning_rate": 1.735680286234546e-05,
      "loss": 0.8985,
      "step": 5106
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4101444482803345,
      "learning_rate": 1.7356163945947676e-05,
      "loss": 0.3385,
      "step": 5107
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.531510591506958,
      "learning_rate": 1.7355525029549885e-05,
      "loss": 0.4033,
      "step": 5108
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0077993869781494,
      "learning_rate": 1.7354886113152096e-05,
      "loss": 0.942,
      "step": 5109
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2979187965393066,
      "learning_rate": 1.7354247196754304e-05,
      "loss": 1.2636,
      "step": 5110
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1425423622131348,
      "learning_rate": 1.7353608280356516e-05,
      "loss": 0.9696,
      "step": 5111
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.051717758178711,
      "learning_rate": 1.7352969363958727e-05,
      "loss": 0.8186,
      "step": 5112
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.674504041671753,
      "learning_rate": 1.735233044756094e-05,
      "loss": 1.1703,
      "step": 5113
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1544792652130127,
      "learning_rate": 1.7351691531163147e-05,
      "loss": 0.8332,
      "step": 5114
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.9842023849487305,
      "learning_rate": 1.735105261476536e-05,
      "loss": 1.2856,
      "step": 5115
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8717525005340576,
      "learning_rate": 1.735041369836757e-05,
      "loss": 0.9261,
      "step": 5116
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4309191703796387,
      "learning_rate": 1.7349774781969782e-05,
      "loss": 1.0372,
      "step": 5117
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.550532579421997,
      "learning_rate": 1.734913586557199e-05,
      "loss": 1.275,
      "step": 5118
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.387352705001831,
      "learning_rate": 1.73484969491742e-05,
      "loss": 1.2421,
      "step": 5119
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.76906156539917,
      "learning_rate": 1.7347858032776413e-05,
      "loss": 0.9694,
      "step": 5120
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.926713943481445,
      "learning_rate": 1.7347219116378625e-05,
      "loss": 1.5222,
      "step": 5121
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8211352825164795,
      "learning_rate": 1.7346580199980833e-05,
      "loss": 1.4426,
      "step": 5122
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.649917483329773,
      "learning_rate": 1.7345941283583044e-05,
      "loss": 0.8406,
      "step": 5123
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7702460289001465,
      "learning_rate": 1.7345302367185256e-05,
      "loss": 1.1274,
      "step": 5124
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.126004695892334,
      "learning_rate": 1.7344663450787467e-05,
      "loss": 0.947,
      "step": 5125
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6643441915512085,
      "learning_rate": 1.7344024534389676e-05,
      "loss": 0.9899,
      "step": 5126
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0819060802459717,
      "learning_rate": 1.7343385617991887e-05,
      "loss": 0.9277,
      "step": 5127
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0922298431396484,
      "learning_rate": 1.73427467015941e-05,
      "loss": 1.0037,
      "step": 5128
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3392159938812256,
      "learning_rate": 1.734210778519631e-05,
      "loss": 1.0571,
      "step": 5129
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8166072368621826,
      "learning_rate": 1.734146886879852e-05,
      "loss": 0.8854,
      "step": 5130
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8225528001785278,
      "learning_rate": 1.734082995240073e-05,
      "loss": 0.7862,
      "step": 5131
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9017744064331055,
      "learning_rate": 1.734019103600294e-05,
      "loss": 0.9326,
      "step": 5132
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9133374691009521,
      "learning_rate": 1.7339552119605153e-05,
      "loss": 0.9137,
      "step": 5133
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8705168962478638,
      "learning_rate": 1.733891320320736e-05,
      "loss": 0.97,
      "step": 5134
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0718932151794434,
      "learning_rate": 1.7338274286809573e-05,
      "loss": 0.9543,
      "step": 5135
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.528745412826538,
      "learning_rate": 1.7337635370411784e-05,
      "loss": 1.0308,
      "step": 5136
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.148458480834961,
      "learning_rate": 1.7336996454013996e-05,
      "loss": 1.0133,
      "step": 5137
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8386605978012085,
      "learning_rate": 1.7336357537616204e-05,
      "loss": 0.4534,
      "step": 5138
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.205655574798584,
      "learning_rate": 1.7335718621218415e-05,
      "loss": 1.83,
      "step": 5139
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1408867835998535,
      "learning_rate": 1.7335079704820627e-05,
      "loss": 1.1785,
      "step": 5140
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0849525928497314,
      "learning_rate": 1.733444078842284e-05,
      "loss": 0.7857,
      "step": 5141
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.059093475341797,
      "learning_rate": 1.7333801872025047e-05,
      "loss": 0.9531,
      "step": 5142
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0662245750427246,
      "learning_rate": 1.7333162955627258e-05,
      "loss": 0.8641,
      "step": 5143
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.139282703399658,
      "learning_rate": 1.7332524039229466e-05,
      "loss": 1.2126,
      "step": 5144
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5862255096435547,
      "learning_rate": 1.733188512283168e-05,
      "loss": 1.0368,
      "step": 5145
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2337777614593506,
      "learning_rate": 1.733124620643389e-05,
      "loss": 0.9885,
      "step": 5146
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7895280122756958,
      "learning_rate": 1.73306072900361e-05,
      "loss": 0.9039,
      "step": 5147
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0817084312438965,
      "learning_rate": 1.732996837363831e-05,
      "loss": 0.838,
      "step": 5148
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2896628379821777,
      "learning_rate": 1.732932945724052e-05,
      "loss": 0.9271,
      "step": 5149
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.061385154724121,
      "learning_rate": 1.7328690540842732e-05,
      "loss": 0.9815,
      "step": 5150
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.893636703491211,
      "learning_rate": 1.7328051624444944e-05,
      "loss": 0.9421,
      "step": 5151
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1707699298858643,
      "learning_rate": 1.7327412708047152e-05,
      "loss": 1.0216,
      "step": 5152
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9381130933761597,
      "learning_rate": 1.7326773791649364e-05,
      "loss": 0.8663,
      "step": 5153
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.7008752822875977,
      "learning_rate": 1.7326134875251575e-05,
      "loss": 1.0109,
      "step": 5154
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.340151786804199,
      "learning_rate": 1.7325495958853787e-05,
      "loss": 0.9201,
      "step": 5155
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.393138885498047,
      "learning_rate": 1.7324857042455995e-05,
      "loss": 0.9506,
      "step": 5156
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1109025478363037,
      "learning_rate": 1.7324218126058206e-05,
      "loss": 0.9777,
      "step": 5157
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.488269805908203,
      "learning_rate": 1.7323579209660418e-05,
      "loss": 1.0992,
      "step": 5158
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.225032091140747,
      "learning_rate": 1.732294029326263e-05,
      "loss": 1.1548,
      "step": 5159
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.6200950145721436,
      "learning_rate": 1.7322301376864838e-05,
      "loss": 0.9001,
      "step": 5160
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.6432549953460693,
      "learning_rate": 1.732166246046705e-05,
      "loss": 0.9851,
      "step": 5161
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.4114139080047607,
      "learning_rate": 1.732102354406926e-05,
      "loss": 0.8987,
      "step": 5162
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7141928672790527,
      "learning_rate": 1.7320384627671472e-05,
      "loss": 0.8876,
      "step": 5163
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6076078414916992,
      "learning_rate": 1.731974571127368e-05,
      "loss": 0.3557,
      "step": 5164
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.205857753753662,
      "learning_rate": 1.7319106794875892e-05,
      "loss": 1.1597,
      "step": 5165
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6054296493530273,
      "learning_rate": 1.7318467878478104e-05,
      "loss": 0.3937,
      "step": 5166
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0587880611419678,
      "learning_rate": 1.7317828962080315e-05,
      "loss": 0.9044,
      "step": 5167
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9199086427688599,
      "learning_rate": 1.7317190045682523e-05,
      "loss": 0.915,
      "step": 5168
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4369239807128906,
      "learning_rate": 1.7316551129284735e-05,
      "loss": 1.0468,
      "step": 5169
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.810289144515991,
      "learning_rate": 1.7315912212886946e-05,
      "loss": 0.6358,
      "step": 5170
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6707675457000732,
      "learning_rate": 1.7315273296489158e-05,
      "loss": 0.3839,
      "step": 5171
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8769983053207397,
      "learning_rate": 1.7314634380091366e-05,
      "loss": 0.9324,
      "step": 5172
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.0476861000061035,
      "learning_rate": 1.7313995463693578e-05,
      "loss": 1.1486,
      "step": 5173
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.014551877975464,
      "learning_rate": 1.731335654729579e-05,
      "loss": 0.9026,
      "step": 5174
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9727094173431396,
      "learning_rate": 1.7312717630898e-05,
      "loss": 0.9967,
      "step": 5175
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0638537406921387,
      "learning_rate": 1.731207871450021e-05,
      "loss": 0.8245,
      "step": 5176
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2639453411102295,
      "learning_rate": 1.731143979810242e-05,
      "loss": 1.0102,
      "step": 5177
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2117960453033447,
      "learning_rate": 1.7310800881704632e-05,
      "loss": 1.303,
      "step": 5178
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3900163173675537,
      "learning_rate": 1.7310161965306843e-05,
      "loss": 0.9093,
      "step": 5179
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.96608829498291,
      "learning_rate": 1.730952304890905e-05,
      "loss": 1.3247,
      "step": 5180
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.28682279586792,
      "learning_rate": 1.7308884132511263e-05,
      "loss": 0.8543,
      "step": 5181
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2885003089904785,
      "learning_rate": 1.730824521611347e-05,
      "loss": 0.9141,
      "step": 5182
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8489325046539307,
      "learning_rate": 1.7307606299715686e-05,
      "loss": 1.141,
      "step": 5183
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.427290439605713,
      "learning_rate": 1.7306967383317894e-05,
      "loss": 0.9912,
      "step": 5184
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.679060935974121,
      "learning_rate": 1.7306328466920106e-05,
      "loss": 0.9927,
      "step": 5185
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7465683221817017,
      "learning_rate": 1.7305689550522314e-05,
      "loss": 1.0097,
      "step": 5186
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.3147079944610596,
      "learning_rate": 1.7305050634124526e-05,
      "loss": 0.9607,
      "step": 5187
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.9383299350738525,
      "learning_rate": 1.7304411717726737e-05,
      "loss": 1.1725,
      "step": 5188
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.842836856842041,
      "learning_rate": 1.730377280132895e-05,
      "loss": 0.8605,
      "step": 5189
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1439576148986816,
      "learning_rate": 1.7303133884931157e-05,
      "loss": 1.2203,
      "step": 5190
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2046358585357666,
      "learning_rate": 1.730249496853337e-05,
      "loss": 0.8812,
      "step": 5191
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7220479249954224,
      "learning_rate": 1.730185605213558e-05,
      "loss": 0.9595,
      "step": 5192
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0410499572753906,
      "learning_rate": 1.730121713573779e-05,
      "loss": 0.8972,
      "step": 5193
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.324369430541992,
      "learning_rate": 1.730057821934e-05,
      "loss": 1.0879,
      "step": 5194
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3645904064178467,
      "learning_rate": 1.729993930294221e-05,
      "loss": 1.1775,
      "step": 5195
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7009022235870361,
      "learning_rate": 1.7299300386544423e-05,
      "loss": 0.8853,
      "step": 5196
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.275360107421875,
      "learning_rate": 1.7298661470146634e-05,
      "loss": 1.0312,
      "step": 5197
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7405909299850464,
      "learning_rate": 1.7298022553748843e-05,
      "loss": 0.925,
      "step": 5198
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9206409454345703,
      "learning_rate": 1.7297383637351054e-05,
      "loss": 0.9303,
      "step": 5199
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.209156513214111,
      "learning_rate": 1.7296744720953266e-05,
      "loss": 1.2614,
      "step": 5200
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.963028073310852,
      "learning_rate": 1.7296105804555477e-05,
      "loss": 1.0005,
      "step": 5201
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5413148403167725,
      "learning_rate": 1.7295466888157685e-05,
      "loss": 0.9332,
      "step": 5202
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7201427221298218,
      "learning_rate": 1.7294827971759897e-05,
      "loss": 0.9437,
      "step": 5203
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.515066623687744,
      "learning_rate": 1.729418905536211e-05,
      "loss": 0.9353,
      "step": 5204
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7907673120498657,
      "learning_rate": 1.729355013896432e-05,
      "loss": 0.8955,
      "step": 5205
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8943662643432617,
      "learning_rate": 1.7292911222566528e-05,
      "loss": 0.8312,
      "step": 5206
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7130571603775024,
      "learning_rate": 1.729227230616874e-05,
      "loss": 0.9659,
      "step": 5207
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.4556360244750977,
      "learning_rate": 1.729163338977095e-05,
      "loss": 1.0981,
      "step": 5208
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.835838317871094,
      "learning_rate": 1.7290994473373163e-05,
      "loss": 1.1884,
      "step": 5209
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.810407876968384,
      "learning_rate": 1.729035555697537e-05,
      "loss": 1.0172,
      "step": 5210
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.998654365539551,
      "learning_rate": 1.7289716640577582e-05,
      "loss": 1.1002,
      "step": 5211
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.660823106765747,
      "learning_rate": 1.7289077724179794e-05,
      "loss": 0.3014,
      "step": 5212
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5005013942718506,
      "learning_rate": 1.7288438807782006e-05,
      "loss": 1.0173,
      "step": 5213
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.892974615097046,
      "learning_rate": 1.7287799891384214e-05,
      "loss": 0.8267,
      "step": 5214
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.144027233123779,
      "learning_rate": 1.7287160974986425e-05,
      "loss": 1.3455,
      "step": 5215
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.0090880393981934,
      "learning_rate": 1.7286522058588633e-05,
      "loss": 0.8455,
      "step": 5216
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.537186622619629,
      "learning_rate": 1.728588314219085e-05,
      "loss": 0.9559,
      "step": 5217
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.000725030899048,
      "learning_rate": 1.7285244225793057e-05,
      "loss": 1.1604,
      "step": 5218
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.457799434661865,
      "learning_rate": 1.7284605309395268e-05,
      "loss": 1.159,
      "step": 5219
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9459058046340942,
      "learning_rate": 1.7283966392997476e-05,
      "loss": 0.9455,
      "step": 5220
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0482192039489746,
      "learning_rate": 1.7283327476599688e-05,
      "loss": 0.9645,
      "step": 5221
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8920124769210815,
      "learning_rate": 1.72826885602019e-05,
      "loss": 0.8789,
      "step": 5222
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3063900470733643,
      "learning_rate": 1.7282049643804108e-05,
      "loss": 1.0964,
      "step": 5223
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.960160732269287,
      "learning_rate": 1.728141072740632e-05,
      "loss": 1.1387,
      "step": 5224
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7756654024124146,
      "learning_rate": 1.728077181100853e-05,
      "loss": 0.8772,
      "step": 5225
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.904351830482483,
      "learning_rate": 1.7280132894610742e-05,
      "loss": 0.9002,
      "step": 5226
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7900848388671875,
      "learning_rate": 1.727949397821295e-05,
      "loss": 0.9668,
      "step": 5227
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.074549436569214,
      "learning_rate": 1.7278855061815162e-05,
      "loss": 1.0875,
      "step": 5228
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.462996482849121,
      "learning_rate": 1.7278216145417373e-05,
      "loss": 0.8239,
      "step": 5229
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9744471311569214,
      "learning_rate": 1.7277577229019585e-05,
      "loss": 0.8492,
      "step": 5230
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9858256578445435,
      "learning_rate": 1.7276938312621793e-05,
      "loss": 1.0026,
      "step": 5231
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.707298517227173,
      "learning_rate": 1.7276299396224005e-05,
      "loss": 0.6945,
      "step": 5232
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.134387969970703,
      "learning_rate": 1.7275660479826216e-05,
      "loss": 1.0764,
      "step": 5233
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9977343082427979,
      "learning_rate": 1.7275021563428428e-05,
      "loss": 0.9962,
      "step": 5234
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.2740612030029297,
      "learning_rate": 1.7274382647030636e-05,
      "loss": 1.0904,
      "step": 5235
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.349954128265381,
      "learning_rate": 1.7273743730632847e-05,
      "loss": 1.7084,
      "step": 5236
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3019661903381348,
      "learning_rate": 1.727310481423506e-05,
      "loss": 0.9213,
      "step": 5237
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.066070556640625,
      "learning_rate": 1.727246589783727e-05,
      "loss": 1.0897,
      "step": 5238
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8592469692230225,
      "learning_rate": 1.727182698143948e-05,
      "loss": 0.9253,
      "step": 5239
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8504899740219116,
      "learning_rate": 1.727118806504169e-05,
      "loss": 0.8254,
      "step": 5240
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8500438928604126,
      "learning_rate": 1.7270549148643902e-05,
      "loss": 0.4214,
      "step": 5241
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9970296621322632,
      "learning_rate": 1.7269910232246113e-05,
      "loss": 0.9084,
      "step": 5242
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.032172203063965,
      "learning_rate": 1.726927131584832e-05,
      "loss": 0.9913,
      "step": 5243
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.44653058052063,
      "learning_rate": 1.7268632399450533e-05,
      "loss": 1.0204,
      "step": 5244
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.109950065612793,
      "learning_rate": 1.7267993483052745e-05,
      "loss": 1.5015,
      "step": 5245
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2680823802948,
      "learning_rate": 1.7267354566654956e-05,
      "loss": 0.9067,
      "step": 5246
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.35380744934082,
      "learning_rate": 1.7266715650257164e-05,
      "loss": 1.7362,
      "step": 5247
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3004422187805176,
      "learning_rate": 1.7266076733859376e-05,
      "loss": 1.0155,
      "step": 5248
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.113382577896118,
      "learning_rate": 1.7265437817461584e-05,
      "loss": 1.2615,
      "step": 5249
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.237556457519531,
      "learning_rate": 1.72647989010638e-05,
      "loss": 1.0444,
      "step": 5250
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9033700227737427,
      "learning_rate": 1.7264159984666007e-05,
      "loss": 0.9021,
      "step": 5251
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.839021921157837,
      "learning_rate": 1.726352106826822e-05,
      "loss": 0.977,
      "step": 5252
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.072941780090332,
      "learning_rate": 1.7262882151870427e-05,
      "loss": 0.9503,
      "step": 5253
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.594712734222412,
      "learning_rate": 1.726224323547264e-05,
      "loss": 0.8033,
      "step": 5254
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.730924606323242,
      "learning_rate": 1.726160431907485e-05,
      "loss": 1.093,
      "step": 5255
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.9962339401245117,
      "learning_rate": 1.726096540267706e-05,
      "loss": 1.3278,
      "step": 5256
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.999565362930298,
      "learning_rate": 1.726032648627927e-05,
      "loss": 1.249,
      "step": 5257
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2272799015045166,
      "learning_rate": 1.725968756988148e-05,
      "loss": 0.9168,
      "step": 5258
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3867108821868896,
      "learning_rate": 1.7259048653483693e-05,
      "loss": 0.9145,
      "step": 5259
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8878448009490967,
      "learning_rate": 1.7258409737085904e-05,
      "loss": 0.9938,
      "step": 5260
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.964977502822876,
      "learning_rate": 1.7257770820688112e-05,
      "loss": 0.9757,
      "step": 5261
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5719761848449707,
      "learning_rate": 1.7257131904290324e-05,
      "loss": 1.0677,
      "step": 5262
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.0524685382843018,
      "learning_rate": 1.7256492987892536e-05,
      "loss": 0.9003,
      "step": 5263
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1778454780578613,
      "learning_rate": 1.7255854071494747e-05,
      "loss": 0.8546,
      "step": 5264
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.280003309249878,
      "learning_rate": 1.7255215155096955e-05,
      "loss": 1.0147,
      "step": 5265
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1496939659118652,
      "learning_rate": 1.7254576238699167e-05,
      "loss": 0.92,
      "step": 5266
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.09165096282959,
      "learning_rate": 1.7253937322301378e-05,
      "loss": 0.7861,
      "step": 5267
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6771446466445923,
      "learning_rate": 1.725329840590359e-05,
      "loss": 0.3452,
      "step": 5268
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.707000255584717,
      "learning_rate": 1.7252659489505798e-05,
      "loss": 1.0165,
      "step": 5269
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.685445785522461,
      "learning_rate": 1.725202057310801e-05,
      "loss": 1.2786,
      "step": 5270
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1510167121887207,
      "learning_rate": 1.725138165671022e-05,
      "loss": 0.9548,
      "step": 5271
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4886717796325684,
      "learning_rate": 1.7250742740312433e-05,
      "loss": 0.9447,
      "step": 5272
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8157026767730713,
      "learning_rate": 1.725010382391464e-05,
      "loss": 0.8773,
      "step": 5273
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.24797248840332,
      "learning_rate": 1.7249464907516852e-05,
      "loss": 1.2758,
      "step": 5274
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.07944917678833,
      "learning_rate": 1.7248825991119064e-05,
      "loss": 0.9557,
      "step": 5275
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.105710029602051,
      "learning_rate": 1.7248187074721275e-05,
      "loss": 1.2261,
      "step": 5276
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3297173976898193,
      "learning_rate": 1.7247548158323484e-05,
      "loss": 0.9945,
      "step": 5277
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.211610794067383,
      "learning_rate": 1.7246909241925695e-05,
      "loss": 1.0192,
      "step": 5278
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.100102663040161,
      "learning_rate": 1.7246270325527907e-05,
      "loss": 1.3399,
      "step": 5279
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5125259160995483,
      "learning_rate": 1.7245631409130118e-05,
      "loss": 0.3193,
      "step": 5280
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.6597886085510254,
      "learning_rate": 1.7244992492732326e-05,
      "loss": 1.2493,
      "step": 5281
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0355308055877686,
      "learning_rate": 1.7244353576334538e-05,
      "loss": 0.9672,
      "step": 5282
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.39621114730835,
      "learning_rate": 1.7243714659936746e-05,
      "loss": 1.6745,
      "step": 5283
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.122502326965332,
      "learning_rate": 1.724307574353896e-05,
      "loss": 0.9102,
      "step": 5284
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.9008302688598633,
      "learning_rate": 1.724243682714117e-05,
      "loss": 0.8585,
      "step": 5285
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.2798664569854736,
      "learning_rate": 1.724179791074338e-05,
      "loss": 0.8699,
      "step": 5286
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.011046886444092,
      "learning_rate": 1.724115899434559e-05,
      "loss": 0.8623,
      "step": 5287
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5855824947357178,
      "learning_rate": 1.72405200779478e-05,
      "loss": 1.2373,
      "step": 5288
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6776812076568604,
      "learning_rate": 1.7239881161550012e-05,
      "loss": 0.8983,
      "step": 5289
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.155757188796997,
      "learning_rate": 1.7239242245152224e-05,
      "loss": 1.016,
      "step": 5290
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.59403395652771,
      "learning_rate": 1.7238603328754432e-05,
      "loss": 1.1938,
      "step": 5291
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.9434025287628174,
      "learning_rate": 1.7237964412356643e-05,
      "loss": 1.1252,
      "step": 5292
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7570492029190063,
      "learning_rate": 1.7237325495958855e-05,
      "loss": 1.0774,
      "step": 5293
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8222522735595703,
      "learning_rate": 1.7236686579561066e-05,
      "loss": 0.8075,
      "step": 5294
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.001194477081299,
      "learning_rate": 1.7236047663163275e-05,
      "loss": 0.375,
      "step": 5295
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.203518867492676,
      "learning_rate": 1.7235408746765486e-05,
      "loss": 0.9422,
      "step": 5296
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8862026929855347,
      "learning_rate": 1.7234769830367698e-05,
      "loss": 0.8471,
      "step": 5297
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4132453203201294,
      "learning_rate": 1.723413091396991e-05,
      "loss": 0.4885,
      "step": 5298
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1017870903015137,
      "learning_rate": 1.7233491997572117e-05,
      "loss": 0.8291,
      "step": 5299
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.392958641052246,
      "learning_rate": 1.723285308117433e-05,
      "loss": 1.9311,
      "step": 5300
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6047704219818115,
      "learning_rate": 1.723221416477654e-05,
      "loss": 0.849,
      "step": 5301
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.7691447734832764,
      "learning_rate": 1.7231575248378752e-05,
      "loss": 0.8554,
      "step": 5302
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6172679662704468,
      "learning_rate": 1.723093633198096e-05,
      "loss": 0.3687,
      "step": 5303
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1951587200164795,
      "learning_rate": 1.723029741558317e-05,
      "loss": 0.8405,
      "step": 5304
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7637447118759155,
      "learning_rate": 1.7229658499185383e-05,
      "loss": 1.0307,
      "step": 5305
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7748851776123047,
      "learning_rate": 1.7229019582787595e-05,
      "loss": 0.4461,
      "step": 5306
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5280277729034424,
      "learning_rate": 1.7228380666389803e-05,
      "loss": 0.9456,
      "step": 5307
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1953766345977783,
      "learning_rate": 1.7227741749992014e-05,
      "loss": 1.0188,
      "step": 5308
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8281654119491577,
      "learning_rate": 1.7227102833594226e-05,
      "loss": 0.9679,
      "step": 5309
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.6670122146606445,
      "learning_rate": 1.7226463917196438e-05,
      "loss": 1.795,
      "step": 5310
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.341217517852783,
      "learning_rate": 1.7225825000798646e-05,
      "loss": 0.9156,
      "step": 5311
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8064264059066772,
      "learning_rate": 1.7225186084400857e-05,
      "loss": 0.7679,
      "step": 5312
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.454101800918579,
      "learning_rate": 1.722454716800307e-05,
      "loss": 1.0539,
      "step": 5313
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9306451082229614,
      "learning_rate": 1.722390825160528e-05,
      "loss": 0.9693,
      "step": 5314
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.067157745361328,
      "learning_rate": 1.722326933520749e-05,
      "loss": 1.3567,
      "step": 5315
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5330426692962646,
      "learning_rate": 1.72226304188097e-05,
      "loss": 0.983,
      "step": 5316
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.145801067352295,
      "learning_rate": 1.722199150241191e-05,
      "loss": 0.8511,
      "step": 5317
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.443831205368042,
      "learning_rate": 1.7221352586014123e-05,
      "loss": 1.0295,
      "step": 5318
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3112196922302246,
      "learning_rate": 1.722071366961633e-05,
      "loss": 0.9575,
      "step": 5319
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.827402949333191,
      "learning_rate": 1.7220074753218543e-05,
      "loss": 0.9561,
      "step": 5320
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.184501886367798,
      "learning_rate": 1.721943583682075e-05,
      "loss": 1.0158,
      "step": 5321
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1005587577819824,
      "learning_rate": 1.7218796920422966e-05,
      "loss": 0.8993,
      "step": 5322
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0721282958984375,
      "learning_rate": 1.7218158004025174e-05,
      "loss": 0.9958,
      "step": 5323
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9772672653198242,
      "learning_rate": 1.7217519087627386e-05,
      "loss": 0.9361,
      "step": 5324
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7199487686157227,
      "learning_rate": 1.7216880171229594e-05,
      "loss": 0.783,
      "step": 5325
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.907625913619995,
      "learning_rate": 1.7216241254831805e-05,
      "loss": 0.8126,
      "step": 5326
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4614923000335693,
      "learning_rate": 1.7215602338434017e-05,
      "loss": 1.1128,
      "step": 5327
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.224819660186768,
      "learning_rate": 1.721496342203623e-05,
      "loss": 1.3505,
      "step": 5328
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.559044599533081,
      "learning_rate": 1.7214324505638437e-05,
      "loss": 1.1934,
      "step": 5329
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1273036003112793,
      "learning_rate": 1.7213685589240648e-05,
      "loss": 0.9272,
      "step": 5330
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9938700199127197,
      "learning_rate": 1.721304667284286e-05,
      "loss": 0.894,
      "step": 5331
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.55035662651062,
      "learning_rate": 1.721240775644507e-05,
      "loss": 0.8751,
      "step": 5332
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1733527183532715,
      "learning_rate": 1.721176884004728e-05,
      "loss": 1.2617,
      "step": 5333
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.69669508934021,
      "learning_rate": 1.721112992364949e-05,
      "loss": 1.2502,
      "step": 5334
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8022202253341675,
      "learning_rate": 1.7210491007251703e-05,
      "loss": 1.0036,
      "step": 5335
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6212503910064697,
      "learning_rate": 1.7209852090853914e-05,
      "loss": 0.8793,
      "step": 5336
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0337841510772705,
      "learning_rate": 1.7209213174456122e-05,
      "loss": 0.8401,
      "step": 5337
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9497383832931519,
      "learning_rate": 1.7208574258058334e-05,
      "loss": 0.9673,
      "step": 5338
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.53511381149292,
      "learning_rate": 1.7207935341660545e-05,
      "loss": 1.0304,
      "step": 5339
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8798019886016846,
      "learning_rate": 1.7207296425262757e-05,
      "loss": 0.8502,
      "step": 5340
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6544585227966309,
      "learning_rate": 1.7206657508864965e-05,
      "loss": 1.0316,
      "step": 5341
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.917165517807007,
      "learning_rate": 1.7206018592467177e-05,
      "loss": 1.0018,
      "step": 5342
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9273693561553955,
      "learning_rate": 1.7205379676069388e-05,
      "loss": 1.1164,
      "step": 5343
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9971919059753418,
      "learning_rate": 1.72047407596716e-05,
      "loss": 1.1949,
      "step": 5344
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8732925653457642,
      "learning_rate": 1.7204101843273808e-05,
      "loss": 0.8906,
      "step": 5345
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4296224117279053,
      "learning_rate": 1.720346292687602e-05,
      "loss": 0.7541,
      "step": 5346
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0323145389556885,
      "learning_rate": 1.720282401047823e-05,
      "loss": 0.9704,
      "step": 5347
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9696197509765625,
      "learning_rate": 1.7202185094080442e-05,
      "loss": 0.8549,
      "step": 5348
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.022226572036743,
      "learning_rate": 1.720154617768265e-05,
      "loss": 0.4244,
      "step": 5349
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.933384656906128,
      "learning_rate": 1.7200907261284862e-05,
      "loss": 0.588,
      "step": 5350
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8180245161056519,
      "learning_rate": 1.7200268344887074e-05,
      "loss": 0.8987,
      "step": 5351
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.0229644775390625,
      "learning_rate": 1.7199629428489285e-05,
      "loss": 1.4952,
      "step": 5352
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.518893003463745,
      "learning_rate": 1.7198990512091493e-05,
      "loss": 0.9357,
      "step": 5353
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0474936962127686,
      "learning_rate": 1.7198351595693705e-05,
      "loss": 0.9644,
      "step": 5354
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.160251140594482,
      "learning_rate": 1.7197712679295913e-05,
      "loss": 1.069,
      "step": 5355
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.039822578430176,
      "learning_rate": 1.7197073762898128e-05,
      "loss": 1.0365,
      "step": 5356
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9309439659118652,
      "learning_rate": 1.7196434846500336e-05,
      "loss": 1.0588,
      "step": 5357
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8428698778152466,
      "learning_rate": 1.7195795930102548e-05,
      "loss": 0.9761,
      "step": 5358
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.951329708099365,
      "learning_rate": 1.7195157013704756e-05,
      "loss": 1.8684,
      "step": 5359
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9536243677139282,
      "learning_rate": 1.719451809730697e-05,
      "loss": 0.8303,
      "step": 5360
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0171473026275635,
      "learning_rate": 1.719387918090918e-05,
      "loss": 0.9167,
      "step": 5361
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7667330503463745,
      "learning_rate": 1.719324026451139e-05,
      "loss": 0.9351,
      "step": 5362
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9087013006210327,
      "learning_rate": 1.71926013481136e-05,
      "loss": 0.8169,
      "step": 5363
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.462759971618652,
      "learning_rate": 1.719196243171581e-05,
      "loss": 0.6326,
      "step": 5364
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.511801242828369,
      "learning_rate": 1.7191323515318022e-05,
      "loss": 1.2676,
      "step": 5365
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6335138082504272,
      "learning_rate": 1.7190684598920233e-05,
      "loss": 0.4686,
      "step": 5366
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3701093196868896,
      "learning_rate": 1.719004568252244e-05,
      "loss": 0.9744,
      "step": 5367
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0684926509857178,
      "learning_rate": 1.7189406766124653e-05,
      "loss": 0.9399,
      "step": 5368
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8407721519470215,
      "learning_rate": 1.7188767849726865e-05,
      "loss": 1.0169,
      "step": 5369
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.9343349933624268,
      "learning_rate": 1.7188128933329076e-05,
      "loss": 1.1238,
      "step": 5370
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.860990047454834,
      "learning_rate": 1.7187490016931284e-05,
      "loss": 0.9877,
      "step": 5371
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6738944053649902,
      "learning_rate": 1.7186851100533496e-05,
      "loss": 0.7334,
      "step": 5372
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.90225887298584,
      "learning_rate": 1.7186212184135707e-05,
      "loss": 1.3809,
      "step": 5373
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1240394115448,
      "learning_rate": 1.718557326773792e-05,
      "loss": 0.8639,
      "step": 5374
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.182493209838867,
      "learning_rate": 1.7184934351340127e-05,
      "loss": 1.0473,
      "step": 5375
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.04567813873291,
      "learning_rate": 1.718429543494234e-05,
      "loss": 1.0149,
      "step": 5376
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.577638626098633,
      "learning_rate": 1.718365651854455e-05,
      "loss": 0.7996,
      "step": 5377
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7543489933013916,
      "learning_rate": 1.7183017602146762e-05,
      "loss": 0.9078,
      "step": 5378
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.759965419769287,
      "learning_rate": 1.718237868574897e-05,
      "loss": 0.8706,
      "step": 5379
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7813446521759033,
      "learning_rate": 1.718173976935118e-05,
      "loss": 1.0137,
      "step": 5380
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0742104053497314,
      "learning_rate": 1.7181100852953393e-05,
      "loss": 1.0289,
      "step": 5381
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.614692211151123,
      "learning_rate": 1.7180461936555605e-05,
      "loss": 1.0846,
      "step": 5382
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.875706434249878,
      "learning_rate": 1.7179823020157813e-05,
      "loss": 0.8434,
      "step": 5383
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.023684501647949,
      "learning_rate": 1.7179184103760024e-05,
      "loss": 1.2022,
      "step": 5384
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.081437826156616,
      "learning_rate": 1.7178545187362236e-05,
      "loss": 0.9315,
      "step": 5385
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6939588785171509,
      "learning_rate": 1.7177906270964447e-05,
      "loss": 0.9207,
      "step": 5386
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.1138391494750977,
      "learning_rate": 1.7177267354566656e-05,
      "loss": 1.1783,
      "step": 5387
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.190934658050537,
      "learning_rate": 1.7176628438168867e-05,
      "loss": 0.9231,
      "step": 5388
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.684385061264038,
      "learning_rate": 1.717598952177108e-05,
      "loss": 0.9623,
      "step": 5389
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.0471367835998535,
      "learning_rate": 1.717535060537329e-05,
      "loss": 1.4213,
      "step": 5390
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.43025279045105,
      "learning_rate": 1.71747116889755e-05,
      "loss": 0.878,
      "step": 5391
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.019896030426025,
      "learning_rate": 1.717407277257771e-05,
      "loss": 1.0916,
      "step": 5392
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3800861835479736,
      "learning_rate": 1.7173433856179918e-05,
      "loss": 1.267,
      "step": 5393
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3311285972595215,
      "learning_rate": 1.7172794939782133e-05,
      "loss": 1.1944,
      "step": 5394
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.622098922729492,
      "learning_rate": 1.717215602338434e-05,
      "loss": 1.4812,
      "step": 5395
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.391176462173462,
      "learning_rate": 1.7171517106986553e-05,
      "loss": 0.9789,
      "step": 5396
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.258502960205078,
      "learning_rate": 1.717087819058876e-05,
      "loss": 0.9356,
      "step": 5397
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7653290033340454,
      "learning_rate": 1.7170239274190972e-05,
      "loss": 0.8834,
      "step": 5398
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.413951873779297,
      "learning_rate": 1.7169600357793184e-05,
      "loss": 1.5259,
      "step": 5399
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9886846542358398,
      "learning_rate": 1.7168961441395395e-05,
      "loss": 1.0586,
      "step": 5400
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4405925273895264,
      "learning_rate": 1.7168322524997604e-05,
      "loss": 1.0055,
      "step": 5401
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.77754282951355,
      "learning_rate": 1.7167683608599815e-05,
      "loss": 1.3554,
      "step": 5402
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.8451004028320312,
      "learning_rate": 1.7167044692202027e-05,
      "loss": 1.7411,
      "step": 5403
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.186930179595947,
      "learning_rate": 1.7166405775804238e-05,
      "loss": 0.7543,
      "step": 5404
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6966784000396729,
      "learning_rate": 1.7165766859406446e-05,
      "loss": 0.8715,
      "step": 5405
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.16827130317688,
      "learning_rate": 1.7165127943008658e-05,
      "loss": 1.0168,
      "step": 5406
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9717810153961182,
      "learning_rate": 1.716448902661087e-05,
      "loss": 1.0336,
      "step": 5407
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5933467149734497,
      "learning_rate": 1.716385011021308e-05,
      "loss": 0.8129,
      "step": 5408
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.967341661453247,
      "learning_rate": 1.716321119381529e-05,
      "loss": 0.9263,
      "step": 5409
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.551882743835449,
      "learning_rate": 1.71625722774175e-05,
      "loss": 1.0786,
      "step": 5410
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9248621463775635,
      "learning_rate": 1.7161933361019712e-05,
      "loss": 0.8734,
      "step": 5411
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.645519733428955,
      "learning_rate": 1.7161294444621924e-05,
      "loss": 1.265,
      "step": 5412
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7319903373718262,
      "learning_rate": 1.7160655528224132e-05,
      "loss": 0.8141,
      "step": 5413
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9059711694717407,
      "learning_rate": 1.7160016611826344e-05,
      "loss": 0.9779,
      "step": 5414
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2406222820281982,
      "learning_rate": 1.7159377695428555e-05,
      "loss": 1.0406,
      "step": 5415
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.989214539527893,
      "learning_rate": 1.7158738779030767e-05,
      "loss": 0.8603,
      "step": 5416
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.139329433441162,
      "learning_rate": 1.7158099862632975e-05,
      "loss": 1.1991,
      "step": 5417
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2349319458007812,
      "learning_rate": 1.7157460946235186e-05,
      "loss": 0.9344,
      "step": 5418
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.157634973526001,
      "learning_rate": 1.7156822029837398e-05,
      "loss": 0.9511,
      "step": 5419
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.090384006500244,
      "learning_rate": 1.715618311343961e-05,
      "loss": 0.9381,
      "step": 5420
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1047730445861816,
      "learning_rate": 1.7155544197041818e-05,
      "loss": 0.9151,
      "step": 5421
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.5823161602020264,
      "learning_rate": 1.715490528064403e-05,
      "loss": 1.0413,
      "step": 5422
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4385125637054443,
      "learning_rate": 1.715426636424624e-05,
      "loss": 0.8488,
      "step": 5423
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.927575707435608,
      "learning_rate": 1.7153627447848452e-05,
      "loss": 0.9668,
      "step": 5424
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.608491063117981,
      "learning_rate": 1.715298853145066e-05,
      "loss": 0.9628,
      "step": 5425
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2409985065460205,
      "learning_rate": 1.7152349615052872e-05,
      "loss": 1.1392,
      "step": 5426
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4978134632110596,
      "learning_rate": 1.7151710698655084e-05,
      "loss": 0.5439,
      "step": 5427
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.815671920776367,
      "learning_rate": 1.7151071782257295e-05,
      "loss": 1.4399,
      "step": 5428
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.35003662109375,
      "learning_rate": 1.7150432865859503e-05,
      "loss": 0.9764,
      "step": 5429
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8999820947647095,
      "learning_rate": 1.7149793949461715e-05,
      "loss": 0.8779,
      "step": 5430
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.501399278640747,
      "learning_rate": 1.7149155033063923e-05,
      "loss": 0.8943,
      "step": 5431
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.164447546005249,
      "learning_rate": 1.7148516116666138e-05,
      "loss": 0.8662,
      "step": 5432
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9676551818847656,
      "learning_rate": 1.7147877200268346e-05,
      "loss": 0.9833,
      "step": 5433
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.237997531890869,
      "learning_rate": 1.7147238283870558e-05,
      "loss": 1.4136,
      "step": 5434
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.066882610321045,
      "learning_rate": 1.7146599367472766e-05,
      "loss": 0.9442,
      "step": 5435
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1138720512390137,
      "learning_rate": 1.7145960451074977e-05,
      "loss": 0.8989,
      "step": 5436
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7723952531814575,
      "learning_rate": 1.714532153467719e-05,
      "loss": 0.8046,
      "step": 5437
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.858180046081543,
      "learning_rate": 1.71446826182794e-05,
      "loss": 0.9511,
      "step": 5438
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.243570566177368,
      "learning_rate": 1.714404370188161e-05,
      "loss": 0.9035,
      "step": 5439
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.885155439376831,
      "learning_rate": 1.714340478548382e-05,
      "loss": 0.9533,
      "step": 5440
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.228663206100464,
      "learning_rate": 1.714276586908603e-05,
      "loss": 1.1699,
      "step": 5441
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9667893648147583,
      "learning_rate": 1.7142126952688243e-05,
      "loss": 0.7239,
      "step": 5442
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.730944871902466,
      "learning_rate": 1.714148803629045e-05,
      "loss": 0.9635,
      "step": 5443
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2009811401367188,
      "learning_rate": 1.7140849119892663e-05,
      "loss": 1.2232,
      "step": 5444
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.258506774902344,
      "learning_rate": 1.7140210203494874e-05,
      "loss": 1.4507,
      "step": 5445
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1921989917755127,
      "learning_rate": 1.7139571287097086e-05,
      "loss": 0.9329,
      "step": 5446
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.968085527420044,
      "learning_rate": 1.7138932370699294e-05,
      "loss": 0.9674,
      "step": 5447
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8237804174423218,
      "learning_rate": 1.7138293454301506e-05,
      "loss": 0.3798,
      "step": 5448
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.605760097503662,
      "learning_rate": 1.7137654537903717e-05,
      "loss": 1.3023,
      "step": 5449
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.371676445007324,
      "learning_rate": 1.713701562150593e-05,
      "loss": 1.0451,
      "step": 5450
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.297588586807251,
      "learning_rate": 1.7136376705108137e-05,
      "loss": 0.9374,
      "step": 5451
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.071507692337036,
      "learning_rate": 1.713573778871035e-05,
      "loss": 0.9846,
      "step": 5452
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.255396842956543,
      "learning_rate": 1.713509887231256e-05,
      "loss": 1.0887,
      "step": 5453
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8233064413070679,
      "learning_rate": 1.713445995591477e-05,
      "loss": 0.9885,
      "step": 5454
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0336413383483887,
      "learning_rate": 1.713382103951698e-05,
      "loss": 0.9166,
      "step": 5455
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5601409673690796,
      "learning_rate": 1.713318212311919e-05,
      "loss": 0.4412,
      "step": 5456
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.7620038986206055,
      "learning_rate": 1.7132543206721403e-05,
      "loss": 1.4511,
      "step": 5457
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8459392786026,
      "learning_rate": 1.7131904290323614e-05,
      "loss": 0.9139,
      "step": 5458
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.362426280975342,
      "learning_rate": 1.7131265373925823e-05,
      "loss": 1.3208,
      "step": 5459
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.398300886154175,
      "learning_rate": 1.7130626457528034e-05,
      "loss": 0.8825,
      "step": 5460
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8028926849365234,
      "learning_rate": 1.7129987541130246e-05,
      "loss": 0.7126,
      "step": 5461
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0870327949523926,
      "learning_rate": 1.7129348624732457e-05,
      "loss": 1.1671,
      "step": 5462
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.204716205596924,
      "learning_rate": 1.7128709708334665e-05,
      "loss": 1.2308,
      "step": 5463
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0510194301605225,
      "learning_rate": 1.7128070791936877e-05,
      "loss": 0.9816,
      "step": 5464
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3009469509124756,
      "learning_rate": 1.7127431875539085e-05,
      "loss": 1.1255,
      "step": 5465
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5978093147277832,
      "learning_rate": 1.71267929591413e-05,
      "loss": 0.8612,
      "step": 5466
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2215158939361572,
      "learning_rate": 1.7126154042743508e-05,
      "loss": 0.8832,
      "step": 5467
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.613010406494141,
      "learning_rate": 1.712551512634572e-05,
      "loss": 1.288,
      "step": 5468
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3374462127685547,
      "learning_rate": 1.7124876209947928e-05,
      "loss": 0.9851,
      "step": 5469
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8489736318588257,
      "learning_rate": 1.7124237293550143e-05,
      "loss": 1.0349,
      "step": 5470
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9725404977798462,
      "learning_rate": 1.712359837715235e-05,
      "loss": 0.8753,
      "step": 5471
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.646946668624878,
      "learning_rate": 1.7122959460754562e-05,
      "loss": 0.8452,
      "step": 5472
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5766324996948242,
      "learning_rate": 1.712232054435677e-05,
      "loss": 0.9405,
      "step": 5473
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.698883295059204,
      "learning_rate": 1.7121681627958982e-05,
      "loss": 1.4398,
      "step": 5474
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.124320030212402,
      "learning_rate": 1.7121042711561194e-05,
      "loss": 1.8799,
      "step": 5475
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.404698133468628,
      "learning_rate": 1.7120403795163405e-05,
      "loss": 1.0267,
      "step": 5476
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8002567291259766,
      "learning_rate": 1.7119764878765613e-05,
      "loss": 0.7061,
      "step": 5477
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.755446195602417,
      "learning_rate": 1.7119125962367825e-05,
      "loss": 0.5934,
      "step": 5478
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6684733629226685,
      "learning_rate": 1.7118487045970037e-05,
      "loss": 0.7715,
      "step": 5479
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9858180284500122,
      "learning_rate": 1.7117848129572248e-05,
      "loss": 0.9231,
      "step": 5480
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2098944187164307,
      "learning_rate": 1.7117209213174456e-05,
      "loss": 1.1073,
      "step": 5481
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4251428842544556,
      "learning_rate": 1.7116570296776668e-05,
      "loss": 0.4556,
      "step": 5482
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3855786323547363,
      "learning_rate": 1.711593138037888e-05,
      "loss": 1.1165,
      "step": 5483
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6128268241882324,
      "learning_rate": 1.711529246398109e-05,
      "loss": 0.9924,
      "step": 5484
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.956814765930176,
      "learning_rate": 1.71146535475833e-05,
      "loss": 0.9121,
      "step": 5485
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.346379041671753,
      "learning_rate": 1.711401463118551e-05,
      "loss": 0.9491,
      "step": 5486
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.906766414642334,
      "learning_rate": 1.7113375714787722e-05,
      "loss": 1.0726,
      "step": 5487
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0898640155792236,
      "learning_rate": 1.7112736798389934e-05,
      "loss": 1.0116,
      "step": 5488
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.441709518432617,
      "learning_rate": 1.7112097881992142e-05,
      "loss": 0.9987,
      "step": 5489
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.694204330444336,
      "learning_rate": 1.7111458965594353e-05,
      "loss": 0.9222,
      "step": 5490
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7496949434280396,
      "learning_rate": 1.7110820049196565e-05,
      "loss": 0.952,
      "step": 5491
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8809301853179932,
      "learning_rate": 1.7110181132798776e-05,
      "loss": 1.0433,
      "step": 5492
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.157740354537964,
      "learning_rate": 1.7109542216400985e-05,
      "loss": 0.946,
      "step": 5493
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5830411911010742,
      "learning_rate": 1.7108903300003196e-05,
      "loss": 0.4807,
      "step": 5494
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.623257875442505,
      "learning_rate": 1.7108264383605408e-05,
      "loss": 0.9983,
      "step": 5495
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.561306357383728,
      "learning_rate": 1.710762546720762e-05,
      "loss": 0.8814,
      "step": 5496
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.216207265853882,
      "learning_rate": 1.7106986550809827e-05,
      "loss": 0.9429,
      "step": 5497
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7769328355789185,
      "learning_rate": 1.710634763441204e-05,
      "loss": 1.0053,
      "step": 5498
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6395268440246582,
      "learning_rate": 1.710570871801425e-05,
      "loss": 0.895,
      "step": 5499
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5728743076324463,
      "learning_rate": 1.7105069801616462e-05,
      "loss": 0.4139,
      "step": 5500
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9758594036102295,
      "learning_rate": 1.710443088521867e-05,
      "loss": 0.8953,
      "step": 5501
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9590122699737549,
      "learning_rate": 1.7103791968820882e-05,
      "loss": 0.8792,
      "step": 5502
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7053507566452026,
      "learning_rate": 1.710315305242309e-05,
      "loss": 0.9088,
      "step": 5503
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.329745054244995,
      "learning_rate": 1.7102514136025305e-05,
      "loss": 0.9749,
      "step": 5504
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0963294506073,
      "learning_rate": 1.7101875219627513e-05,
      "loss": 1.0124,
      "step": 5505
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.409748077392578,
      "learning_rate": 1.7101236303229725e-05,
      "loss": 1.0278,
      "step": 5506
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9035037755966187,
      "learning_rate": 1.7100597386831933e-05,
      "loss": 1.0152,
      "step": 5507
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1405551433563232,
      "learning_rate": 1.7099958470434144e-05,
      "loss": 0.9125,
      "step": 5508
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9908417463302612,
      "learning_rate": 1.7099319554036356e-05,
      "loss": 0.8955,
      "step": 5509
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.812958240509033,
      "learning_rate": 1.7098680637638567e-05,
      "loss": 1.2139,
      "step": 5510
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.722844123840332,
      "learning_rate": 1.7098041721240776e-05,
      "loss": 1.3951,
      "step": 5511
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.381432294845581,
      "learning_rate": 1.7097402804842987e-05,
      "loss": 0.8831,
      "step": 5512
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1453471183776855,
      "learning_rate": 1.70967638884452e-05,
      "loss": 0.9777,
      "step": 5513
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9188674688339233,
      "learning_rate": 1.709612497204741e-05,
      "loss": 0.9873,
      "step": 5514
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4532577991485596,
      "learning_rate": 1.709548605564962e-05,
      "loss": 0.8855,
      "step": 5515
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8489093780517578,
      "learning_rate": 1.709484713925183e-05,
      "loss": 0.9041,
      "step": 5516
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.75338876247406,
      "learning_rate": 1.709420822285404e-05,
      "loss": 1.0054,
      "step": 5517
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.694981098175049,
      "learning_rate": 1.7093569306456253e-05,
      "loss": 0.6092,
      "step": 5518
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9387681484222412,
      "learning_rate": 1.709293039005846e-05,
      "loss": 0.8275,
      "step": 5519
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.836892008781433,
      "learning_rate": 1.7092291473660673e-05,
      "loss": 0.9372,
      "step": 5520
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.07861065864563,
      "learning_rate": 1.7091652557262884e-05,
      "loss": 1.0599,
      "step": 5521
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6370389461517334,
      "learning_rate": 1.7091013640865096e-05,
      "loss": 1.0482,
      "step": 5522
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1607604026794434,
      "learning_rate": 1.7090374724467304e-05,
      "loss": 1.3058,
      "step": 5523
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.063577890396118,
      "learning_rate": 1.7089735808069515e-05,
      "loss": 1.3068,
      "step": 5524
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.576493263244629,
      "learning_rate": 1.7089096891671727e-05,
      "loss": 1.4676,
      "step": 5525
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.199613094329834,
      "learning_rate": 1.708845797527394e-05,
      "loss": 1.2631,
      "step": 5526
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.237856388092041,
      "learning_rate": 1.7087819058876147e-05,
      "loss": 0.7139,
      "step": 5527
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.360032796859741,
      "learning_rate": 1.7087180142478358e-05,
      "loss": 0.99,
      "step": 5528
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.821285367012024,
      "learning_rate": 1.708654122608057e-05,
      "loss": 1.0444,
      "step": 5529
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.013585090637207,
      "learning_rate": 1.708590230968278e-05,
      "loss": 0.9704,
      "step": 5530
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8304802179336548,
      "learning_rate": 1.708526339328499e-05,
      "loss": 0.9791,
      "step": 5531
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0139358043670654,
      "learning_rate": 1.70846244768872e-05,
      "loss": 1.02,
      "step": 5532
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.033352851867676,
      "learning_rate": 1.7083985560489413e-05,
      "loss": 1.0001,
      "step": 5533
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6618698835372925,
      "learning_rate": 1.7083346644091624e-05,
      "loss": 0.4142,
      "step": 5534
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9944210052490234,
      "learning_rate": 1.7082707727693832e-05,
      "loss": 1.0649,
      "step": 5535
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1573846340179443,
      "learning_rate": 1.7082068811296044e-05,
      "loss": 0.8232,
      "step": 5536
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.402390241622925,
      "learning_rate": 1.7081429894898255e-05,
      "loss": 0.9691,
      "step": 5537
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.973407506942749,
      "learning_rate": 1.7080790978500467e-05,
      "loss": 1.0355,
      "step": 5538
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9711369276046753,
      "learning_rate": 1.7080152062102675e-05,
      "loss": 0.8969,
      "step": 5539
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.429307460784912,
      "learning_rate": 1.7079513145704887e-05,
      "loss": 0.8505,
      "step": 5540
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8210270404815674,
      "learning_rate": 1.7078874229307095e-05,
      "loss": 1.0182,
      "step": 5541
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.16019868850708,
      "learning_rate": 1.707823531290931e-05,
      "loss": 0.91,
      "step": 5542
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0587527751922607,
      "learning_rate": 1.7077596396511518e-05,
      "loss": 0.8214,
      "step": 5543
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9451167583465576,
      "learning_rate": 1.707695748011373e-05,
      "loss": 0.8997,
      "step": 5544
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9698208570480347,
      "learning_rate": 1.7076318563715938e-05,
      "loss": 1.3284,
      "step": 5545
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.638164758682251,
      "learning_rate": 1.707567964731815e-05,
      "loss": 0.9085,
      "step": 5546
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5346928834915161,
      "learning_rate": 1.707504073092036e-05,
      "loss": 0.8763,
      "step": 5547
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0558366775512695,
      "learning_rate": 1.7074401814522572e-05,
      "loss": 0.8407,
      "step": 5548
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7065544128417969,
      "learning_rate": 1.707376289812478e-05,
      "loss": 0.9091,
      "step": 5549
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.853294610977173,
      "learning_rate": 1.7073123981726992e-05,
      "loss": 0.802,
      "step": 5550
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9255449771881104,
      "learning_rate": 1.7072485065329204e-05,
      "loss": 1.0002,
      "step": 5551
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.516808032989502,
      "learning_rate": 1.7071846148931415e-05,
      "loss": 1.1229,
      "step": 5552
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9707616567611694,
      "learning_rate": 1.7071207232533623e-05,
      "loss": 0.8933,
      "step": 5553
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5283924341201782,
      "learning_rate": 1.7070568316135835e-05,
      "loss": 0.8801,
      "step": 5554
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.530027151107788,
      "learning_rate": 1.7069929399738046e-05,
      "loss": 0.884,
      "step": 5555
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9397790431976318,
      "learning_rate": 1.7069290483340258e-05,
      "loss": 0.8565,
      "step": 5556
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9779994487762451,
      "learning_rate": 1.7068651566942466e-05,
      "loss": 1.003,
      "step": 5557
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6589748859405518,
      "learning_rate": 1.7068012650544678e-05,
      "loss": 1.0036,
      "step": 5558
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6659045219421387,
      "learning_rate": 1.706737373414689e-05,
      "loss": 0.7971,
      "step": 5559
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4685847759246826,
      "learning_rate": 1.70667348177491e-05,
      "loss": 0.7434,
      "step": 5560
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.051677703857422,
      "learning_rate": 1.706609590135131e-05,
      "loss": 1.0791,
      "step": 5561
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6501635313034058,
      "learning_rate": 1.706545698495352e-05,
      "loss": 0.9251,
      "step": 5562
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.911197304725647,
      "learning_rate": 1.7064818068555732e-05,
      "loss": 1.0382,
      "step": 5563
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1157352924346924,
      "learning_rate": 1.7064179152157943e-05,
      "loss": 0.9358,
      "step": 5564
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7372565269470215,
      "learning_rate": 1.706354023576015e-05,
      "loss": 1.0249,
      "step": 5565
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6737912893295288,
      "learning_rate": 1.7062901319362363e-05,
      "loss": 0.9911,
      "step": 5566
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.051693916320801,
      "learning_rate": 1.7062262402964575e-05,
      "loss": 0.8663,
      "step": 5567
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1752357482910156,
      "learning_rate": 1.7061623486566786e-05,
      "loss": 0.9,
      "step": 5568
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4925068616867065,
      "learning_rate": 1.7060984570168994e-05,
      "loss": 0.8891,
      "step": 5569
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8892455101013184,
      "learning_rate": 1.7060345653771206e-05,
      "loss": 0.9885,
      "step": 5570
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.009356737136841,
      "learning_rate": 1.7059706737373418e-05,
      "loss": 0.7959,
      "step": 5571
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.028639793395996,
      "learning_rate": 1.705906782097563e-05,
      "loss": 1.2893,
      "step": 5572
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.843369722366333,
      "learning_rate": 1.7058428904577837e-05,
      "loss": 0.8771,
      "step": 5573
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.772435188293457,
      "learning_rate": 1.705778998818005e-05,
      "loss": 0.908,
      "step": 5574
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.2002360820770264,
      "learning_rate": 1.7057151071782257e-05,
      "loss": 0.886,
      "step": 5575
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8537111282348633,
      "learning_rate": 1.7056512155384472e-05,
      "loss": 1.0621,
      "step": 5576
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7316992282867432,
      "learning_rate": 1.705587323898668e-05,
      "loss": 1.2023,
      "step": 5577
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.062540054321289,
      "learning_rate": 1.705523432258889e-05,
      "loss": 1.2009,
      "step": 5578
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.463801860809326,
      "learning_rate": 1.70545954061911e-05,
      "loss": 1.004,
      "step": 5579
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6945574283599854,
      "learning_rate": 1.705395648979331e-05,
      "loss": 0.9538,
      "step": 5580
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.67690908908844,
      "learning_rate": 1.7053317573395523e-05,
      "loss": 0.7837,
      "step": 5581
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.239534378051758,
      "learning_rate": 1.7052678656997734e-05,
      "loss": 1.0706,
      "step": 5582
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5245728492736816,
      "learning_rate": 1.7052039740599943e-05,
      "loss": 0.3461,
      "step": 5583
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.317350149154663,
      "learning_rate": 1.7051400824202154e-05,
      "loss": 0.967,
      "step": 5584
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.199466705322266,
      "learning_rate": 1.7050761907804366e-05,
      "loss": 1.6061,
      "step": 5585
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9056665897369385,
      "learning_rate": 1.7050122991406577e-05,
      "loss": 0.9061,
      "step": 5586
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.9639270305633545,
      "learning_rate": 1.7049484075008785e-05,
      "loss": 1.031,
      "step": 5587
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6633882522583008,
      "learning_rate": 1.7048845158610997e-05,
      "loss": 0.8892,
      "step": 5588
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.093656063079834,
      "learning_rate": 1.704820624221321e-05,
      "loss": 0.7863,
      "step": 5589
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8752344846725464,
      "learning_rate": 1.704756732581542e-05,
      "loss": 0.9501,
      "step": 5590
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7966331243515015,
      "learning_rate": 1.7046928409417628e-05,
      "loss": 0.8835,
      "step": 5591
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6065988540649414,
      "learning_rate": 1.704628949301984e-05,
      "loss": 0.8362,
      "step": 5592
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.9096782207489014,
      "learning_rate": 1.704565057662205e-05,
      "loss": 1.0955,
      "step": 5593
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9831880331039429,
      "learning_rate": 1.7045011660224263e-05,
      "loss": 0.8381,
      "step": 5594
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9206557273864746,
      "learning_rate": 1.704437274382647e-05,
      "loss": 0.843,
      "step": 5595
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.080294609069824,
      "learning_rate": 1.7043733827428682e-05,
      "loss": 0.9037,
      "step": 5596
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.221470594406128,
      "learning_rate": 1.7043094911030894e-05,
      "loss": 0.8378,
      "step": 5597
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1560616493225098,
      "learning_rate": 1.7042455994633106e-05,
      "loss": 0.9023,
      "step": 5598
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0847630500793457,
      "learning_rate": 1.7041817078235314e-05,
      "loss": 0.9203,
      "step": 5599
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3506360054016113,
      "learning_rate": 1.7041178161837525e-05,
      "loss": 1.0752,
      "step": 5600
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.503155469894409,
      "learning_rate": 1.7040539245439737e-05,
      "loss": 0.8981,
      "step": 5601
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.242492437362671,
      "learning_rate": 1.703990032904195e-05,
      "loss": 0.8176,
      "step": 5602
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.347795486450195,
      "learning_rate": 1.7039261412644157e-05,
      "loss": 1.8527,
      "step": 5603
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1755077838897705,
      "learning_rate": 1.7038622496246368e-05,
      "loss": 1.0,
      "step": 5604
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7416034936904907,
      "learning_rate": 1.703798357984858e-05,
      "loss": 0.8762,
      "step": 5605
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2595958709716797,
      "learning_rate": 1.703734466345079e-05,
      "loss": 0.8492,
      "step": 5606
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.770848751068115,
      "learning_rate": 1.7036705747053e-05,
      "loss": 1.4138,
      "step": 5607
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9386587142944336,
      "learning_rate": 1.7036066830655208e-05,
      "loss": 0.9952,
      "step": 5608
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0257818698883057,
      "learning_rate": 1.7035427914257422e-05,
      "loss": 1.0885,
      "step": 5609
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2487292289733887,
      "learning_rate": 1.703478899785963e-05,
      "loss": 1.0096,
      "step": 5610
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8692785501480103,
      "learning_rate": 1.7034150081461842e-05,
      "loss": 0.9845,
      "step": 5611
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.399524688720703,
      "learning_rate": 1.703351116506405e-05,
      "loss": 0.9655,
      "step": 5612
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7064965963363647,
      "learning_rate": 1.7032872248666262e-05,
      "loss": 1.0334,
      "step": 5613
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.5437779426574707,
      "learning_rate": 1.7032233332268473e-05,
      "loss": 0.6419,
      "step": 5614
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6316397190093994,
      "learning_rate": 1.7031594415870685e-05,
      "loss": 0.7244,
      "step": 5615
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9945930242538452,
      "learning_rate": 1.7030955499472893e-05,
      "loss": 1.0279,
      "step": 5616
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.139918804168701,
      "learning_rate": 1.7030316583075105e-05,
      "loss": 1.1839,
      "step": 5617
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.428191661834717,
      "learning_rate": 1.7029677666677316e-05,
      "loss": 0.9299,
      "step": 5618
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.811864852905273,
      "learning_rate": 1.7029038750279528e-05,
      "loss": 1.4433,
      "step": 5619
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0011234283447266,
      "learning_rate": 1.7028399833881736e-05,
      "loss": 0.9834,
      "step": 5620
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8861281871795654,
      "learning_rate": 1.7027760917483947e-05,
      "loss": 1.0831,
      "step": 5621
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4671385288238525,
      "learning_rate": 1.702712200108616e-05,
      "loss": 0.9646,
      "step": 5622
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8584234714508057,
      "learning_rate": 1.702648308468837e-05,
      "loss": 0.8744,
      "step": 5623
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.292440891265869,
      "learning_rate": 1.702584416829058e-05,
      "loss": 0.8268,
      "step": 5624
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9748132228851318,
      "learning_rate": 1.702520525189279e-05,
      "loss": 0.9447,
      "step": 5625
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.035184144973755,
      "learning_rate": 1.7024566335495002e-05,
      "loss": 0.8418,
      "step": 5626
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9536035060882568,
      "learning_rate": 1.7023927419097213e-05,
      "loss": 0.9023,
      "step": 5627
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.220726728439331,
      "learning_rate": 1.702328850269942e-05,
      "loss": 0.8103,
      "step": 5628
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9888039827346802,
      "learning_rate": 1.7022649586301633e-05,
      "loss": 0.9677,
      "step": 5629
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8194535970687866,
      "learning_rate": 1.7022010669903845e-05,
      "loss": 0.9445,
      "step": 5630
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.149606943130493,
      "learning_rate": 1.7021371753506056e-05,
      "loss": 0.9929,
      "step": 5631
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0375022888183594,
      "learning_rate": 1.7020732837108264e-05,
      "loss": 0.8464,
      "step": 5632
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.880854845046997,
      "learning_rate": 1.7020093920710476e-05,
      "loss": 0.9494,
      "step": 5633
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.184521198272705,
      "learning_rate": 1.7019455004312687e-05,
      "loss": 0.8517,
      "step": 5634
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.309720754623413,
      "learning_rate": 1.70188160879149e-05,
      "loss": 0.7701,
      "step": 5635
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.339298963546753,
      "learning_rate": 1.7018177171517107e-05,
      "loss": 0.904,
      "step": 5636
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2290713787078857,
      "learning_rate": 1.701753825511932e-05,
      "loss": 0.9221,
      "step": 5637
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8288092613220215,
      "learning_rate": 1.701689933872153e-05,
      "loss": 0.9248,
      "step": 5638
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.444822311401367,
      "learning_rate": 1.7016260422323742e-05,
      "loss": 1.5107,
      "step": 5639
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3817617893218994,
      "learning_rate": 1.701562150592595e-05,
      "loss": 1.2723,
      "step": 5640
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.058250904083252,
      "learning_rate": 1.701498258952816e-05,
      "loss": 0.9798,
      "step": 5641
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0762550830841064,
      "learning_rate": 1.701434367313037e-05,
      "loss": 1.1929,
      "step": 5642
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.208287000656128,
      "learning_rate": 1.7013704756732585e-05,
      "loss": 1.0269,
      "step": 5643
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.850783348083496,
      "learning_rate": 1.7013065840334793e-05,
      "loss": 0.9208,
      "step": 5644
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.851045608520508,
      "learning_rate": 1.7012426923937004e-05,
      "loss": 0.9009,
      "step": 5645
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8154162168502808,
      "learning_rate": 1.7011788007539212e-05,
      "loss": 0.931,
      "step": 5646
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3088483810424805,
      "learning_rate": 1.7011149091141427e-05,
      "loss": 0.8836,
      "step": 5647
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.692562580108643,
      "learning_rate": 1.7010510174743636e-05,
      "loss": 1.4835,
      "step": 5648
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9371777772903442,
      "learning_rate": 1.7009871258345847e-05,
      "loss": 1.0093,
      "step": 5649
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.8970794677734375,
      "learning_rate": 1.7009232341948055e-05,
      "loss": 0.5811,
      "step": 5650
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.295509338378906,
      "learning_rate": 1.7008593425550267e-05,
      "loss": 1.2569,
      "step": 5651
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9210882186889648,
      "learning_rate": 1.7007954509152478e-05,
      "loss": 0.7833,
      "step": 5652
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.139007806777954,
      "learning_rate": 1.700731559275469e-05,
      "loss": 0.9464,
      "step": 5653
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5767476558685303,
      "learning_rate": 1.7006676676356898e-05,
      "loss": 0.8865,
      "step": 5654
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.7465577125549316,
      "learning_rate": 1.700603775995911e-05,
      "loss": 0.8716,
      "step": 5655
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.028013229370117,
      "learning_rate": 1.700539884356132e-05,
      "loss": 1.3296,
      "step": 5656
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2177534103393555,
      "learning_rate": 1.7004759927163533e-05,
      "loss": 1.3528,
      "step": 5657
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9891505241394043,
      "learning_rate": 1.700412101076574e-05,
      "loss": 0.385,
      "step": 5658
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.041496992111206,
      "learning_rate": 1.7003482094367952e-05,
      "loss": 0.9862,
      "step": 5659
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6483304500579834,
      "learning_rate": 1.7002843177970164e-05,
      "loss": 0.8299,
      "step": 5660
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.656147480010986,
      "learning_rate": 1.7002204261572375e-05,
      "loss": 1.4988,
      "step": 5661
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.20302152633667,
      "learning_rate": 1.7001565345174584e-05,
      "loss": 0.9589,
      "step": 5662
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0136220455169678,
      "learning_rate": 1.7000926428776795e-05,
      "loss": 1.0228,
      "step": 5663
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0674350261688232,
      "learning_rate": 1.7000287512379007e-05,
      "loss": 0.9789,
      "step": 5664
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.042484760284424,
      "learning_rate": 1.6999648595981218e-05,
      "loss": 0.8484,
      "step": 5665
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.181990385055542,
      "learning_rate": 1.6999009679583426e-05,
      "loss": 1.0671,
      "step": 5666
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8522509336471558,
      "learning_rate": 1.6998370763185638e-05,
      "loss": 1.0322,
      "step": 5667
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.924290657043457,
      "learning_rate": 1.699773184678785e-05,
      "loss": 0.7752,
      "step": 5668
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.819620132446289,
      "learning_rate": 1.699709293039006e-05,
      "loss": 1.0126,
      "step": 5669
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5637954473495483,
      "learning_rate": 1.699645401399227e-05,
      "loss": 0.4288,
      "step": 5670
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.375672817230225,
      "learning_rate": 1.699581509759448e-05,
      "loss": 0.515,
      "step": 5671
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0440776348114014,
      "learning_rate": 1.6995176181196692e-05,
      "loss": 0.9267,
      "step": 5672
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.545915126800537,
      "learning_rate": 1.6994537264798904e-05,
      "loss": 0.8801,
      "step": 5673
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.479898452758789,
      "learning_rate": 1.6993898348401112e-05,
      "loss": 1.2385,
      "step": 5674
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9651703834533691,
      "learning_rate": 1.6993259432003324e-05,
      "loss": 0.9166,
      "step": 5675
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.288362979888916,
      "learning_rate": 1.6992620515605535e-05,
      "loss": 0.8463,
      "step": 5676
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8946229219436646,
      "learning_rate": 1.6991981599207747e-05,
      "loss": 0.9464,
      "step": 5677
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.091153383255005,
      "learning_rate": 1.6991342682809955e-05,
      "loss": 1.0075,
      "step": 5678
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7734311819076538,
      "learning_rate": 1.6990703766412166e-05,
      "loss": 0.9478,
      "step": 5679
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8954830169677734,
      "learning_rate": 1.6990064850014375e-05,
      "loss": 0.8732,
      "step": 5680
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7012112140655518,
      "learning_rate": 1.698942593361659e-05,
      "loss": 0.8781,
      "step": 5681
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.049954414367676,
      "learning_rate": 1.6988787017218798e-05,
      "loss": 1.6531,
      "step": 5682
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.099660873413086,
      "learning_rate": 1.698814810082101e-05,
      "loss": 0.9694,
      "step": 5683
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4976696968078613,
      "learning_rate": 1.6987509184423217e-05,
      "loss": 1.1729,
      "step": 5684
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.360403537750244,
      "learning_rate": 1.698687026802543e-05,
      "loss": 0.9886,
      "step": 5685
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.304133415222168,
      "learning_rate": 1.698623135162764e-05,
      "loss": 1.5828,
      "step": 5686
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0935490131378174,
      "learning_rate": 1.6985592435229852e-05,
      "loss": 0.8808,
      "step": 5687
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.977879285812378,
      "learning_rate": 1.698495351883206e-05,
      "loss": 0.4072,
      "step": 5688
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.361799478530884,
      "learning_rate": 1.698431460243427e-05,
      "loss": 1.0838,
      "step": 5689
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7800861597061157,
      "learning_rate": 1.6983675686036483e-05,
      "loss": 0.8013,
      "step": 5690
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8227403163909912,
      "learning_rate": 1.6983036769638695e-05,
      "loss": 0.9972,
      "step": 5691
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6444172859191895,
      "learning_rate": 1.6982397853240903e-05,
      "loss": 0.7932,
      "step": 5692
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7308896780014038,
      "learning_rate": 1.6981758936843114e-05,
      "loss": 0.8917,
      "step": 5693
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9949041604995728,
      "learning_rate": 1.6981120020445326e-05,
      "loss": 1.0345,
      "step": 5694
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6942901611328125,
      "learning_rate": 1.6980481104047538e-05,
      "loss": 0.8863,
      "step": 5695
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.979793667793274,
      "learning_rate": 1.6979842187649746e-05,
      "loss": 0.9974,
      "step": 5696
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.762925386428833,
      "learning_rate": 1.6979203271251957e-05,
      "loss": 1.3858,
      "step": 5697
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.516833782196045,
      "learning_rate": 1.697856435485417e-05,
      "loss": 1.7333,
      "step": 5698
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.647772789001465,
      "learning_rate": 1.697792543845638e-05,
      "loss": 0.914,
      "step": 5699
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2539873123168945,
      "learning_rate": 1.697728652205859e-05,
      "loss": 1.2932,
      "step": 5700
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.098215103149414,
      "learning_rate": 1.69766476056608e-05,
      "loss": 0.9364,
      "step": 5701
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.052857875823975,
      "learning_rate": 1.697600868926301e-05,
      "loss": 1.1774,
      "step": 5702
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4406397342681885,
      "learning_rate": 1.6975369772865223e-05,
      "loss": 1.0562,
      "step": 5703
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5059794187545776,
      "learning_rate": 1.697473085646743e-05,
      "loss": 0.8689,
      "step": 5704
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7866640090942383,
      "learning_rate": 1.6974091940069643e-05,
      "loss": 0.829,
      "step": 5705
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9278225898742676,
      "learning_rate": 1.6973453023671854e-05,
      "loss": 0.8303,
      "step": 5706
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.5883851051330566,
      "learning_rate": 1.6972814107274066e-05,
      "loss": 0.9642,
      "step": 5707
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6416527032852173,
      "learning_rate": 1.6972175190876274e-05,
      "loss": 0.7893,
      "step": 5708
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.3301167488098145,
      "learning_rate": 1.6971536274478486e-05,
      "loss": 1.7395,
      "step": 5709
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0334665775299072,
      "learning_rate": 1.6970897358080697e-05,
      "loss": 0.8746,
      "step": 5710
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.218015670776367,
      "learning_rate": 1.697025844168291e-05,
      "loss": 1.0489,
      "step": 5711
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.5154876708984375,
      "learning_rate": 1.6969619525285117e-05,
      "loss": 1.0917,
      "step": 5712
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.970531463623047,
      "learning_rate": 1.696898060888733e-05,
      "loss": 1.113,
      "step": 5713
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3169543743133545,
      "learning_rate": 1.696834169248954e-05,
      "loss": 1.0077,
      "step": 5714
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4540441036224365,
      "learning_rate": 1.696770277609175e-05,
      "loss": 0.9146,
      "step": 5715
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.791718602180481,
      "learning_rate": 1.696706385969396e-05,
      "loss": 0.4011,
      "step": 5716
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.272698402404785,
      "learning_rate": 1.696642494329617e-05,
      "loss": 0.8706,
      "step": 5717
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.429903984069824,
      "learning_rate": 1.696578602689838e-05,
      "loss": 1.6713,
      "step": 5718
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.732614278793335,
      "learning_rate": 1.6965147110500594e-05,
      "loss": 1.2141,
      "step": 5719
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0979390144348145,
      "learning_rate": 1.6964508194102803e-05,
      "loss": 1.0349,
      "step": 5720
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1934549808502197,
      "learning_rate": 1.6963869277705014e-05,
      "loss": 1.037,
      "step": 5721
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.511950492858887,
      "learning_rate": 1.6963230361307222e-05,
      "loss": 1.5405,
      "step": 5722
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3281219005584717,
      "learning_rate": 1.6962591444909434e-05,
      "loss": 0.8534,
      "step": 5723
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0153627395629883,
      "learning_rate": 1.6961952528511645e-05,
      "loss": 1.032,
      "step": 5724
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6514524221420288,
      "learning_rate": 1.6961313612113857e-05,
      "loss": 0.7847,
      "step": 5725
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.645910263061523,
      "learning_rate": 1.6960674695716065e-05,
      "loss": 1.9437,
      "step": 5726
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8545119762420654,
      "learning_rate": 1.6960035779318277e-05,
      "loss": 0.9172,
      "step": 5727
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6876221895217896,
      "learning_rate": 1.6959396862920488e-05,
      "loss": 0.8979,
      "step": 5728
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9101213216781616,
      "learning_rate": 1.69587579465227e-05,
      "loss": 1.0677,
      "step": 5729
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4008495807647705,
      "learning_rate": 1.6958119030124908e-05,
      "loss": 1.0053,
      "step": 5730
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3774538040161133,
      "learning_rate": 1.695748011372712e-05,
      "loss": 0.7939,
      "step": 5731
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.117967367172241,
      "learning_rate": 1.695684119732933e-05,
      "loss": 1.1206,
      "step": 5732
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.444792747497559,
      "learning_rate": 1.6956202280931542e-05,
      "loss": 1.6244,
      "step": 5733
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6571189165115356,
      "learning_rate": 1.695556336453375e-05,
      "loss": 0.8779,
      "step": 5734
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.457759857177734,
      "learning_rate": 1.6954924448135962e-05,
      "loss": 1.2137,
      "step": 5735
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8410532474517822,
      "learning_rate": 1.6954285531738174e-05,
      "loss": 0.945,
      "step": 5736
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.629940152168274,
      "learning_rate": 1.6953646615340385e-05,
      "loss": 0.9372,
      "step": 5737
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9331560134887695,
      "learning_rate": 1.6953007698942593e-05,
      "loss": 0.8985,
      "step": 5738
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.741236686706543,
      "learning_rate": 1.6952368782544805e-05,
      "loss": 1.1066,
      "step": 5739
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.168407917022705,
      "learning_rate": 1.6951729866147017e-05,
      "loss": 0.9165,
      "step": 5740
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.5113179683685303,
      "learning_rate": 1.6951090949749228e-05,
      "loss": 0.6921,
      "step": 5741
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.36336612701416,
      "learning_rate": 1.6950452033351436e-05,
      "loss": 1.4049,
      "step": 5742
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6949450969696045,
      "learning_rate": 1.6949813116953648e-05,
      "loss": 1.0381,
      "step": 5743
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.105848789215088,
      "learning_rate": 1.694917420055586e-05,
      "loss": 1.3472,
      "step": 5744
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7838492393493652,
      "learning_rate": 1.694853528415807e-05,
      "loss": 0.934,
      "step": 5745
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.248676061630249,
      "learning_rate": 1.694789636776028e-05,
      "loss": 0.9341,
      "step": 5746
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7740908861160278,
      "learning_rate": 1.694725745136249e-05,
      "loss": 0.9466,
      "step": 5747
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0465009212493896,
      "learning_rate": 1.6946618534964702e-05,
      "loss": 0.9179,
      "step": 5748
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.253445625305176,
      "learning_rate": 1.6945979618566914e-05,
      "loss": 0.8876,
      "step": 5749
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8427504301071167,
      "learning_rate": 1.6945340702169122e-05,
      "loss": 0.3458,
      "step": 5750
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0987536907196045,
      "learning_rate": 1.6944701785771333e-05,
      "loss": 1.0625,
      "step": 5751
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.19671893119812,
      "learning_rate": 1.694406286937354e-05,
      "loss": 0.9874,
      "step": 5752
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.01609206199646,
      "learning_rate": 1.6943423952975756e-05,
      "loss": 1.2201,
      "step": 5753
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9823096990585327,
      "learning_rate": 1.6942785036577965e-05,
      "loss": 0.9404,
      "step": 5754
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.710299015045166,
      "learning_rate": 1.6942146120180176e-05,
      "loss": 1.5118,
      "step": 5755
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9980939626693726,
      "learning_rate": 1.6941507203782384e-05,
      "loss": 0.998,
      "step": 5756
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7116808891296387,
      "learning_rate": 1.6940868287384596e-05,
      "loss": 0.8827,
      "step": 5757
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3186209201812744,
      "learning_rate": 1.6940229370986807e-05,
      "loss": 1.2604,
      "step": 5758
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0818774700164795,
      "learning_rate": 1.693959045458902e-05,
      "loss": 0.913,
      "step": 5759
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9609328508377075,
      "learning_rate": 1.6938951538191227e-05,
      "loss": 0.7695,
      "step": 5760
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3047268390655518,
      "learning_rate": 1.693831262179344e-05,
      "loss": 0.8985,
      "step": 5761
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9552513360977173,
      "learning_rate": 1.693767370539565e-05,
      "loss": 0.8991,
      "step": 5762
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9024782180786133,
      "learning_rate": 1.6937034788997862e-05,
      "loss": 1.0208,
      "step": 5763
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.988936424255371,
      "learning_rate": 1.693639587260007e-05,
      "loss": 0.8318,
      "step": 5764
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6028612852096558,
      "learning_rate": 1.693575695620228e-05,
      "loss": 0.4143,
      "step": 5765
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.180955410003662,
      "learning_rate": 1.6935118039804493e-05,
      "loss": 1.1976,
      "step": 5766
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.550363779067993,
      "learning_rate": 1.6934479123406705e-05,
      "loss": 0.888,
      "step": 5767
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.785672664642334,
      "learning_rate": 1.6933840207008913e-05,
      "loss": 0.9243,
      "step": 5768
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1972272396087646,
      "learning_rate": 1.6933201290611124e-05,
      "loss": 0.8119,
      "step": 5769
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.2125747203826904,
      "learning_rate": 1.6932562374213336e-05,
      "loss": 1.0323,
      "step": 5770
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.737372398376465,
      "learning_rate": 1.6931923457815547e-05,
      "loss": 0.8858,
      "step": 5771
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.130966901779175,
      "learning_rate": 1.6931284541417756e-05,
      "loss": 0.9207,
      "step": 5772
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.7540245056152344,
      "learning_rate": 1.6930645625019967e-05,
      "loss": 0.7003,
      "step": 5773
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7182899713516235,
      "learning_rate": 1.693000670862218e-05,
      "loss": 0.8667,
      "step": 5774
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.11236310005188,
      "learning_rate": 1.692936779222439e-05,
      "loss": 1.107,
      "step": 5775
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1758131980895996,
      "learning_rate": 1.69287288758266e-05,
      "loss": 0.8783,
      "step": 5776
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.478323936462402,
      "learning_rate": 1.692808995942881e-05,
      "loss": 1.0218,
      "step": 5777
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8851121664047241,
      "learning_rate": 1.692745104303102e-05,
      "loss": 0.9268,
      "step": 5778
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.233647584915161,
      "learning_rate": 1.6926812126633233e-05,
      "loss": 1.0701,
      "step": 5779
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.660714030265808,
      "learning_rate": 1.692617321023544e-05,
      "loss": 0.9961,
      "step": 5780
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.338945150375366,
      "learning_rate": 1.6925534293837653e-05,
      "loss": 0.9904,
      "step": 5781
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.301825523376465,
      "learning_rate": 1.6924895377439864e-05,
      "loss": 0.9562,
      "step": 5782
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3854568004608154,
      "learning_rate": 1.6924256461042076e-05,
      "loss": 0.8687,
      "step": 5783
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.3343441486358643,
      "learning_rate": 1.6923617544644284e-05,
      "loss": 0.4687,
      "step": 5784
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1512362957000732,
      "learning_rate": 1.6922978628246495e-05,
      "loss": 1.0392,
      "step": 5785
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0119786262512207,
      "learning_rate": 1.6922339711848707e-05,
      "loss": 1.0077,
      "step": 5786
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.0206634998321533,
      "learning_rate": 1.692170079545092e-05,
      "loss": 1.2783,
      "step": 5787
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.960322856903076,
      "learning_rate": 1.6921061879053127e-05,
      "loss": 1.4796,
      "step": 5788
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.412785768508911,
      "learning_rate": 1.6920422962655338e-05,
      "loss": 1.1914,
      "step": 5789
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.421473503112793,
      "learning_rate": 1.6919784046257546e-05,
      "loss": 0.8787,
      "step": 5790
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.293405532836914,
      "learning_rate": 1.691914512985976e-05,
      "loss": 0.9343,
      "step": 5791
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9561409950256348,
      "learning_rate": 1.691850621346197e-05,
      "loss": 1.0888,
      "step": 5792
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8851224184036255,
      "learning_rate": 1.691786729706418e-05,
      "loss": 1.0137,
      "step": 5793
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.805359363555908,
      "learning_rate": 1.691722838066639e-05,
      "loss": 0.9977,
      "step": 5794
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.076048374176025,
      "learning_rate": 1.69165894642686e-05,
      "loss": 1.2938,
      "step": 5795
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.0495548248291016,
      "learning_rate": 1.6915950547870812e-05,
      "loss": 1.1855,
      "step": 5796
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.552786350250244,
      "learning_rate": 1.6915311631473024e-05,
      "loss": 1.1704,
      "step": 5797
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.463305473327637,
      "learning_rate": 1.6914672715075232e-05,
      "loss": 2.0088,
      "step": 5798
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9485443830490112,
      "learning_rate": 1.6914033798677444e-05,
      "loss": 0.9529,
      "step": 5799
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4267853498458862,
      "learning_rate": 1.6913394882279655e-05,
      "loss": 0.4179,
      "step": 5800
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.519962787628174,
      "learning_rate": 1.6912755965881867e-05,
      "loss": 0.8205,
      "step": 5801
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8516064882278442,
      "learning_rate": 1.6912117049484075e-05,
      "loss": 0.5981,
      "step": 5802
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9602439403533936,
      "learning_rate": 1.6911478133086286e-05,
      "loss": 0.6531,
      "step": 5803
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8823484182357788,
      "learning_rate": 1.6910839216688498e-05,
      "loss": 0.9071,
      "step": 5804
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0589792728424072,
      "learning_rate": 1.691020030029071e-05,
      "loss": 0.8841,
      "step": 5805
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.883999228477478,
      "learning_rate": 1.6909561383892918e-05,
      "loss": 0.9768,
      "step": 5806
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.5123841762542725,
      "learning_rate": 1.690892246749513e-05,
      "loss": 0.9465,
      "step": 5807
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0641531944274902,
      "learning_rate": 1.690828355109734e-05,
      "loss": 0.9991,
      "step": 5808
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.494948387145996,
      "learning_rate": 1.6907644634699552e-05,
      "loss": 0.623,
      "step": 5809
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.264791011810303,
      "learning_rate": 1.690700571830176e-05,
      "loss": 0.8898,
      "step": 5810
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3056889772415161,
      "learning_rate": 1.6906366801903972e-05,
      "loss": 0.3692,
      "step": 5811
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.412435531616211,
      "learning_rate": 1.6905727885506184e-05,
      "loss": 1.1768,
      "step": 5812
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.80289888381958,
      "learning_rate": 1.6905088969108395e-05,
      "loss": 0.9336,
      "step": 5813
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.204329013824463,
      "learning_rate": 1.6904450052710603e-05,
      "loss": 1.0277,
      "step": 5814
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3870997428894043,
      "learning_rate": 1.6903811136312815e-05,
      "loss": 0.8616,
      "step": 5815
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.217442035675049,
      "learning_rate": 1.6903172219915026e-05,
      "loss": 1.501,
      "step": 5816
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7531683444976807,
      "learning_rate": 1.6902533303517238e-05,
      "loss": 0.3748,
      "step": 5817
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8829535245895386,
      "learning_rate": 1.6901894387119446e-05,
      "loss": 1.309,
      "step": 5818
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.534684658050537,
      "learning_rate": 1.6901255470721658e-05,
      "loss": 0.5269,
      "step": 5819
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8337470293045044,
      "learning_rate": 1.690061655432387e-05,
      "loss": 0.8429,
      "step": 5820
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.607871651649475,
      "learning_rate": 1.689997763792608e-05,
      "loss": 0.8912,
      "step": 5821
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9317513704299927,
      "learning_rate": 1.689933872152829e-05,
      "loss": 0.8793,
      "step": 5822
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9111602306365967,
      "learning_rate": 1.68986998051305e-05,
      "loss": 1.3434,
      "step": 5823
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.6164374351501465,
      "learning_rate": 1.689806088873271e-05,
      "loss": 1.7813,
      "step": 5824
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.921014785766602,
      "learning_rate": 1.6897421972334923e-05,
      "loss": 1.6188,
      "step": 5825
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9146515130996704,
      "learning_rate": 1.689678305593713e-05,
      "loss": 0.9835,
      "step": 5826
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.137193202972412,
      "learning_rate": 1.6896144139539343e-05,
      "loss": 0.9708,
      "step": 5827
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.8396735191345215,
      "learning_rate": 1.689550522314155e-05,
      "loss": 1.047,
      "step": 5828
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.206543207168579,
      "learning_rate": 1.6894866306743766e-05,
      "loss": 0.9787,
      "step": 5829
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.243196725845337,
      "learning_rate": 1.6894227390345974e-05,
      "loss": 0.9457,
      "step": 5830
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9390778541564941,
      "learning_rate": 1.6893588473948186e-05,
      "loss": 0.9998,
      "step": 5831
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.7359488010406494,
      "learning_rate": 1.6892949557550394e-05,
      "loss": 0.833,
      "step": 5832
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.434828758239746,
      "learning_rate": 1.6892310641152606e-05,
      "loss": 0.8531,
      "step": 5833
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.101654052734375,
      "learning_rate": 1.6891671724754817e-05,
      "loss": 0.9696,
      "step": 5834
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4392080307006836,
      "learning_rate": 1.689103280835703e-05,
      "loss": 0.5052,
      "step": 5835
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.999517798423767,
      "learning_rate": 1.6890393891959237e-05,
      "loss": 0.8448,
      "step": 5836
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1109962463378906,
      "learning_rate": 1.688975497556145e-05,
      "loss": 0.9295,
      "step": 5837
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.706552505493164,
      "learning_rate": 1.688911605916366e-05,
      "loss": 0.9177,
      "step": 5838
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5047993659973145,
      "learning_rate": 1.688847714276587e-05,
      "loss": 0.4447,
      "step": 5839
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9383748769760132,
      "learning_rate": 1.688783822636808e-05,
      "loss": 0.8701,
      "step": 5840
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0408694744110107,
      "learning_rate": 1.688719930997029e-05,
      "loss": 1.0927,
      "step": 5841
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.6946685314178467,
      "learning_rate": 1.6886560393572503e-05,
      "loss": 1.2061,
      "step": 5842
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7329497337341309,
      "learning_rate": 1.6885921477174714e-05,
      "loss": 0.8415,
      "step": 5843
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1262645721435547,
      "learning_rate": 1.6885282560776923e-05,
      "loss": 0.9388,
      "step": 5844
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.905272960662842,
      "learning_rate": 1.6884643644379134e-05,
      "loss": 0.9681,
      "step": 5845
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9147874116897583,
      "learning_rate": 1.6884004727981346e-05,
      "loss": 0.9944,
      "step": 5846
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9266045093536377,
      "learning_rate": 1.6883365811583557e-05,
      "loss": 0.9797,
      "step": 5847
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1405999660491943,
      "learning_rate": 1.6882726895185765e-05,
      "loss": 0.9302,
      "step": 5848
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6491303443908691,
      "learning_rate": 1.6882087978787977e-05,
      "loss": 0.4556,
      "step": 5849
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8460010290145874,
      "learning_rate": 1.688144906239019e-05,
      "loss": 0.9105,
      "step": 5850
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3839402198791504,
      "learning_rate": 1.68808101459924e-05,
      "loss": 1.1721,
      "step": 5851
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.0448157787323,
      "learning_rate": 1.6880171229594608e-05,
      "loss": 1.2303,
      "step": 5852
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7083630561828613,
      "learning_rate": 1.687953231319682e-05,
      "loss": 1.2283,
      "step": 5853
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.733670473098755,
      "learning_rate": 1.687889339679903e-05,
      "loss": 0.9521,
      "step": 5854
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4966835975646973,
      "learning_rate": 1.6878254480401243e-05,
      "loss": 1.1751,
      "step": 5855
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.7286500930786133,
      "learning_rate": 1.687761556400345e-05,
      "loss": 1.1047,
      "step": 5856
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6483372449874878,
      "learning_rate": 1.6876976647605662e-05,
      "loss": 0.869,
      "step": 5857
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.088249683380127,
      "learning_rate": 1.6876337731207874e-05,
      "loss": 0.9498,
      "step": 5858
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7046674489974976,
      "learning_rate": 1.6875698814810086e-05,
      "loss": 0.9113,
      "step": 5859
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.654099941253662,
      "learning_rate": 1.6875059898412294e-05,
      "loss": 1.2438,
      "step": 5860
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.8633031845092773,
      "learning_rate": 1.6874420982014505e-05,
      "loss": 0.7792,
      "step": 5861
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.765152454376221,
      "learning_rate": 1.6873782065616713e-05,
      "loss": 1.3951,
      "step": 5862
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2481493949890137,
      "learning_rate": 1.687314314921893e-05,
      "loss": 1.007,
      "step": 5863
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0834591388702393,
      "learning_rate": 1.6872504232821137e-05,
      "loss": 0.6849,
      "step": 5864
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.617562770843506,
      "learning_rate": 1.6871865316423348e-05,
      "loss": 1.0745,
      "step": 5865
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.001816749572754,
      "learning_rate": 1.6871226400025556e-05,
      "loss": 0.923,
      "step": 5866
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2525532245635986,
      "learning_rate": 1.6870587483627768e-05,
      "loss": 1.0427,
      "step": 5867
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.873846411705017,
      "learning_rate": 1.686994856722998e-05,
      "loss": 0.998,
      "step": 5868
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.637041091918945,
      "learning_rate": 1.686930965083219e-05,
      "loss": 2.1762,
      "step": 5869
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6035361289978027,
      "learning_rate": 1.68686707344344e-05,
      "loss": 0.7925,
      "step": 5870
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.255488157272339,
      "learning_rate": 1.686803181803661e-05,
      "loss": 1.0255,
      "step": 5871
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.865027904510498,
      "learning_rate": 1.6867392901638822e-05,
      "loss": 1.1554,
      "step": 5872
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.314765691757202,
      "learning_rate": 1.6866753985241034e-05,
      "loss": 1.0382,
      "step": 5873
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9222818613052368,
      "learning_rate": 1.6866115068843242e-05,
      "loss": 0.9198,
      "step": 5874
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8104809522628784,
      "learning_rate": 1.6865476152445453e-05,
      "loss": 0.9837,
      "step": 5875
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5185871124267578,
      "learning_rate": 1.6864837236047665e-05,
      "loss": 0.8492,
      "step": 5876
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1166183948516846,
      "learning_rate": 1.6864198319649876e-05,
      "loss": 0.9724,
      "step": 5877
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4499775171279907,
      "learning_rate": 1.6863559403252085e-05,
      "loss": 0.456,
      "step": 5878
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.553471326828003,
      "learning_rate": 1.6862920486854296e-05,
      "loss": 0.9122,
      "step": 5879
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8864691257476807,
      "learning_rate": 1.6862281570456508e-05,
      "loss": 0.8981,
      "step": 5880
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6069215536117554,
      "learning_rate": 1.686164265405872e-05,
      "loss": 0.9711,
      "step": 5881
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.355985164642334,
      "learning_rate": 1.6861003737660927e-05,
      "loss": 0.8513,
      "step": 5882
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.281398296356201,
      "learning_rate": 1.686036482126314e-05,
      "loss": 0.9241,
      "step": 5883
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.747255802154541,
      "learning_rate": 1.685972590486535e-05,
      "loss": 0.8731,
      "step": 5884
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.036342620849609,
      "learning_rate": 1.6859086988467562e-05,
      "loss": 1.1284,
      "step": 5885
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.870415210723877,
      "learning_rate": 1.685844807206977e-05,
      "loss": 0.9251,
      "step": 5886
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2348697185516357,
      "learning_rate": 1.6857809155671982e-05,
      "loss": 1.1153,
      "step": 5887
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1379361152648926,
      "learning_rate": 1.6857170239274193e-05,
      "loss": 0.729,
      "step": 5888
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.586530089378357,
      "learning_rate": 1.6856531322876405e-05,
      "loss": 0.8287,
      "step": 5889
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.236250400543213,
      "learning_rate": 1.6855892406478613e-05,
      "loss": 0.8726,
      "step": 5890
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.235778570175171,
      "learning_rate": 1.6855253490080825e-05,
      "loss": 1.2331,
      "step": 5891
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.348383665084839,
      "learning_rate": 1.6854614573683036e-05,
      "loss": 0.8452,
      "step": 5892
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.312907695770264,
      "learning_rate": 1.6853975657285248e-05,
      "loss": 1.265,
      "step": 5893
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.091106414794922,
      "learning_rate": 1.6853336740887456e-05,
      "loss": 0.949,
      "step": 5894
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3571767807006836,
      "learning_rate": 1.6852697824489667e-05,
      "loss": 1.1114,
      "step": 5895
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7450863122940063,
      "learning_rate": 1.685205890809188e-05,
      "loss": 0.9056,
      "step": 5896
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.874335289001465,
      "learning_rate": 1.685141999169409e-05,
      "loss": 1.7107,
      "step": 5897
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.46333384513855,
      "learning_rate": 1.68507810752963e-05,
      "loss": 0.8341,
      "step": 5898
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6660847663879395,
      "learning_rate": 1.685014215889851e-05,
      "loss": 0.9389,
      "step": 5899
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1116888523101807,
      "learning_rate": 1.684950324250072e-05,
      "loss": 0.9372,
      "step": 5900
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.155876398086548,
      "learning_rate": 1.6848864326102933e-05,
      "loss": 1.2291,
      "step": 5901
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0593631267547607,
      "learning_rate": 1.684822540970514e-05,
      "loss": 0.9522,
      "step": 5902
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1489078998565674,
      "learning_rate": 1.6847586493307353e-05,
      "loss": 0.6112,
      "step": 5903
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.390748977661133,
      "learning_rate": 1.684694757690956e-05,
      "loss": 1.0911,
      "step": 5904
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0890376567840576,
      "learning_rate": 1.6846308660511773e-05,
      "loss": 0.972,
      "step": 5905
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1290862560272217,
      "learning_rate": 1.6845669744113984e-05,
      "loss": 0.9447,
      "step": 5906
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9335672855377197,
      "learning_rate": 1.6845030827716196e-05,
      "loss": 0.8828,
      "step": 5907
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.359414577484131,
      "learning_rate": 1.6844391911318404e-05,
      "loss": 0.6169,
      "step": 5908
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.354708671569824,
      "learning_rate": 1.6843752994920615e-05,
      "loss": 1.8864,
      "step": 5909
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.387722969055176,
      "learning_rate": 1.6843114078522827e-05,
      "loss": 1.1929,
      "step": 5910
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.925810694694519,
      "learning_rate": 1.684247516212504e-05,
      "loss": 0.921,
      "step": 5911
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.0637636184692383,
      "learning_rate": 1.6841836245727247e-05,
      "loss": 0.6726,
      "step": 5912
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7300283908843994,
      "learning_rate": 1.6841197329329458e-05,
      "loss": 0.9225,
      "step": 5913
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8224507570266724,
      "learning_rate": 1.684055841293167e-05,
      "loss": 0.9944,
      "step": 5914
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6752750873565674,
      "learning_rate": 1.683991949653388e-05,
      "loss": 1.1994,
      "step": 5915
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8859906196594238,
      "learning_rate": 1.683928058013609e-05,
      "loss": 1.0014,
      "step": 5916
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.356161117553711,
      "learning_rate": 1.68386416637383e-05,
      "loss": 1.6536,
      "step": 5917
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.638657808303833,
      "learning_rate": 1.6838002747340513e-05,
      "loss": 1.038,
      "step": 5918
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4584925174713135,
      "learning_rate": 1.6837363830942724e-05,
      "loss": 0.8916,
      "step": 5919
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1463136672973633,
      "learning_rate": 1.6836724914544932e-05,
      "loss": 0.845,
      "step": 5920
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.234257698059082,
      "learning_rate": 1.6836085998147144e-05,
      "loss": 0.8375,
      "step": 5921
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9303786754608154,
      "learning_rate": 1.6835447081749355e-05,
      "loss": 0.7984,
      "step": 5922
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8667641878128052,
      "learning_rate": 1.6834808165351567e-05,
      "loss": 1.2231,
      "step": 5923
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.563840866088867,
      "learning_rate": 1.6834169248953775e-05,
      "loss": 0.9173,
      "step": 5924
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.225254535675049,
      "learning_rate": 1.6833530332555987e-05,
      "loss": 0.9294,
      "step": 5925
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5369154214859009,
      "learning_rate": 1.6832891416158198e-05,
      "loss": 0.7687,
      "step": 5926
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.797014832496643,
      "learning_rate": 1.683225249976041e-05,
      "loss": 0.8212,
      "step": 5927
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8799904584884644,
      "learning_rate": 1.6831613583362618e-05,
      "loss": 0.9793,
      "step": 5928
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.121408939361572,
      "learning_rate": 1.683097466696483e-05,
      "loss": 0.9052,
      "step": 5929
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4565998315811157,
      "learning_rate": 1.683033575056704e-05,
      "loss": 0.8979,
      "step": 5930
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6070460081100464,
      "learning_rate": 1.6829696834169253e-05,
      "loss": 0.3627,
      "step": 5931
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.033885955810547,
      "learning_rate": 1.682905791777146e-05,
      "loss": 0.9074,
      "step": 5932
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9556031227111816,
      "learning_rate": 1.6828419001373672e-05,
      "loss": 0.7371,
      "step": 5933
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.914750337600708,
      "learning_rate": 1.682778008497588e-05,
      "loss": 1.1444,
      "step": 5934
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.317988157272339,
      "learning_rate": 1.6827141168578095e-05,
      "loss": 0.9401,
      "step": 5935
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.5348761081695557,
      "learning_rate": 1.6826502252180304e-05,
      "loss": 0.9213,
      "step": 5936
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5896326303482056,
      "learning_rate": 1.6825863335782515e-05,
      "loss": 0.9375,
      "step": 5937
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.008188486099243,
      "learning_rate": 1.6825224419384723e-05,
      "loss": 0.9232,
      "step": 5938
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6727805137634277,
      "learning_rate": 1.6824585502986938e-05,
      "loss": 0.8748,
      "step": 5939
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.0981945991516113,
      "learning_rate": 1.6823946586589146e-05,
      "loss": 1.1158,
      "step": 5940
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.254506587982178,
      "learning_rate": 1.6823307670191358e-05,
      "loss": 1.5696,
      "step": 5941
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.136817216873169,
      "learning_rate": 1.6822668753793566e-05,
      "loss": 0.884,
      "step": 5942
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1699161529541016,
      "learning_rate": 1.6822029837395778e-05,
      "loss": 1.2536,
      "step": 5943
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.388813018798828,
      "learning_rate": 1.682139092099799e-05,
      "loss": 0.9794,
      "step": 5944
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7037954330444336,
      "learning_rate": 1.68207520046002e-05,
      "loss": 0.9084,
      "step": 5945
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0565714836120605,
      "learning_rate": 1.682011308820241e-05,
      "loss": 1.0897,
      "step": 5946
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3605458736419678,
      "learning_rate": 1.681947417180462e-05,
      "loss": 0.8888,
      "step": 5947
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0007853507995605,
      "learning_rate": 1.6818835255406832e-05,
      "loss": 0.9514,
      "step": 5948
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6538342237472534,
      "learning_rate": 1.6818196339009043e-05,
      "loss": 0.9726,
      "step": 5949
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1722412109375,
      "learning_rate": 1.681755742261125e-05,
      "loss": 0.9676,
      "step": 5950
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.6569182872772217,
      "learning_rate": 1.6816918506213463e-05,
      "loss": 1.2067,
      "step": 5951
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.786990761756897,
      "learning_rate": 1.6816279589815675e-05,
      "loss": 0.9259,
      "step": 5952
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.648777961730957,
      "learning_rate": 1.6815640673417886e-05,
      "loss": 0.8762,
      "step": 5953
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.9019882678985596,
      "learning_rate": 1.6815001757020094e-05,
      "loss": 1.1369,
      "step": 5954
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8146634101867676,
      "learning_rate": 1.6814362840622306e-05,
      "loss": 0.8859,
      "step": 5955
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7560794353485107,
      "learning_rate": 1.6813723924224518e-05,
      "loss": 0.3951,
      "step": 5956
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.217226505279541,
      "learning_rate": 1.681308500782673e-05,
      "loss": 0.9611,
      "step": 5957
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.374038219451904,
      "learning_rate": 1.6812446091428937e-05,
      "loss": 1.3365,
      "step": 5958
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7437814474105835,
      "learning_rate": 1.681180717503115e-05,
      "loss": 0.7716,
      "step": 5959
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4181020259857178,
      "learning_rate": 1.681116825863336e-05,
      "loss": 0.3738,
      "step": 5960
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7512564659118652,
      "learning_rate": 1.6810529342235572e-05,
      "loss": 0.9271,
      "step": 5961
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.705444812774658,
      "learning_rate": 1.680989042583778e-05,
      "loss": 1.6336,
      "step": 5962
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.402648687362671,
      "learning_rate": 1.680925150943999e-05,
      "loss": 0.8449,
      "step": 5963
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7653669118881226,
      "learning_rate": 1.6808612593042203e-05,
      "loss": 0.9073,
      "step": 5964
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8025306463241577,
      "learning_rate": 1.6807973676644415e-05,
      "loss": 0.9809,
      "step": 5965
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5874673128128052,
      "learning_rate": 1.6807334760246623e-05,
      "loss": 1.3039,
      "step": 5966
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9817795753479004,
      "learning_rate": 1.6806695843848834e-05,
      "loss": 1.0236,
      "step": 5967
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.9639089107513428,
      "learning_rate": 1.6806056927451046e-05,
      "loss": 0.9844,
      "step": 5968
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4497973918914795,
      "learning_rate": 1.6805418011053257e-05,
      "loss": 0.8753,
      "step": 5969
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.8345348834991455,
      "learning_rate": 1.6804779094655466e-05,
      "loss": 1.1237,
      "step": 5970
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5659470558166504,
      "learning_rate": 1.6804140178257677e-05,
      "loss": 0.8366,
      "step": 5971
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.369213104248047,
      "learning_rate": 1.6803501261859885e-05,
      "loss": 1.4164,
      "step": 5972
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.687640905380249,
      "learning_rate": 1.68028623454621e-05,
      "loss": 0.8665,
      "step": 5973
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3375792503356934,
      "learning_rate": 1.680222342906431e-05,
      "loss": 1.1022,
      "step": 5974
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0698330402374268,
      "learning_rate": 1.680158451266652e-05,
      "loss": 0.991,
      "step": 5975
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.212163925170898,
      "learning_rate": 1.6800945596268728e-05,
      "loss": 1.8226,
      "step": 5976
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6453725099563599,
      "learning_rate": 1.680030667987094e-05,
      "loss": 0.889,
      "step": 5977
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2942397594451904,
      "learning_rate": 1.679966776347315e-05,
      "loss": 0.8712,
      "step": 5978
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.144063949584961,
      "learning_rate": 1.6799028847075363e-05,
      "loss": 1.2027,
      "step": 5979
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9357234239578247,
      "learning_rate": 1.679838993067757e-05,
      "loss": 1.028,
      "step": 5980
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4932395219802856,
      "learning_rate": 1.6797751014279782e-05,
      "loss": 0.935,
      "step": 5981
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9402894973754883,
      "learning_rate": 1.6797112097881994e-05,
      "loss": 0.9146,
      "step": 5982
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8199567794799805,
      "learning_rate": 1.6796473181484206e-05,
      "loss": 0.9183,
      "step": 5983
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.172600269317627,
      "learning_rate": 1.6795834265086414e-05,
      "loss": 0.8645,
      "step": 5984
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.806976079940796,
      "learning_rate": 1.6795195348688625e-05,
      "loss": 0.9554,
      "step": 5985
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5120776891708374,
      "learning_rate": 1.6794556432290837e-05,
      "loss": 0.8379,
      "step": 5986
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6819589138031006,
      "learning_rate": 1.679391751589305e-05,
      "loss": 0.8233,
      "step": 5987
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.247080326080322,
      "learning_rate": 1.6793278599495257e-05,
      "loss": 1.5647,
      "step": 5988
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.215360403060913,
      "learning_rate": 1.6792639683097468e-05,
      "loss": 0.949,
      "step": 5989
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0672459602355957,
      "learning_rate": 1.679200076669968e-05,
      "loss": 0.8057,
      "step": 5990
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7330652475357056,
      "learning_rate": 1.679136185030189e-05,
      "loss": 0.9736,
      "step": 5991
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2661454677581787,
      "learning_rate": 1.67907229339041e-05,
      "loss": 0.9375,
      "step": 5992
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4619078636169434,
      "learning_rate": 1.679008401750631e-05,
      "loss": 1.0424,
      "step": 5993
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6848317384719849,
      "learning_rate": 1.6789445101108522e-05,
      "loss": 0.4664,
      "step": 5994
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.06076717376709,
      "learning_rate": 1.678880618471073e-05,
      "loss": 0.9859,
      "step": 5995
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.03892183303833,
      "learning_rate": 1.6788167268312942e-05,
      "loss": 1.0053,
      "step": 5996
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4919517040252686,
      "learning_rate": 1.6787528351915154e-05,
      "loss": 0.3926,
      "step": 5997
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.896596074104309,
      "learning_rate": 1.6786889435517365e-05,
      "loss": 1.0737,
      "step": 5998
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.135463237762451,
      "learning_rate": 1.6786250519119573e-05,
      "loss": 0.9944,
      "step": 5999
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5134984254837036,
      "learning_rate": 1.6785611602721785e-05,
      "loss": 0.9028,
      "step": 6000
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9912219047546387,
      "learning_rate": 1.6784972686323993e-05,
      "loss": 1.1093,
      "step": 6001
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6987954378128052,
      "learning_rate": 1.6784333769926208e-05,
      "loss": 1.0041,
      "step": 6002
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.1770365238189697,
      "learning_rate": 1.6783694853528416e-05,
      "loss": 0.5438,
      "step": 6003
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.279750823974609,
      "learning_rate": 1.6783055937130628e-05,
      "loss": 1.6296,
      "step": 6004
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8644561767578125,
      "learning_rate": 1.6782417020732836e-05,
      "loss": 0.9817,
      "step": 6005
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.568636178970337,
      "learning_rate": 1.678177810433505e-05,
      "loss": 1.1391,
      "step": 6006
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.724799633026123,
      "learning_rate": 1.678113918793726e-05,
      "loss": 1.0181,
      "step": 6007
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.05668830871582,
      "learning_rate": 1.678050027153947e-05,
      "loss": 1.7983,
      "step": 6008
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.836711049079895,
      "learning_rate": 1.677986135514168e-05,
      "loss": 1.2506,
      "step": 6009
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.966507077217102,
      "learning_rate": 1.677922243874389e-05,
      "loss": 0.9675,
      "step": 6010
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.752920150756836,
      "learning_rate": 1.6778583522346102e-05,
      "loss": 1.8273,
      "step": 6011
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3947644233703613,
      "learning_rate": 1.6777944605948313e-05,
      "loss": 0.829,
      "step": 6012
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7417513132095337,
      "learning_rate": 1.677730568955052e-05,
      "loss": 0.8556,
      "step": 6013
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0021491050720215,
      "learning_rate": 1.6776666773152733e-05,
      "loss": 0.9562,
      "step": 6014
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3735145330429077,
      "learning_rate": 1.6776027856754945e-05,
      "loss": 0.8709,
      "step": 6015
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4778159856796265,
      "learning_rate": 1.6775388940357156e-05,
      "loss": 0.879,
      "step": 6016
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5905077457427979,
      "learning_rate": 1.6774750023959364e-05,
      "loss": 0.7029,
      "step": 6017
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4805245399475098,
      "learning_rate": 1.6774111107561576e-05,
      "loss": 0.8764,
      "step": 6018
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9552417993545532,
      "learning_rate": 1.6773472191163787e-05,
      "loss": 0.9721,
      "step": 6019
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.147677183151245,
      "learning_rate": 1.6772833274766e-05,
      "loss": 0.9108,
      "step": 6020
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.712675094604492,
      "learning_rate": 1.6772194358368207e-05,
      "loss": 0.685,
      "step": 6021
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5584893226623535,
      "learning_rate": 1.677155544197042e-05,
      "loss": 0.9904,
      "step": 6022
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.147937059402466,
      "learning_rate": 1.677091652557263e-05,
      "loss": 0.9794,
      "step": 6023
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9518327713012695,
      "learning_rate": 1.6770277609174842e-05,
      "loss": 0.9751,
      "step": 6024
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.79219913482666,
      "learning_rate": 1.676963869277705e-05,
      "loss": 0.8517,
      "step": 6025
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7557719945907593,
      "learning_rate": 1.676899977637926e-05,
      "loss": 0.8822,
      "step": 6026
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.1883246898651123,
      "learning_rate": 1.6768360859981473e-05,
      "loss": 0.9236,
      "step": 6027
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7530850172042847,
      "learning_rate": 1.6767721943583685e-05,
      "loss": 0.9423,
      "step": 6028
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9933468103408813,
      "learning_rate": 1.6767083027185893e-05,
      "loss": 0.8399,
      "step": 6029
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4983386993408203,
      "learning_rate": 1.6766444110788104e-05,
      "loss": 1.1449,
      "step": 6030
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.889339804649353,
      "learning_rate": 1.6765805194390316e-05,
      "loss": 1.1299,
      "step": 6031
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0810201168060303,
      "learning_rate": 1.6765166277992527e-05,
      "loss": 1.3297,
      "step": 6032
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.30963134765625,
      "learning_rate": 1.6764527361594736e-05,
      "loss": 1.3025,
      "step": 6033
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.16898512840271,
      "learning_rate": 1.6763888445196947e-05,
      "loss": 0.7986,
      "step": 6034
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.96791934967041,
      "learning_rate": 1.676324952879916e-05,
      "loss": 1.3018,
      "step": 6035
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9012129306793213,
      "learning_rate": 1.676261061240137e-05,
      "loss": 0.4658,
      "step": 6036
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.002439498901367,
      "learning_rate": 1.6761971696003578e-05,
      "loss": 1.1786,
      "step": 6037
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9071000814437866,
      "learning_rate": 1.676133277960579e-05,
      "loss": 1.0208,
      "step": 6038
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.749577760696411,
      "learning_rate": 1.6760693863207998e-05,
      "loss": 1.0938,
      "step": 6039
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6742148399353027,
      "learning_rate": 1.6760054946810213e-05,
      "loss": 1.1244,
      "step": 6040
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.53493595123291,
      "learning_rate": 1.675941603041242e-05,
      "loss": 1.4561,
      "step": 6041
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9927418231964111,
      "learning_rate": 1.6758777114014633e-05,
      "loss": 0.9456,
      "step": 6042
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.916022539138794,
      "learning_rate": 1.675813819761684e-05,
      "loss": 1.1204,
      "step": 6043
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9136457443237305,
      "learning_rate": 1.6757499281219052e-05,
      "loss": 0.8918,
      "step": 6044
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.5760462284088135,
      "learning_rate": 1.6756860364821264e-05,
      "loss": 1.1616,
      "step": 6045
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5998599529266357,
      "learning_rate": 1.6756221448423475e-05,
      "loss": 0.8811,
      "step": 6046
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9762238264083862,
      "learning_rate": 1.6755582532025684e-05,
      "loss": 0.972,
      "step": 6047
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.58066725730896,
      "learning_rate": 1.6754943615627895e-05,
      "loss": 0.3344,
      "step": 6048
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0836873054504395,
      "learning_rate": 1.6754304699230107e-05,
      "loss": 0.9538,
      "step": 6049
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.704711675643921,
      "learning_rate": 1.6753665782832318e-05,
      "loss": 0.9189,
      "step": 6050
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7363314628601074,
      "learning_rate": 1.6753026866434526e-05,
      "loss": 0.973,
      "step": 6051
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.801379919052124,
      "learning_rate": 1.6752387950036738e-05,
      "loss": 0.9509,
      "step": 6052
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.851241111755371,
      "learning_rate": 1.675174903363895e-05,
      "loss": 0.8526,
      "step": 6053
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7187553644180298,
      "learning_rate": 1.675111011724116e-05,
      "loss": 0.9373,
      "step": 6054
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.018712043762207,
      "learning_rate": 1.675047120084337e-05,
      "loss": 0.9995,
      "step": 6055
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.04180908203125,
      "learning_rate": 1.674983228444558e-05,
      "loss": 0.8159,
      "step": 6056
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.57437801361084,
      "learning_rate": 1.6749193368047792e-05,
      "loss": 0.6786,
      "step": 6057
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.021223068237305,
      "learning_rate": 1.6748554451650004e-05,
      "loss": 1.4519,
      "step": 6058
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8754727840423584,
      "learning_rate": 1.6747915535252212e-05,
      "loss": 1.1195,
      "step": 6059
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2061665058135986,
      "learning_rate": 1.6747276618854424e-05,
      "loss": 0.947,
      "step": 6060
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.724734902381897,
      "learning_rate": 1.6746637702456635e-05,
      "loss": 0.797,
      "step": 6061
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7161157131195068,
      "learning_rate": 1.6745998786058847e-05,
      "loss": 0.8301,
      "step": 6062
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0374159812927246,
      "learning_rate": 1.6745359869661055e-05,
      "loss": 1.0041,
      "step": 6063
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6061418056488037,
      "learning_rate": 1.6744720953263266e-05,
      "loss": 0.8726,
      "step": 6064
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.987210273742676,
      "learning_rate": 1.6744082036865478e-05,
      "loss": 1.1117,
      "step": 6065
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3084999322891235,
      "learning_rate": 1.674344312046769e-05,
      "loss": 0.8076,
      "step": 6066
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8082083463668823,
      "learning_rate": 1.6742804204069898e-05,
      "loss": 0.8935,
      "step": 6067
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9267336130142212,
      "learning_rate": 1.674216528767211e-05,
      "loss": 0.923,
      "step": 6068
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1744534969329834,
      "learning_rate": 1.674152637127432e-05,
      "loss": 1.0401,
      "step": 6069
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.643428325653076,
      "learning_rate": 1.6740887454876532e-05,
      "loss": 1.0049,
      "step": 6070
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6680002212524414,
      "learning_rate": 1.674024853847874e-05,
      "loss": 1.0943,
      "step": 6071
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.805922508239746,
      "learning_rate": 1.6739609622080952e-05,
      "loss": 1.329,
      "step": 6072
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.000511646270752,
      "learning_rate": 1.6738970705683164e-05,
      "loss": 1.0636,
      "step": 6073
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.105933427810669,
      "learning_rate": 1.6738331789285375e-05,
      "loss": 0.9334,
      "step": 6074
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8158057928085327,
      "learning_rate": 1.6737692872887583e-05,
      "loss": 0.8271,
      "step": 6075
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5904877185821533,
      "learning_rate": 1.6737053956489795e-05,
      "loss": 0.8166,
      "step": 6076
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8441206216812134,
      "learning_rate": 1.6736415040092003e-05,
      "loss": 0.8438,
      "step": 6077
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.116340160369873,
      "learning_rate": 1.6735776123694218e-05,
      "loss": 0.932,
      "step": 6078
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.014698028564453,
      "learning_rate": 1.6735137207296426e-05,
      "loss": 0.7725,
      "step": 6079
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4407286643981934,
      "learning_rate": 1.6734498290898638e-05,
      "loss": 1.0422,
      "step": 6080
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.605949401855469,
      "learning_rate": 1.6733859374500846e-05,
      "loss": 1.2908,
      "step": 6081
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.689202308654785,
      "learning_rate": 1.6733220458103057e-05,
      "loss": 1.2195,
      "step": 6082
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7648180723190308,
      "learning_rate": 1.673258154170527e-05,
      "loss": 0.9673,
      "step": 6083
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0159854888916016,
      "learning_rate": 1.673194262530748e-05,
      "loss": 1.0549,
      "step": 6084
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.959564208984375,
      "learning_rate": 1.673130370890969e-05,
      "loss": 0.9072,
      "step": 6085
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.138930559158325,
      "learning_rate": 1.67306647925119e-05,
      "loss": 0.7851,
      "step": 6086
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.683973789215088,
      "learning_rate": 1.673002587611411e-05,
      "loss": 0.9641,
      "step": 6087
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8462871313095093,
      "learning_rate": 1.6729386959716323e-05,
      "loss": 0.9326,
      "step": 6088
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.183276653289795,
      "learning_rate": 1.672874804331853e-05,
      "loss": 1.0884,
      "step": 6089
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9450969696044922,
      "learning_rate": 1.6728109126920743e-05,
      "loss": 1.0269,
      "step": 6090
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9785778522491455,
      "learning_rate": 1.6727470210522954e-05,
      "loss": 1.07,
      "step": 6091
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.886328935623169,
      "learning_rate": 1.6726831294125166e-05,
      "loss": 0.8504,
      "step": 6092
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9434024095535278,
      "learning_rate": 1.6726192377727374e-05,
      "loss": 0.9698,
      "step": 6093
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8309165239334106,
      "learning_rate": 1.6725553461329586e-05,
      "loss": 0.8671,
      "step": 6094
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.4285807609558105,
      "learning_rate": 1.6724914544931797e-05,
      "loss": 1.0999,
      "step": 6095
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.101293563842773,
      "learning_rate": 1.672427562853401e-05,
      "loss": 1.577,
      "step": 6096
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8808382749557495,
      "learning_rate": 1.6723636712136217e-05,
      "loss": 0.9245,
      "step": 6097
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8277943134307861,
      "learning_rate": 1.672299779573843e-05,
      "loss": 1.0254,
      "step": 6098
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6644350290298462,
      "learning_rate": 1.672235887934064e-05,
      "loss": 0.9652,
      "step": 6099
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.578160047531128,
      "learning_rate": 1.672171996294285e-05,
      "loss": 0.9886,
      "step": 6100
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.061612367630005,
      "learning_rate": 1.672108104654506e-05,
      "loss": 0.9849,
      "step": 6101
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.104228973388672,
      "learning_rate": 1.672044213014727e-05,
      "loss": 0.9711,
      "step": 6102
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.993016242980957,
      "learning_rate": 1.6719803213749483e-05,
      "loss": 1.0953,
      "step": 6103
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6882309913635254,
      "learning_rate": 1.6719164297351694e-05,
      "loss": 0.8402,
      "step": 6104
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.35930871963501,
      "learning_rate": 1.6718525380953903e-05,
      "loss": 1.1843,
      "step": 6105
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0817208290100098,
      "learning_rate": 1.6717886464556114e-05,
      "loss": 0.8762,
      "step": 6106
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6283440589904785,
      "learning_rate": 1.6717247548158326e-05,
      "loss": 0.9639,
      "step": 6107
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.490810394287109,
      "learning_rate": 1.6716608631760537e-05,
      "loss": 1.2188,
      "step": 6108
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.387073278427124,
      "learning_rate": 1.6715969715362745e-05,
      "loss": 1.043,
      "step": 6109
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.1382927894592285,
      "learning_rate": 1.6715330798964957e-05,
      "loss": 0.8589,
      "step": 6110
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0309622287750244,
      "learning_rate": 1.6714691882567165e-05,
      "loss": 1.0401,
      "step": 6111
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0859322547912598,
      "learning_rate": 1.671405296616938e-05,
      "loss": 0.8883,
      "step": 6112
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9316905736923218,
      "learning_rate": 1.6713414049771588e-05,
      "loss": 0.9767,
      "step": 6113
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3867053985595703,
      "learning_rate": 1.67127751333738e-05,
      "loss": 1.1618,
      "step": 6114
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8421509265899658,
      "learning_rate": 1.6712136216976008e-05,
      "loss": 0.9173,
      "step": 6115
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9150564670562744,
      "learning_rate": 1.671149730057822e-05,
      "loss": 0.9468,
      "step": 6116
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7197386026382446,
      "learning_rate": 1.671085838418043e-05,
      "loss": 0.9364,
      "step": 6117
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.706268072128296,
      "learning_rate": 1.6710219467782642e-05,
      "loss": 0.857,
      "step": 6118
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9795769453048706,
      "learning_rate": 1.670958055138485e-05,
      "loss": 0.9984,
      "step": 6119
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.312452554702759,
      "learning_rate": 1.6708941634987062e-05,
      "loss": 1.0649,
      "step": 6120
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.181312322616577,
      "learning_rate": 1.6708302718589274e-05,
      "loss": 0.938,
      "step": 6121
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.948221206665039,
      "learning_rate": 1.6707663802191485e-05,
      "loss": 0.8828,
      "step": 6122
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8311244249343872,
      "learning_rate": 1.6707024885793693e-05,
      "loss": 0.8136,
      "step": 6123
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0299072265625,
      "learning_rate": 1.6706385969395905e-05,
      "loss": 0.8906,
      "step": 6124
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7801392078399658,
      "learning_rate": 1.6705747052998117e-05,
      "loss": 0.8333,
      "step": 6125
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7824760675430298,
      "learning_rate": 1.6705108136600328e-05,
      "loss": 0.8946,
      "step": 6126
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6217234134674072,
      "learning_rate": 1.6704469220202536e-05,
      "loss": 1.1234,
      "step": 6127
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6256983280181885,
      "learning_rate": 1.6703830303804748e-05,
      "loss": 0.8829,
      "step": 6128
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.326671600341797,
      "learning_rate": 1.670319138740696e-05,
      "loss": 1.1681,
      "step": 6129
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.887677788734436,
      "learning_rate": 1.670255247100917e-05,
      "loss": 0.98,
      "step": 6130
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.141375780105591,
      "learning_rate": 1.670191355461138e-05,
      "loss": 1.0178,
      "step": 6131
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.885848045349121,
      "learning_rate": 1.670127463821359e-05,
      "loss": 0.993,
      "step": 6132
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.469827175140381,
      "learning_rate": 1.6700635721815802e-05,
      "loss": 1.0696,
      "step": 6133
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0873875617980957,
      "learning_rate": 1.6699996805418014e-05,
      "loss": 1.328,
      "step": 6134
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2475850582122803,
      "learning_rate": 1.6699357889020222e-05,
      "loss": 1.0786,
      "step": 6135
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7753390073776245,
      "learning_rate": 1.6698718972622433e-05,
      "loss": 0.8606,
      "step": 6136
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9281578063964844,
      "learning_rate": 1.6698080056224645e-05,
      "loss": 1.0595,
      "step": 6137
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.70768404006958,
      "learning_rate": 1.6697441139826856e-05,
      "loss": 1.5726,
      "step": 6138
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.032597541809082,
      "learning_rate": 1.6696802223429065e-05,
      "loss": 1.1138,
      "step": 6139
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4381639957427979,
      "learning_rate": 1.6696163307031276e-05,
      "loss": 0.8168,
      "step": 6140
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1627349853515625,
      "learning_rate": 1.6695524390633488e-05,
      "loss": 1.0218,
      "step": 6141
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6520411968231201,
      "learning_rate": 1.66948854742357e-05,
      "loss": 0.9083,
      "step": 6142
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.44564151763916,
      "learning_rate": 1.6694246557837907e-05,
      "loss": 1.6396,
      "step": 6143
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0452558994293213,
      "learning_rate": 1.669360764144012e-05,
      "loss": 1.0315,
      "step": 6144
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.02850604057312,
      "learning_rate": 1.669296872504233e-05,
      "loss": 0.9305,
      "step": 6145
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7981935739517212,
      "learning_rate": 1.6692329808644542e-05,
      "loss": 0.9369,
      "step": 6146
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8712080717086792,
      "learning_rate": 1.669169089224675e-05,
      "loss": 0.9351,
      "step": 6147
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7725954055786133,
      "learning_rate": 1.6691051975848962e-05,
      "loss": 0.8552,
      "step": 6148
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7221760749816895,
      "learning_rate": 1.669041305945117e-05,
      "loss": 0.8857,
      "step": 6149
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.026561975479126,
      "learning_rate": 1.6689774143053385e-05,
      "loss": 0.8641,
      "step": 6150
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7177468538284302,
      "learning_rate": 1.6689135226655593e-05,
      "loss": 0.9803,
      "step": 6151
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3989503383636475,
      "learning_rate": 1.6688496310257805e-05,
      "loss": 1.3108,
      "step": 6152
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8951796293258667,
      "learning_rate": 1.6687857393860013e-05,
      "loss": 0.972,
      "step": 6153
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8980792760849,
      "learning_rate": 1.6687218477462224e-05,
      "loss": 0.9544,
      "step": 6154
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.758489966392517,
      "learning_rate": 1.6686579561064436e-05,
      "loss": 0.8969,
      "step": 6155
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.42976188659668,
      "learning_rate": 1.6685940644666647e-05,
      "loss": 1.0566,
      "step": 6156
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.080498695373535,
      "learning_rate": 1.6685301728268856e-05,
      "loss": 0.9021,
      "step": 6157
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.1710097789764404,
      "learning_rate": 1.6684662811871067e-05,
      "loss": 1.1322,
      "step": 6158
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9305577278137207,
      "learning_rate": 1.668402389547328e-05,
      "loss": 0.972,
      "step": 6159
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.759845018386841,
      "learning_rate": 1.668338497907549e-05,
      "loss": 1.147,
      "step": 6160
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8342124223709106,
      "learning_rate": 1.66827460626777e-05,
      "loss": 0.843,
      "step": 6161
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.496245861053467,
      "learning_rate": 1.668210714627991e-05,
      "loss": 0.9043,
      "step": 6162
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9469261169433594,
      "learning_rate": 1.668146822988212e-05,
      "loss": 0.9247,
      "step": 6163
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2852983474731445,
      "learning_rate": 1.6680829313484333e-05,
      "loss": 0.8551,
      "step": 6164
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5658676624298096,
      "learning_rate": 1.668019039708654e-05,
      "loss": 1.0939,
      "step": 6165
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.09379506111145,
      "learning_rate": 1.6679551480688753e-05,
      "loss": 1.0004,
      "step": 6166
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1990106105804443,
      "learning_rate": 1.6678912564290964e-05,
      "loss": 0.9758,
      "step": 6167
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3183822631835938,
      "learning_rate": 1.6678273647893176e-05,
      "loss": 0.9959,
      "step": 6168
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9743144512176514,
      "learning_rate": 1.6677634731495384e-05,
      "loss": 0.9299,
      "step": 6169
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5388948917388916,
      "learning_rate": 1.6676995815097595e-05,
      "loss": 0.9235,
      "step": 6170
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.029674768447876,
      "learning_rate": 1.6676356898699807e-05,
      "loss": 0.5971,
      "step": 6171
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.916032314300537,
      "learning_rate": 1.667571798230202e-05,
      "loss": 0.681,
      "step": 6172
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.0401647090911865,
      "learning_rate": 1.6675079065904227e-05,
      "loss": 1.2186,
      "step": 6173
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8787810802459717,
      "learning_rate": 1.6674440149506438e-05,
      "loss": 0.9328,
      "step": 6174
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7255512475967407,
      "learning_rate": 1.667380123310865e-05,
      "loss": 0.929,
      "step": 6175
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.443817615509033,
      "learning_rate": 1.667316231671086e-05,
      "loss": 1.1044,
      "step": 6176
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.085141897201538,
      "learning_rate": 1.667252340031307e-05,
      "loss": 0.9758,
      "step": 6177
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8497623205184937,
      "learning_rate": 1.667188448391528e-05,
      "loss": 1.0547,
      "step": 6178
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.541311264038086,
      "learning_rate": 1.6671245567517493e-05,
      "loss": 0.4058,
      "step": 6179
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.270662784576416,
      "learning_rate": 1.6670606651119704e-05,
      "loss": 1.6516,
      "step": 6180
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2111637592315674,
      "learning_rate": 1.6669967734721912e-05,
      "loss": 0.9432,
      "step": 6181
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6021558046340942,
      "learning_rate": 1.6669328818324124e-05,
      "loss": 0.8585,
      "step": 6182
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.507539987564087,
      "learning_rate": 1.6668689901926335e-05,
      "loss": 1.2289,
      "step": 6183
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.010122060775757,
      "learning_rate": 1.6668050985528547e-05,
      "loss": 0.9868,
      "step": 6184
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.8709003925323486,
      "learning_rate": 1.6667412069130755e-05,
      "loss": 1.2866,
      "step": 6185
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3268046379089355,
      "learning_rate": 1.6666773152732967e-05,
      "loss": 1.0077,
      "step": 6186
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3116085529327393,
      "learning_rate": 1.6666134236335175e-05,
      "loss": 0.9353,
      "step": 6187
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.035675048828125,
      "learning_rate": 1.666549531993739e-05,
      "loss": 0.9695,
      "step": 6188
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5781418085098267,
      "learning_rate": 1.6664856403539598e-05,
      "loss": 0.9027,
      "step": 6189
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1153981685638428,
      "learning_rate": 1.666421748714181e-05,
      "loss": 0.9678,
      "step": 6190
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6899657249450684,
      "learning_rate": 1.6663578570744018e-05,
      "loss": 0.6578,
      "step": 6191
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4293649196624756,
      "learning_rate": 1.666293965434623e-05,
      "loss": 1.0719,
      "step": 6192
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.016643762588501,
      "learning_rate": 1.666230073794844e-05,
      "loss": 1.1282,
      "step": 6193
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7265846729278564,
      "learning_rate": 1.6661661821550652e-05,
      "loss": 0.9114,
      "step": 6194
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3700058460235596,
      "learning_rate": 1.666102290515286e-05,
      "loss": 0.9049,
      "step": 6195
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.066624402999878,
      "learning_rate": 1.6660383988755072e-05,
      "loss": 0.9943,
      "step": 6196
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3156118392944336,
      "learning_rate": 1.6659745072357284e-05,
      "loss": 1.0395,
      "step": 6197
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9086917638778687,
      "learning_rate": 1.6659106155959495e-05,
      "loss": 1.0575,
      "step": 6198
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.434636116027832,
      "learning_rate": 1.6658467239561703e-05,
      "loss": 1.3424,
      "step": 6199
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7460263967514038,
      "learning_rate": 1.6657828323163915e-05,
      "loss": 0.8599,
      "step": 6200
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0215795040130615,
      "learning_rate": 1.6657189406766126e-05,
      "loss": 0.9441,
      "step": 6201
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.101458787918091,
      "learning_rate": 1.6656550490368338e-05,
      "loss": 0.8744,
      "step": 6202
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5287511348724365,
      "learning_rate": 1.6655911573970546e-05,
      "loss": 1.0808,
      "step": 6203
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.734940767288208,
      "learning_rate": 1.6655272657572758e-05,
      "loss": 0.9811,
      "step": 6204
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8993099927902222,
      "learning_rate": 1.665463374117497e-05,
      "loss": 1.0592,
      "step": 6205
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8746899366378784,
      "learning_rate": 1.665399482477718e-05,
      "loss": 0.8241,
      "step": 6206
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5325030088424683,
      "learning_rate": 1.665335590837939e-05,
      "loss": 0.4846,
      "step": 6207
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8967558145523071,
      "learning_rate": 1.66527169919816e-05,
      "loss": 0.8652,
      "step": 6208
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.722762107849121,
      "learning_rate": 1.6652078075583812e-05,
      "loss": 0.8792,
      "step": 6209
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.1107332706451416,
      "learning_rate": 1.6651439159186023e-05,
      "loss": 1.2583,
      "step": 6210
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.331730842590332,
      "learning_rate": 1.665080024278823e-05,
      "loss": 1.048,
      "step": 6211
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.7290596961975098,
      "learning_rate": 1.6650161326390443e-05,
      "loss": 0.7892,
      "step": 6212
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.0405778884887695,
      "learning_rate": 1.6649522409992655e-05,
      "loss": 0.725,
      "step": 6213
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7447305917739868,
      "learning_rate": 1.6648883493594866e-05,
      "loss": 0.903,
      "step": 6214
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5103445053100586,
      "learning_rate": 1.6648244577197074e-05,
      "loss": 0.662,
      "step": 6215
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.504610538482666,
      "learning_rate": 1.6647605660799286e-05,
      "loss": 0.4145,
      "step": 6216
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1033833026885986,
      "learning_rate": 1.6646966744401498e-05,
      "loss": 1.062,
      "step": 6217
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.090761661529541,
      "learning_rate": 1.664632782800371e-05,
      "loss": 0.9833,
      "step": 6218
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.903008222579956,
      "learning_rate": 1.6645688911605917e-05,
      "loss": 0.909,
      "step": 6219
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.241161346435547,
      "learning_rate": 1.664504999520813e-05,
      "loss": 1.1641,
      "step": 6220
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3338921070098877,
      "learning_rate": 1.6644411078810337e-05,
      "loss": 1.1713,
      "step": 6221
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.388305187225342,
      "learning_rate": 1.6643772162412552e-05,
      "loss": 0.9043,
      "step": 6222
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9166401624679565,
      "learning_rate": 1.664313324601476e-05,
      "loss": 0.905,
      "step": 6223
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9782276153564453,
      "learning_rate": 1.664249432961697e-05,
      "loss": 0.9367,
      "step": 6224
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1860594749450684,
      "learning_rate": 1.664185541321918e-05,
      "loss": 0.9398,
      "step": 6225
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6517579555511475,
      "learning_rate": 1.664121649682139e-05,
      "loss": 0.9033,
      "step": 6226
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3030970096588135,
      "learning_rate": 1.6640577580423603e-05,
      "loss": 0.9997,
      "step": 6227
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.816715121269226,
      "learning_rate": 1.6639938664025814e-05,
      "loss": 0.8205,
      "step": 6228
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.161287307739258,
      "learning_rate": 1.6639299747628023e-05,
      "loss": 0.9967,
      "step": 6229
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.8909850120544434,
      "learning_rate": 1.6638660831230234e-05,
      "loss": 0.8101,
      "step": 6230
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9522783756256104,
      "learning_rate": 1.6638021914832446e-05,
      "loss": 1.2639,
      "step": 6231
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8718771934509277,
      "learning_rate": 1.6637382998434657e-05,
      "loss": 0.9837,
      "step": 6232
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.870023250579834,
      "learning_rate": 1.6636744082036865e-05,
      "loss": 0.9326,
      "step": 6233
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3618645668029785,
      "learning_rate": 1.6636105165639077e-05,
      "loss": 1.0546,
      "step": 6234
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9307734966278076,
      "learning_rate": 1.663546624924129e-05,
      "loss": 1.1759,
      "step": 6235
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.2708911895751953,
      "learning_rate": 1.66348273328435e-05,
      "loss": 0.9363,
      "step": 6236
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.30946946144104,
      "learning_rate": 1.6634188416445708e-05,
      "loss": 1.0889,
      "step": 6237
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.471500873565674,
      "learning_rate": 1.663354950004792e-05,
      "loss": 1.0293,
      "step": 6238
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2259325981140137,
      "learning_rate": 1.663291058365013e-05,
      "loss": 1.0815,
      "step": 6239
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2941110134124756,
      "learning_rate": 1.6632271667252343e-05,
      "loss": 1.0988,
      "step": 6240
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1800665855407715,
      "learning_rate": 1.663163275085455e-05,
      "loss": 0.9593,
      "step": 6241
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0325558185577393,
      "learning_rate": 1.6630993834456762e-05,
      "loss": 0.9634,
      "step": 6242
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8577479124069214,
      "learning_rate": 1.6630354918058974e-05,
      "loss": 1.0881,
      "step": 6243
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.831324577331543,
      "learning_rate": 1.6629716001661186e-05,
      "loss": 1.2818,
      "step": 6244
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8511680364608765,
      "learning_rate": 1.6629077085263394e-05,
      "loss": 0.8809,
      "step": 6245
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.724534034729004,
      "learning_rate": 1.6628438168865605e-05,
      "loss": 1.087,
      "step": 6246
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.057819843292236,
      "learning_rate": 1.6627799252467817e-05,
      "loss": 0.9007,
      "step": 6247
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.0559234619140625,
      "learning_rate": 1.662716033607003e-05,
      "loss": 1.4375,
      "step": 6248
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8365238904953003,
      "learning_rate": 1.6626521419672237e-05,
      "loss": 0.9356,
      "step": 6249
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.82436466217041,
      "learning_rate": 1.6625882503274448e-05,
      "loss": 1.5497,
      "step": 6250
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.080291748046875,
      "learning_rate": 1.662524358687666e-05,
      "loss": 1.4431,
      "step": 6251
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.470880508422852,
      "learning_rate": 1.662460467047887e-05,
      "loss": 1.9995,
      "step": 6252
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.613790988922119,
      "learning_rate": 1.662396575408108e-05,
      "loss": 1.0107,
      "step": 6253
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.202150583267212,
      "learning_rate": 1.662332683768329e-05,
      "loss": 1.0371,
      "step": 6254
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.820694923400879,
      "learning_rate": 1.6622687921285502e-05,
      "loss": 1.4993,
      "step": 6255
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.494964361190796,
      "learning_rate": 1.6622049004887714e-05,
      "loss": 1.0865,
      "step": 6256
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7554713487625122,
      "learning_rate": 1.6621410088489922e-05,
      "loss": 1.0402,
      "step": 6257
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9025529623031616,
      "learning_rate": 1.6620771172092134e-05,
      "loss": 0.9626,
      "step": 6258
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.303219318389893,
      "learning_rate": 1.6620132255694342e-05,
      "loss": 1.2897,
      "step": 6259
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.990222454071045,
      "learning_rate": 1.6619493339296557e-05,
      "loss": 1.0144,
      "step": 6260
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0391147136688232,
      "learning_rate": 1.6618854422898765e-05,
      "loss": 0.92,
      "step": 6261
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7985502481460571,
      "learning_rate": 1.6618215506500976e-05,
      "loss": 1.017,
      "step": 6262
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9790700674057007,
      "learning_rate": 1.6617576590103185e-05,
      "loss": 0.8589,
      "step": 6263
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4769307374954224,
      "learning_rate": 1.6616937673705396e-05,
      "loss": 0.8761,
      "step": 6264
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7808048725128174,
      "learning_rate": 1.6616298757307608e-05,
      "loss": 1.0651,
      "step": 6265
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5116090774536133,
      "learning_rate": 1.661565984090982e-05,
      "loss": 0.8971,
      "step": 6266
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.510253429412842,
      "learning_rate": 1.6615020924512027e-05,
      "loss": 0.6223,
      "step": 6267
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.3357439041137695,
      "learning_rate": 1.661438200811424e-05,
      "loss": 1.6446,
      "step": 6268
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9874637126922607,
      "learning_rate": 1.661374309171645e-05,
      "loss": 1.2658,
      "step": 6269
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0061211585998535,
      "learning_rate": 1.6613104175318662e-05,
      "loss": 0.8778,
      "step": 6270
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.627542018890381,
      "learning_rate": 1.661246525892087e-05,
      "loss": 1.1909,
      "step": 6271
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.927456259727478,
      "learning_rate": 1.6611826342523082e-05,
      "loss": 0.8892,
      "step": 6272
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.859529972076416,
      "learning_rate": 1.6611187426125293e-05,
      "loss": 1.0941,
      "step": 6273
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3560116291046143,
      "learning_rate": 1.6610548509727505e-05,
      "loss": 0.9887,
      "step": 6274
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.722589373588562,
      "learning_rate": 1.6609909593329713e-05,
      "loss": 0.8977,
      "step": 6275
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.060271739959717,
      "learning_rate": 1.6609270676931925e-05,
      "loss": 1.2605,
      "step": 6276
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8437681198120117,
      "learning_rate": 1.6608631760534136e-05,
      "loss": 0.8762,
      "step": 6277
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8478437662124634,
      "learning_rate": 1.6607992844136348e-05,
      "loss": 1.0008,
      "step": 6278
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9601593017578125,
      "learning_rate": 1.6607353927738556e-05,
      "loss": 0.8746,
      "step": 6279
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7279824018478394,
      "learning_rate": 1.6606715011340767e-05,
      "loss": 0.9908,
      "step": 6280
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.066748142242432,
      "learning_rate": 1.660607609494298e-05,
      "loss": 1.8194,
      "step": 6281
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8298635482788086,
      "learning_rate": 1.660543717854519e-05,
      "loss": 0.9381,
      "step": 6282
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0169012546539307,
      "learning_rate": 1.66047982621474e-05,
      "loss": 1.008,
      "step": 6283
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1387083530426025,
      "learning_rate": 1.660415934574961e-05,
      "loss": 1.1173,
      "step": 6284
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.753182291984558,
      "learning_rate": 1.6603520429351822e-05,
      "loss": 0.9983,
      "step": 6285
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7834240198135376,
      "learning_rate": 1.6602881512954033e-05,
      "loss": 1.0102,
      "step": 6286
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.7168667316436768,
      "learning_rate": 1.660224259655624e-05,
      "loss": 1.1021,
      "step": 6287
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.130527019500732,
      "learning_rate": 1.6601603680158453e-05,
      "loss": 1.4861,
      "step": 6288
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.26042366027832,
      "learning_rate": 1.6600964763760665e-05,
      "loss": 1.7208,
      "step": 6289
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.015171051025391,
      "learning_rate": 1.6600325847362876e-05,
      "loss": 1.1014,
      "step": 6290
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.323721408843994,
      "learning_rate": 1.6599686930965084e-05,
      "loss": 0.8676,
      "step": 6291
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0513508319854736,
      "learning_rate": 1.6599048014567296e-05,
      "loss": 0.9894,
      "step": 6292
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.987874150276184,
      "learning_rate": 1.6598409098169504e-05,
      "loss": 0.9384,
      "step": 6293
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.485299587249756,
      "learning_rate": 1.659777018177172e-05,
      "loss": 1.1218,
      "step": 6294
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.803294062614441,
      "learning_rate": 1.6597131265373927e-05,
      "loss": 0.7453,
      "step": 6295
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0807840824127197,
      "learning_rate": 1.659649234897614e-05,
      "loss": 1.0464,
      "step": 6296
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.23313307762146,
      "learning_rate": 1.6595853432578347e-05,
      "loss": 0.8628,
      "step": 6297
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.8698840141296387,
      "learning_rate": 1.659521451618056e-05,
      "loss": 0.7543,
      "step": 6298
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.257991075515747,
      "learning_rate": 1.659457559978277e-05,
      "loss": 0.9148,
      "step": 6299
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.889984130859375,
      "learning_rate": 1.659393668338498e-05,
      "loss": 0.9559,
      "step": 6300
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8363863229751587,
      "learning_rate": 1.659329776698719e-05,
      "loss": 1.0275,
      "step": 6301
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.360682725906372,
      "learning_rate": 1.65926588505894e-05,
      "loss": 0.9037,
      "step": 6302
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.611570119857788,
      "learning_rate": 1.6592019934191613e-05,
      "loss": 0.8263,
      "step": 6303
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7884082794189453,
      "learning_rate": 1.6591381017793824e-05,
      "loss": 0.9459,
      "step": 6304
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7280259132385254,
      "learning_rate": 1.6590742101396032e-05,
      "loss": 0.9535,
      "step": 6305
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.146286964416504,
      "learning_rate": 1.6590103184998244e-05,
      "loss": 1.0507,
      "step": 6306
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7773138284683228,
      "learning_rate": 1.6589464268600455e-05,
      "loss": 0.9217,
      "step": 6307
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.3278417587280273,
      "learning_rate": 1.6588825352202667e-05,
      "loss": 0.9094,
      "step": 6308
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.491007328033447,
      "learning_rate": 1.6588186435804875e-05,
      "loss": 1.1428,
      "step": 6309
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0249862670898438,
      "learning_rate": 1.6587547519407087e-05,
      "loss": 1.037,
      "step": 6310
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.581948757171631,
      "learning_rate": 1.6586908603009298e-05,
      "loss": 1.0158,
      "step": 6311
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1259777545928955,
      "learning_rate": 1.658626968661151e-05,
      "loss": 0.8276,
      "step": 6312
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.057036876678467,
      "learning_rate": 1.6585630770213718e-05,
      "loss": 0.9782,
      "step": 6313
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.046267509460449,
      "learning_rate": 1.658499185381593e-05,
      "loss": 1.2224,
      "step": 6314
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.061962366104126,
      "learning_rate": 1.658435293741814e-05,
      "loss": 0.9949,
      "step": 6315
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.887074589729309,
      "learning_rate": 1.6583714021020353e-05,
      "loss": 0.9104,
      "step": 6316
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7901474237442017,
      "learning_rate": 1.658307510462256e-05,
      "loss": 0.9779,
      "step": 6317
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6940555572509766,
      "learning_rate": 1.6582436188224772e-05,
      "loss": 1.2519,
      "step": 6318
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.155797958374023,
      "learning_rate": 1.6581797271826984e-05,
      "loss": 1.8887,
      "step": 6319
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9359101057052612,
      "learning_rate": 1.6581158355429195e-05,
      "loss": 1.1112,
      "step": 6320
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0801451206207275,
      "learning_rate": 1.6580519439031404e-05,
      "loss": 0.9528,
      "step": 6321
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6614384651184082,
      "learning_rate": 1.6579880522633615e-05,
      "loss": 0.9017,
      "step": 6322
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.020162582397461,
      "learning_rate": 1.6579241606235827e-05,
      "loss": 0.9806,
      "step": 6323
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.49855637550354,
      "learning_rate": 1.6578602689838038e-05,
      "loss": 0.9212,
      "step": 6324
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.791114091873169,
      "learning_rate": 1.6577963773440246e-05,
      "loss": 1.006,
      "step": 6325
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.447408676147461,
      "learning_rate": 1.6577324857042458e-05,
      "loss": 1.2197,
      "step": 6326
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5844815969467163,
      "learning_rate": 1.657668594064467e-05,
      "loss": 0.9234,
      "step": 6327
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.325287342071533,
      "learning_rate": 1.657604702424688e-05,
      "loss": 1.3992,
      "step": 6328
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7380892038345337,
      "learning_rate": 1.657540810784909e-05,
      "loss": 0.9369,
      "step": 6329
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9298051595687866,
      "learning_rate": 1.65747691914513e-05,
      "loss": 1.0841,
      "step": 6330
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6306325197219849,
      "learning_rate": 1.657413027505351e-05,
      "loss": 0.8303,
      "step": 6331
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0115623474121094,
      "learning_rate": 1.6573491358655724e-05,
      "loss": 0.9731,
      "step": 6332
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9801676273345947,
      "learning_rate": 1.6572852442257932e-05,
      "loss": 1.2917,
      "step": 6333
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.533902406692505,
      "learning_rate": 1.6572213525860143e-05,
      "loss": 1.1182,
      "step": 6334
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3196003437042236,
      "learning_rate": 1.657157460946235e-05,
      "loss": 0.9766,
      "step": 6335
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2330307960510254,
      "learning_rate": 1.6570935693064563e-05,
      "loss": 1.0237,
      "step": 6336
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7202132940292358,
      "learning_rate": 1.6570296776666775e-05,
      "loss": 0.989,
      "step": 6337
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1590261459350586,
      "learning_rate": 1.6569657860268986e-05,
      "loss": 0.9231,
      "step": 6338
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.420508861541748,
      "learning_rate": 1.6569018943871194e-05,
      "loss": 1.4968,
      "step": 6339
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4551745653152466,
      "learning_rate": 1.6568380027473406e-05,
      "loss": 0.812,
      "step": 6340
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1701712608337402,
      "learning_rate": 1.6567741111075618e-05,
      "loss": 0.8669,
      "step": 6341
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8425697088241577,
      "learning_rate": 1.656710219467783e-05,
      "loss": 0.9567,
      "step": 6342
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1623740196228027,
      "learning_rate": 1.6566463278280037e-05,
      "loss": 0.9349,
      "step": 6343
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1482443809509277,
      "learning_rate": 1.656582436188225e-05,
      "loss": 0.9322,
      "step": 6344
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3345189094543457,
      "learning_rate": 1.656518544548446e-05,
      "loss": 1.1024,
      "step": 6345
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2759861946105957,
      "learning_rate": 1.6564546529086672e-05,
      "loss": 0.9755,
      "step": 6346
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8118153810501099,
      "learning_rate": 1.656390761268888e-05,
      "loss": 0.9835,
      "step": 6347
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9123106002807617,
      "learning_rate": 1.656326869629109e-05,
      "loss": 0.8772,
      "step": 6348
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5988438129425049,
      "learning_rate": 1.6562629779893303e-05,
      "loss": 0.856,
      "step": 6349
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.680053234100342,
      "learning_rate": 1.6561990863495515e-05,
      "loss": 1.1398,
      "step": 6350
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.172353744506836,
      "learning_rate": 1.6561351947097723e-05,
      "loss": 1.1094,
      "step": 6351
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.787919044494629,
      "learning_rate": 1.6560713030699934e-05,
      "loss": 1.3798,
      "step": 6352
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.102750062942505,
      "learning_rate": 1.6560074114302146e-05,
      "loss": 1.0088,
      "step": 6353
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.89889395236969,
      "learning_rate": 1.6559435197904357e-05,
      "loss": 0.9489,
      "step": 6354
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1261062622070312,
      "learning_rate": 1.6558796281506566e-05,
      "loss": 1.0377,
      "step": 6355
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.982999801635742,
      "learning_rate": 1.6558157365108777e-05,
      "loss": 0.9711,
      "step": 6356
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5067241191864014,
      "learning_rate": 1.655751844871099e-05,
      "loss": 0.8793,
      "step": 6357
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8558074235916138,
      "learning_rate": 1.65568795323132e-05,
      "loss": 0.9111,
      "step": 6358
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8311007022857666,
      "learning_rate": 1.655624061591541e-05,
      "loss": 0.9094,
      "step": 6359
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.73526930809021,
      "learning_rate": 1.655560169951762e-05,
      "loss": 0.8679,
      "step": 6360
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9026994705200195,
      "learning_rate": 1.655496278311983e-05,
      "loss": 0.8937,
      "step": 6361
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0213263034820557,
      "learning_rate": 1.6554323866722043e-05,
      "loss": 0.8414,
      "step": 6362
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7204570770263672,
      "learning_rate": 1.655368495032425e-05,
      "loss": 0.9873,
      "step": 6363
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.044346332550049,
      "learning_rate": 1.6553046033926463e-05,
      "loss": 1.287,
      "step": 6364
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.191057205200195,
      "learning_rate": 1.6552407117528674e-05,
      "loss": 1.4931,
      "step": 6365
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.98305344581604,
      "learning_rate": 1.6551768201130886e-05,
      "loss": 0.82,
      "step": 6366
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.028842449188232,
      "learning_rate": 1.6551129284733094e-05,
      "loss": 0.8946,
      "step": 6367
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2073352336883545,
      "learning_rate": 1.6550490368335306e-05,
      "loss": 1.1343,
      "step": 6368
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0705361366271973,
      "learning_rate": 1.6549851451937514e-05,
      "loss": 0.9461,
      "step": 6369
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3908934593200684,
      "learning_rate": 1.654921253553973e-05,
      "loss": 0.9277,
      "step": 6370
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2364180088043213,
      "learning_rate": 1.6548573619141937e-05,
      "loss": 0.8811,
      "step": 6371
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.5570454597473145,
      "learning_rate": 1.654793470274415e-05,
      "loss": 1.0329,
      "step": 6372
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3476927280426025,
      "learning_rate": 1.6547295786346357e-05,
      "loss": 1.1553,
      "step": 6373
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.86349356174469,
      "learning_rate": 1.6546656869948568e-05,
      "loss": 0.8724,
      "step": 6374
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9505363702774048,
      "learning_rate": 1.654601795355078e-05,
      "loss": 0.9845,
      "step": 6375
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0470187664031982,
      "learning_rate": 1.654537903715299e-05,
      "loss": 1.2252,
      "step": 6376
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9403645992279053,
      "learning_rate": 1.65447401207552e-05,
      "loss": 0.9668,
      "step": 6377
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.3329524993896484,
      "learning_rate": 1.654410120435741e-05,
      "loss": 1.0656,
      "step": 6378
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0745792388916016,
      "learning_rate": 1.6543462287959622e-05,
      "loss": 0.8661,
      "step": 6379
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9106634855270386,
      "learning_rate": 1.654282337156183e-05,
      "loss": 0.8876,
      "step": 6380
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.077629566192627,
      "learning_rate": 1.6542184455164042e-05,
      "loss": 0.9711,
      "step": 6381
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9215971231460571,
      "learning_rate": 1.6541545538766254e-05,
      "loss": 0.9465,
      "step": 6382
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.151127576828003,
      "learning_rate": 1.6540906622368465e-05,
      "loss": 0.9734,
      "step": 6383
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7647563219070435,
      "learning_rate": 1.6540267705970673e-05,
      "loss": 0.8724,
      "step": 6384
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9985910654067993,
      "learning_rate": 1.6539628789572885e-05,
      "loss": 0.9789,
      "step": 6385
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.5303115844726562,
      "learning_rate": 1.6538989873175096e-05,
      "loss": 1.0878,
      "step": 6386
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.136260986328125,
      "learning_rate": 1.6538350956777308e-05,
      "loss": 0.8379,
      "step": 6387
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9540287256240845,
      "learning_rate": 1.6537712040379516e-05,
      "loss": 0.9697,
      "step": 6388
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.075416326522827,
      "learning_rate": 1.6537073123981728e-05,
      "loss": 0.993,
      "step": 6389
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.437821388244629,
      "learning_rate": 1.653643420758394e-05,
      "loss": 1.0793,
      "step": 6390
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.257448196411133,
      "learning_rate": 1.653579529118615e-05,
      "loss": 1.2949,
      "step": 6391
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9683419466018677,
      "learning_rate": 1.653515637478836e-05,
      "loss": 0.9373,
      "step": 6392
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1562349796295166,
      "learning_rate": 1.653451745839057e-05,
      "loss": 0.9686,
      "step": 6393
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8265695571899414,
      "learning_rate": 1.6533878541992782e-05,
      "loss": 0.9076,
      "step": 6394
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1287412643432617,
      "learning_rate": 1.6533239625594994e-05,
      "loss": 0.9178,
      "step": 6395
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.310258865356445,
      "learning_rate": 1.6532600709197202e-05,
      "loss": 1.3293,
      "step": 6396
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5297913551330566,
      "learning_rate": 1.6531961792799413e-05,
      "loss": 0.8491,
      "step": 6397
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.989208221435547,
      "learning_rate": 1.653132287640162e-05,
      "loss": 0.9911,
      "step": 6398
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.532543182373047,
      "learning_rate": 1.6530683960003836e-05,
      "loss": 1.6577,
      "step": 6399
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2785277366638184,
      "learning_rate": 1.6530045043606045e-05,
      "loss": 1.3431,
      "step": 6400
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4348045587539673,
      "learning_rate": 1.6529406127208256e-05,
      "loss": 0.5095,
      "step": 6401
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.479346513748169,
      "learning_rate": 1.6528767210810464e-05,
      "loss": 0.9092,
      "step": 6402
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8173768520355225,
      "learning_rate": 1.6528128294412676e-05,
      "loss": 0.9105,
      "step": 6403
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.9047017097473145,
      "learning_rate": 1.6527489378014887e-05,
      "loss": 0.5781,
      "step": 6404
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9925529956817627,
      "learning_rate": 1.65268504616171e-05,
      "loss": 0.9259,
      "step": 6405
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2707085609436035,
      "learning_rate": 1.6526211545219307e-05,
      "loss": 1.0428,
      "step": 6406
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2465808391571045,
      "learning_rate": 1.652557262882152e-05,
      "loss": 0.9298,
      "step": 6407
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.724565267562866,
      "learning_rate": 1.652493371242373e-05,
      "loss": 0.9923,
      "step": 6408
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.668568730354309,
      "learning_rate": 1.6524294796025942e-05,
      "loss": 0.9053,
      "step": 6409
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2431459426879883,
      "learning_rate": 1.652365587962815e-05,
      "loss": 0.9916,
      "step": 6410
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6908353567123413,
      "learning_rate": 1.652301696323036e-05,
      "loss": 0.3819,
      "step": 6411
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.249549388885498,
      "learning_rate": 1.6522378046832573e-05,
      "loss": 1.1863,
      "step": 6412
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7501622438430786,
      "learning_rate": 1.6521739130434785e-05,
      "loss": 0.8598,
      "step": 6413
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.238259792327881,
      "learning_rate": 1.6521100214036993e-05,
      "loss": 0.9646,
      "step": 6414
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.1281700134277344,
      "learning_rate": 1.6520461297639204e-05,
      "loss": 0.9992,
      "step": 6415
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1802916526794434,
      "learning_rate": 1.6519822381241416e-05,
      "loss": 0.9921,
      "step": 6416
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.683387279510498,
      "learning_rate": 1.6519183464843627e-05,
      "loss": 0.925,
      "step": 6417
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.860331058502197,
      "learning_rate": 1.6518544548445836e-05,
      "loss": 1.5997,
      "step": 6418
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.271817684173584,
      "learning_rate": 1.6517905632048047e-05,
      "loss": 1.0043,
      "step": 6419
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.7038495540618896,
      "learning_rate": 1.651726671565026e-05,
      "loss": 0.9974,
      "step": 6420
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6907647848129272,
      "learning_rate": 1.651662779925247e-05,
      "loss": 0.914,
      "step": 6421
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.954350233078003,
      "learning_rate": 1.651598888285468e-05,
      "loss": 1.3118,
      "step": 6422
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4044156074523926,
      "learning_rate": 1.651534996645689e-05,
      "loss": 1.0117,
      "step": 6423
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7444828748703003,
      "learning_rate": 1.65147110500591e-05,
      "loss": 0.8293,
      "step": 6424
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.172377586364746,
      "learning_rate": 1.6514072133661313e-05,
      "loss": 1.1531,
      "step": 6425
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8833887577056885,
      "learning_rate": 1.651343321726352e-05,
      "loss": 0.9243,
      "step": 6426
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7432453632354736,
      "learning_rate": 1.6512794300865733e-05,
      "loss": 1.0215,
      "step": 6427
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.7734365463256836,
      "learning_rate": 1.6512155384467944e-05,
      "loss": 0.9092,
      "step": 6428
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6123298406600952,
      "learning_rate": 1.6511516468070156e-05,
      "loss": 0.8976,
      "step": 6429
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.705250263214111,
      "learning_rate": 1.6510877551672364e-05,
      "loss": 1.6316,
      "step": 6430
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.895049810409546,
      "learning_rate": 1.6510238635274575e-05,
      "loss": 1.2125,
      "step": 6431
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9354231357574463,
      "learning_rate": 1.6509599718876787e-05,
      "loss": 0.9488,
      "step": 6432
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0535454750061035,
      "learning_rate": 1.6508960802479e-05,
      "loss": 0.9021,
      "step": 6433
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2865102291107178,
      "learning_rate": 1.6508321886081207e-05,
      "loss": 1.0537,
      "step": 6434
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8743737936019897,
      "learning_rate": 1.6507682969683418e-05,
      "loss": 0.9899,
      "step": 6435
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7680660486221313,
      "learning_rate": 1.6507044053285626e-05,
      "loss": 0.9823,
      "step": 6436
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.342749118804932,
      "learning_rate": 1.650640513688784e-05,
      "loss": 1.6077,
      "step": 6437
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7327721118927002,
      "learning_rate": 1.650576622049005e-05,
      "loss": 0.8742,
      "step": 6438
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5097174644470215,
      "learning_rate": 1.650512730409226e-05,
      "loss": 1.0095,
      "step": 6439
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.512202024459839,
      "learning_rate": 1.650448838769447e-05,
      "loss": 1.0283,
      "step": 6440
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.998976230621338,
      "learning_rate": 1.650384947129668e-05,
      "loss": 1.0179,
      "step": 6441
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.255617141723633,
      "learning_rate": 1.6503210554898892e-05,
      "loss": 1.6155,
      "step": 6442
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9264036417007446,
      "learning_rate": 1.6502571638501104e-05,
      "loss": 0.897,
      "step": 6443
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0901267528533936,
      "learning_rate": 1.6501932722103312e-05,
      "loss": 0.8673,
      "step": 6444
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7343692779541016,
      "learning_rate": 1.6501293805705524e-05,
      "loss": 0.8804,
      "step": 6445
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.180267333984375,
      "learning_rate": 1.6500654889307735e-05,
      "loss": 1.4628,
      "step": 6446
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.851306438446045,
      "learning_rate": 1.6500015972909947e-05,
      "loss": 0.8526,
      "step": 6447
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6195205450057983,
      "learning_rate": 1.6499377056512155e-05,
      "loss": 0.871,
      "step": 6448
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.07822847366333,
      "learning_rate": 1.6498738140114366e-05,
      "loss": 0.8155,
      "step": 6449
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2386715412139893,
      "learning_rate": 1.6498099223716578e-05,
      "loss": 0.9842,
      "step": 6450
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.984500765800476,
      "learning_rate": 1.649746030731879e-05,
      "loss": 0.8966,
      "step": 6451
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.996232271194458,
      "learning_rate": 1.6496821390920998e-05,
      "loss": 1.0073,
      "step": 6452
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.202777624130249,
      "learning_rate": 1.649618247452321e-05,
      "loss": 1.0123,
      "step": 6453
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7307426929473877,
      "learning_rate": 1.649554355812542e-05,
      "loss": 0.3871,
      "step": 6454
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7381757497787476,
      "learning_rate": 1.6494904641727632e-05,
      "loss": 0.9546,
      "step": 6455
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9956988096237183,
      "learning_rate": 1.649426572532984e-05,
      "loss": 0.9936,
      "step": 6456
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4151010513305664,
      "learning_rate": 1.6493626808932052e-05,
      "loss": 0.9397,
      "step": 6457
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.206422328948975,
      "learning_rate": 1.6492987892534264e-05,
      "loss": 1.2421,
      "step": 6458
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.158803939819336,
      "learning_rate": 1.6492348976136475e-05,
      "loss": 1.7258,
      "step": 6459
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.2506070137023926,
      "learning_rate": 1.6491710059738683e-05,
      "loss": 1.3042,
      "step": 6460
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.106783866882324,
      "learning_rate": 1.6491071143340895e-05,
      "loss": 0.9536,
      "step": 6461
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.193230628967285,
      "learning_rate": 1.6490432226943106e-05,
      "loss": 1.007,
      "step": 6462
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.58000111579895,
      "learning_rate": 1.6489793310545318e-05,
      "loss": 0.9646,
      "step": 6463
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.8784327507019043,
      "learning_rate": 1.6489154394147526e-05,
      "loss": 1.0258,
      "step": 6464
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.852804660797119,
      "learning_rate": 1.6488515477749738e-05,
      "loss": 1.5008,
      "step": 6465
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.664863586425781,
      "learning_rate": 1.648787656135195e-05,
      "loss": 1.195,
      "step": 6466
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.222330331802368,
      "learning_rate": 1.648723764495416e-05,
      "loss": 1.0009,
      "step": 6467
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.169482707977295,
      "learning_rate": 1.648659872855637e-05,
      "loss": 0.881,
      "step": 6468
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.124394655227661,
      "learning_rate": 1.648595981215858e-05,
      "loss": 1.0259,
      "step": 6469
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8101783990859985,
      "learning_rate": 1.648532089576079e-05,
      "loss": 0.942,
      "step": 6470
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.813167691230774,
      "learning_rate": 1.6484681979363003e-05,
      "loss": 1.041,
      "step": 6471
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.339627742767334,
      "learning_rate": 1.648404306296521e-05,
      "loss": 1.106,
      "step": 6472
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.046701431274414,
      "learning_rate": 1.6483404146567423e-05,
      "loss": 0.8818,
      "step": 6473
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.967745065689087,
      "learning_rate": 1.648276523016963e-05,
      "loss": 1.0262,
      "step": 6474
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.285746097564697,
      "learning_rate": 1.6482126313771846e-05,
      "loss": 1.4031,
      "step": 6475
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.623408317565918,
      "learning_rate": 1.6481487397374054e-05,
      "loss": 1.2053,
      "step": 6476
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7394330501556396,
      "learning_rate": 1.6480848480976266e-05,
      "loss": 0.9464,
      "step": 6477
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7937527894973755,
      "learning_rate": 1.6480209564578474e-05,
      "loss": 0.865,
      "step": 6478
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.183732748031616,
      "learning_rate": 1.6479570648180686e-05,
      "loss": 0.9664,
      "step": 6479
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.948476791381836,
      "learning_rate": 1.6478931731782897e-05,
      "loss": 0.8124,
      "step": 6480
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.143065929412842,
      "learning_rate": 1.647829281538511e-05,
      "loss": 0.9149,
      "step": 6481
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7120568752288818,
      "learning_rate": 1.6477653898987317e-05,
      "loss": 0.8942,
      "step": 6482
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.824378252029419,
      "learning_rate": 1.647701498258953e-05,
      "loss": 0.9808,
      "step": 6483
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.251465320587158,
      "learning_rate": 1.647637606619174e-05,
      "loss": 1.1504,
      "step": 6484
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.7519443035125732,
      "learning_rate": 1.647573714979395e-05,
      "loss": 1.0735,
      "step": 6485
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7630877494812012,
      "learning_rate": 1.647509823339616e-05,
      "loss": 0.8413,
      "step": 6486
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5023369789123535,
      "learning_rate": 1.647445931699837e-05,
      "loss": 0.9455,
      "step": 6487
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4459917545318604,
      "learning_rate": 1.6473820400600583e-05,
      "loss": 1.0614,
      "step": 6488
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.3834049701690674,
      "learning_rate": 1.6473181484202794e-05,
      "loss": 1.6704,
      "step": 6489
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.467520713806152,
      "learning_rate": 1.6472542567805003e-05,
      "loss": 1.4588,
      "step": 6490
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.699416399002075,
      "learning_rate": 1.6471903651407214e-05,
      "loss": 0.846,
      "step": 6491
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9100842475891113,
      "learning_rate": 1.6471264735009426e-05,
      "loss": 1.0272,
      "step": 6492
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8524234294891357,
      "learning_rate": 1.6470625818611637e-05,
      "loss": 1.0463,
      "step": 6493
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9839383363723755,
      "learning_rate": 1.6469986902213845e-05,
      "loss": 0.9609,
      "step": 6494
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3910727500915527,
      "learning_rate": 1.6469347985816057e-05,
      "loss": 0.9329,
      "step": 6495
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3002266883850098,
      "learning_rate": 1.646870906941827e-05,
      "loss": 1.1226,
      "step": 6496
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.415066719055176,
      "learning_rate": 1.646807015302048e-05,
      "loss": 0.8195,
      "step": 6497
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.867490530014038,
      "learning_rate": 1.6467431236622688e-05,
      "loss": 0.8119,
      "step": 6498
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1421287059783936,
      "learning_rate": 1.64667923202249e-05,
      "loss": 0.9826,
      "step": 6499
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.880234718322754,
      "learning_rate": 1.646615340382711e-05,
      "loss": 1.7601,
      "step": 6500
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.499041199684143,
      "learning_rate": 1.6465514487429323e-05,
      "loss": 0.4492,
      "step": 6501
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.953055739402771,
      "learning_rate": 1.646487557103153e-05,
      "loss": 0.8831,
      "step": 6502
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9394738674163818,
      "learning_rate": 1.6464236654633742e-05,
      "loss": 0.8326,
      "step": 6503
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4760093688964844,
      "learning_rate": 1.6463597738235954e-05,
      "loss": 1.2351,
      "step": 6504
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.042555809020996,
      "learning_rate": 1.6462958821838166e-05,
      "loss": 1.0629,
      "step": 6505
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.891001582145691,
      "learning_rate": 1.6462319905440374e-05,
      "loss": 0.797,
      "step": 6506
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6516143083572388,
      "learning_rate": 1.6461680989042585e-05,
      "loss": 0.4685,
      "step": 6507
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5313092470169067,
      "learning_rate": 1.6461042072644793e-05,
      "loss": 0.8606,
      "step": 6508
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.483985424041748,
      "learning_rate": 1.646040315624701e-05,
      "loss": 1.1246,
      "step": 6509
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.303320407867432,
      "learning_rate": 1.6459764239849217e-05,
      "loss": 1.3364,
      "step": 6510
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7848154306411743,
      "learning_rate": 1.6459125323451428e-05,
      "loss": 0.9462,
      "step": 6511
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5886642932891846,
      "learning_rate": 1.6458486407053636e-05,
      "loss": 0.9922,
      "step": 6512
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.59148108959198,
      "learning_rate": 1.6457847490655848e-05,
      "loss": 0.3599,
      "step": 6513
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.276716709136963,
      "learning_rate": 1.645720857425806e-05,
      "loss": 1.0749,
      "step": 6514
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.035231113433838,
      "learning_rate": 1.645656965786027e-05,
      "loss": 1.1714,
      "step": 6515
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.041491985321045,
      "learning_rate": 1.645593074146248e-05,
      "loss": 1.0248,
      "step": 6516
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2105655670166016,
      "learning_rate": 1.645529182506469e-05,
      "loss": 1.0025,
      "step": 6517
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.15415096282959,
      "learning_rate": 1.6454652908666902e-05,
      "loss": 1.0618,
      "step": 6518
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.78826379776001,
      "learning_rate": 1.6454013992269114e-05,
      "loss": 0.7669,
      "step": 6519
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8904441595077515,
      "learning_rate": 1.6453375075871322e-05,
      "loss": 1.2805,
      "step": 6520
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5279757976531982,
      "learning_rate": 1.6452736159473533e-05,
      "loss": 1.1313,
      "step": 6521
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6647977828979492,
      "learning_rate": 1.6452097243075745e-05,
      "loss": 0.7696,
      "step": 6522
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.874078631401062,
      "learning_rate": 1.6451458326677956e-05,
      "loss": 0.8269,
      "step": 6523
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.577022075653076,
      "learning_rate": 1.6450819410280165e-05,
      "loss": 0.8609,
      "step": 6524
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6984565258026123,
      "learning_rate": 1.6450180493882376e-05,
      "loss": 0.951,
      "step": 6525
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6469802856445312,
      "learning_rate": 1.6449541577484588e-05,
      "loss": 1.1254,
      "step": 6526
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.419586181640625,
      "learning_rate": 1.64489026610868e-05,
      "loss": 1.2217,
      "step": 6527
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.049027681350708,
      "learning_rate": 1.6448263744689007e-05,
      "loss": 0.968,
      "step": 6528
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.924316644668579,
      "learning_rate": 1.644762482829122e-05,
      "loss": 1.032,
      "step": 6529
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.073904037475586,
      "learning_rate": 1.644698591189343e-05,
      "loss": 0.9639,
      "step": 6530
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7491836547851562,
      "learning_rate": 1.6446346995495642e-05,
      "loss": 1.0299,
      "step": 6531
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.802531361579895,
      "learning_rate": 1.644570807909785e-05,
      "loss": 1.001,
      "step": 6532
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.998535394668579,
      "learning_rate": 1.6445069162700062e-05,
      "loss": 1.2598,
      "step": 6533
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7444416284561157,
      "learning_rate": 1.6444430246302273e-05,
      "loss": 0.9465,
      "step": 6534
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7256970405578613,
      "learning_rate": 1.6443791329904485e-05,
      "loss": 0.8621,
      "step": 6535
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.978779673576355,
      "learning_rate": 1.6443152413506693e-05,
      "loss": 0.7884,
      "step": 6536
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4033281803131104,
      "learning_rate": 1.6442513497108905e-05,
      "loss": 0.9413,
      "step": 6537
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1464133262634277,
      "learning_rate": 1.6441874580711116e-05,
      "loss": 0.9039,
      "step": 6538
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1347415447235107,
      "learning_rate": 1.6441235664313328e-05,
      "loss": 0.9697,
      "step": 6539
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5751421451568604,
      "learning_rate": 1.6440596747915536e-05,
      "loss": 1.1909,
      "step": 6540
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.267876148223877,
      "learning_rate": 1.6439957831517747e-05,
      "loss": 0.9182,
      "step": 6541
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2760863304138184,
      "learning_rate": 1.643931891511996e-05,
      "loss": 1.1491,
      "step": 6542
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2710907459259033,
      "learning_rate": 1.643867999872217e-05,
      "loss": 1.0171,
      "step": 6543
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6190850734710693,
      "learning_rate": 1.643804108232438e-05,
      "loss": 0.8814,
      "step": 6544
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.239229679107666,
      "learning_rate": 1.643740216592659e-05,
      "loss": 1.1736,
      "step": 6545
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6320688724517822,
      "learning_rate": 1.64367632495288e-05,
      "loss": 0.9599,
      "step": 6546
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.401966094970703,
      "learning_rate": 1.6436124333131013e-05,
      "loss": 0.9939,
      "step": 6547
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8283761739730835,
      "learning_rate": 1.643548541673322e-05,
      "loss": 1.1132,
      "step": 6548
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7002756595611572,
      "learning_rate": 1.6434846500335433e-05,
      "loss": 0.8696,
      "step": 6549
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.754116415977478,
      "learning_rate": 1.643420758393764e-05,
      "loss": 0.9613,
      "step": 6550
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6488784551620483,
      "learning_rate": 1.6433568667539853e-05,
      "loss": 0.4007,
      "step": 6551
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1586337089538574,
      "learning_rate": 1.6432929751142064e-05,
      "loss": 1.1443,
      "step": 6552
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.48182487487793,
      "learning_rate": 1.6432290834744276e-05,
      "loss": 1.3797,
      "step": 6553
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9892858266830444,
      "learning_rate": 1.6431651918346484e-05,
      "loss": 0.956,
      "step": 6554
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3469133377075195,
      "learning_rate": 1.6431013001948695e-05,
      "loss": 0.9098,
      "step": 6555
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.050544500350952,
      "learning_rate": 1.6430374085550907e-05,
      "loss": 0.9444,
      "step": 6556
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9253751039505005,
      "learning_rate": 1.642973516915312e-05,
      "loss": 1.1013,
      "step": 6557
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.269583225250244,
      "learning_rate": 1.6429096252755327e-05,
      "loss": 1.0923,
      "step": 6558
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6867061853408813,
      "learning_rate": 1.6428457336357538e-05,
      "loss": 0.8579,
      "step": 6559
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2106025218963623,
      "learning_rate": 1.642781841995975e-05,
      "loss": 0.9072,
      "step": 6560
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1804146766662598,
      "learning_rate": 1.642717950356196e-05,
      "loss": 0.9013,
      "step": 6561
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6190896034240723,
      "learning_rate": 1.642654058716417e-05,
      "loss": 0.8129,
      "step": 6562
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8504712581634521,
      "learning_rate": 1.642590167076638e-05,
      "loss": 0.9438,
      "step": 6563
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2951653003692627,
      "learning_rate": 1.6425262754368593e-05,
      "loss": 1.244,
      "step": 6564
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.382337808609009,
      "learning_rate": 1.6424623837970804e-05,
      "loss": 0.8325,
      "step": 6565
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.7287890911102295,
      "learning_rate": 1.6423984921573012e-05,
      "loss": 1.1732,
      "step": 6566
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.317453622817993,
      "learning_rate": 1.6423346005175224e-05,
      "loss": 0.9641,
      "step": 6567
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.308664560317993,
      "learning_rate": 1.6422707088777435e-05,
      "loss": 1.2029,
      "step": 6568
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6338706016540527,
      "learning_rate": 1.6422068172379647e-05,
      "loss": 0.5346,
      "step": 6569
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.284845352172852,
      "learning_rate": 1.6421429255981855e-05,
      "loss": 0.808,
      "step": 6570
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6127313375473022,
      "learning_rate": 1.6420790339584067e-05,
      "loss": 0.4346,
      "step": 6571
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.326788902282715,
      "learning_rate": 1.6420151423186278e-05,
      "loss": 1.1387,
      "step": 6572
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.1402435302734375,
      "learning_rate": 1.641951250678849e-05,
      "loss": 1.5442,
      "step": 6573
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.945221424102783,
      "learning_rate": 1.6418873590390698e-05,
      "loss": 1.3357,
      "step": 6574
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.025693893432617,
      "learning_rate": 1.641823467399291e-05,
      "loss": 1.1969,
      "step": 6575
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.118985652923584,
      "learning_rate": 1.641759575759512e-05,
      "loss": 1.0929,
      "step": 6576
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.0796000957489014,
      "learning_rate": 1.6416956841197333e-05,
      "loss": 0.8819,
      "step": 6577
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7169537544250488,
      "learning_rate": 1.641631792479954e-05,
      "loss": 0.9059,
      "step": 6578
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.100555181503296,
      "learning_rate": 1.6415679008401752e-05,
      "loss": 1.0622,
      "step": 6579
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2153749465942383,
      "learning_rate": 1.641504009200396e-05,
      "loss": 0.9422,
      "step": 6580
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0735769271850586,
      "learning_rate": 1.6414401175606175e-05,
      "loss": 0.9499,
      "step": 6581
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6807183027267456,
      "learning_rate": 1.6413762259208384e-05,
      "loss": 0.9462,
      "step": 6582
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.423185110092163,
      "learning_rate": 1.6413123342810595e-05,
      "loss": 0.7978,
      "step": 6583
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.857511043548584,
      "learning_rate": 1.6412484426412803e-05,
      "loss": 0.9694,
      "step": 6584
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2317469120025635,
      "learning_rate": 1.6411845510015015e-05,
      "loss": 1.1086,
      "step": 6585
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.483100652694702,
      "learning_rate": 1.6411206593617226e-05,
      "loss": 0.7395,
      "step": 6586
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7816115617752075,
      "learning_rate": 1.6410567677219438e-05,
      "loss": 0.9274,
      "step": 6587
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.674256443977356,
      "learning_rate": 1.6409928760821646e-05,
      "loss": 0.3876,
      "step": 6588
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7139254808425903,
      "learning_rate": 1.6409289844423858e-05,
      "loss": 1.047,
      "step": 6589
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6386144161224365,
      "learning_rate": 1.640865092802607e-05,
      "loss": 1.0509,
      "step": 6590
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6588361263275146,
      "learning_rate": 1.640801201162828e-05,
      "loss": 0.8642,
      "step": 6591
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5100879669189453,
      "learning_rate": 1.640737309523049e-05,
      "loss": 0.8715,
      "step": 6592
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8406387567520142,
      "learning_rate": 1.64067341788327e-05,
      "loss": 0.9456,
      "step": 6593
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.646665334701538,
      "learning_rate": 1.6406095262434912e-05,
      "loss": 0.8713,
      "step": 6594
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.978297710418701,
      "learning_rate": 1.6405456346037123e-05,
      "loss": 1.1954,
      "step": 6595
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.935407280921936,
      "learning_rate": 1.640481742963933e-05,
      "loss": 0.9409,
      "step": 6596
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4915382862091064,
      "learning_rate": 1.6404178513241543e-05,
      "loss": 0.9014,
      "step": 6597
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.053370237350464,
      "learning_rate": 1.6403539596843755e-05,
      "loss": 0.9888,
      "step": 6598
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.933061122894287,
      "learning_rate": 1.6402900680445966e-05,
      "loss": 0.9121,
      "step": 6599
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.723228931427002,
      "learning_rate": 1.6402261764048174e-05,
      "loss": 1.2717,
      "step": 6600
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7380043268203735,
      "learning_rate": 1.6401622847650386e-05,
      "loss": 0.8791,
      "step": 6601
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9153997898101807,
      "learning_rate": 1.6400983931252598e-05,
      "loss": 1.1321,
      "step": 6602
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.5839085578918457,
      "learning_rate": 1.640034501485481e-05,
      "loss": 1.4617,
      "step": 6603
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5823724269866943,
      "learning_rate": 1.6399706098457017e-05,
      "loss": 1.1056,
      "step": 6604
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9027975797653198,
      "learning_rate": 1.639906718205923e-05,
      "loss": 0.8436,
      "step": 6605
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.21882963180542,
      "learning_rate": 1.639842826566144e-05,
      "loss": 1.0813,
      "step": 6606
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.7298033237457275,
      "learning_rate": 1.6397789349263652e-05,
      "loss": 0.523,
      "step": 6607
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8142513036727905,
      "learning_rate": 1.639715043286586e-05,
      "loss": 0.4143,
      "step": 6608
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8650388717651367,
      "learning_rate": 1.639651151646807e-05,
      "loss": 0.648,
      "step": 6609
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8368107080459595,
      "learning_rate": 1.6395872600070283e-05,
      "loss": 0.891,
      "step": 6610
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5628814697265625,
      "learning_rate": 1.6395233683672495e-05,
      "loss": 0.8626,
      "step": 6611
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6693986654281616,
      "learning_rate": 1.6394594767274703e-05,
      "loss": 0.9527,
      "step": 6612
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.097959041595459,
      "learning_rate": 1.6393955850876914e-05,
      "loss": 0.9487,
      "step": 6613
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6450817584991455,
      "learning_rate": 1.6393316934479126e-05,
      "loss": 0.9893,
      "step": 6614
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.733609914779663,
      "learning_rate": 1.6392678018081337e-05,
      "loss": 1.0074,
      "step": 6615
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0804781913757324,
      "learning_rate": 1.6392039101683546e-05,
      "loss": 0.9592,
      "step": 6616
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7364033460617065,
      "learning_rate": 1.6391400185285757e-05,
      "loss": 0.9624,
      "step": 6617
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.620857000350952,
      "learning_rate": 1.6390761268887965e-05,
      "loss": 0.934,
      "step": 6618
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8493613004684448,
      "learning_rate": 1.639012235249018e-05,
      "loss": 0.8919,
      "step": 6619
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.904891014099121,
      "learning_rate": 1.638948343609239e-05,
      "loss": 1.002,
      "step": 6620
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.0389564037323,
      "learning_rate": 1.63888445196946e-05,
      "loss": 0.7342,
      "step": 6621
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.921622633934021,
      "learning_rate": 1.6388205603296808e-05,
      "loss": 0.971,
      "step": 6622
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0807814598083496,
      "learning_rate": 1.638756668689902e-05,
      "loss": 1.1368,
      "step": 6623
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7164463996887207,
      "learning_rate": 1.638692777050123e-05,
      "loss": 0.3753,
      "step": 6624
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.074495792388916,
      "learning_rate": 1.6386288854103443e-05,
      "loss": 1.3273,
      "step": 6625
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9715003967285156,
      "learning_rate": 1.638564993770565e-05,
      "loss": 0.9366,
      "step": 6626
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.869826316833496,
      "learning_rate": 1.6385011021307862e-05,
      "loss": 0.4046,
      "step": 6627
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.369743585586548,
      "learning_rate": 1.6384372104910074e-05,
      "loss": 0.9718,
      "step": 6628
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7350592613220215,
      "learning_rate": 1.6383733188512286e-05,
      "loss": 0.8851,
      "step": 6629
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.1161179542541504,
      "learning_rate": 1.6383094272114494e-05,
      "loss": 1.1807,
      "step": 6630
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7877018451690674,
      "learning_rate": 1.6382455355716705e-05,
      "loss": 1.0235,
      "step": 6631
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.9161839485168457,
      "learning_rate": 1.6381816439318917e-05,
      "loss": 1.1618,
      "step": 6632
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0437424182891846,
      "learning_rate": 1.638117752292113e-05,
      "loss": 0.9158,
      "step": 6633
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3714170455932617,
      "learning_rate": 1.6380538606523337e-05,
      "loss": 0.8442,
      "step": 6634
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.548703670501709,
      "learning_rate": 1.6379899690125548e-05,
      "loss": 0.9557,
      "step": 6635
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7387151718139648,
      "learning_rate": 1.637926077372776e-05,
      "loss": 0.927,
      "step": 6636
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.0540645122528076,
      "learning_rate": 1.637862185732997e-05,
      "loss": 1.039,
      "step": 6637
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.316680908203125,
      "learning_rate": 1.637798294093218e-05,
      "loss": 1.3955,
      "step": 6638
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8437554836273193,
      "learning_rate": 1.637734402453439e-05,
      "loss": 0.8493,
      "step": 6639
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.574190616607666,
      "learning_rate": 1.6376705108136602e-05,
      "loss": 0.8672,
      "step": 6640
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.4062998294830322,
      "learning_rate": 1.6376066191738814e-05,
      "loss": 0.9779,
      "step": 6641
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.898200511932373,
      "learning_rate": 1.6375427275341022e-05,
      "loss": 0.9648,
      "step": 6642
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7807724475860596,
      "learning_rate": 1.6374788358943234e-05,
      "loss": 0.9548,
      "step": 6643
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.017261028289795,
      "learning_rate": 1.6374149442545445e-05,
      "loss": 1.3147,
      "step": 6644
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.801287055015564,
      "learning_rate": 1.6373510526147657e-05,
      "loss": 0.898,
      "step": 6645
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6205270290374756,
      "learning_rate": 1.6372871609749865e-05,
      "loss": 0.3919,
      "step": 6646
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8268511295318604,
      "learning_rate": 1.6372232693352076e-05,
      "loss": 0.9188,
      "step": 6647
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.9430766105651855,
      "learning_rate": 1.6371593776954288e-05,
      "loss": 1.263,
      "step": 6648
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.97238302230835,
      "learning_rate": 1.63709548605565e-05,
      "loss": 1.7462,
      "step": 6649
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.883544921875,
      "learning_rate": 1.6370315944158708e-05,
      "loss": 0.9163,
      "step": 6650
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0152978897094727,
      "learning_rate": 1.636967702776092e-05,
      "loss": 0.9568,
      "step": 6651
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.581229329109192,
      "learning_rate": 1.636903811136313e-05,
      "loss": 0.8955,
      "step": 6652
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.584151029586792,
      "learning_rate": 1.6368399194965342e-05,
      "loss": 1.3822,
      "step": 6653
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8924566507339478,
      "learning_rate": 1.636776027856755e-05,
      "loss": 0.8939,
      "step": 6654
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8611003160476685,
      "learning_rate": 1.6367121362169762e-05,
      "loss": 1.2363,
      "step": 6655
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8965691328048706,
      "learning_rate": 1.636648244577197e-05,
      "loss": 1.133,
      "step": 6656
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0304017066955566,
      "learning_rate": 1.6365843529374185e-05,
      "loss": 0.9792,
      "step": 6657
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9210606813430786,
      "learning_rate": 1.6365204612976393e-05,
      "loss": 0.8918,
      "step": 6658
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7688939571380615,
      "learning_rate": 1.6364565696578605e-05,
      "loss": 0.5218,
      "step": 6659
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.326582431793213,
      "learning_rate": 1.6363926780180813e-05,
      "loss": 0.7064,
      "step": 6660
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.02017879486084,
      "learning_rate": 1.6363287863783025e-05,
      "loss": 1.8275,
      "step": 6661
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3482792377471924,
      "learning_rate": 1.6362648947385236e-05,
      "loss": 1.0596,
      "step": 6662
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5787698030471802,
      "learning_rate": 1.6362010030987448e-05,
      "loss": 0.8402,
      "step": 6663
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.281104564666748,
      "learning_rate": 1.6361371114589656e-05,
      "loss": 1.1201,
      "step": 6664
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.327444553375244,
      "learning_rate": 1.6360732198191867e-05,
      "loss": 0.9443,
      "step": 6665
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7690982818603516,
      "learning_rate": 1.636009328179408e-05,
      "loss": 1.0835,
      "step": 6666
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8602403402328491,
      "learning_rate": 1.635945436539629e-05,
      "loss": 0.9308,
      "step": 6667
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.887752652168274,
      "learning_rate": 1.63588154489985e-05,
      "loss": 1.0427,
      "step": 6668
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8284416198730469,
      "learning_rate": 1.635817653260071e-05,
      "loss": 0.9473,
      "step": 6669
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1094775199890137,
      "learning_rate": 1.6357537616202922e-05,
      "loss": 0.9025,
      "step": 6670
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.823305130004883,
      "learning_rate": 1.6356898699805133e-05,
      "loss": 1.0999,
      "step": 6671
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.551185131072998,
      "learning_rate": 1.635625978340734e-05,
      "loss": 1.5889,
      "step": 6672
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.877203106880188,
      "learning_rate": 1.6355620867009553e-05,
      "loss": 0.9659,
      "step": 6673
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.8429741859436035,
      "learning_rate": 1.6354981950611765e-05,
      "loss": 0.8981,
      "step": 6674
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8173426389694214,
      "learning_rate": 1.6354343034213976e-05,
      "loss": 0.8162,
      "step": 6675
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8323731422424316,
      "learning_rate": 1.6353704117816184e-05,
      "loss": 1.0106,
      "step": 6676
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8101181983947754,
      "learning_rate": 1.6353065201418396e-05,
      "loss": 0.9256,
      "step": 6677
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.105541229248047,
      "learning_rate": 1.6352426285020607e-05,
      "loss": 1.0524,
      "step": 6678
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0351312160491943,
      "learning_rate": 1.635178736862282e-05,
      "loss": 0.8569,
      "step": 6679
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.504009485244751,
      "learning_rate": 1.6351148452225027e-05,
      "loss": 0.6761,
      "step": 6680
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5317678451538086,
      "learning_rate": 1.635050953582724e-05,
      "loss": 0.9796,
      "step": 6681
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.225803375244141,
      "learning_rate": 1.634987061942945e-05,
      "loss": 1.3342,
      "step": 6682
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.2738037109375,
      "learning_rate": 1.634923170303166e-05,
      "loss": 1.3762,
      "step": 6683
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.770728826522827,
      "learning_rate": 1.634859278663387e-05,
      "loss": 1.0634,
      "step": 6684
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4152798652648926,
      "learning_rate": 1.634795387023608e-05,
      "loss": 0.4609,
      "step": 6685
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9638164043426514,
      "learning_rate": 1.6347314953838293e-05,
      "loss": 0.9485,
      "step": 6686
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.618670105934143,
      "learning_rate": 1.6346676037440504e-05,
      "loss": 0.9661,
      "step": 6687
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.9765617847442627,
      "learning_rate": 1.6346037121042713e-05,
      "loss": 0.6703,
      "step": 6688
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8275266885757446,
      "learning_rate": 1.6345398204644924e-05,
      "loss": 0.8348,
      "step": 6689
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8341894149780273,
      "learning_rate": 1.6344759288247132e-05,
      "loss": 1.0055,
      "step": 6690
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.598222494125366,
      "learning_rate": 1.6344120371849347e-05,
      "loss": 1.2912,
      "step": 6691
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0576064586639404,
      "learning_rate": 1.6343481455451555e-05,
      "loss": 0.9656,
      "step": 6692
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.6354312896728516,
      "learning_rate": 1.6342842539053767e-05,
      "loss": 0.9852,
      "step": 6693
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7695263624191284,
      "learning_rate": 1.6342203622655975e-05,
      "loss": 0.9177,
      "step": 6694
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1357648372650146,
      "learning_rate": 1.6341564706258187e-05,
      "loss": 0.9181,
      "step": 6695
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.586383819580078,
      "learning_rate": 1.6340925789860398e-05,
      "loss": 1.4193,
      "step": 6696
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7656590938568115,
      "learning_rate": 1.634028687346261e-05,
      "loss": 1.0327,
      "step": 6697
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.006535768508911,
      "learning_rate": 1.6339647957064818e-05,
      "loss": 1.0284,
      "step": 6698
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.543102979660034,
      "learning_rate": 1.633900904066703e-05,
      "loss": 0.9872,
      "step": 6699
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.496680736541748,
      "learning_rate": 1.633837012426924e-05,
      "loss": 1.0688,
      "step": 6700
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0583362579345703,
      "learning_rate": 1.6337731207871453e-05,
      "loss": 0.9928,
      "step": 6701
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.477125644683838,
      "learning_rate": 1.633709229147366e-05,
      "loss": 0.9211,
      "step": 6702
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3974345922470093,
      "learning_rate": 1.6336453375075872e-05,
      "loss": 0.5068,
      "step": 6703
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1578094959259033,
      "learning_rate": 1.6335814458678084e-05,
      "loss": 0.9831,
      "step": 6704
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.317256212234497,
      "learning_rate": 1.6335175542280295e-05,
      "loss": 0.7635,
      "step": 6705
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.523637294769287,
      "learning_rate": 1.6334536625882504e-05,
      "loss": 1.1915,
      "step": 6706
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2367899417877197,
      "learning_rate": 1.6333897709484715e-05,
      "loss": 0.9134,
      "step": 6707
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5087875127792358,
      "learning_rate": 1.6333258793086927e-05,
      "loss": 0.877,
      "step": 6708
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8521281480789185,
      "learning_rate": 1.6332619876689138e-05,
      "loss": 0.9683,
      "step": 6709
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.160231590270996,
      "learning_rate": 1.6331980960291346e-05,
      "loss": 0.9425,
      "step": 6710
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0034754276275635,
      "learning_rate": 1.6331342043893558e-05,
      "loss": 0.9853,
      "step": 6711
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.986652851104736,
      "learning_rate": 1.633070312749577e-05,
      "loss": 1.5756,
      "step": 6712
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9615986347198486,
      "learning_rate": 1.633006421109798e-05,
      "loss": 0.849,
      "step": 6713
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5999224185943604,
      "learning_rate": 1.632942529470019e-05,
      "loss": 1.2779,
      "step": 6714
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.243210554122925,
      "learning_rate": 1.63287863783024e-05,
      "loss": 0.9559,
      "step": 6715
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.197634696960449,
      "learning_rate": 1.6328147461904612e-05,
      "loss": 1.031,
      "step": 6716
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.744501113891602,
      "learning_rate": 1.6327508545506824e-05,
      "loss": 1.2445,
      "step": 6717
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.665861129760742,
      "learning_rate": 1.6326869629109032e-05,
      "loss": 0.8631,
      "step": 6718
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9235609769821167,
      "learning_rate": 1.6326230712711243e-05,
      "loss": 1.2114,
      "step": 6719
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.597759962081909,
      "learning_rate": 1.6325591796313455e-05,
      "loss": 0.7014,
      "step": 6720
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6699855327606201,
      "learning_rate": 1.6324952879915667e-05,
      "loss": 0.861,
      "step": 6721
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6273828744888306,
      "learning_rate": 1.6324313963517875e-05,
      "loss": 0.887,
      "step": 6722
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9416484832763672,
      "learning_rate": 1.6323675047120086e-05,
      "loss": 0.9287,
      "step": 6723
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9366059303283691,
      "learning_rate": 1.6323036130722298e-05,
      "loss": 0.9737,
      "step": 6724
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9675278663635254,
      "learning_rate": 1.632239721432451e-05,
      "loss": 0.9331,
      "step": 6725
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8811018466949463,
      "learning_rate": 1.6321758297926718e-05,
      "loss": 0.9594,
      "step": 6726
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.921157717704773,
      "learning_rate": 1.632111938152893e-05,
      "loss": 0.8522,
      "step": 6727
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.902667999267578,
      "learning_rate": 1.6320480465131137e-05,
      "loss": 0.8616,
      "step": 6728
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.906927466392517,
      "learning_rate": 1.6319841548733352e-05,
      "loss": 0.7878,
      "step": 6729
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.84007728099823,
      "learning_rate": 1.631920263233556e-05,
      "loss": 0.9176,
      "step": 6730
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.8229827880859375,
      "learning_rate": 1.6318563715937772e-05,
      "loss": 1.7006,
      "step": 6731
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.49607515335083,
      "learning_rate": 1.631792479953998e-05,
      "loss": 0.9285,
      "step": 6732
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.742531776428223,
      "learning_rate": 1.631728588314219e-05,
      "loss": 1.0389,
      "step": 6733
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2653591632843018,
      "learning_rate": 1.6316646966744403e-05,
      "loss": 0.8869,
      "step": 6734
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.118873357772827,
      "learning_rate": 1.6316008050346615e-05,
      "loss": 1.1421,
      "step": 6735
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1761510372161865,
      "learning_rate": 1.6315369133948823e-05,
      "loss": 0.947,
      "step": 6736
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.212217092514038,
      "learning_rate": 1.6314730217551034e-05,
      "loss": 0.9801,
      "step": 6737
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.554738998413086,
      "learning_rate": 1.6314091301153246e-05,
      "loss": 0.9488,
      "step": 6738
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7037999629974365,
      "learning_rate": 1.6313452384755457e-05,
      "loss": 1.1336,
      "step": 6739
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.528472423553467,
      "learning_rate": 1.6312813468357666e-05,
      "loss": 1.0242,
      "step": 6740
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9983327388763428,
      "learning_rate": 1.6312174551959877e-05,
      "loss": 0.967,
      "step": 6741
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1077640056610107,
      "learning_rate": 1.631153563556209e-05,
      "loss": 1.0285,
      "step": 6742
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7738698720932007,
      "learning_rate": 1.63108967191643e-05,
      "loss": 0.9412,
      "step": 6743
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.03306245803833,
      "learning_rate": 1.631025780276651e-05,
      "loss": 0.9321,
      "step": 6744
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5557698011398315,
      "learning_rate": 1.630961888636872e-05,
      "loss": 0.8208,
      "step": 6745
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.9411113262176514,
      "learning_rate": 1.630897996997093e-05,
      "loss": 0.6912,
      "step": 6746
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.947158932685852,
      "learning_rate": 1.6308341053573143e-05,
      "loss": 1.0666,
      "step": 6747
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8873740434646606,
      "learning_rate": 1.630770213717535e-05,
      "loss": 1.0064,
      "step": 6748
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7058638334274292,
      "learning_rate": 1.6307063220777563e-05,
      "loss": 0.8998,
      "step": 6749
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.130236864089966,
      "learning_rate": 1.6306424304379774e-05,
      "loss": 0.9997,
      "step": 6750
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.082688570022583,
      "learning_rate": 1.6305785387981986e-05,
      "loss": 0.9574,
      "step": 6751
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8758187294006348,
      "learning_rate": 1.6305146471584194e-05,
      "loss": 0.7897,
      "step": 6752
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.074378490447998,
      "learning_rate": 1.6304507555186406e-05,
      "loss": 0.9666,
      "step": 6753
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.747744083404541,
      "learning_rate": 1.6303868638788617e-05,
      "loss": 1.4422,
      "step": 6754
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.229533672332764,
      "learning_rate": 1.630322972239083e-05,
      "loss": 1.2587,
      "step": 6755
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1148836612701416,
      "learning_rate": 1.6302590805993037e-05,
      "loss": 1.282,
      "step": 6756
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.841969609260559,
      "learning_rate": 1.630195188959525e-05,
      "loss": 1.1586,
      "step": 6757
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.465489625930786,
      "learning_rate": 1.630131297319746e-05,
      "loss": 1.0372,
      "step": 6758
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6603655815124512,
      "learning_rate": 1.630067405679967e-05,
      "loss": 0.8787,
      "step": 6759
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.542113184928894,
      "learning_rate": 1.630003514040188e-05,
      "loss": 0.8946,
      "step": 6760
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1429295539855957,
      "learning_rate": 1.629939622400409e-05,
      "loss": 1.0308,
      "step": 6761
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.240789890289307,
      "learning_rate": 1.62987573076063e-05,
      "loss": 1.6939,
      "step": 6762
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5786811113357544,
      "learning_rate": 1.6298118391208514e-05,
      "loss": 0.9697,
      "step": 6763
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.112022638320923,
      "learning_rate": 1.6297479474810722e-05,
      "loss": 0.784,
      "step": 6764
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5507161617279053,
      "learning_rate": 1.6296840558412934e-05,
      "loss": 0.4355,
      "step": 6765
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0080387592315674,
      "learning_rate": 1.6296201642015142e-05,
      "loss": 0.9644,
      "step": 6766
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0914533138275146,
      "learning_rate": 1.6295562725617354e-05,
      "loss": 0.9687,
      "step": 6767
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.254121780395508,
      "learning_rate": 1.6294923809219565e-05,
      "loss": 0.7986,
      "step": 6768
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8956842422485352,
      "learning_rate": 1.6294284892821773e-05,
      "loss": 0.9187,
      "step": 6769
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.1025493144989014,
      "learning_rate": 1.6293645976423985e-05,
      "loss": 1.0124,
      "step": 6770
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.5480246543884277,
      "learning_rate": 1.6293007060026196e-05,
      "loss": 0.6576,
      "step": 6771
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.262542724609375,
      "learning_rate": 1.6292368143628408e-05,
      "loss": 1.0198,
      "step": 6772
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8675349950790405,
      "learning_rate": 1.6291729227230616e-05,
      "loss": 1.0363,
      "step": 6773
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7007449865341187,
      "learning_rate": 1.6291090310832828e-05,
      "loss": 0.9837,
      "step": 6774
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8449422121047974,
      "learning_rate": 1.629045139443504e-05,
      "loss": 0.9067,
      "step": 6775
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0985426902770996,
      "learning_rate": 1.628981247803725e-05,
      "loss": 0.854,
      "step": 6776
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.697457194328308,
      "learning_rate": 1.628917356163946e-05,
      "loss": 0.9327,
      "step": 6777
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4071866273880005,
      "learning_rate": 1.628853464524167e-05,
      "loss": 0.5031,
      "step": 6778
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.965600848197937,
      "learning_rate": 1.6287895728843882e-05,
      "loss": 1.1711,
      "step": 6779
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8331794738769531,
      "learning_rate": 1.6287256812446094e-05,
      "loss": 0.8056,
      "step": 6780
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.4781947135925293,
      "learning_rate": 1.6286617896048302e-05,
      "loss": 0.9506,
      "step": 6781
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.294234275817871,
      "learning_rate": 1.6285978979650513e-05,
      "loss": 0.9281,
      "step": 6782
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.1990721225738525,
      "learning_rate": 1.6285340063252725e-05,
      "loss": 0.9641,
      "step": 6783
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7638521194458008,
      "learning_rate": 1.6284701146854936e-05,
      "loss": 0.8812,
      "step": 6784
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.594762086868286,
      "learning_rate": 1.6284062230457145e-05,
      "loss": 0.7221,
      "step": 6785
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9264672994613647,
      "learning_rate": 1.6283423314059356e-05,
      "loss": 0.9321,
      "step": 6786
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1144778728485107,
      "learning_rate": 1.6282784397661568e-05,
      "loss": 0.946,
      "step": 6787
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.927331447601318,
      "learning_rate": 1.628214548126378e-05,
      "loss": 1.7541,
      "step": 6788
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1490204334259033,
      "learning_rate": 1.6281506564865987e-05,
      "loss": 0.7958,
      "step": 6789
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7507466077804565,
      "learning_rate": 1.62808676484682e-05,
      "loss": 0.8331,
      "step": 6790
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8510804176330566,
      "learning_rate": 1.628022873207041e-05,
      "loss": 0.9379,
      "step": 6791
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7912501096725464,
      "learning_rate": 1.6279589815672622e-05,
      "loss": 0.8908,
      "step": 6792
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1562278270721436,
      "learning_rate": 1.627895089927483e-05,
      "loss": 1.1997,
      "step": 6793
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.398465871810913,
      "learning_rate": 1.6278311982877042e-05,
      "loss": 1.0245,
      "step": 6794
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1853151321411133,
      "learning_rate": 1.627767306647925e-05,
      "loss": 1.1194,
      "step": 6795
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7983784675598145,
      "learning_rate": 1.6277034150081465e-05,
      "loss": 0.8776,
      "step": 6796
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3244848251342773,
      "learning_rate": 1.6276395233683673e-05,
      "loss": 1.3384,
      "step": 6797
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.172518253326416,
      "learning_rate": 1.6275756317285885e-05,
      "loss": 0.9827,
      "step": 6798
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.742322564125061,
      "learning_rate": 1.6275117400888093e-05,
      "loss": 0.9003,
      "step": 6799
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0901553630828857,
      "learning_rate": 1.6274478484490304e-05,
      "loss": 0.8698,
      "step": 6800
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9987642765045166,
      "learning_rate": 1.6273839568092516e-05,
      "loss": 1.076,
      "step": 6801
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.208479642868042,
      "learning_rate": 1.6273200651694727e-05,
      "loss": 1.2456,
      "step": 6802
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.916596531867981,
      "learning_rate": 1.6272561735296936e-05,
      "loss": 0.8691,
      "step": 6803
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.38509202003479,
      "learning_rate": 1.6271922818899147e-05,
      "loss": 1.1345,
      "step": 6804
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.102433919906616,
      "learning_rate": 1.627128390250136e-05,
      "loss": 1.3282,
      "step": 6805
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5100873708724976,
      "learning_rate": 1.627064498610357e-05,
      "loss": 0.8442,
      "step": 6806
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9370959997177124,
      "learning_rate": 1.627000606970578e-05,
      "loss": 0.9677,
      "step": 6807
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.697508811950684,
      "learning_rate": 1.626936715330799e-05,
      "loss": 1.3896,
      "step": 6808
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.477963924407959,
      "learning_rate": 1.62687282369102e-05,
      "loss": 1.3754,
      "step": 6809
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2310121059417725,
      "learning_rate": 1.6268089320512413e-05,
      "loss": 0.9658,
      "step": 6810
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.873423099517822,
      "learning_rate": 1.626745040411462e-05,
      "loss": 1.9292,
      "step": 6811
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8107303380966187,
      "learning_rate": 1.6266811487716833e-05,
      "loss": 1.0037,
      "step": 6812
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0789482593536377,
      "learning_rate": 1.6266172571319044e-05,
      "loss": 0.9933,
      "step": 6813
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.607635021209717,
      "learning_rate": 1.6265533654921256e-05,
      "loss": 1.3983,
      "step": 6814
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.5175349712371826,
      "learning_rate": 1.6264894738523464e-05,
      "loss": 0.7995,
      "step": 6815
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1564600467681885,
      "learning_rate": 1.6264255822125675e-05,
      "loss": 1.0675,
      "step": 6816
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.037172794342041,
      "learning_rate": 1.6263616905727887e-05,
      "loss": 0.965,
      "step": 6817
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.545989990234375,
      "learning_rate": 1.62629779893301e-05,
      "loss": 1.3109,
      "step": 6818
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.567753553390503,
      "learning_rate": 1.6262339072932307e-05,
      "loss": 0.9452,
      "step": 6819
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1507506370544434,
      "learning_rate": 1.6261700156534518e-05,
      "loss": 1.1678,
      "step": 6820
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9380593299865723,
      "learning_rate": 1.626106124013673e-05,
      "loss": 0.8488,
      "step": 6821
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9773294925689697,
      "learning_rate": 1.626042232373894e-05,
      "loss": 0.9847,
      "step": 6822
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.496368646621704,
      "learning_rate": 1.625978340734115e-05,
      "loss": 0.6452,
      "step": 6823
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.138049840927124,
      "learning_rate": 1.625914449094336e-05,
      "loss": 0.905,
      "step": 6824
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1985349655151367,
      "learning_rate": 1.6258505574545573e-05,
      "loss": 0.8468,
      "step": 6825
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.9377403259277344,
      "learning_rate": 1.6257866658147784e-05,
      "loss": 1.1989,
      "step": 6826
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.6269001960754395,
      "learning_rate": 1.6257227741749992e-05,
      "loss": 1.6424,
      "step": 6827
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1765329837799072,
      "learning_rate": 1.6256588825352204e-05,
      "loss": 0.986,
      "step": 6828
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.721854567527771,
      "learning_rate": 1.6255949908954415e-05,
      "loss": 0.9081,
      "step": 6829
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2994437217712402,
      "learning_rate": 1.6255310992556627e-05,
      "loss": 1.0061,
      "step": 6830
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9081352949142456,
      "learning_rate": 1.6254672076158835e-05,
      "loss": 1.2646,
      "step": 6831
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2963218688964844,
      "learning_rate": 1.6254033159761047e-05,
      "loss": 0.846,
      "step": 6832
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.973229169845581,
      "learning_rate": 1.6253394243363255e-05,
      "loss": 1.0475,
      "step": 6833
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8043783903121948,
      "learning_rate": 1.625275532696547e-05,
      "loss": 0.8998,
      "step": 6834
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8297629356384277,
      "learning_rate": 1.6252116410567678e-05,
      "loss": 0.8849,
      "step": 6835
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8341784477233887,
      "learning_rate": 1.625147749416989e-05,
      "loss": 0.9767,
      "step": 6836
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8075668811798096,
      "learning_rate": 1.6250838577772098e-05,
      "loss": 1.0691,
      "step": 6837
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7613651752471924,
      "learning_rate": 1.625019966137431e-05,
      "loss": 0.8866,
      "step": 6838
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.4271352291107178,
      "learning_rate": 1.624956074497652e-05,
      "loss": 1.0054,
      "step": 6839
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.897237777709961,
      "learning_rate": 1.6248921828578732e-05,
      "loss": 0.9011,
      "step": 6840
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.220170736312866,
      "learning_rate": 1.624828291218094e-05,
      "loss": 0.864,
      "step": 6841
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.328364849090576,
      "learning_rate": 1.6247643995783152e-05,
      "loss": 1.9183,
      "step": 6842
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.162158966064453,
      "learning_rate": 1.6247005079385364e-05,
      "loss": 1.0905,
      "step": 6843
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.5303239822387695,
      "learning_rate": 1.6246366162987575e-05,
      "loss": 1.4,
      "step": 6844
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.832442760467529,
      "learning_rate": 1.6245727246589783e-05,
      "loss": 1.0017,
      "step": 6845
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6656255722045898,
      "learning_rate": 1.6245088330191995e-05,
      "loss": 0.3777,
      "step": 6846
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.224721670150757,
      "learning_rate": 1.6244449413794206e-05,
      "loss": 0.8548,
      "step": 6847
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.211031198501587,
      "learning_rate": 1.6243810497396418e-05,
      "loss": 1.3617,
      "step": 6848
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2487733364105225,
      "learning_rate": 1.6243171580998626e-05,
      "loss": 0.9329,
      "step": 6849
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0336060523986816,
      "learning_rate": 1.6242532664600838e-05,
      "loss": 1.2793,
      "step": 6850
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.374565601348877,
      "learning_rate": 1.624189374820305e-05,
      "loss": 1.0511,
      "step": 6851
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9682773351669312,
      "learning_rate": 1.624125483180526e-05,
      "loss": 1.2573,
      "step": 6852
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.489306926727295,
      "learning_rate": 1.624061591540747e-05,
      "loss": 1.1333,
      "step": 6853
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.560053825378418,
      "learning_rate": 1.623997699900968e-05,
      "loss": 1.0982,
      "step": 6854
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5355865955352783,
      "learning_rate": 1.6239338082611892e-05,
      "loss": 0.4778,
      "step": 6855
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7882540225982666,
      "learning_rate": 1.6238699166214103e-05,
      "loss": 0.9136,
      "step": 6856
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4964332580566406,
      "learning_rate": 1.623806024981631e-05,
      "loss": 0.8428,
      "step": 6857
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7693108320236206,
      "learning_rate": 1.6237421333418523e-05,
      "loss": 0.8488,
      "step": 6858
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.237185001373291,
      "learning_rate": 1.6236782417020735e-05,
      "loss": 0.9617,
      "step": 6859
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7108755111694336,
      "learning_rate": 1.6236143500622946e-05,
      "loss": 0.9324,
      "step": 6860
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7515324354171753,
      "learning_rate": 1.6235504584225154e-05,
      "loss": 0.9181,
      "step": 6861
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.023502826690674,
      "learning_rate": 1.6234865667827366e-05,
      "loss": 1.0718,
      "step": 6862
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5335469245910645,
      "learning_rate": 1.6234226751429578e-05,
      "loss": 0.8166,
      "step": 6863
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.358292818069458,
      "learning_rate": 1.623358783503179e-05,
      "loss": 0.975,
      "step": 6864
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6701734066009521,
      "learning_rate": 1.6232948918633997e-05,
      "loss": 0.9125,
      "step": 6865
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.733935832977295,
      "learning_rate": 1.623231000223621e-05,
      "loss": 2.0464,
      "step": 6866
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.501582145690918,
      "learning_rate": 1.6231671085838417e-05,
      "loss": 0.4768,
      "step": 6867
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.63988196849823,
      "learning_rate": 1.6231032169440632e-05,
      "loss": 0.3747,
      "step": 6868
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.106794834136963,
      "learning_rate": 1.623039325304284e-05,
      "loss": 1.0326,
      "step": 6869
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.27236008644104,
      "learning_rate": 1.622975433664505e-05,
      "loss": 1.4473,
      "step": 6870
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1693084239959717,
      "learning_rate": 1.622911542024726e-05,
      "loss": 1.3547,
      "step": 6871
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4436591863632202,
      "learning_rate": 1.622847650384947e-05,
      "loss": 0.8467,
      "step": 6872
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.198760747909546,
      "learning_rate": 1.6227837587451683e-05,
      "loss": 1.0543,
      "step": 6873
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1477766036987305,
      "learning_rate": 1.6227198671053894e-05,
      "loss": 0.5762,
      "step": 6874
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.109314203262329,
      "learning_rate": 1.6226559754656103e-05,
      "loss": 0.9189,
      "step": 6875
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1613752841949463,
      "learning_rate": 1.6225920838258314e-05,
      "loss": 0.8315,
      "step": 6876
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1884570121765137,
      "learning_rate": 1.6225281921860526e-05,
      "loss": 0.9696,
      "step": 6877
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2438559532165527,
      "learning_rate": 1.6224643005462737e-05,
      "loss": 1.1135,
      "step": 6878
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7251319885253906,
      "learning_rate": 1.6224004089064945e-05,
      "loss": 1.0315,
      "step": 6879
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0064384937286377,
      "learning_rate": 1.6223365172667157e-05,
      "loss": 0.8734,
      "step": 6880
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.445988416671753,
      "learning_rate": 1.622272625626937e-05,
      "loss": 0.9381,
      "step": 6881
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7260414361953735,
      "learning_rate": 1.622208733987158e-05,
      "loss": 1.0321,
      "step": 6882
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.9684176445007324,
      "learning_rate": 1.6221448423473788e-05,
      "loss": 0.8271,
      "step": 6883
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.604257345199585,
      "learning_rate": 1.6220809507076e-05,
      "loss": 0.8478,
      "step": 6884
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.757899522781372,
      "learning_rate": 1.622017059067821e-05,
      "loss": 1.2182,
      "step": 6885
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.4713265895843506,
      "learning_rate": 1.6219531674280423e-05,
      "loss": 0.8827,
      "step": 6886
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7984874248504639,
      "learning_rate": 1.621889275788263e-05,
      "loss": 0.8054,
      "step": 6887
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.63650643825531,
      "learning_rate": 1.6218253841484842e-05,
      "loss": 1.0143,
      "step": 6888
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1473278999328613,
      "learning_rate": 1.6217614925087054e-05,
      "loss": 1.1009,
      "step": 6889
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0116302967071533,
      "learning_rate": 1.6216976008689266e-05,
      "loss": 0.8734,
      "step": 6890
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7801131010055542,
      "learning_rate": 1.6216337092291474e-05,
      "loss": 0.9417,
      "step": 6891
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4972888231277466,
      "learning_rate": 1.6215698175893685e-05,
      "loss": 0.8698,
      "step": 6892
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.292487144470215,
      "learning_rate": 1.6215059259495897e-05,
      "loss": 0.9293,
      "step": 6893
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6578466892242432,
      "learning_rate": 1.621442034309811e-05,
      "loss": 0.9239,
      "step": 6894
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.590512752532959,
      "learning_rate": 1.6213781426700317e-05,
      "loss": 1.0703,
      "step": 6895
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7608280181884766,
      "learning_rate": 1.6213142510302528e-05,
      "loss": 0.4045,
      "step": 6896
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9199222326278687,
      "learning_rate": 1.621250359390474e-05,
      "loss": 0.9455,
      "step": 6897
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7155009508132935,
      "learning_rate": 1.621186467750695e-05,
      "loss": 0.9631,
      "step": 6898
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7825814485549927,
      "learning_rate": 1.621122576110916e-05,
      "loss": 0.8574,
      "step": 6899
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1784303188323975,
      "learning_rate": 1.621058684471137e-05,
      "loss": 0.966,
      "step": 6900
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.132535934448242,
      "learning_rate": 1.6209947928313582e-05,
      "loss": 0.9739,
      "step": 6901
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0097410678863525,
      "learning_rate": 1.6209309011915794e-05,
      "loss": 0.9831,
      "step": 6902
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.446451187133789,
      "learning_rate": 1.6208670095518002e-05,
      "loss": 0.9697,
      "step": 6903
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.919358491897583,
      "learning_rate": 1.6208031179120214e-05,
      "loss": 1.1617,
      "step": 6904
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.439936876296997,
      "learning_rate": 1.6207392262722422e-05,
      "loss": 0.9699,
      "step": 6905
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7384957075119019,
      "learning_rate": 1.6206753346324637e-05,
      "loss": 0.8653,
      "step": 6906
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.871006965637207,
      "learning_rate": 1.6206114429926845e-05,
      "loss": 0.9863,
      "step": 6907
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.010640859603882,
      "learning_rate": 1.6205475513529056e-05,
      "loss": 0.9819,
      "step": 6908
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.344726085662842,
      "learning_rate": 1.6204836597131265e-05,
      "loss": 0.9884,
      "step": 6909
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.8694474697113037,
      "learning_rate": 1.6204197680733476e-05,
      "loss": 1.2065,
      "step": 6910
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8364231586456299,
      "learning_rate": 1.6203558764335688e-05,
      "loss": 0.8947,
      "step": 6911
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2751810550689697,
      "learning_rate": 1.62029198479379e-05,
      "loss": 0.9621,
      "step": 6912
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8350387811660767,
      "learning_rate": 1.6202280931540107e-05,
      "loss": 1.0431,
      "step": 6913
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.1643755435943604,
      "learning_rate": 1.620164201514232e-05,
      "loss": 0.9054,
      "step": 6914
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9616844654083252,
      "learning_rate": 1.620100309874453e-05,
      "loss": 0.9234,
      "step": 6915
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4626966714859009,
      "learning_rate": 1.6200364182346742e-05,
      "loss": 0.8543,
      "step": 6916
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.738409161567688,
      "learning_rate": 1.619972526594895e-05,
      "loss": 0.4106,
      "step": 6917
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3173367977142334,
      "learning_rate": 1.6199086349551162e-05,
      "loss": 0.3751,
      "step": 6918
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.074218988418579,
      "learning_rate": 1.6198447433153373e-05,
      "loss": 0.8815,
      "step": 6919
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.275252103805542,
      "learning_rate": 1.6197808516755585e-05,
      "loss": 1.0089,
      "step": 6920
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3593862056732178,
      "learning_rate": 1.6197169600357793e-05,
      "loss": 0.7106,
      "step": 6921
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.453946590423584,
      "learning_rate": 1.6196530683960005e-05,
      "loss": 0.9345,
      "step": 6922
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0785272121429443,
      "learning_rate": 1.6195891767562216e-05,
      "loss": 0.9989,
      "step": 6923
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6792219877243042,
      "learning_rate": 1.6195252851164428e-05,
      "loss": 0.9461,
      "step": 6924
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7671988010406494,
      "learning_rate": 1.6194613934766636e-05,
      "loss": 0.9587,
      "step": 6925
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0802664756774902,
      "learning_rate": 1.6193975018368847e-05,
      "loss": 0.838,
      "step": 6926
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.232982635498047,
      "learning_rate": 1.619333610197106e-05,
      "loss": 1.3243,
      "step": 6927
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8275071382522583,
      "learning_rate": 1.619269718557327e-05,
      "loss": 0.8476,
      "step": 6928
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9418545961380005,
      "learning_rate": 1.619205826917548e-05,
      "loss": 0.9483,
      "step": 6929
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7690988779067993,
      "learning_rate": 1.619141935277769e-05,
      "loss": 0.9329,
      "step": 6930
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5196151733398438,
      "learning_rate": 1.6190780436379902e-05,
      "loss": 0.8316,
      "step": 6931
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.875416278839111,
      "learning_rate": 1.6190141519982113e-05,
      "loss": 1.2132,
      "step": 6932
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.866960048675537,
      "learning_rate": 1.618950260358432e-05,
      "loss": 1.0169,
      "step": 6933
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7615138292312622,
      "learning_rate": 1.6188863687186533e-05,
      "loss": 0.9861,
      "step": 6934
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.019277572631836,
      "learning_rate": 1.6188224770788745e-05,
      "loss": 1.6432,
      "step": 6935
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.988661766052246,
      "learning_rate": 1.6187585854390956e-05,
      "loss": 0.899,
      "step": 6936
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.3776378631591797,
      "learning_rate": 1.6186946937993164e-05,
      "loss": 1.4035,
      "step": 6937
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2300195693969727,
      "learning_rate": 1.6186308021595376e-05,
      "loss": 0.9609,
      "step": 6938
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6543470621109009,
      "learning_rate": 1.6185669105197584e-05,
      "loss": 1.0523,
      "step": 6939
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.016758441925049,
      "learning_rate": 1.61850301887998e-05,
      "loss": 0.9357,
      "step": 6940
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.873218297958374,
      "learning_rate": 1.6184391272402007e-05,
      "loss": 1.0091,
      "step": 6941
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.379628896713257,
      "learning_rate": 1.618375235600422e-05,
      "loss": 1.0752,
      "step": 6942
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0825552940368652,
      "learning_rate": 1.6183113439606427e-05,
      "loss": 1.0489,
      "step": 6943
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.056366443634033,
      "learning_rate": 1.618247452320864e-05,
      "loss": 1.2193,
      "step": 6944
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.357987880706787,
      "learning_rate": 1.618183560681085e-05,
      "loss": 1.4619,
      "step": 6945
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.983461380004883,
      "learning_rate": 1.618119669041306e-05,
      "loss": 0.8634,
      "step": 6946
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.667811155319214,
      "learning_rate": 1.618055777401527e-05,
      "loss": 0.9296,
      "step": 6947
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.6463258266448975,
      "learning_rate": 1.617991885761748e-05,
      "loss": 1.0086,
      "step": 6948
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8104370832443237,
      "learning_rate": 1.6179279941219693e-05,
      "loss": 0.9906,
      "step": 6949
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7832571268081665,
      "learning_rate": 1.6178641024821904e-05,
      "loss": 0.8078,
      "step": 6950
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1683340072631836,
      "learning_rate": 1.6178002108424112e-05,
      "loss": 0.9317,
      "step": 6951
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.737170934677124,
      "learning_rate": 1.6177363192026324e-05,
      "loss": 0.9934,
      "step": 6952
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.277827739715576,
      "learning_rate": 1.6176724275628535e-05,
      "loss": 1.5292,
      "step": 6953
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5970447063446045,
      "learning_rate": 1.6176085359230747e-05,
      "loss": 1.0675,
      "step": 6954
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.311631679534912,
      "learning_rate": 1.6175446442832955e-05,
      "loss": 1.8346,
      "step": 6955
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.86002779006958,
      "learning_rate": 1.6174807526435167e-05,
      "loss": 0.871,
      "step": 6956
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3892719745635986,
      "learning_rate": 1.6174168610037378e-05,
      "loss": 0.9581,
      "step": 6957
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0767552852630615,
      "learning_rate": 1.617352969363959e-05,
      "loss": 0.9189,
      "step": 6958
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.2031633853912354,
      "learning_rate": 1.6172890777241798e-05,
      "loss": 0.7096,
      "step": 6959
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7189689874649048,
      "learning_rate": 1.617225186084401e-05,
      "loss": 0.9081,
      "step": 6960
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5174791812896729,
      "learning_rate": 1.617161294444622e-05,
      "loss": 0.8077,
      "step": 6961
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8208684921264648,
      "learning_rate": 1.6170974028048433e-05,
      "loss": 0.9173,
      "step": 6962
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0717475414276123,
      "learning_rate": 1.617033511165064e-05,
      "loss": 1.0699,
      "step": 6963
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.002635955810547,
      "learning_rate": 1.6169696195252852e-05,
      "loss": 0.8754,
      "step": 6964
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1870059967041016,
      "learning_rate": 1.6169057278855064e-05,
      "loss": 1.016,
      "step": 6965
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.813111305236816,
      "learning_rate": 1.6168418362457275e-05,
      "loss": 1.55,
      "step": 6966
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.591619968414307,
      "learning_rate": 1.6167779446059484e-05,
      "loss": 1.6186,
      "step": 6967
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7590428590774536,
      "learning_rate": 1.6167140529661695e-05,
      "loss": 0.9218,
      "step": 6968
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0027313232421875,
      "learning_rate": 1.6166501613263907e-05,
      "loss": 0.9538,
      "step": 6969
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8498092889785767,
      "learning_rate": 1.6165862696866118e-05,
      "loss": 0.9239,
      "step": 6970
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.276279926300049,
      "learning_rate": 1.6165223780468326e-05,
      "loss": 1.0419,
      "step": 6971
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5759127140045166,
      "learning_rate": 1.6164584864070538e-05,
      "loss": 0.7459,
      "step": 6972
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9041011333465576,
      "learning_rate": 1.616394594767275e-05,
      "loss": 0.8464,
      "step": 6973
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1002302169799805,
      "learning_rate": 1.616330703127496e-05,
      "loss": 0.8873,
      "step": 6974
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.322406053543091,
      "learning_rate": 1.616266811487717e-05,
      "loss": 0.9955,
      "step": 6975
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0218255519866943,
      "learning_rate": 1.616202919847938e-05,
      "loss": 1.0369,
      "step": 6976
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9857975244522095,
      "learning_rate": 1.616139028208159e-05,
      "loss": 0.8858,
      "step": 6977
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.233015537261963,
      "learning_rate": 1.6160751365683804e-05,
      "loss": 0.9934,
      "step": 6978
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5710667371749878,
      "learning_rate": 1.6160112449286012e-05,
      "loss": 0.8688,
      "step": 6979
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.582974433898926,
      "learning_rate": 1.6159473532888223e-05,
      "loss": 0.7773,
      "step": 6980
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.457093238830566,
      "learning_rate": 1.615883461649043e-05,
      "loss": 1.4609,
      "step": 6981
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0395684242248535,
      "learning_rate": 1.6158195700092643e-05,
      "loss": 0.7471,
      "step": 6982
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.422386646270752,
      "learning_rate": 1.6157556783694855e-05,
      "loss": 1.0448,
      "step": 6983
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9774515628814697,
      "learning_rate": 1.6156917867297066e-05,
      "loss": 0.9058,
      "step": 6984
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3277106285095215,
      "learning_rate": 1.6156278950899274e-05,
      "loss": 1.1011,
      "step": 6985
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.400667667388916,
      "learning_rate": 1.6155640034501486e-05,
      "loss": 1.6649,
      "step": 6986
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.861943006515503,
      "learning_rate": 1.6155001118103698e-05,
      "loss": 0.9019,
      "step": 6987
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1134071350097656,
      "learning_rate": 1.615436220170591e-05,
      "loss": 0.952,
      "step": 6988
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.2712836265563965,
      "learning_rate": 1.6153723285308117e-05,
      "loss": 1.2699,
      "step": 6989
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.192450523376465,
      "learning_rate": 1.615308436891033e-05,
      "loss": 1.3917,
      "step": 6990
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4951313734054565,
      "learning_rate": 1.615244545251254e-05,
      "loss": 0.8014,
      "step": 6991
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8775365352630615,
      "learning_rate": 1.6151806536114752e-05,
      "loss": 0.9199,
      "step": 6992
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.128427505493164,
      "learning_rate": 1.615116761971696e-05,
      "loss": 1.3554,
      "step": 6993
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.957291841506958,
      "learning_rate": 1.615052870331917e-05,
      "loss": 0.9276,
      "step": 6994
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3108301162719727,
      "learning_rate": 1.6149889786921383e-05,
      "loss": 0.9039,
      "step": 6995
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.891043663024902,
      "learning_rate": 1.6149250870523595e-05,
      "loss": 1.3401,
      "step": 6996
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9983264207839966,
      "learning_rate": 1.6148611954125803e-05,
      "loss": 1.0331,
      "step": 6997
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.430047512054443,
      "learning_rate": 1.6147973037728014e-05,
      "loss": 1.0949,
      "step": 6998
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.763291835784912,
      "learning_rate": 1.6147334121330226e-05,
      "loss": 0.893,
      "step": 6999
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.6745097637176514,
      "learning_rate": 1.6146695204932437e-05,
      "loss": 0.9945,
      "step": 7000
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7195119857788086,
      "learning_rate": 1.6146056288534646e-05,
      "loss": 0.4889,
      "step": 7001
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9654285907745361,
      "learning_rate": 1.6145417372136857e-05,
      "loss": 0.9394,
      "step": 7002
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.6470088958740234,
      "learning_rate": 1.614477845573907e-05,
      "loss": 1.1739,
      "step": 7003
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.860987186431885,
      "learning_rate": 1.614413953934128e-05,
      "loss": 0.953,
      "step": 7004
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7010750770568848,
      "learning_rate": 1.614350062294349e-05,
      "loss": 1.2041,
      "step": 7005
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5018595457077026,
      "learning_rate": 1.61428617065457e-05,
      "loss": 0.8372,
      "step": 7006
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5815308094024658,
      "learning_rate": 1.614222279014791e-05,
      "loss": 0.8992,
      "step": 7007
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.46390962600708,
      "learning_rate": 1.6141583873750123e-05,
      "loss": 1.1534,
      "step": 7008
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7932746410369873,
      "learning_rate": 1.614094495735233e-05,
      "loss": 0.8554,
      "step": 7009
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7026312351226807,
      "learning_rate": 1.6140306040954543e-05,
      "loss": 0.9632,
      "step": 7010
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.321834087371826,
      "learning_rate": 1.6139667124556754e-05,
      "loss": 1.2938,
      "step": 7011
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.149564266204834,
      "learning_rate": 1.6139028208158966e-05,
      "loss": 0.988,
      "step": 7012
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.4068918228149414,
      "learning_rate": 1.6138389291761174e-05,
      "loss": 1.0226,
      "step": 7013
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2466979026794434,
      "learning_rate": 1.6137750375363386e-05,
      "loss": 1.0039,
      "step": 7014
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.591889500617981,
      "learning_rate": 1.6137111458965594e-05,
      "loss": 0.5337,
      "step": 7015
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.258786201477051,
      "learning_rate": 1.613647254256781e-05,
      "loss": 0.9614,
      "step": 7016
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.4626786708831787,
      "learning_rate": 1.6135833626170017e-05,
      "loss": 1.2225,
      "step": 7017
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1561005115509033,
      "learning_rate": 1.613519470977223e-05,
      "loss": 1.0904,
      "step": 7018
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1867294311523438,
      "learning_rate": 1.6134555793374437e-05,
      "loss": 0.946,
      "step": 7019
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5239241123199463,
      "learning_rate": 1.6133916876976648e-05,
      "loss": 0.8515,
      "step": 7020
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.661111831665039,
      "learning_rate": 1.613327796057886e-05,
      "loss": 0.8866,
      "step": 7021
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1698622703552246,
      "learning_rate": 1.613263904418107e-05,
      "loss": 0.9757,
      "step": 7022
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0391268730163574,
      "learning_rate": 1.613200012778328e-05,
      "loss": 0.9273,
      "step": 7023
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5293099880218506,
      "learning_rate": 1.613136121138549e-05,
      "loss": 1.262,
      "step": 7024
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.378967523574829,
      "learning_rate": 1.6130722294987702e-05,
      "loss": 0.6684,
      "step": 7025
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.335232734680176,
      "learning_rate": 1.6130083378589914e-05,
      "loss": 1.3261,
      "step": 7026
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.788229465484619,
      "learning_rate": 1.6129444462192122e-05,
      "loss": 1.2774,
      "step": 7027
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.060944080352783,
      "learning_rate": 1.6128805545794334e-05,
      "loss": 0.9814,
      "step": 7028
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0102906227111816,
      "learning_rate": 1.6128166629396545e-05,
      "loss": 0.9203,
      "step": 7029
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.164600372314453,
      "learning_rate": 1.6127527712998757e-05,
      "loss": 0.98,
      "step": 7030
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.061230182647705,
      "learning_rate": 1.6126888796600965e-05,
      "loss": 1.2519,
      "step": 7031
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5041141510009766,
      "learning_rate": 1.6126249880203176e-05,
      "loss": 0.934,
      "step": 7032
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8051447868347168,
      "learning_rate": 1.6125610963805388e-05,
      "loss": 0.8393,
      "step": 7033
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.07342529296875,
      "learning_rate": 1.61249720474076e-05,
      "loss": 1.0749,
      "step": 7034
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9847413301467896,
      "learning_rate": 1.6124333131009808e-05,
      "loss": 0.9743,
      "step": 7035
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.4148640632629395,
      "learning_rate": 1.612369421461202e-05,
      "loss": 0.856,
      "step": 7036
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6342945098876953,
      "learning_rate": 1.612305529821423e-05,
      "loss": 0.9423,
      "step": 7037
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.444486141204834,
      "learning_rate": 1.6122416381816442e-05,
      "loss": 0.8249,
      "step": 7038
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8033279180526733,
      "learning_rate": 1.612177746541865e-05,
      "loss": 0.8569,
      "step": 7039
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5022521018981934,
      "learning_rate": 1.6121138549020862e-05,
      "loss": 0.8986,
      "step": 7040
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8753931522369385,
      "learning_rate": 1.6120499632623074e-05,
      "loss": 0.9619,
      "step": 7041
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.295060873031616,
      "learning_rate": 1.6119860716225285e-05,
      "loss": 1.1469,
      "step": 7042
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.970810890197754,
      "learning_rate": 1.6119221799827493e-05,
      "loss": 1.003,
      "step": 7043
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3432540893554688,
      "learning_rate": 1.6118582883429705e-05,
      "loss": 0.8349,
      "step": 7044
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.153616428375244,
      "learning_rate": 1.6117943967031916e-05,
      "loss": 0.5897,
      "step": 7045
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.12779426574707,
      "learning_rate": 1.6117305050634128e-05,
      "loss": 1.8576,
      "step": 7046
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9501298666000366,
      "learning_rate": 1.6116666134236336e-05,
      "loss": 1.3598,
      "step": 7047
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.747679591178894,
      "learning_rate": 1.6116027217838548e-05,
      "loss": 0.9817,
      "step": 7048
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8654842376708984,
      "learning_rate": 1.6115388301440756e-05,
      "loss": 0.8944,
      "step": 7049
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7407262325286865,
      "learning_rate": 1.611474938504297e-05,
      "loss": 0.9284,
      "step": 7050
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.4739139080047607,
      "learning_rate": 1.611411046864518e-05,
      "loss": 1.0552,
      "step": 7051
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8866976499557495,
      "learning_rate": 1.611347155224739e-05,
      "loss": 0.9964,
      "step": 7052
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5993636846542358,
      "learning_rate": 1.61128326358496e-05,
      "loss": 0.955,
      "step": 7053
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.87539529800415,
      "learning_rate": 1.611219371945181e-05,
      "loss": 1.236,
      "step": 7054
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8891912698745728,
      "learning_rate": 1.6111554803054022e-05,
      "loss": 0.9565,
      "step": 7055
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9956039190292358,
      "learning_rate": 1.6110915886656233e-05,
      "loss": 1.0881,
      "step": 7056
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.751185655593872,
      "learning_rate": 1.611027697025844e-05,
      "loss": 1.0212,
      "step": 7057
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9629039764404297,
      "learning_rate": 1.6109638053860653e-05,
      "loss": 0.8994,
      "step": 7058
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8762341737747192,
      "learning_rate": 1.6108999137462865e-05,
      "loss": 0.8739,
      "step": 7059
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.104433059692383,
      "learning_rate": 1.6108360221065076e-05,
      "loss": 0.8888,
      "step": 7060
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5222747325897217,
      "learning_rate": 1.6107721304667284e-05,
      "loss": 0.7891,
      "step": 7061
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.368332862854004,
      "learning_rate": 1.6107082388269496e-05,
      "loss": 0.8139,
      "step": 7062
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.7733893394470215,
      "learning_rate": 1.6106443471871707e-05,
      "loss": 1.5124,
      "step": 7063
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3579256534576416,
      "learning_rate": 1.610580455547392e-05,
      "loss": 1.0266,
      "step": 7064
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.5282206535339355,
      "learning_rate": 1.6105165639076127e-05,
      "loss": 0.8504,
      "step": 7065
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.5477800369262695,
      "learning_rate": 1.610452672267834e-05,
      "loss": 0.9298,
      "step": 7066
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6178076267242432,
      "learning_rate": 1.610388780628055e-05,
      "loss": 0.9088,
      "step": 7067
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.12080979347229,
      "learning_rate": 1.610324888988276e-05,
      "loss": 0.7876,
      "step": 7068
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.684839129447937,
      "learning_rate": 1.610260997348497e-05,
      "loss": 0.9785,
      "step": 7069
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8574098348617554,
      "learning_rate": 1.610197105708718e-05,
      "loss": 0.9185,
      "step": 7070
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8078532218933105,
      "learning_rate": 1.6101332140689393e-05,
      "loss": 1.0,
      "step": 7071
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.410591721534729,
      "learning_rate": 1.6100693224291604e-05,
      "loss": 0.3994,
      "step": 7072
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5723166465759277,
      "learning_rate": 1.6100054307893813e-05,
      "loss": 1.1622,
      "step": 7073
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.981269359588623,
      "learning_rate": 1.6099415391496024e-05,
      "loss": 0.8065,
      "step": 7074
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7819594144821167,
      "learning_rate": 1.6098776475098236e-05,
      "loss": 0.9771,
      "step": 7075
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.757843255996704,
      "learning_rate": 1.6098137558700447e-05,
      "loss": 0.9739,
      "step": 7076
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2004945278167725,
      "learning_rate": 1.6097498642302655e-05,
      "loss": 0.9772,
      "step": 7077
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1811094284057617,
      "learning_rate": 1.6096859725904867e-05,
      "loss": 1.0051,
      "step": 7078
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.644460916519165,
      "learning_rate": 1.609622080950708e-05,
      "loss": 1.0689,
      "step": 7079
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7277251482009888,
      "learning_rate": 1.609558189310929e-05,
      "loss": 0.4044,
      "step": 7080
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5067930221557617,
      "learning_rate": 1.6094942976711498e-05,
      "loss": 1.0494,
      "step": 7081
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.361064910888672,
      "learning_rate": 1.609430406031371e-05,
      "loss": 0.9855,
      "step": 7082
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7209609746932983,
      "learning_rate": 1.609366514391592e-05,
      "loss": 0.9683,
      "step": 7083
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6892783641815186,
      "learning_rate": 1.6093026227518133e-05,
      "loss": 0.8668,
      "step": 7084
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5346450805664062,
      "learning_rate": 1.609238731112034e-05,
      "loss": 1.0704,
      "step": 7085
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7094717025756836,
      "learning_rate": 1.6091748394722553e-05,
      "loss": 0.8521,
      "step": 7086
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.791977405548096,
      "learning_rate": 1.609110947832476e-05,
      "loss": 1.5744,
      "step": 7087
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.1279571056365967,
      "learning_rate": 1.6090470561926976e-05,
      "loss": 1.2219,
      "step": 7088
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.085341453552246,
      "learning_rate": 1.6089831645529184e-05,
      "loss": 0.9401,
      "step": 7089
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.075110912322998,
      "learning_rate": 1.6089192729131395e-05,
      "loss": 0.9663,
      "step": 7090
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.288782119750977,
      "learning_rate": 1.6088553812733604e-05,
      "loss": 1.4772,
      "step": 7091
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7508608102798462,
      "learning_rate": 1.6087914896335815e-05,
      "loss": 1.0258,
      "step": 7092
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0085792541503906,
      "learning_rate": 1.6087275979938027e-05,
      "loss": 0.9531,
      "step": 7093
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.987173318862915,
      "learning_rate": 1.6086637063540238e-05,
      "loss": 0.9072,
      "step": 7094
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.04848575592041,
      "learning_rate": 1.6085998147142446e-05,
      "loss": 0.8741,
      "step": 7095
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7458672523498535,
      "learning_rate": 1.6085359230744658e-05,
      "loss": 0.9291,
      "step": 7096
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8549449443817139,
      "learning_rate": 1.608472031434687e-05,
      "loss": 0.8255,
      "step": 7097
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.759840488433838,
      "learning_rate": 1.608408139794908e-05,
      "loss": 0.8355,
      "step": 7098
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9980573654174805,
      "learning_rate": 1.608344248155129e-05,
      "loss": 1.2432,
      "step": 7099
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0769214630126953,
      "learning_rate": 1.60828035651535e-05,
      "loss": 0.9142,
      "step": 7100
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.339144706726074,
      "learning_rate": 1.6082164648755712e-05,
      "loss": 1.0941,
      "step": 7101
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.697190999984741,
      "learning_rate": 1.6081525732357924e-05,
      "loss": 1.0757,
      "step": 7102
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1243457794189453,
      "learning_rate": 1.6080886815960132e-05,
      "loss": 0.9229,
      "step": 7103
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.629070520401001,
      "learning_rate": 1.6080247899562343e-05,
      "loss": 0.8973,
      "step": 7104
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9793745279312134,
      "learning_rate": 1.6079608983164555e-05,
      "loss": 0.4132,
      "step": 7105
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6902308464050293,
      "learning_rate": 1.6078970066766767e-05,
      "loss": 0.8607,
      "step": 7106
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.887136459350586,
      "learning_rate": 1.6078331150368975e-05,
      "loss": 0.8605,
      "step": 7107
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.4788198471069336,
      "learning_rate": 1.6077692233971186e-05,
      "loss": 0.9611,
      "step": 7108
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.950669288635254,
      "learning_rate": 1.6077053317573398e-05,
      "loss": 0.9043,
      "step": 7109
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.173480987548828,
      "learning_rate": 1.607641440117561e-05,
      "loss": 1.1849,
      "step": 7110
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0770344734191895,
      "learning_rate": 1.6075775484777818e-05,
      "loss": 1.0588,
      "step": 7111
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9968708753585815,
      "learning_rate": 1.607513656838003e-05,
      "loss": 0.9487,
      "step": 7112
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.302522659301758,
      "learning_rate": 1.607449765198224e-05,
      "loss": 1.1738,
      "step": 7113
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.5821313858032227,
      "learning_rate": 1.6073858735584452e-05,
      "loss": 0.8701,
      "step": 7114
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3863344192504883,
      "learning_rate": 1.607321981918666e-05,
      "loss": 1.0528,
      "step": 7115
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.7660772800445557,
      "learning_rate": 1.6072580902788872e-05,
      "loss": 0.8884,
      "step": 7116
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6903358697891235,
      "learning_rate": 1.6071941986391083e-05,
      "loss": 0.9423,
      "step": 7117
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.65762710571289,
      "learning_rate": 1.6071303069993295e-05,
      "loss": 1.5512,
      "step": 7118
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8180047273635864,
      "learning_rate": 1.6070664153595503e-05,
      "loss": 0.993,
      "step": 7119
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.473217248916626,
      "learning_rate": 1.6070025237197715e-05,
      "loss": 1.056,
      "step": 7120
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9934985637664795,
      "learning_rate": 1.6069386320799926e-05,
      "loss": 0.9355,
      "step": 7121
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.670044422149658,
      "learning_rate": 1.6068747404402138e-05,
      "loss": 1.5418,
      "step": 7122
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.1970601081848145,
      "learning_rate": 1.6068108488004346e-05,
      "loss": 1.1076,
      "step": 7123
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1767728328704834,
      "learning_rate": 1.6067469571606557e-05,
      "loss": 1.2,
      "step": 7124
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.43546724319458,
      "learning_rate": 1.6066830655208766e-05,
      "loss": 1.0934,
      "step": 7125
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9186186790466309,
      "learning_rate": 1.606619173881098e-05,
      "loss": 1.3486,
      "step": 7126
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8268121480941772,
      "learning_rate": 1.606555282241319e-05,
      "loss": 0.9214,
      "step": 7127
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8674578666687012,
      "learning_rate": 1.60649139060154e-05,
      "loss": 0.8014,
      "step": 7128
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.3056371212005615,
      "learning_rate": 1.606427498961761e-05,
      "loss": 1.0637,
      "step": 7129
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8655617237091064,
      "learning_rate": 1.606363607321982e-05,
      "loss": 0.9364,
      "step": 7130
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3056142330169678,
      "learning_rate": 1.606299715682203e-05,
      "loss": 1.0376,
      "step": 7131
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6664544343948364,
      "learning_rate": 1.6062358240424243e-05,
      "loss": 1.0346,
      "step": 7132
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.678911566734314,
      "learning_rate": 1.606171932402645e-05,
      "loss": 0.8768,
      "step": 7133
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.4820523262023926,
      "learning_rate": 1.6061080407628663e-05,
      "loss": 1.1132,
      "step": 7134
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9095793962478638,
      "learning_rate": 1.6060441491230874e-05,
      "loss": 0.9091,
      "step": 7135
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8606103658676147,
      "learning_rate": 1.6059802574833086e-05,
      "loss": 0.9265,
      "step": 7136
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2684314250946045,
      "learning_rate": 1.6059163658435294e-05,
      "loss": 0.9885,
      "step": 7137
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8676961660385132,
      "learning_rate": 1.6058524742037506e-05,
      "loss": 0.82,
      "step": 7138
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.6492629051208496,
      "learning_rate": 1.6057885825639717e-05,
      "loss": 0.8089,
      "step": 7139
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9389736652374268,
      "learning_rate": 1.605724690924193e-05,
      "loss": 0.9415,
      "step": 7140
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6886805295944214,
      "learning_rate": 1.6056607992844137e-05,
      "loss": 0.9616,
      "step": 7141
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.905017614364624,
      "learning_rate": 1.605596907644635e-05,
      "loss": 0.6651,
      "step": 7142
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.723792552947998,
      "learning_rate": 1.605533016004856e-05,
      "loss": 0.8874,
      "step": 7143
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7639483213424683,
      "learning_rate": 1.605469124365077e-05,
      "loss": 0.9312,
      "step": 7144
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.682036876678467,
      "learning_rate": 1.605405232725298e-05,
      "loss": 0.6323,
      "step": 7145
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8983855247497559,
      "learning_rate": 1.605341341085519e-05,
      "loss": 0.9291,
      "step": 7146
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9270424842834473,
      "learning_rate": 1.6052774494457403e-05,
      "loss": 0.8819,
      "step": 7147
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6898066997528076,
      "learning_rate": 1.6052135578059614e-05,
      "loss": 0.7612,
      "step": 7148
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.270644426345825,
      "learning_rate": 1.6051496661661822e-05,
      "loss": 0.9619,
      "step": 7149
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.207080125808716,
      "learning_rate": 1.6050857745264034e-05,
      "loss": 0.9362,
      "step": 7150
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.247200012207031,
      "learning_rate": 1.6050218828866246e-05,
      "loss": 1.0494,
      "step": 7151
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8212206363677979,
      "learning_rate": 1.6049579912468457e-05,
      "loss": 0.9248,
      "step": 7152
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3276731967926025,
      "learning_rate": 1.6048940996070665e-05,
      "loss": 0.901,
      "step": 7153
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.740556240081787,
      "learning_rate": 1.6048302079672873e-05,
      "loss": 0.988,
      "step": 7154
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7735774517059326,
      "learning_rate": 1.604766316327509e-05,
      "loss": 0.9931,
      "step": 7155
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9896750450134277,
      "learning_rate": 1.6047024246877297e-05,
      "loss": 1.0455,
      "step": 7156
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.342284917831421,
      "learning_rate": 1.6046385330479508e-05,
      "loss": 1.0941,
      "step": 7157
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9061217308044434,
      "learning_rate": 1.6045746414081716e-05,
      "loss": 0.8882,
      "step": 7158
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.497391700744629,
      "learning_rate": 1.6045107497683928e-05,
      "loss": 1.1084,
      "step": 7159
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2473504543304443,
      "learning_rate": 1.604446858128614e-05,
      "loss": 0.7418,
      "step": 7160
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5037875175476074,
      "learning_rate": 1.604382966488835e-05,
      "loss": 1.0966,
      "step": 7161
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.829027533531189,
      "learning_rate": 1.604319074849056e-05,
      "loss": 0.8226,
      "step": 7162
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.6404385566711426,
      "learning_rate": 1.604255183209277e-05,
      "loss": 1.1527,
      "step": 7163
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7395408153533936,
      "learning_rate": 1.6041912915694982e-05,
      "loss": 0.9758,
      "step": 7164
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.3449506759643555,
      "learning_rate": 1.6041273999297194e-05,
      "loss": 1.3403,
      "step": 7165
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.7833516597747803,
      "learning_rate": 1.6040635082899402e-05,
      "loss": 1.305,
      "step": 7166
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.263030767440796,
      "learning_rate": 1.6039996166501613e-05,
      "loss": 0.8297,
      "step": 7167
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.455833911895752,
      "learning_rate": 1.6039357250103825e-05,
      "loss": 1.0499,
      "step": 7168
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.279935359954834,
      "learning_rate": 1.6038718333706036e-05,
      "loss": 1.3991,
      "step": 7169
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3457837104797363,
      "learning_rate": 1.6038079417308245e-05,
      "loss": 0.8765,
      "step": 7170
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0127155780792236,
      "learning_rate": 1.6037440500910456e-05,
      "loss": 0.8883,
      "step": 7171
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1815402507781982,
      "learning_rate": 1.6036801584512668e-05,
      "loss": 0.9983,
      "step": 7172
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7551395893096924,
      "learning_rate": 1.603616266811488e-05,
      "loss": 0.8693,
      "step": 7173
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7461116313934326,
      "learning_rate": 1.6035523751717087e-05,
      "loss": 0.9779,
      "step": 7174
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.9947144985198975,
      "learning_rate": 1.60348848353193e-05,
      "loss": 1.0991,
      "step": 7175
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.080904006958008,
      "learning_rate": 1.603424591892151e-05,
      "loss": 1.333,
      "step": 7176
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.306826591491699,
      "learning_rate": 1.6033607002523722e-05,
      "loss": 1.096,
      "step": 7177
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.974327564239502,
      "learning_rate": 1.603296808612593e-05,
      "loss": 0.3897,
      "step": 7178
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7216829061508179,
      "learning_rate": 1.6032329169728142e-05,
      "loss": 0.8841,
      "step": 7179
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9916653633117676,
      "learning_rate": 1.6031690253330353e-05,
      "loss": 1.0573,
      "step": 7180
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2435078620910645,
      "learning_rate": 1.6031051336932565e-05,
      "loss": 0.9744,
      "step": 7181
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7835988998413086,
      "learning_rate": 1.6030412420534773e-05,
      "loss": 0.9529,
      "step": 7182
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9151721000671387,
      "learning_rate": 1.6029773504136985e-05,
      "loss": 1.0508,
      "step": 7183
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1384105682373047,
      "learning_rate": 1.6029134587739196e-05,
      "loss": 1.0458,
      "step": 7184
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.948890209197998,
      "learning_rate": 1.6028495671341408e-05,
      "loss": 0.9595,
      "step": 7185
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1959540843963623,
      "learning_rate": 1.6027856754943616e-05,
      "loss": 0.9177,
      "step": 7186
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.776064395904541,
      "learning_rate": 1.6027217838545827e-05,
      "loss": 1.1229,
      "step": 7187
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9229137897491455,
      "learning_rate": 1.602657892214804e-05,
      "loss": 0.9071,
      "step": 7188
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7127954959869385,
      "learning_rate": 1.602594000575025e-05,
      "loss": 1.0189,
      "step": 7189
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8717365264892578,
      "learning_rate": 1.602530108935246e-05,
      "loss": 0.8749,
      "step": 7190
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0604865550994873,
      "learning_rate": 1.602466217295467e-05,
      "loss": 1.2649,
      "step": 7191
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7160142660140991,
      "learning_rate": 1.602402325655688e-05,
      "loss": 0.9135,
      "step": 7192
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5862820148468018,
      "learning_rate": 1.6023384340159093e-05,
      "loss": 0.8222,
      "step": 7193
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.366283655166626,
      "learning_rate": 1.60227454237613e-05,
      "loss": 0.7824,
      "step": 7194
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8787946701049805,
      "learning_rate": 1.6022106507363513e-05,
      "loss": 0.8081,
      "step": 7195
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1797499656677246,
      "learning_rate": 1.602146759096572e-05,
      "loss": 0.8668,
      "step": 7196
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.045271396636963,
      "learning_rate": 1.6020828674567933e-05,
      "loss": 0.9742,
      "step": 7197
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8904768228530884,
      "learning_rate": 1.6020189758170144e-05,
      "loss": 1.4247,
      "step": 7198
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.04136323928833,
      "learning_rate": 1.6019550841772356e-05,
      "loss": 0.8086,
      "step": 7199
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.6093268394470215,
      "learning_rate": 1.6018911925374564e-05,
      "loss": 0.705,
      "step": 7200
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8506765365600586,
      "learning_rate": 1.6018273008976775e-05,
      "loss": 0.8614,
      "step": 7201
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8313173055648804,
      "learning_rate": 1.6017634092578987e-05,
      "loss": 0.9149,
      "step": 7202
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.598099708557129,
      "learning_rate": 1.60169951761812e-05,
      "loss": 0.9289,
      "step": 7203
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.4374032020568848,
      "learning_rate": 1.6016356259783407e-05,
      "loss": 0.8675,
      "step": 7204
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.4675815105438232,
      "learning_rate": 1.6015717343385618e-05,
      "loss": 0.8744,
      "step": 7205
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.089871644973755,
      "learning_rate": 1.601507842698783e-05,
      "loss": 0.98,
      "step": 7206
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6490933895111084,
      "learning_rate": 1.601443951059004e-05,
      "loss": 0.8432,
      "step": 7207
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8077601194381714,
      "learning_rate": 1.601380059419225e-05,
      "loss": 0.4061,
      "step": 7208
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6546909809112549,
      "learning_rate": 1.601316167779446e-05,
      "loss": 0.4507,
      "step": 7209
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2051146030426025,
      "learning_rate": 1.6012522761396673e-05,
      "loss": 1.0469,
      "step": 7210
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2366695404052734,
      "learning_rate": 1.6011883844998884e-05,
      "loss": 0.875,
      "step": 7211
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1023237705230713,
      "learning_rate": 1.6011244928601092e-05,
      "loss": 1.0395,
      "step": 7212
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.6456832885742188,
      "learning_rate": 1.6010606012203304e-05,
      "loss": 1.1173,
      "step": 7213
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.611101746559143,
      "learning_rate": 1.6009967095805515e-05,
      "loss": 0.412,
      "step": 7214
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0299384593963623,
      "learning_rate": 1.6009328179407727e-05,
      "loss": 1.0048,
      "step": 7215
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.462947130203247,
      "learning_rate": 1.6008689263009935e-05,
      "loss": 1.1108,
      "step": 7216
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8774170875549316,
      "learning_rate": 1.6008050346612147e-05,
      "loss": 1.0082,
      "step": 7217
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8518545627593994,
      "learning_rate": 1.6007411430214358e-05,
      "loss": 1.127,
      "step": 7218
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1578733921051025,
      "learning_rate": 1.600677251381657e-05,
      "loss": 0.9087,
      "step": 7219
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5458052158355713,
      "learning_rate": 1.6006133597418778e-05,
      "loss": 0.8596,
      "step": 7220
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.255833625793457,
      "learning_rate": 1.600549468102099e-05,
      "loss": 0.4498,
      "step": 7221
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.419527292251587,
      "learning_rate": 1.60048557646232e-05,
      "loss": 0.8621,
      "step": 7222
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8410499095916748,
      "learning_rate": 1.6004216848225413e-05,
      "loss": 0.9442,
      "step": 7223
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1649467945098877,
      "learning_rate": 1.600357793182762e-05,
      "loss": 1.0263,
      "step": 7224
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8197473287582397,
      "learning_rate": 1.6002939015429832e-05,
      "loss": 0.3699,
      "step": 7225
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.333134174346924,
      "learning_rate": 1.600230009903204e-05,
      "loss": 1.8554,
      "step": 7226
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.746734857559204,
      "learning_rate": 1.6001661182634255e-05,
      "loss": 1.0864,
      "step": 7227
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.155377149581909,
      "learning_rate": 1.6001022266236464e-05,
      "loss": 0.9977,
      "step": 7228
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.702451467514038,
      "learning_rate": 1.6000383349838675e-05,
      "loss": 0.8638,
      "step": 7229
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.183697700500488,
      "learning_rate": 1.5999744433440883e-05,
      "loss": 0.9093,
      "step": 7230
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.751394271850586,
      "learning_rate": 1.5999105517043095e-05,
      "loss": 0.7973,
      "step": 7231
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9594347476959229,
      "learning_rate": 1.5998466600645306e-05,
      "loss": 0.8657,
      "step": 7232
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.702329158782959,
      "learning_rate": 1.5997827684247518e-05,
      "loss": 0.9372,
      "step": 7233
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.675732135772705,
      "learning_rate": 1.5997188767849726e-05,
      "loss": 0.8619,
      "step": 7234
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9772822856903076,
      "learning_rate": 1.5996549851451938e-05,
      "loss": 1.0407,
      "step": 7235
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.447875499725342,
      "learning_rate": 1.599591093505415e-05,
      "loss": 1.0306,
      "step": 7236
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7505264282226562,
      "learning_rate": 1.599527201865636e-05,
      "loss": 0.8457,
      "step": 7237
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.423609256744385,
      "learning_rate": 1.599463310225857e-05,
      "loss": 1.3762,
      "step": 7238
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.134052276611328,
      "learning_rate": 1.599399418586078e-05,
      "loss": 0.9989,
      "step": 7239
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.860381841659546,
      "learning_rate": 1.5993355269462992e-05,
      "loss": 1.2023,
      "step": 7240
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0245895385742188,
      "learning_rate": 1.5992716353065203e-05,
      "loss": 1.0095,
      "step": 7241
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.178748607635498,
      "learning_rate": 1.599207743666741e-05,
      "loss": 0.9457,
      "step": 7242
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.7174649238586426,
      "learning_rate": 1.5991438520269623e-05,
      "loss": 0.9432,
      "step": 7243
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5315568447113037,
      "learning_rate": 1.5990799603871835e-05,
      "loss": 0.8943,
      "step": 7244
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.267502784729004,
      "learning_rate": 1.5990160687474046e-05,
      "loss": 0.5083,
      "step": 7245
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6880595684051514,
      "learning_rate": 1.5989521771076254e-05,
      "loss": 0.8382,
      "step": 7246
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8057515621185303,
      "learning_rate": 1.5988882854678466e-05,
      "loss": 1.0131,
      "step": 7247
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.455840587615967,
      "learning_rate": 1.5988243938280678e-05,
      "loss": 0.9612,
      "step": 7248
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7017593383789062,
      "learning_rate": 1.598760502188289e-05,
      "loss": 0.8281,
      "step": 7249
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3188321590423584,
      "learning_rate": 1.5986966105485097e-05,
      "loss": 1.047,
      "step": 7250
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.391056537628174,
      "learning_rate": 1.598632718908731e-05,
      "loss": 0.6164,
      "step": 7251
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8566395044326782,
      "learning_rate": 1.598568827268952e-05,
      "loss": 0.9695,
      "step": 7252
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8359606266021729,
      "learning_rate": 1.5985049356291732e-05,
      "loss": 0.9692,
      "step": 7253
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.833665370941162,
      "learning_rate": 1.598441043989394e-05,
      "loss": 0.9762,
      "step": 7254
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.124476671218872,
      "learning_rate": 1.598377152349615e-05,
      "loss": 1.0022,
      "step": 7255
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.747668981552124,
      "learning_rate": 1.5983132607098363e-05,
      "loss": 0.6514,
      "step": 7256
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3960070610046387,
      "learning_rate": 1.5982493690700575e-05,
      "loss": 1.0221,
      "step": 7257
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7759495973587036,
      "learning_rate": 1.5981854774302783e-05,
      "loss": 1.02,
      "step": 7258
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.414414405822754,
      "learning_rate": 1.5981215857904994e-05,
      "loss": 1.133,
      "step": 7259
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.6059916019439697,
      "learning_rate": 1.5980576941507206e-05,
      "loss": 0.5496,
      "step": 7260
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3550362586975098,
      "learning_rate": 1.5979938025109417e-05,
      "loss": 0.9485,
      "step": 7261
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8681621551513672,
      "learning_rate": 1.5979299108711626e-05,
      "loss": 1.2073,
      "step": 7262
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.784496784210205,
      "learning_rate": 1.5978660192313837e-05,
      "loss": 0.3799,
      "step": 7263
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0124123096466064,
      "learning_rate": 1.5978021275916045e-05,
      "loss": 1.2158,
      "step": 7264
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8458706140518188,
      "learning_rate": 1.597738235951826e-05,
      "loss": 0.8904,
      "step": 7265
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2529139518737793,
      "learning_rate": 1.597674344312047e-05,
      "loss": 0.9616,
      "step": 7266
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6152849197387695,
      "learning_rate": 1.597610452672268e-05,
      "loss": 1.0564,
      "step": 7267
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.374579429626465,
      "learning_rate": 1.5975465610324888e-05,
      "loss": 1.3948,
      "step": 7268
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.434541702270508,
      "learning_rate": 1.59748266939271e-05,
      "loss": 0.8045,
      "step": 7269
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1869044303894043,
      "learning_rate": 1.597418777752931e-05,
      "loss": 0.9393,
      "step": 7270
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9507495164871216,
      "learning_rate": 1.5973548861131523e-05,
      "loss": 1.3401,
      "step": 7271
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4240033626556396,
      "learning_rate": 1.597290994473373e-05,
      "loss": 1.0115,
      "step": 7272
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2243857383728027,
      "learning_rate": 1.5972271028335942e-05,
      "loss": 1.0165,
      "step": 7273
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.46803617477417,
      "learning_rate": 1.5971632111938154e-05,
      "loss": 0.8184,
      "step": 7274
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.787715196609497,
      "learning_rate": 1.5970993195540366e-05,
      "loss": 1.137,
      "step": 7275
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3345749378204346,
      "learning_rate": 1.5970354279142574e-05,
      "loss": 0.9542,
      "step": 7276
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.177572250366211,
      "learning_rate": 1.5969715362744785e-05,
      "loss": 1.1193,
      "step": 7277
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.375424385070801,
      "learning_rate": 1.5969076446346997e-05,
      "loss": 1.0962,
      "step": 7278
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.690269947052002,
      "learning_rate": 1.596843752994921e-05,
      "loss": 0.9193,
      "step": 7279
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3896963596343994,
      "learning_rate": 1.5967798613551417e-05,
      "loss": 0.4477,
      "step": 7280
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.074915647506714,
      "learning_rate": 1.5967159697153628e-05,
      "loss": 1.0837,
      "step": 7281
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.692239761352539,
      "learning_rate": 1.596652078075584e-05,
      "loss": 0.7944,
      "step": 7282
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.750677227973938,
      "learning_rate": 1.596588186435805e-05,
      "loss": 0.9041,
      "step": 7283
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.313385009765625,
      "learning_rate": 1.596524294796026e-05,
      "loss": 1.1847,
      "step": 7284
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.872950553894043,
      "learning_rate": 1.596460403156247e-05,
      "loss": 0.7646,
      "step": 7285
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4899604320526123,
      "learning_rate": 1.5963965115164682e-05,
      "loss": 0.7789,
      "step": 7286
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.723358392715454,
      "learning_rate": 1.5963326198766894e-05,
      "loss": 0.8805,
      "step": 7287
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6211004257202148,
      "learning_rate": 1.5962687282369102e-05,
      "loss": 0.9171,
      "step": 7288
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.366959810256958,
      "learning_rate": 1.5962048365971314e-05,
      "loss": 1.0801,
      "step": 7289
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5345011949539185,
      "learning_rate": 1.5961409449573525e-05,
      "loss": 0.9474,
      "step": 7290
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9449520111083984,
      "learning_rate": 1.5960770533175737e-05,
      "loss": 0.9716,
      "step": 7291
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8223719596862793,
      "learning_rate": 1.5960131616777945e-05,
      "loss": 1.1514,
      "step": 7292
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.304020404815674,
      "learning_rate": 1.5959492700380156e-05,
      "loss": 1.0866,
      "step": 7293
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1532890796661377,
      "learning_rate": 1.5958853783982368e-05,
      "loss": 0.9004,
      "step": 7294
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3943090438842773,
      "learning_rate": 1.595821486758458e-05,
      "loss": 0.4537,
      "step": 7295
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7736321687698364,
      "learning_rate": 1.5957575951186788e-05,
      "loss": 0.941,
      "step": 7296
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0137460231781006,
      "learning_rate": 1.5956937034789e-05,
      "loss": 0.9437,
      "step": 7297
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.463264226913452,
      "learning_rate": 1.595629811839121e-05,
      "loss": 0.845,
      "step": 7298
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9938346147537231,
      "learning_rate": 1.5955659201993422e-05,
      "loss": 0.8605,
      "step": 7299
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.705967426300049,
      "learning_rate": 1.595502028559563e-05,
      "loss": 0.9743,
      "step": 7300
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5475125312805176,
      "learning_rate": 1.5954381369197842e-05,
      "loss": 0.8152,
      "step": 7301
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0446932315826416,
      "learning_rate": 1.595374245280005e-05,
      "loss": 0.9559,
      "step": 7302
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.199789524078369,
      "learning_rate": 1.5953103536402265e-05,
      "loss": 1.0491,
      "step": 7303
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7057712078094482,
      "learning_rate": 1.5952464620004473e-05,
      "loss": 0.9401,
      "step": 7304
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.399829626083374,
      "learning_rate": 1.5951825703606685e-05,
      "loss": 1.0102,
      "step": 7305
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.028569459915161,
      "learning_rate": 1.5951186787208893e-05,
      "loss": 0.918,
      "step": 7306
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.00028133392334,
      "learning_rate": 1.5950547870811105e-05,
      "loss": 0.9678,
      "step": 7307
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.3718202114105225,
      "learning_rate": 1.5949908954413316e-05,
      "loss": 1.3486,
      "step": 7308
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.6883509159088135,
      "learning_rate": 1.5949270038015528e-05,
      "loss": 1.1033,
      "step": 7309
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.772452712059021,
      "learning_rate": 1.5948631121617736e-05,
      "loss": 0.97,
      "step": 7310
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.834849238395691,
      "learning_rate": 1.5947992205219947e-05,
      "loss": 0.9658,
      "step": 7311
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0803611278533936,
      "learning_rate": 1.594735328882216e-05,
      "loss": 1.0841,
      "step": 7312
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.222277879714966,
      "learning_rate": 1.594671437242437e-05,
      "loss": 1.0183,
      "step": 7313
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6960911750793457,
      "learning_rate": 1.594607545602658e-05,
      "loss": 0.8531,
      "step": 7314
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.520127296447754,
      "learning_rate": 1.594543653962879e-05,
      "loss": 1.146,
      "step": 7315
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9519823789596558,
      "learning_rate": 1.5944797623231002e-05,
      "loss": 0.8807,
      "step": 7316
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.135924816131592,
      "learning_rate": 1.5944158706833213e-05,
      "loss": 1.0131,
      "step": 7317
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5273796319961548,
      "learning_rate": 1.594351979043542e-05,
      "loss": 0.3793,
      "step": 7318
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.7632598876953125,
      "learning_rate": 1.5942880874037633e-05,
      "loss": 1.0917,
      "step": 7319
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.785542368888855,
      "learning_rate": 1.5942241957639845e-05,
      "loss": 0.8974,
      "step": 7320
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.706200122833252,
      "learning_rate": 1.5941603041242056e-05,
      "loss": 0.9676,
      "step": 7321
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9371980428695679,
      "learning_rate": 1.5940964124844264e-05,
      "loss": 1.2836,
      "step": 7322
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1681113243103027,
      "learning_rate": 1.5940325208446476e-05,
      "loss": 0.6874,
      "step": 7323
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.126624822616577,
      "learning_rate": 1.5939686292048687e-05,
      "loss": 0.9239,
      "step": 7324
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0807077884674072,
      "learning_rate": 1.59390473756509e-05,
      "loss": 1.0388,
      "step": 7325
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6012448072433472,
      "learning_rate": 1.5938408459253107e-05,
      "loss": 0.9104,
      "step": 7326
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8504140377044678,
      "learning_rate": 1.593776954285532e-05,
      "loss": 0.7514,
      "step": 7327
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.155569553375244,
      "learning_rate": 1.593713062645753e-05,
      "loss": 1.6631,
      "step": 7328
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1530020236968994,
      "learning_rate": 1.593649171005974e-05,
      "loss": 0.7276,
      "step": 7329
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.614508032798767,
      "learning_rate": 1.593585279366195e-05,
      "loss": 0.4535,
      "step": 7330
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4631682634353638,
      "learning_rate": 1.593521387726416e-05,
      "loss": 0.3836,
      "step": 7331
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5826034545898438,
      "learning_rate": 1.5934574960866373e-05,
      "loss": 0.7614,
      "step": 7332
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.057852029800415,
      "learning_rate": 1.5933936044468584e-05,
      "loss": 0.6906,
      "step": 7333
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.2996745109558105,
      "learning_rate": 1.5933297128070793e-05,
      "loss": 1.5318,
      "step": 7334
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0133981704711914,
      "learning_rate": 1.5932658211673004e-05,
      "loss": 1.0098,
      "step": 7335
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6048834323883057,
      "learning_rate": 1.5932019295275212e-05,
      "loss": 1.1323,
      "step": 7336
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8961231708526611,
      "learning_rate": 1.5931380378877427e-05,
      "loss": 1.0652,
      "step": 7337
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.95908784866333,
      "learning_rate": 1.5930741462479635e-05,
      "loss": 0.9888,
      "step": 7338
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1618969440460205,
      "learning_rate": 1.5930102546081847e-05,
      "loss": 1.021,
      "step": 7339
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8400696516036987,
      "learning_rate": 1.5929463629684055e-05,
      "loss": 1.0511,
      "step": 7340
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6304476261138916,
      "learning_rate": 1.5928824713286267e-05,
      "loss": 0.8361,
      "step": 7341
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5436793565750122,
      "learning_rate": 1.5928185796888478e-05,
      "loss": 0.9049,
      "step": 7342
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.974191427230835,
      "learning_rate": 1.592754688049069e-05,
      "loss": 0.6542,
      "step": 7343
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6637994050979614,
      "learning_rate": 1.5926907964092898e-05,
      "loss": 0.4155,
      "step": 7344
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8075164556503296,
      "learning_rate": 1.592626904769511e-05,
      "loss": 0.8751,
      "step": 7345
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7917046546936035,
      "learning_rate": 1.592563013129732e-05,
      "loss": 0.9091,
      "step": 7346
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.018785238265991,
      "learning_rate": 1.5924991214899533e-05,
      "loss": 0.8625,
      "step": 7347
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8389711380004883,
      "learning_rate": 1.592435229850174e-05,
      "loss": 0.8984,
      "step": 7348
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.755401134490967,
      "learning_rate": 1.5923713382103952e-05,
      "loss": 1.2538,
      "step": 7349
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.931844711303711,
      "learning_rate": 1.5923074465706164e-05,
      "loss": 1.2844,
      "step": 7350
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7595982551574707,
      "learning_rate": 1.5922435549308375e-05,
      "loss": 0.8916,
      "step": 7351
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.0290982723236084,
      "learning_rate": 1.5921796632910584e-05,
      "loss": 0.7182,
      "step": 7352
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9242860078811646,
      "learning_rate": 1.5921157716512795e-05,
      "loss": 1.0609,
      "step": 7353
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.1311328411102295,
      "learning_rate": 1.5920518800115007e-05,
      "loss": 0.978,
      "step": 7354
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6710948944091797,
      "learning_rate": 1.5919879883717218e-05,
      "loss": 0.8657,
      "step": 7355
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.184767246246338,
      "learning_rate": 1.5919240967319426e-05,
      "loss": 0.6696,
      "step": 7356
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6677114963531494,
      "learning_rate": 1.5918602050921638e-05,
      "loss": 1.0337,
      "step": 7357
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8997137546539307,
      "learning_rate": 1.591796313452385e-05,
      "loss": 0.8584,
      "step": 7358
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0009028911590576,
      "learning_rate": 1.591732421812606e-05,
      "loss": 1.0276,
      "step": 7359
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0362372398376465,
      "learning_rate": 1.591668530172827e-05,
      "loss": 1.0196,
      "step": 7360
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7804861068725586,
      "learning_rate": 1.591604638533048e-05,
      "loss": 0.9806,
      "step": 7361
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.7792320251464844,
      "learning_rate": 1.5915407468932692e-05,
      "loss": 1.0744,
      "step": 7362
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7312040328979492,
      "learning_rate": 1.5914768552534904e-05,
      "loss": 1.0196,
      "step": 7363
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2822916507720947,
      "learning_rate": 1.5914129636137112e-05,
      "loss": 0.9868,
      "step": 7364
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7781620025634766,
      "learning_rate": 1.5913490719739323e-05,
      "loss": 0.8416,
      "step": 7365
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9030503034591675,
      "learning_rate": 1.5912851803341535e-05,
      "loss": 1.0051,
      "step": 7366
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8491737842559814,
      "learning_rate": 1.5912212886943747e-05,
      "loss": 0.9374,
      "step": 7367
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8618274927139282,
      "learning_rate": 1.5911573970545955e-05,
      "loss": 0.9262,
      "step": 7368
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2841546535491943,
      "learning_rate": 1.5910935054148166e-05,
      "loss": 1.4203,
      "step": 7369
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.034984588623047,
      "learning_rate": 1.5910296137750378e-05,
      "loss": 0.8977,
      "step": 7370
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4904886484146118,
      "learning_rate": 1.590965722135259e-05,
      "loss": 0.792,
      "step": 7371
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.783860206604004,
      "learning_rate": 1.5909018304954798e-05,
      "loss": 0.981,
      "step": 7372
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0669026374816895,
      "learning_rate": 1.590837938855701e-05,
      "loss": 0.936,
      "step": 7373
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7812776565551758,
      "learning_rate": 1.5907740472159217e-05,
      "loss": 0.4548,
      "step": 7374
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0572197437286377,
      "learning_rate": 1.5907101555761432e-05,
      "loss": 0.9525,
      "step": 7375
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7491716146469116,
      "learning_rate": 1.590646263936364e-05,
      "loss": 0.4416,
      "step": 7376
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6081938743591309,
      "learning_rate": 1.5905823722965852e-05,
      "loss": 0.8749,
      "step": 7377
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2427315711975098,
      "learning_rate": 1.590518480656806e-05,
      "loss": 0.9926,
      "step": 7378
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9159094095230103,
      "learning_rate": 1.590454589017027e-05,
      "loss": 0.9107,
      "step": 7379
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.142406463623047,
      "learning_rate": 1.5903906973772483e-05,
      "loss": 0.6111,
      "step": 7380
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.069993495941162,
      "learning_rate": 1.5903268057374695e-05,
      "loss": 0.9951,
      "step": 7381
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.328369379043579,
      "learning_rate": 1.5902629140976903e-05,
      "loss": 0.9055,
      "step": 7382
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5036253929138184,
      "learning_rate": 1.5901990224579114e-05,
      "loss": 0.9535,
      "step": 7383
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6339982748031616,
      "learning_rate": 1.5901351308181326e-05,
      "loss": 0.8563,
      "step": 7384
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.054708957672119,
      "learning_rate": 1.5900712391783537e-05,
      "loss": 0.9604,
      "step": 7385
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.844789505004883,
      "learning_rate": 1.5900073475385746e-05,
      "loss": 1.1806,
      "step": 7386
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.762325406074524,
      "learning_rate": 1.5899434558987957e-05,
      "loss": 0.9397,
      "step": 7387
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6968780755996704,
      "learning_rate": 1.589879564259017e-05,
      "loss": 0.9334,
      "step": 7388
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4070632457733154,
      "learning_rate": 1.589815672619238e-05,
      "loss": 0.8317,
      "step": 7389
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.765309453010559,
      "learning_rate": 1.589751780979459e-05,
      "loss": 0.9619,
      "step": 7390
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.727105736732483,
      "learning_rate": 1.58968788933968e-05,
      "loss": 1.0165,
      "step": 7391
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6176187992095947,
      "learning_rate": 1.589623997699901e-05,
      "loss": 0.9415,
      "step": 7392
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.599658966064453,
      "learning_rate": 1.5895601060601223e-05,
      "loss": 1.0459,
      "step": 7393
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7422693967819214,
      "learning_rate": 1.589496214420343e-05,
      "loss": 0.9895,
      "step": 7394
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1445610523223877,
      "learning_rate": 1.5894323227805643e-05,
      "loss": 0.8386,
      "step": 7395
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8623250722885132,
      "learning_rate": 1.5893684311407854e-05,
      "loss": 1.0292,
      "step": 7396
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4671645164489746,
      "learning_rate": 1.5893045395010066e-05,
      "loss": 1.2393,
      "step": 7397
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7810304164886475,
      "learning_rate": 1.5892406478612274e-05,
      "loss": 0.9776,
      "step": 7398
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.067711353302002,
      "learning_rate": 1.5891767562214486e-05,
      "loss": 1.0214,
      "step": 7399
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.021117210388184,
      "learning_rate": 1.5891128645816697e-05,
      "loss": 1.3694,
      "step": 7400
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9529783725738525,
      "learning_rate": 1.589048972941891e-05,
      "loss": 1.1028,
      "step": 7401
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8598095178604126,
      "learning_rate": 1.5889850813021117e-05,
      "loss": 0.8049,
      "step": 7402
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9034719467163086,
      "learning_rate": 1.588921189662333e-05,
      "loss": 0.5389,
      "step": 7403
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7829877138137817,
      "learning_rate": 1.588857298022554e-05,
      "loss": 0.9364,
      "step": 7404
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9420809745788574,
      "learning_rate": 1.588793406382775e-05,
      "loss": 0.8715,
      "step": 7405
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7546100616455078,
      "learning_rate": 1.588729514742996e-05,
      "loss": 0.9759,
      "step": 7406
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4846848249435425,
      "learning_rate": 1.588665623103217e-05,
      "loss": 0.9041,
      "step": 7407
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.265113830566406,
      "learning_rate": 1.588601731463438e-05,
      "loss": 1.4738,
      "step": 7408
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9586186408996582,
      "learning_rate": 1.5885378398236594e-05,
      "loss": 1.0923,
      "step": 7409
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3388595581054688,
      "learning_rate": 1.5884739481838802e-05,
      "loss": 1.3507,
      "step": 7410
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.275418758392334,
      "learning_rate": 1.5884100565441014e-05,
      "loss": 1.0069,
      "step": 7411
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.500677108764648,
      "learning_rate": 1.5883461649043222e-05,
      "loss": 0.7876,
      "step": 7412
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9255987405776978,
      "learning_rate": 1.5882822732645437e-05,
      "loss": 0.8597,
      "step": 7413
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4699110984802246,
      "learning_rate": 1.5882183816247645e-05,
      "loss": 0.8292,
      "step": 7414
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.772747755050659,
      "learning_rate": 1.5881544899849857e-05,
      "loss": 0.9758,
      "step": 7415
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8837263584136963,
      "learning_rate": 1.5880905983452065e-05,
      "loss": 0.7288,
      "step": 7416
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.569263458251953,
      "learning_rate": 1.5880267067054276e-05,
      "loss": 0.9107,
      "step": 7417
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.1421895027160645,
      "learning_rate": 1.5879628150656488e-05,
      "loss": 0.9661,
      "step": 7418
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.8735735416412354,
      "learning_rate": 1.58789892342587e-05,
      "loss": 1.0639,
      "step": 7419
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.51540470123291,
      "learning_rate": 1.5878350317860908e-05,
      "loss": 0.9343,
      "step": 7420
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6738674640655518,
      "learning_rate": 1.587771140146312e-05,
      "loss": 0.4604,
      "step": 7421
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.479297637939453,
      "learning_rate": 1.587707248506533e-05,
      "loss": 1.0992,
      "step": 7422
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7876954078674316,
      "learning_rate": 1.5876433568667542e-05,
      "loss": 0.8719,
      "step": 7423
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9256582260131836,
      "learning_rate": 1.587579465226975e-05,
      "loss": 1.076,
      "step": 7424
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.046416997909546,
      "learning_rate": 1.5875155735871962e-05,
      "loss": 0.9128,
      "step": 7425
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5697301626205444,
      "learning_rate": 1.5874516819474174e-05,
      "loss": 0.8551,
      "step": 7426
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.3566997051239014,
      "learning_rate": 1.5873877903076385e-05,
      "loss": 0.9033,
      "step": 7427
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6075029373168945,
      "learning_rate": 1.5873238986678593e-05,
      "loss": 0.9888,
      "step": 7428
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9221136569976807,
      "learning_rate": 1.5872600070280805e-05,
      "loss": 0.8816,
      "step": 7429
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.617578506469727,
      "learning_rate": 1.5871961153883016e-05,
      "loss": 1.2585,
      "step": 7430
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6593838930130005,
      "learning_rate": 1.5871322237485228e-05,
      "loss": 0.9353,
      "step": 7431
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4789351224899292,
      "learning_rate": 1.5870683321087436e-05,
      "loss": 0.4612,
      "step": 7432
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3512701988220215,
      "learning_rate": 1.5870044404689648e-05,
      "loss": 0.998,
      "step": 7433
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9431817531585693,
      "learning_rate": 1.586940548829186e-05,
      "loss": 1.0261,
      "step": 7434
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7407649755477905,
      "learning_rate": 1.586876657189407e-05,
      "loss": 0.9312,
      "step": 7435
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7910643815994263,
      "learning_rate": 1.586812765549628e-05,
      "loss": 0.9928,
      "step": 7436
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8250905275344849,
      "learning_rate": 1.586748873909849e-05,
      "loss": 0.9389,
      "step": 7437
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9174147844314575,
      "learning_rate": 1.5866849822700702e-05,
      "loss": 0.9842,
      "step": 7438
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8849222660064697,
      "learning_rate": 1.5866210906302914e-05,
      "loss": 1.1912,
      "step": 7439
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.268594980239868,
      "learning_rate": 1.5865571989905122e-05,
      "loss": 1.0198,
      "step": 7440
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8330868482589722,
      "learning_rate": 1.5864933073507333e-05,
      "loss": 1.1299,
      "step": 7441
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5469425916671753,
      "learning_rate": 1.5864294157109545e-05,
      "loss": 0.9471,
      "step": 7442
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.257050037384033,
      "learning_rate": 1.5863655240711756e-05,
      "loss": 1.0325,
      "step": 7443
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8963509798049927,
      "learning_rate": 1.5863016324313965e-05,
      "loss": 0.9409,
      "step": 7444
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8882009983062744,
      "learning_rate": 1.5862377407916176e-05,
      "loss": 0.9563,
      "step": 7445
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8235678672790527,
      "learning_rate": 1.5861738491518384e-05,
      "loss": 0.8792,
      "step": 7446
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.2056143283843994,
      "learning_rate": 1.58610995751206e-05,
      "loss": 1.0054,
      "step": 7447
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7085661888122559,
      "learning_rate": 1.5860460658722807e-05,
      "loss": 0.8993,
      "step": 7448
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.195753574371338,
      "learning_rate": 1.585982174232502e-05,
      "loss": 0.924,
      "step": 7449
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.930527687072754,
      "learning_rate": 1.5859182825927227e-05,
      "loss": 1.0431,
      "step": 7450
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.944509506225586,
      "learning_rate": 1.585854390952944e-05,
      "loss": 0.9592,
      "step": 7451
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.569934844970703,
      "learning_rate": 1.585790499313165e-05,
      "loss": 0.9682,
      "step": 7452
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8888654708862305,
      "learning_rate": 1.585726607673386e-05,
      "loss": 1.3135,
      "step": 7453
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.52583909034729,
      "learning_rate": 1.585662716033607e-05,
      "loss": 1.0376,
      "step": 7454
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8358793258666992,
      "learning_rate": 1.585598824393828e-05,
      "loss": 0.9011,
      "step": 7455
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.729950189590454,
      "learning_rate": 1.5855349327540493e-05,
      "loss": 0.943,
      "step": 7456
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.645286202430725,
      "learning_rate": 1.5854710411142704e-05,
      "loss": 0.4239,
      "step": 7457
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.7018487453460693,
      "learning_rate": 1.5854071494744913e-05,
      "loss": 0.9185,
      "step": 7458
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6663492918014526,
      "learning_rate": 1.5853432578347124e-05,
      "loss": 0.9866,
      "step": 7459
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6561543941497803,
      "learning_rate": 1.5852793661949336e-05,
      "loss": 0.9559,
      "step": 7460
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.881376266479492,
      "learning_rate": 1.5852154745551547e-05,
      "loss": 1.0808,
      "step": 7461
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5154623985290527,
      "learning_rate": 1.5851515829153755e-05,
      "loss": 0.3773,
      "step": 7462
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6913092136383057,
      "learning_rate": 1.5850876912755967e-05,
      "loss": 1.1806,
      "step": 7463
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3299951553344727,
      "learning_rate": 1.585023799635818e-05,
      "loss": 1.0204,
      "step": 7464
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7626279592514038,
      "learning_rate": 1.584959907996039e-05,
      "loss": 1.0032,
      "step": 7465
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.109349489212036,
      "learning_rate": 1.5848960163562598e-05,
      "loss": 0.7662,
      "step": 7466
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.148763656616211,
      "learning_rate": 1.584832124716481e-05,
      "loss": 0.5394,
      "step": 7467
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.484311580657959,
      "learning_rate": 1.584768233076702e-05,
      "loss": 0.992,
      "step": 7468
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0651614665985107,
      "learning_rate": 1.5847043414369233e-05,
      "loss": 1.0457,
      "step": 7469
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.02140998840332,
      "learning_rate": 1.584640449797144e-05,
      "loss": 1.109,
      "step": 7470
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4032576084136963,
      "learning_rate": 1.5845765581573653e-05,
      "loss": 0.5381,
      "step": 7471
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9046571254730225,
      "learning_rate": 1.5845126665175864e-05,
      "loss": 1.1996,
      "step": 7472
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7063007354736328,
      "learning_rate": 1.5844487748778076e-05,
      "loss": 0.9948,
      "step": 7473
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5174469947814941,
      "learning_rate": 1.5843848832380284e-05,
      "loss": 0.8396,
      "step": 7474
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.1879143714904785,
      "learning_rate": 1.5843209915982495e-05,
      "loss": 1.3264,
      "step": 7475
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8105368614196777,
      "learning_rate": 1.5842570999584707e-05,
      "loss": 1.047,
      "step": 7476
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.0515265464782715,
      "learning_rate": 1.584193208318692e-05,
      "loss": 1.4406,
      "step": 7477
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8508940935134888,
      "learning_rate": 1.5841293166789127e-05,
      "loss": 0.8448,
      "step": 7478
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2111411094665527,
      "learning_rate": 1.5840654250391338e-05,
      "loss": 0.828,
      "step": 7479
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8368924856185913,
      "learning_rate": 1.584001533399355e-05,
      "loss": 1.0422,
      "step": 7480
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.700432538986206,
      "learning_rate": 1.583937641759576e-05,
      "loss": 0.8708,
      "step": 7481
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.530921459197998,
      "learning_rate": 1.583873750119797e-05,
      "loss": 0.936,
      "step": 7482
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7140740156173706,
      "learning_rate": 1.583809858480018e-05,
      "loss": 0.9058,
      "step": 7483
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.202800273895264,
      "learning_rate": 1.583745966840239e-05,
      "loss": 1.2474,
      "step": 7484
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9058902263641357,
      "learning_rate": 1.5836820752004604e-05,
      "loss": 0.8785,
      "step": 7485
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.175416946411133,
      "learning_rate": 1.5836181835606812e-05,
      "loss": 1.6862,
      "step": 7486
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9057835340499878,
      "learning_rate": 1.5835542919209024e-05,
      "loss": 0.931,
      "step": 7487
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.616089940071106,
      "learning_rate": 1.5834904002811232e-05,
      "loss": 0.8653,
      "step": 7488
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4105063676834106,
      "learning_rate": 1.5834265086413443e-05,
      "loss": 0.4225,
      "step": 7489
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.080331325531006,
      "learning_rate": 1.5833626170015655e-05,
      "loss": 0.9412,
      "step": 7490
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5335988998413086,
      "learning_rate": 1.5832987253617867e-05,
      "loss": 0.8479,
      "step": 7491
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8595409393310547,
      "learning_rate": 1.5832348337220075e-05,
      "loss": 0.9419,
      "step": 7492
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.9411091804504395,
      "learning_rate": 1.5831709420822286e-05,
      "loss": 2.024,
      "step": 7493
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9154224395751953,
      "learning_rate": 1.5831070504424498e-05,
      "loss": 1.0213,
      "step": 7494
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.7841248512268066,
      "learning_rate": 1.583043158802671e-05,
      "loss": 1.0084,
      "step": 7495
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.495266914367676,
      "learning_rate": 1.5829792671628918e-05,
      "loss": 1.3031,
      "step": 7496
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.713418483734131,
      "learning_rate": 1.582915375523113e-05,
      "loss": 1.6132,
      "step": 7497
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3117830753326416,
      "learning_rate": 1.582851483883334e-05,
      "loss": 0.9459,
      "step": 7498
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.431297779083252,
      "learning_rate": 1.5827875922435552e-05,
      "loss": 0.9528,
      "step": 7499
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.097470283508301,
      "learning_rate": 1.582723700603776e-05,
      "loss": 0.8072,
      "step": 7500
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7032139301300049,
      "learning_rate": 1.5826598089639972e-05,
      "loss": 0.7796,
      "step": 7501
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.110682725906372,
      "learning_rate": 1.5825959173242183e-05,
      "loss": 0.9252,
      "step": 7502
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.582856297492981,
      "learning_rate": 1.5825320256844395e-05,
      "loss": 0.8534,
      "step": 7503
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.4747281074523926,
      "learning_rate": 1.5824681340446603e-05,
      "loss": 1.0663,
      "step": 7504
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.851348638534546,
      "learning_rate": 1.5824042424048815e-05,
      "loss": 0.9894,
      "step": 7505
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.078688621520996,
      "learning_rate": 1.5823403507651026e-05,
      "loss": 1.007,
      "step": 7506
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.091841220855713,
      "learning_rate": 1.5822764591253238e-05,
      "loss": 2.0121,
      "step": 7507
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.620648145675659,
      "learning_rate": 1.5822125674855446e-05,
      "loss": 1.0198,
      "step": 7508
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6938209533691406,
      "learning_rate": 1.5821486758457657e-05,
      "loss": 0.885,
      "step": 7509
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.809520959854126,
      "learning_rate": 1.582084784205987e-05,
      "loss": 0.8673,
      "step": 7510
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.759514570236206,
      "learning_rate": 1.582020892566208e-05,
      "loss": 0.8875,
      "step": 7511
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.680905818939209,
      "learning_rate": 1.581957000926429e-05,
      "loss": 0.8926,
      "step": 7512
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1561639308929443,
      "learning_rate": 1.58189310928665e-05,
      "loss": 0.8303,
      "step": 7513
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.283581733703613,
      "learning_rate": 1.5818292176468712e-05,
      "loss": 1.2073,
      "step": 7514
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.869003415107727,
      "learning_rate": 1.5817653260070923e-05,
      "loss": 0.9214,
      "step": 7515
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6899343729019165,
      "learning_rate": 1.581701434367313e-05,
      "loss": 0.8925,
      "step": 7516
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.765272855758667,
      "learning_rate": 1.5816375427275343e-05,
      "loss": 1.2127,
      "step": 7517
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.21051549911499,
      "learning_rate": 1.581573651087755e-05,
      "loss": 1.4241,
      "step": 7518
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.647959589958191,
      "learning_rate": 1.5815097594479766e-05,
      "loss": 0.864,
      "step": 7519
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5933756828308105,
      "learning_rate": 1.5814458678081974e-05,
      "loss": 1.1477,
      "step": 7520
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.906100034713745,
      "learning_rate": 1.5813819761684186e-05,
      "loss": 0.9973,
      "step": 7521
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3242392539978027,
      "learning_rate": 1.5813180845286394e-05,
      "loss": 1.0583,
      "step": 7522
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3254668712615967,
      "learning_rate": 1.5812541928888606e-05,
      "loss": 0.9232,
      "step": 7523
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6897332668304443,
      "learning_rate": 1.5811903012490817e-05,
      "loss": 0.8693,
      "step": 7524
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4493281841278076,
      "learning_rate": 1.581126409609303e-05,
      "loss": 1.0,
      "step": 7525
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6601717472076416,
      "learning_rate": 1.5810625179695237e-05,
      "loss": 1.1057,
      "step": 7526
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.63126540184021,
      "learning_rate": 1.580998626329745e-05,
      "loss": 0.9119,
      "step": 7527
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.716712236404419,
      "learning_rate": 1.580934734689966e-05,
      "loss": 0.8343,
      "step": 7528
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0337986946105957,
      "learning_rate": 1.580870843050187e-05,
      "loss": 0.9995,
      "step": 7529
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7055553197860718,
      "learning_rate": 1.580806951410408e-05,
      "loss": 0.9732,
      "step": 7530
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.887148380279541,
      "learning_rate": 1.580743059770629e-05,
      "loss": 0.9158,
      "step": 7531
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6354146003723145,
      "learning_rate": 1.5806791681308503e-05,
      "loss": 0.8958,
      "step": 7532
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.413444995880127,
      "learning_rate": 1.5806152764910714e-05,
      "loss": 0.9618,
      "step": 7533
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.232449769973755,
      "learning_rate": 1.5805513848512922e-05,
      "loss": 1.0194,
      "step": 7534
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9206299781799316,
      "learning_rate": 1.5804874932115134e-05,
      "loss": 0.8544,
      "step": 7535
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0243380069732666,
      "learning_rate": 1.5804236015717346e-05,
      "loss": 0.9747,
      "step": 7536
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.00618052482605,
      "learning_rate": 1.5803597099319557e-05,
      "loss": 1.0097,
      "step": 7537
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1686394214630127,
      "learning_rate": 1.5802958182921765e-05,
      "loss": 0.9424,
      "step": 7538
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6029164791107178,
      "learning_rate": 1.5802319266523977e-05,
      "loss": 1.0857,
      "step": 7539
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0475668907165527,
      "learning_rate": 1.580168035012619e-05,
      "loss": 0.9629,
      "step": 7540
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7918516397476196,
      "learning_rate": 1.5801041433728397e-05,
      "loss": 0.8682,
      "step": 7541
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6060643196105957,
      "learning_rate": 1.5800402517330608e-05,
      "loss": 0.743,
      "step": 7542
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.727208375930786,
      "learning_rate": 1.579976360093282e-05,
      "loss": 0.6831,
      "step": 7543
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6290912628173828,
      "learning_rate": 1.579912468453503e-05,
      "loss": 0.9479,
      "step": 7544
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.258824348449707,
      "learning_rate": 1.579848576813724e-05,
      "loss": 1.6367,
      "step": 7545
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.5583977699279785,
      "learning_rate": 1.579784685173945e-05,
      "loss": 1.2662,
      "step": 7546
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5140419006347656,
      "learning_rate": 1.5797207935341662e-05,
      "loss": 0.8156,
      "step": 7547
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9451152086257935,
      "learning_rate": 1.5796569018943874e-05,
      "loss": 0.9418,
      "step": 7548
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2429122924804688,
      "learning_rate": 1.5795930102546082e-05,
      "loss": 1.0358,
      "step": 7549
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.3262386322021484,
      "learning_rate": 1.5795291186148294e-05,
      "loss": 0.8252,
      "step": 7550
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3100948333740234,
      "learning_rate": 1.5794652269750502e-05,
      "loss": 1.0142,
      "step": 7551
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7290068864822388,
      "learning_rate": 1.5794013353352717e-05,
      "loss": 0.8723,
      "step": 7552
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2566354274749756,
      "learning_rate": 1.5793374436954925e-05,
      "loss": 1.2838,
      "step": 7553
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.709228754043579,
      "learning_rate": 1.5792735520557136e-05,
      "loss": 0.9117,
      "step": 7554
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.154569625854492,
      "learning_rate": 1.5792096604159345e-05,
      "loss": 1.0064,
      "step": 7555
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.981339454650879,
      "learning_rate": 1.5791457687761556e-05,
      "loss": 1.7342,
      "step": 7556
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4861505031585693,
      "learning_rate": 1.5790818771363768e-05,
      "loss": 0.9924,
      "step": 7557
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.099694728851318,
      "learning_rate": 1.579017985496598e-05,
      "loss": 1.4467,
      "step": 7558
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.577518939971924,
      "learning_rate": 1.5789540938568187e-05,
      "loss": 1.317,
      "step": 7559
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0831940174102783,
      "learning_rate": 1.57889020221704e-05,
      "loss": 0.9061,
      "step": 7560
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8971084356307983,
      "learning_rate": 1.578826310577261e-05,
      "loss": 0.969,
      "step": 7561
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2785587310791016,
      "learning_rate": 1.5787624189374822e-05,
      "loss": 0.9522,
      "step": 7562
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.7357218265533447,
      "learning_rate": 1.578698527297703e-05,
      "loss": 0.7714,
      "step": 7563
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.1890177726745605,
      "learning_rate": 1.5786346356579242e-05,
      "loss": 1.4803,
      "step": 7564
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5219006538391113,
      "learning_rate": 1.5785707440181453e-05,
      "loss": 1.1163,
      "step": 7565
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.943472146987915,
      "learning_rate": 1.5785068523783665e-05,
      "loss": 0.9088,
      "step": 7566
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.953761100769043,
      "learning_rate": 1.5784429607385873e-05,
      "loss": 0.7869,
      "step": 7567
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5456334352493286,
      "learning_rate": 1.5783790690988085e-05,
      "loss": 0.4724,
      "step": 7568
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.602489471435547,
      "learning_rate": 1.5783151774590296e-05,
      "loss": 1.3153,
      "step": 7569
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8684524297714233,
      "learning_rate": 1.5782512858192508e-05,
      "loss": 0.9522,
      "step": 7570
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9283533096313477,
      "learning_rate": 1.5781873941794716e-05,
      "loss": 0.9042,
      "step": 7571
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0789830684661865,
      "learning_rate": 1.5781235025396927e-05,
      "loss": 0.9809,
      "step": 7572
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9827457666397095,
      "learning_rate": 1.578059610899914e-05,
      "loss": 0.9785,
      "step": 7573
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.807332158088684,
      "learning_rate": 1.577995719260135e-05,
      "loss": 0.3934,
      "step": 7574
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7837954759597778,
      "learning_rate": 1.577931827620356e-05,
      "loss": 0.8788,
      "step": 7575
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0994582176208496,
      "learning_rate": 1.577867935980577e-05,
      "loss": 1.1445,
      "step": 7576
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.876765251159668,
      "learning_rate": 1.5778040443407982e-05,
      "loss": 0.9705,
      "step": 7577
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9569644927978516,
      "learning_rate": 1.5777401527010193e-05,
      "loss": 0.9416,
      "step": 7578
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0337398052215576,
      "learning_rate": 1.57767626106124e-05,
      "loss": 0.8688,
      "step": 7579
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9768543243408203,
      "learning_rate": 1.5776123694214613e-05,
      "loss": 0.8337,
      "step": 7580
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6486846208572388,
      "learning_rate": 1.5775484777816825e-05,
      "loss": 0.3345,
      "step": 7581
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.7709033489227295,
      "learning_rate": 1.5774845861419036e-05,
      "loss": 0.7473,
      "step": 7582
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.277961015701294,
      "learning_rate": 1.5774206945021244e-05,
      "loss": 0.9508,
      "step": 7583
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.5889384746551514,
      "learning_rate": 1.5773568028623456e-05,
      "loss": 0.8506,
      "step": 7584
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.219830274581909,
      "learning_rate": 1.5772929112225664e-05,
      "loss": 1.2188,
      "step": 7585
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.210108995437622,
      "learning_rate": 1.577229019582788e-05,
      "loss": 1.0773,
      "step": 7586
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8556679487228394,
      "learning_rate": 1.5771651279430087e-05,
      "loss": 0.9255,
      "step": 7587
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.876523494720459,
      "learning_rate": 1.57710123630323e-05,
      "loss": 1.6648,
      "step": 7588
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9363770484924316,
      "learning_rate": 1.5770373446634507e-05,
      "loss": 1.0377,
      "step": 7589
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.977641224861145,
      "learning_rate": 1.576973453023672e-05,
      "loss": 1.0001,
      "step": 7590
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.595374822616577,
      "learning_rate": 1.576909561383893e-05,
      "loss": 1.0258,
      "step": 7591
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.416087031364441,
      "learning_rate": 1.576845669744114e-05,
      "loss": 0.944,
      "step": 7592
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.588331937789917,
      "learning_rate": 1.576781778104335e-05,
      "loss": 0.8505,
      "step": 7593
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8266786336898804,
      "learning_rate": 1.576717886464556e-05,
      "loss": 0.9455,
      "step": 7594
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6408121585845947,
      "learning_rate": 1.5766539948247773e-05,
      "loss": 0.8789,
      "step": 7595
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.869612216949463,
      "learning_rate": 1.5765901031849984e-05,
      "loss": 0.9753,
      "step": 7596
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1074600219726562,
      "learning_rate": 1.5765262115452192e-05,
      "loss": 1.3033,
      "step": 7597
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.735891580581665,
      "learning_rate": 1.5764623199054404e-05,
      "loss": 0.8859,
      "step": 7598
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.745902419090271,
      "learning_rate": 1.5763984282656615e-05,
      "loss": 0.8838,
      "step": 7599
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4847298860549927,
      "learning_rate": 1.5763345366258827e-05,
      "loss": 0.4274,
      "step": 7600
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1478512287139893,
      "learning_rate": 1.5762706449861035e-05,
      "loss": 0.8955,
      "step": 7601
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.505284309387207,
      "learning_rate": 1.5762067533463247e-05,
      "loss": 0.9365,
      "step": 7602
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1606831550598145,
      "learning_rate": 1.5761428617065458e-05,
      "loss": 1.315,
      "step": 7603
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.850642442703247,
      "learning_rate": 1.576078970066767e-05,
      "loss": 0.9456,
      "step": 7604
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8206138610839844,
      "learning_rate": 1.5760150784269878e-05,
      "loss": 0.9496,
      "step": 7605
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.752124547958374,
      "learning_rate": 1.575951186787209e-05,
      "loss": 0.8318,
      "step": 7606
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2240312099456787,
      "learning_rate": 1.57588729514743e-05,
      "loss": 0.8795,
      "step": 7607
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1820850372314453,
      "learning_rate": 1.5758234035076513e-05,
      "loss": 1.0124,
      "step": 7608
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9302345514297485,
      "learning_rate": 1.575759511867872e-05,
      "loss": 0.9552,
      "step": 7609
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.393319845199585,
      "learning_rate": 1.5756956202280932e-05,
      "loss": 0.9573,
      "step": 7610
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.8645544052124023,
      "learning_rate": 1.5756317285883144e-05,
      "loss": 1.2473,
      "step": 7611
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9215991497039795,
      "learning_rate": 1.5755678369485355e-05,
      "loss": 0.937,
      "step": 7612
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.009611129760742,
      "learning_rate": 1.5755039453087564e-05,
      "loss": 0.821,
      "step": 7613
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.683447360992432,
      "learning_rate": 1.5754400536689775e-05,
      "loss": 1.3504,
      "step": 7614
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.8681764602661133,
      "learning_rate": 1.5753761620291987e-05,
      "loss": 1.0506,
      "step": 7615
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4982614517211914,
      "learning_rate": 1.5753122703894198e-05,
      "loss": 0.9685,
      "step": 7616
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.257040500640869,
      "learning_rate": 1.5752483787496406e-05,
      "loss": 1.2496,
      "step": 7617
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.6786537170410156,
      "learning_rate": 1.5751844871098618e-05,
      "loss": 0.8923,
      "step": 7618
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.898930549621582,
      "learning_rate": 1.575120595470083e-05,
      "loss": 1.25,
      "step": 7619
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1209633350372314,
      "learning_rate": 1.575056703830304e-05,
      "loss": 1.0908,
      "step": 7620
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.038344621658325,
      "learning_rate": 1.574992812190525e-05,
      "loss": 0.9434,
      "step": 7621
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0677576065063477,
      "learning_rate": 1.574928920550746e-05,
      "loss": 0.8922,
      "step": 7622
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.7772326469421387,
      "learning_rate": 1.574865028910967e-05,
      "loss": 0.8687,
      "step": 7623
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.32620906829834,
      "learning_rate": 1.5748011372711884e-05,
      "loss": 0.9832,
      "step": 7624
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4053170680999756,
      "learning_rate": 1.5747372456314092e-05,
      "loss": 1.0971,
      "step": 7625
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6090478897094727,
      "learning_rate": 1.5746733539916303e-05,
      "loss": 0.8037,
      "step": 7626
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6125143766403198,
      "learning_rate": 1.574609462351851e-05,
      "loss": 0.9316,
      "step": 7627
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.959761142730713,
      "learning_rate": 1.5745455707120723e-05,
      "loss": 0.8621,
      "step": 7628
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.954103708267212,
      "learning_rate": 1.5744816790722935e-05,
      "loss": 0.9545,
      "step": 7629
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9120773077011108,
      "learning_rate": 1.5744177874325146e-05,
      "loss": 1.2526,
      "step": 7630
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.5640616416931152,
      "learning_rate": 1.5743538957927354e-05,
      "loss": 0.9548,
      "step": 7631
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4604902267456055,
      "learning_rate": 1.5742900041529566e-05,
      "loss": 1.0377,
      "step": 7632
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.3345346450805664,
      "learning_rate": 1.5742261125131778e-05,
      "loss": 0.9496,
      "step": 7633
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6118781566619873,
      "learning_rate": 1.574162220873399e-05,
      "loss": 0.8278,
      "step": 7634
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9847259521484375,
      "learning_rate": 1.5740983292336197e-05,
      "loss": 0.965,
      "step": 7635
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.924191951751709,
      "learning_rate": 1.574034437593841e-05,
      "loss": 0.8744,
      "step": 7636
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.3862922191619873,
      "learning_rate": 1.573970545954062e-05,
      "loss": 1.0159,
      "step": 7637
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.950460910797119,
      "learning_rate": 1.5739066543142832e-05,
      "loss": 1.0402,
      "step": 7638
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.051489353179932,
      "learning_rate": 1.573842762674504e-05,
      "loss": 0.9706,
      "step": 7639
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.159580707550049,
      "learning_rate": 1.573778871034725e-05,
      "loss": 0.9777,
      "step": 7640
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.031738758087158,
      "learning_rate": 1.5737149793949463e-05,
      "loss": 0.9636,
      "step": 7641
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.512883186340332,
      "learning_rate": 1.5736510877551675e-05,
      "loss": 0.8675,
      "step": 7642
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6234314441680908,
      "learning_rate": 1.5735871961153883e-05,
      "loss": 0.9781,
      "step": 7643
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.350275039672852,
      "learning_rate": 1.5735233044756094e-05,
      "loss": 1.3117,
      "step": 7644
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.368363857269287,
      "learning_rate": 1.5734594128358306e-05,
      "loss": 0.9161,
      "step": 7645
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.521160840988159,
      "learning_rate": 1.5733955211960517e-05,
      "loss": 0.9504,
      "step": 7646
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.6525774002075195,
      "learning_rate": 1.5733316295562726e-05,
      "loss": 1.6435,
      "step": 7647
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7341889142990112,
      "learning_rate": 1.5732677379164937e-05,
      "loss": 0.7459,
      "step": 7648
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3022899627685547,
      "learning_rate": 1.573203846276715e-05,
      "loss": 0.7998,
      "step": 7649
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0779175758361816,
      "learning_rate": 1.573139954636936e-05,
      "loss": 0.9899,
      "step": 7650
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.252337694168091,
      "learning_rate": 1.573076062997157e-05,
      "loss": 0.982,
      "step": 7651
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4654459953308105,
      "learning_rate": 1.573012171357378e-05,
      "loss": 1.0686,
      "step": 7652
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1792876720428467,
      "learning_rate": 1.572948279717599e-05,
      "loss": 0.8847,
      "step": 7653
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4270312786102295,
      "learning_rate": 1.5728843880778203e-05,
      "loss": 1.1444,
      "step": 7654
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.612210750579834,
      "learning_rate": 1.572820496438041e-05,
      "loss": 0.9267,
      "step": 7655
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.89460027217865,
      "learning_rate": 1.5727566047982623e-05,
      "loss": 1.0218,
      "step": 7656
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.164872646331787,
      "learning_rate": 1.5726927131584834e-05,
      "loss": 1.0733,
      "step": 7657
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8522639274597168,
      "learning_rate": 1.5726288215187046e-05,
      "loss": 0.9334,
      "step": 7658
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.548981189727783,
      "learning_rate": 1.5725649298789254e-05,
      "loss": 1.2062,
      "step": 7659
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8938549757003784,
      "learning_rate": 1.5725010382391466e-05,
      "loss": 0.891,
      "step": 7660
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.53390634059906,
      "learning_rate": 1.5724371465993674e-05,
      "loss": 0.8798,
      "step": 7661
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.5648629665374756,
      "learning_rate": 1.572373254959589e-05,
      "loss": 1.1031,
      "step": 7662
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.549107313156128,
      "learning_rate": 1.5723093633198097e-05,
      "loss": 0.9639,
      "step": 7663
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.024250030517578,
      "learning_rate": 1.572245471680031e-05,
      "loss": 1.2169,
      "step": 7664
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1217575073242188,
      "learning_rate": 1.5721815800402517e-05,
      "loss": 0.9222,
      "step": 7665
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8974711894989014,
      "learning_rate": 1.5721176884004728e-05,
      "loss": 1.0065,
      "step": 7666
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9131118059158325,
      "learning_rate": 1.572053796760694e-05,
      "loss": 0.823,
      "step": 7667
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.512846827507019,
      "learning_rate": 1.571989905120915e-05,
      "loss": 0.9646,
      "step": 7668
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5624765157699585,
      "learning_rate": 1.571926013481136e-05,
      "loss": 0.4077,
      "step": 7669
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1133134365081787,
      "learning_rate": 1.571862121841357e-05,
      "loss": 0.9624,
      "step": 7670
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.7762556076049805,
      "learning_rate": 1.5717982302015782e-05,
      "loss": 1.3425,
      "step": 7671
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7032564878463745,
      "learning_rate": 1.5717343385617994e-05,
      "loss": 0.5529,
      "step": 7672
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.355727672576904,
      "learning_rate": 1.5716704469220202e-05,
      "loss": 1.3194,
      "step": 7673
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7008553743362427,
      "learning_rate": 1.5716065552822414e-05,
      "loss": 0.9046,
      "step": 7674
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.8418686389923096,
      "learning_rate": 1.5715426636424625e-05,
      "loss": 1.2417,
      "step": 7675
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8538976907730103,
      "learning_rate": 1.5714787720026837e-05,
      "loss": 0.9846,
      "step": 7676
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6817641258239746,
      "learning_rate": 1.5714148803629045e-05,
      "loss": 0.5795,
      "step": 7677
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6118640899658203,
      "learning_rate": 1.5713509887231256e-05,
      "loss": 0.9122,
      "step": 7678
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3791449069976807,
      "learning_rate": 1.5712870970833468e-05,
      "loss": 1.0064,
      "step": 7679
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.062209367752075,
      "learning_rate": 1.571223205443568e-05,
      "loss": 1.0925,
      "step": 7680
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.117953777313232,
      "learning_rate": 1.5711593138037888e-05,
      "loss": 1.1283,
      "step": 7681
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8220173120498657,
      "learning_rate": 1.57109542216401e-05,
      "loss": 0.8676,
      "step": 7682
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4666472673416138,
      "learning_rate": 1.571031530524231e-05,
      "loss": 0.4839,
      "step": 7683
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6629000902175903,
      "learning_rate": 1.5709676388844522e-05,
      "loss": 0.8996,
      "step": 7684
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3522731065750122,
      "learning_rate": 1.570903747244673e-05,
      "loss": 0.3448,
      "step": 7685
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3690459728240967,
      "learning_rate": 1.5708398556048942e-05,
      "loss": 0.9925,
      "step": 7686
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5562045574188232,
      "learning_rate": 1.5707759639651154e-05,
      "loss": 0.8823,
      "step": 7687
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7852857112884521,
      "learning_rate": 1.5707120723253365e-05,
      "loss": 0.9734,
      "step": 7688
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.71515953540802,
      "learning_rate": 1.5706481806855573e-05,
      "loss": 0.9938,
      "step": 7689
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.809096097946167,
      "learning_rate": 1.5705842890457785e-05,
      "loss": 0.9034,
      "step": 7690
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8648337125778198,
      "learning_rate": 1.5705203974059996e-05,
      "loss": 0.8324,
      "step": 7691
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4026448726654053,
      "learning_rate": 1.5704565057662208e-05,
      "loss": 1.0509,
      "step": 7692
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5678229331970215,
      "learning_rate": 1.5703926141264416e-05,
      "loss": 0.9718,
      "step": 7693
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6169805526733398,
      "learning_rate": 1.5703287224866628e-05,
      "loss": 0.8835,
      "step": 7694
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7137982845306396,
      "learning_rate": 1.5702648308468836e-05,
      "loss": 0.9568,
      "step": 7695
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.871247410774231,
      "learning_rate": 1.570200939207105e-05,
      "loss": 1.0137,
      "step": 7696
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.8239011764526367,
      "learning_rate": 1.570137047567326e-05,
      "loss": 1.0871,
      "step": 7697
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.550083160400391,
      "learning_rate": 1.570073155927547e-05,
      "loss": 1.265,
      "step": 7698
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5677988529205322,
      "learning_rate": 1.570009264287768e-05,
      "loss": 0.9601,
      "step": 7699
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6729934215545654,
      "learning_rate": 1.569945372647989e-05,
      "loss": 0.988,
      "step": 7700
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.8270771503448486,
      "learning_rate": 1.5698814810082102e-05,
      "loss": 1.2743,
      "step": 7701
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.303771734237671,
      "learning_rate": 1.5698175893684313e-05,
      "loss": 0.9809,
      "step": 7702
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.3077375888824463,
      "learning_rate": 1.569753697728652e-05,
      "loss": 1.1643,
      "step": 7703
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1489760875701904,
      "learning_rate": 1.5696898060888733e-05,
      "loss": 0.9071,
      "step": 7704
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7698715925216675,
      "learning_rate": 1.5696259144490945e-05,
      "loss": 0.9258,
      "step": 7705
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0536417961120605,
      "learning_rate": 1.5695620228093156e-05,
      "loss": 0.9975,
      "step": 7706
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1517343521118164,
      "learning_rate": 1.5694981311695364e-05,
      "loss": 1.0223,
      "step": 7707
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.122650384902954,
      "learning_rate": 1.5694342395297576e-05,
      "loss": 0.9595,
      "step": 7708
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6410794258117676,
      "learning_rate": 1.5693703478899787e-05,
      "loss": 0.8228,
      "step": 7709
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.015205144882202,
      "learning_rate": 1.5693064562502e-05,
      "loss": 0.9641,
      "step": 7710
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.430959224700928,
      "learning_rate": 1.5692425646104207e-05,
      "loss": 1.7016,
      "step": 7711
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.434000015258789,
      "learning_rate": 1.569178672970642e-05,
      "loss": 1.5068,
      "step": 7712
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.194255828857422,
      "learning_rate": 1.569114781330863e-05,
      "loss": 0.9516,
      "step": 7713
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6480543613433838,
      "learning_rate": 1.569050889691084e-05,
      "loss": 0.8872,
      "step": 7714
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6963655948638916,
      "learning_rate": 1.568986998051305e-05,
      "loss": 0.8133,
      "step": 7715
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0652740001678467,
      "learning_rate": 1.568923106411526e-05,
      "loss": 1.0388,
      "step": 7716
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7674026489257812,
      "learning_rate": 1.5688592147717473e-05,
      "loss": 0.954,
      "step": 7717
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9386556148529053,
      "learning_rate": 1.5687953231319684e-05,
      "loss": 0.9695,
      "step": 7718
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.9737207889556885,
      "learning_rate": 1.5687314314921893e-05,
      "loss": 0.9807,
      "step": 7719
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8538551330566406,
      "learning_rate": 1.5686675398524104e-05,
      "loss": 0.9564,
      "step": 7720
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6700023412704468,
      "learning_rate": 1.5686036482126316e-05,
      "loss": 0.794,
      "step": 7721
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.097232341766357,
      "learning_rate": 1.5685397565728527e-05,
      "loss": 1.5279,
      "step": 7722
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.186906337738037,
      "learning_rate": 1.5684758649330735e-05,
      "loss": 0.9978,
      "step": 7723
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3370628356933594,
      "learning_rate": 1.5684119732932947e-05,
      "loss": 1.0592,
      "step": 7724
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4542977809906006,
      "learning_rate": 1.568348081653516e-05,
      "loss": 0.8805,
      "step": 7725
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.759042501449585,
      "learning_rate": 1.568284190013737e-05,
      "loss": 1.0119,
      "step": 7726
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5487395524978638,
      "learning_rate": 1.5682202983739578e-05,
      "loss": 0.7569,
      "step": 7727
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.702052116394043,
      "learning_rate": 1.568156406734179e-05,
      "loss": 0.8292,
      "step": 7728
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2299644947052,
      "learning_rate": 1.5680925150944e-05,
      "loss": 1.1503,
      "step": 7729
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6446640491485596,
      "learning_rate": 1.5680286234546213e-05,
      "loss": 0.6011,
      "step": 7730
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.227996826171875,
      "learning_rate": 1.567964731814842e-05,
      "loss": 1.0913,
      "step": 7731
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1349973678588867,
      "learning_rate": 1.5679008401750633e-05,
      "loss": 0.6452,
      "step": 7732
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.763471007347107,
      "learning_rate": 1.567836948535284e-05,
      "loss": 1.2306,
      "step": 7733
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.632353663444519,
      "learning_rate": 1.5677730568955056e-05,
      "loss": 0.9523,
      "step": 7734
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.735025644302368,
      "learning_rate": 1.5677091652557264e-05,
      "loss": 0.6836,
      "step": 7735
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8555766344070435,
      "learning_rate": 1.5676452736159475e-05,
      "loss": 0.4184,
      "step": 7736
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1961867809295654,
      "learning_rate": 1.5675813819761684e-05,
      "loss": 1.0725,
      "step": 7737
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.9654881954193115,
      "learning_rate": 1.5675174903363895e-05,
      "loss": 0.858,
      "step": 7738
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1277542114257812,
      "learning_rate": 1.5674535986966107e-05,
      "loss": 0.869,
      "step": 7739
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.293029308319092,
      "learning_rate": 1.5673897070568318e-05,
      "loss": 1.0454,
      "step": 7740
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9682832956314087,
      "learning_rate": 1.5673258154170526e-05,
      "loss": 1.0195,
      "step": 7741
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.182429075241089,
      "learning_rate": 1.5672619237772738e-05,
      "loss": 0.9379,
      "step": 7742
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5639094114303589,
      "learning_rate": 1.567198032137495e-05,
      "loss": 0.4773,
      "step": 7743
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8993827104568481,
      "learning_rate": 1.567134140497716e-05,
      "loss": 1.0316,
      "step": 7744
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2239928245544434,
      "learning_rate": 1.567070248857937e-05,
      "loss": 0.889,
      "step": 7745
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8625357151031494,
      "learning_rate": 1.567006357218158e-05,
      "loss": 1.1402,
      "step": 7746
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.083805561065674,
      "learning_rate": 1.5669424655783792e-05,
      "loss": 1.0537,
      "step": 7747
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8005849123001099,
      "learning_rate": 1.5668785739386004e-05,
      "loss": 0.8659,
      "step": 7748
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6143542528152466,
      "learning_rate": 1.5668146822988212e-05,
      "loss": 0.8736,
      "step": 7749
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.005481719970703,
      "learning_rate": 1.5667507906590423e-05,
      "loss": 1.159,
      "step": 7750
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.396078586578369,
      "learning_rate": 1.5666868990192635e-05,
      "loss": 1.1045,
      "step": 7751
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.376060724258423,
      "learning_rate": 1.5666230073794847e-05,
      "loss": 1.0591,
      "step": 7752
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.558213710784912,
      "learning_rate": 1.5665591157397055e-05,
      "loss": 0.4225,
      "step": 7753
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9261945486068726,
      "learning_rate": 1.5664952240999266e-05,
      "loss": 0.9017,
      "step": 7754
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9344818592071533,
      "learning_rate": 1.5664313324601478e-05,
      "loss": 0.8754,
      "step": 7755
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.760817289352417,
      "learning_rate": 1.566367440820369e-05,
      "loss": 0.9403,
      "step": 7756
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.277338981628418,
      "learning_rate": 1.5663035491805898e-05,
      "loss": 1.2182,
      "step": 7757
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.228878498077393,
      "learning_rate": 1.566239657540811e-05,
      "loss": 1.2942,
      "step": 7758
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1135363578796387,
      "learning_rate": 1.566175765901032e-05,
      "loss": 0.9448,
      "step": 7759
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.496624231338501,
      "learning_rate": 1.5661118742612532e-05,
      "loss": 1.0987,
      "step": 7760
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8390185832977295,
      "learning_rate": 1.566047982621474e-05,
      "loss": 0.9347,
      "step": 7761
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.878821849822998,
      "learning_rate": 1.5659840909816952e-05,
      "loss": 1.0207,
      "step": 7762
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1727731227874756,
      "learning_rate": 1.5659201993419163e-05,
      "loss": 1.1595,
      "step": 7763
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4650229215621948,
      "learning_rate": 1.5658563077021375e-05,
      "loss": 0.3866,
      "step": 7764
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.408748149871826,
      "learning_rate": 1.5657924160623583e-05,
      "loss": 1.1294,
      "step": 7765
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8042503595352173,
      "learning_rate": 1.5657285244225795e-05,
      "loss": 0.9397,
      "step": 7766
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7236902713775635,
      "learning_rate": 1.5656646327828006e-05,
      "loss": 0.9484,
      "step": 7767
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5366935729980469,
      "learning_rate": 1.5656007411430218e-05,
      "loss": 0.4447,
      "step": 7768
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6251226663589478,
      "learning_rate": 1.5655368495032426e-05,
      "loss": 0.9396,
      "step": 7769
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.558708667755127,
      "learning_rate": 1.5654729578634637e-05,
      "loss": 0.8681,
      "step": 7770
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1583564281463623,
      "learning_rate": 1.5654090662236846e-05,
      "loss": 1.114,
      "step": 7771
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5331335067749023,
      "learning_rate": 1.565345174583906e-05,
      "loss": 0.8555,
      "step": 7772
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7868237495422363,
      "learning_rate": 1.565281282944127e-05,
      "loss": 0.9346,
      "step": 7773
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6216766834259033,
      "learning_rate": 1.565217391304348e-05,
      "loss": 1.0653,
      "step": 7774
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0162436962127686,
      "learning_rate": 1.565153499664569e-05,
      "loss": 1.0248,
      "step": 7775
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.498835325241089,
      "learning_rate": 1.56508960802479e-05,
      "loss": 1.1494,
      "step": 7776
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.482362985610962,
      "learning_rate": 1.565025716385011e-05,
      "loss": 0.8522,
      "step": 7777
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.048213243484497,
      "learning_rate": 1.5649618247452323e-05,
      "loss": 0.9347,
      "step": 7778
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1159660816192627,
      "learning_rate": 1.564897933105453e-05,
      "loss": 0.9763,
      "step": 7779
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9874595403671265,
      "learning_rate": 1.5648340414656743e-05,
      "loss": 0.9435,
      "step": 7780
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6417306661605835,
      "learning_rate": 1.5647701498258954e-05,
      "loss": 0.3847,
      "step": 7781
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.5801515579223633,
      "learning_rate": 1.5647062581861166e-05,
      "loss": 1.1826,
      "step": 7782
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9183297157287598,
      "learning_rate": 1.5646423665463374e-05,
      "loss": 0.9075,
      "step": 7783
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9534683227539062,
      "learning_rate": 1.5645784749065586e-05,
      "loss": 1.0927,
      "step": 7784
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.826401948928833,
      "learning_rate": 1.5645145832667797e-05,
      "loss": 0.9426,
      "step": 7785
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.758319616317749,
      "learning_rate": 1.564450691627001e-05,
      "loss": 0.9747,
      "step": 7786
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9440234899520874,
      "learning_rate": 1.5643867999872217e-05,
      "loss": 0.85,
      "step": 7787
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0884945392608643,
      "learning_rate": 1.564322908347443e-05,
      "loss": 0.8845,
      "step": 7788
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7647535800933838,
      "learning_rate": 1.564259016707664e-05,
      "loss": 1.051,
      "step": 7789
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1908133029937744,
      "learning_rate": 1.564195125067885e-05,
      "loss": 1.1028,
      "step": 7790
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.018014907836914,
      "learning_rate": 1.564131233428106e-05,
      "loss": 1.0184,
      "step": 7791
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9160528182983398,
      "learning_rate": 1.564067341788327e-05,
      "loss": 0.8745,
      "step": 7792
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1092042922973633,
      "learning_rate": 1.5640034501485483e-05,
      "loss": 0.8642,
      "step": 7793
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6427078247070312,
      "learning_rate": 1.5639395585087694e-05,
      "loss": 0.8602,
      "step": 7794
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.395357608795166,
      "learning_rate": 1.5638756668689902e-05,
      "loss": 0.9644,
      "step": 7795
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6963436603546143,
      "learning_rate": 1.5638117752292114e-05,
      "loss": 0.9269,
      "step": 7796
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.588443398475647,
      "learning_rate": 1.5637478835894326e-05,
      "loss": 0.8823,
      "step": 7797
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.822097897529602,
      "learning_rate": 1.5636839919496537e-05,
      "loss": 0.5562,
      "step": 7798
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.7113232612609863,
      "learning_rate": 1.5636201003098745e-05,
      "loss": 1.4288,
      "step": 7799
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6965996026992798,
      "learning_rate": 1.5635562086700957e-05,
      "loss": 0.8789,
      "step": 7800
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2991678714752197,
      "learning_rate": 1.563492317030317e-05,
      "loss": 0.7462,
      "step": 7801
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.692544937133789,
      "learning_rate": 1.563428425390538e-05,
      "loss": 1.1609,
      "step": 7802
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9746642112731934,
      "learning_rate": 1.5633645337507588e-05,
      "loss": 1.0248,
      "step": 7803
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7730200290679932,
      "learning_rate": 1.56330064211098e-05,
      "loss": 0.9728,
      "step": 7804
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.5680999755859375,
      "learning_rate": 1.5632367504712008e-05,
      "loss": 1.2302,
      "step": 7805
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1324305534362793,
      "learning_rate": 1.5631728588314223e-05,
      "loss": 0.8793,
      "step": 7806
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1210386753082275,
      "learning_rate": 1.563108967191643e-05,
      "loss": 0.9442,
      "step": 7807
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.241044044494629,
      "learning_rate": 1.5630450755518642e-05,
      "loss": 1.3994,
      "step": 7808
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9407094717025757,
      "learning_rate": 1.562981183912085e-05,
      "loss": 0.884,
      "step": 7809
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1491196155548096,
      "learning_rate": 1.5629172922723062e-05,
      "loss": 1.1432,
      "step": 7810
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6397700309753418,
      "learning_rate": 1.5628534006325274e-05,
      "loss": 0.9451,
      "step": 7811
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6687841415405273,
      "learning_rate": 1.5627895089927485e-05,
      "loss": 0.6151,
      "step": 7812
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3651998043060303,
      "learning_rate": 1.5627256173529693e-05,
      "loss": 0.39,
      "step": 7813
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.296952724456787,
      "learning_rate": 1.5626617257131905e-05,
      "loss": 1.1219,
      "step": 7814
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.837916612625122,
      "learning_rate": 1.5625978340734116e-05,
      "loss": 0.9861,
      "step": 7815
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.264280796051025,
      "learning_rate": 1.5625339424336328e-05,
      "loss": 1.5135,
      "step": 7816
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.924153804779053,
      "learning_rate": 1.5624700507938536e-05,
      "loss": 1.9455,
      "step": 7817
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.284090757369995,
      "learning_rate": 1.5624061591540748e-05,
      "loss": 1.0629,
      "step": 7818
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6523429155349731,
      "learning_rate": 1.562342267514296e-05,
      "loss": 0.8879,
      "step": 7819
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.863146424293518,
      "learning_rate": 1.562278375874517e-05,
      "loss": 0.7546,
      "step": 7820
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8533575534820557,
      "learning_rate": 1.562214484234738e-05,
      "loss": 0.7826,
      "step": 7821
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8107575178146362,
      "learning_rate": 1.562150592594959e-05,
      "loss": 0.8607,
      "step": 7822
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9184250831604004,
      "learning_rate": 1.5620867009551802e-05,
      "loss": 0.957,
      "step": 7823
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.766621470451355,
      "learning_rate": 1.5620228093154014e-05,
      "loss": 0.9481,
      "step": 7824
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5995240211486816,
      "learning_rate": 1.5619589176756222e-05,
      "loss": 0.9292,
      "step": 7825
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7462409734725952,
      "learning_rate": 1.5618950260358433e-05,
      "loss": 0.942,
      "step": 7826
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7785652875900269,
      "learning_rate": 1.5618311343960645e-05,
      "loss": 0.9304,
      "step": 7827
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2597193717956543,
      "learning_rate": 1.5617672427562856e-05,
      "loss": 1.188,
      "step": 7828
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.441175937652588,
      "learning_rate": 1.5617033511165065e-05,
      "loss": 1.4342,
      "step": 7829
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.5191261768341064,
      "learning_rate": 1.5616394594767276e-05,
      "loss": 0.8324,
      "step": 7830
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1652960777282715,
      "learning_rate": 1.5615755678369488e-05,
      "loss": 0.8949,
      "step": 7831
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.198326349258423,
      "learning_rate": 1.56151167619717e-05,
      "loss": 0.9627,
      "step": 7832
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6706448793411255,
      "learning_rate": 1.5614477845573907e-05,
      "loss": 0.9445,
      "step": 7833
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9553171396255493,
      "learning_rate": 1.561383892917612e-05,
      "loss": 0.9294,
      "step": 7834
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4133384227752686,
      "learning_rate": 1.561320001277833e-05,
      "loss": 0.8222,
      "step": 7835
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.162997007369995,
      "learning_rate": 1.5612561096380542e-05,
      "loss": 0.927,
      "step": 7836
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.992102026939392,
      "learning_rate": 1.561192217998275e-05,
      "loss": 0.9673,
      "step": 7837
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5715088844299316,
      "learning_rate": 1.561128326358496e-05,
      "loss": 0.836,
      "step": 7838
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2096831798553467,
      "learning_rate": 1.5610644347187173e-05,
      "loss": 1.092,
      "step": 7839
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4575945138931274,
      "learning_rate": 1.5610005430789385e-05,
      "loss": 0.7939,
      "step": 7840
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4294118881225586,
      "learning_rate": 1.5609366514391593e-05,
      "loss": 0.9614,
      "step": 7841
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5579274892807007,
      "learning_rate": 1.5608727597993804e-05,
      "loss": 0.8433,
      "step": 7842
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0655136108398438,
      "learning_rate": 1.5608088681596013e-05,
      "loss": 0.8974,
      "step": 7843
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5347838401794434,
      "learning_rate": 1.5607449765198228e-05,
      "loss": 0.8552,
      "step": 7844
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7197009325027466,
      "learning_rate": 1.5606810848800436e-05,
      "loss": 0.7872,
      "step": 7845
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4361977577209473,
      "learning_rate": 1.5606171932402647e-05,
      "loss": 0.9701,
      "step": 7846
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7314847707748413,
      "learning_rate": 1.5605533016004855e-05,
      "loss": 0.9529,
      "step": 7847
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6640465259552002,
      "learning_rate": 1.5604894099607067e-05,
      "loss": 0.8779,
      "step": 7848
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0374228954315186,
      "learning_rate": 1.560425518320928e-05,
      "loss": 0.8924,
      "step": 7849
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4451731443405151,
      "learning_rate": 1.560361626681149e-05,
      "loss": 0.4269,
      "step": 7850
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.074453353881836,
      "learning_rate": 1.5602977350413698e-05,
      "loss": 1.3551,
      "step": 7851
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.236616611480713,
      "learning_rate": 1.560233843401591e-05,
      "loss": 1.0376,
      "step": 7852
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.859847903251648,
      "learning_rate": 1.560169951761812e-05,
      "loss": 0.933,
      "step": 7853
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8454235792160034,
      "learning_rate": 1.5601060601220333e-05,
      "loss": 0.8877,
      "step": 7854
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.597827672958374,
      "learning_rate": 1.560042168482254e-05,
      "loss": 1.1317,
      "step": 7855
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5967044830322266,
      "learning_rate": 1.5599782768424753e-05,
      "loss": 0.4122,
      "step": 7856
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.91709303855896,
      "learning_rate": 1.5599143852026964e-05,
      "loss": 0.8456,
      "step": 7857
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.187920093536377,
      "learning_rate": 1.5598504935629176e-05,
      "loss": 1.4143,
      "step": 7858
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.966925024986267,
      "learning_rate": 1.5597866019231384e-05,
      "loss": 0.8611,
      "step": 7859
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6815377473831177,
      "learning_rate": 1.5597227102833595e-05,
      "loss": 0.9674,
      "step": 7860
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2832484245300293,
      "learning_rate": 1.5596588186435807e-05,
      "loss": 1.2198,
      "step": 7861
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.867349624633789,
      "learning_rate": 1.559594927003802e-05,
      "loss": 1.1368,
      "step": 7862
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7631514072418213,
      "learning_rate": 1.5595310353640227e-05,
      "loss": 0.986,
      "step": 7863
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7322911024093628,
      "learning_rate": 1.5594671437242438e-05,
      "loss": 0.9901,
      "step": 7864
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6009931564331055,
      "learning_rate": 1.559403252084465e-05,
      "loss": 1.0465,
      "step": 7865
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2279601097106934,
      "learning_rate": 1.559339360444686e-05,
      "loss": 0.9156,
      "step": 7866
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.03360652923584,
      "learning_rate": 1.559275468804907e-05,
      "loss": 1.6217,
      "step": 7867
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8442286252975464,
      "learning_rate": 1.559211577165128e-05,
      "loss": 0.9934,
      "step": 7868
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1311941146850586,
      "learning_rate": 1.5591476855253493e-05,
      "loss": 1.1564,
      "step": 7869
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.398829460144043,
      "learning_rate": 1.5590837938855704e-05,
      "loss": 0.8788,
      "step": 7870
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6147034168243408,
      "learning_rate": 1.5590199022457912e-05,
      "loss": 0.8433,
      "step": 7871
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0652573108673096,
      "learning_rate": 1.5589560106060124e-05,
      "loss": 1.0543,
      "step": 7872
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4679038524627686,
      "learning_rate": 1.5588921189662335e-05,
      "loss": 0.9723,
      "step": 7873
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8192894458770752,
      "learning_rate": 1.5588282273264547e-05,
      "loss": 0.859,
      "step": 7874
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.236274242401123,
      "learning_rate": 1.5587643356866755e-05,
      "loss": 1.2748,
      "step": 7875
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9103641510009766,
      "learning_rate": 1.5587004440468967e-05,
      "loss": 0.9968,
      "step": 7876
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6194664239883423,
      "learning_rate": 1.5586365524071175e-05,
      "loss": 0.926,
      "step": 7877
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.7493233680725098,
      "learning_rate": 1.558572660767339e-05,
      "loss": 1.3486,
      "step": 7878
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.84550142288208,
      "learning_rate": 1.5585087691275598e-05,
      "loss": 1.067,
      "step": 7879
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.356593132019043,
      "learning_rate": 1.558444877487781e-05,
      "loss": 1.0983,
      "step": 7880
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7477030754089355,
      "learning_rate": 1.5583809858480018e-05,
      "loss": 0.9004,
      "step": 7881
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4297540187835693,
      "learning_rate": 1.5583170942082232e-05,
      "loss": 0.9846,
      "step": 7882
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.0077691078186035,
      "learning_rate": 1.558253202568444e-05,
      "loss": 0.9181,
      "step": 7883
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.032360553741455,
      "learning_rate": 1.5581893109286652e-05,
      "loss": 1.0229,
      "step": 7884
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.088233232498169,
      "learning_rate": 1.558125419288886e-05,
      "loss": 1.2608,
      "step": 7885
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8274686336517334,
      "learning_rate": 1.5580615276491072e-05,
      "loss": 0.9281,
      "step": 7886
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.8687314987182617,
      "learning_rate": 1.5579976360093283e-05,
      "loss": 0.911,
      "step": 7887
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.8344228267669678,
      "learning_rate": 1.5579337443695495e-05,
      "loss": 1.3973,
      "step": 7888
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.245433807373047,
      "learning_rate": 1.5578698527297703e-05,
      "loss": 1.3939,
      "step": 7889
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.919474720954895,
      "learning_rate": 1.5578059610899915e-05,
      "loss": 0.9303,
      "step": 7890
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7448922395706177,
      "learning_rate": 1.5577420694502126e-05,
      "loss": 0.9468,
      "step": 7891
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.100302219390869,
      "learning_rate": 1.5576781778104338e-05,
      "loss": 0.8996,
      "step": 7892
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7109456062316895,
      "learning_rate": 1.5576142861706546e-05,
      "loss": 0.8576,
      "step": 7893
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.891364097595215,
      "learning_rate": 1.5575503945308757e-05,
      "loss": 1.2866,
      "step": 7894
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0212042331695557,
      "learning_rate": 1.557486502891097e-05,
      "loss": 0.9776,
      "step": 7895
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.001660108566284,
      "learning_rate": 1.557422611251318e-05,
      "loss": 1.1485,
      "step": 7896
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.518804669380188,
      "learning_rate": 1.557358719611539e-05,
      "loss": 0.7731,
      "step": 7897
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6359766721725464,
      "learning_rate": 1.55729482797176e-05,
      "loss": 0.3827,
      "step": 7898
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.2479047775268555,
      "learning_rate": 1.5572309363319812e-05,
      "loss": 1.1594,
      "step": 7899
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6334855556488037,
      "learning_rate": 1.5571670446922023e-05,
      "loss": 0.4124,
      "step": 7900
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.120112895965576,
      "learning_rate": 1.557103153052423e-05,
      "loss": 1.0817,
      "step": 7901
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6752007007598877,
      "learning_rate": 1.5570392614126443e-05,
      "loss": 1.1029,
      "step": 7902
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1725285053253174,
      "learning_rate": 1.5569753697728655e-05,
      "loss": 0.8134,
      "step": 7903
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.148838520050049,
      "learning_rate": 1.5569114781330866e-05,
      "loss": 0.9148,
      "step": 7904
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.012624740600586,
      "learning_rate": 1.5568475864933074e-05,
      "loss": 0.9319,
      "step": 7905
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8518112897872925,
      "learning_rate": 1.5567836948535286e-05,
      "loss": 0.9408,
      "step": 7906
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.191737651824951,
      "learning_rate": 1.5567198032137497e-05,
      "loss": 0.8559,
      "step": 7907
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.2996387481689453,
      "learning_rate": 1.556655911573971e-05,
      "loss": 0.8261,
      "step": 7908
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.677340030670166,
      "learning_rate": 1.5565920199341917e-05,
      "loss": 0.7624,
      "step": 7909
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.456151485443115,
      "learning_rate": 1.556528128294413e-05,
      "loss": 1.0988,
      "step": 7910
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9871612787246704,
      "learning_rate": 1.556464236654634e-05,
      "loss": 1.0923,
      "step": 7911
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5472155809402466,
      "learning_rate": 1.5564003450148552e-05,
      "loss": 0.3424,
      "step": 7912
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.082754611968994,
      "learning_rate": 1.556336453375076e-05,
      "loss": 0.954,
      "step": 7913
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8392428159713745,
      "learning_rate": 1.556272561735297e-05,
      "loss": 0.9029,
      "step": 7914
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6954889297485352,
      "learning_rate": 1.556208670095518e-05,
      "loss": 0.8412,
      "step": 7915
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8596415519714355,
      "learning_rate": 1.5561447784557395e-05,
      "loss": 1.2895,
      "step": 7916
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.9487829208374023,
      "learning_rate": 1.5560808868159603e-05,
      "loss": 0.7413,
      "step": 7917
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7463778257369995,
      "learning_rate": 1.5560169951761814e-05,
      "loss": 0.9725,
      "step": 7918
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0321545600891113,
      "learning_rate": 1.5559531035364022e-05,
      "loss": 0.8671,
      "step": 7919
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5299774408340454,
      "learning_rate": 1.5558892118966234e-05,
      "loss": 0.839,
      "step": 7920
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6772220134735107,
      "learning_rate": 1.5558253202568446e-05,
      "loss": 0.9949,
      "step": 7921
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.295291900634766,
      "learning_rate": 1.5557614286170657e-05,
      "loss": 1.399,
      "step": 7922
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.944082498550415,
      "learning_rate": 1.5556975369772865e-05,
      "loss": 0.9029,
      "step": 7923
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.851585030555725,
      "learning_rate": 1.5556336453375077e-05,
      "loss": 0.998,
      "step": 7924
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5085405111312866,
      "learning_rate": 1.555569753697729e-05,
      "loss": 0.7861,
      "step": 7925
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1003637313842773,
      "learning_rate": 1.5555058620579497e-05,
      "loss": 0.9421,
      "step": 7926
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7659837007522583,
      "learning_rate": 1.5554419704181708e-05,
      "loss": 0.987,
      "step": 7927
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9165254831314087,
      "learning_rate": 1.555378078778392e-05,
      "loss": 0.9787,
      "step": 7928
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.6332623958587646,
      "learning_rate": 1.555314187138613e-05,
      "loss": 1.0661,
      "step": 7929
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.5945873260498047,
      "learning_rate": 1.555250295498834e-05,
      "loss": 1.1718,
      "step": 7930
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.3650033473968506,
      "learning_rate": 1.555186403859055e-05,
      "loss": 1.0436,
      "step": 7931
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.038552761077881,
      "learning_rate": 1.5551225122192762e-05,
      "loss": 0.9756,
      "step": 7932
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2987358570098877,
      "learning_rate": 1.5550586205794974e-05,
      "loss": 1.0867,
      "step": 7933
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1161932945251465,
      "learning_rate": 1.5549947289397182e-05,
      "loss": 1.0668,
      "step": 7934
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.009143114089966,
      "learning_rate": 1.5549308372999394e-05,
      "loss": 0.8482,
      "step": 7935
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.660521984100342,
      "learning_rate": 1.5548669456601605e-05,
      "loss": 0.9957,
      "step": 7936
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.781876802444458,
      "learning_rate": 1.5548030540203817e-05,
      "loss": 0.9506,
      "step": 7937
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6252394914627075,
      "learning_rate": 1.5547391623806025e-05,
      "loss": 0.8646,
      "step": 7938
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7593798637390137,
      "learning_rate": 1.5546752707408236e-05,
      "loss": 0.9526,
      "step": 7939
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0453569889068604,
      "learning_rate": 1.5546113791010448e-05,
      "loss": 0.89,
      "step": 7940
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9908907413482666,
      "learning_rate": 1.554547487461266e-05,
      "loss": 0.9523,
      "step": 7941
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.929916501045227,
      "learning_rate": 1.5544835958214868e-05,
      "loss": 0.8991,
      "step": 7942
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7951916456222534,
      "learning_rate": 1.554419704181708e-05,
      "loss": 0.9522,
      "step": 7943
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7853569984436035,
      "learning_rate": 1.5543558125419287e-05,
      "loss": 0.867,
      "step": 7944
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.265226364135742,
      "learning_rate": 1.5542919209021502e-05,
      "loss": 0.9223,
      "step": 7945
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.495511054992676,
      "learning_rate": 1.554228029262371e-05,
      "loss": 0.6937,
      "step": 7946
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.0807278156280518,
      "learning_rate": 1.5541641376225922e-05,
      "loss": 1.0303,
      "step": 7947
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.376304626464844,
      "learning_rate": 1.554100245982813e-05,
      "loss": 0.5432,
      "step": 7948
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.618009328842163,
      "learning_rate": 1.5540363543430345e-05,
      "loss": 1.2966,
      "step": 7949
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.127281427383423,
      "learning_rate": 1.5539724627032553e-05,
      "loss": 1.0097,
      "step": 7950
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.871770977973938,
      "learning_rate": 1.5539085710634765e-05,
      "loss": 0.9266,
      "step": 7951
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.38826847076416,
      "learning_rate": 1.5538446794236973e-05,
      "loss": 1.4493,
      "step": 7952
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0581603050231934,
      "learning_rate": 1.5537807877839185e-05,
      "loss": 0.9432,
      "step": 7953
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8114988803863525,
      "learning_rate": 1.5537168961441396e-05,
      "loss": 0.4516,
      "step": 7954
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.173835039138794,
      "learning_rate": 1.5536530045043608e-05,
      "loss": 1.0988,
      "step": 7955
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4929901361465454,
      "learning_rate": 1.5535891128645816e-05,
      "loss": 0.7719,
      "step": 7956
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.792227029800415,
      "learning_rate": 1.5535252212248027e-05,
      "loss": 0.9653,
      "step": 7957
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.239335536956787,
      "learning_rate": 1.553461329585024e-05,
      "loss": 1.3458,
      "step": 7958
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0730843544006348,
      "learning_rate": 1.553397437945245e-05,
      "loss": 1.0556,
      "step": 7959
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0155935287475586,
      "learning_rate": 1.553333546305466e-05,
      "loss": 1.1744,
      "step": 7960
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9323005676269531,
      "learning_rate": 1.553269654665687e-05,
      "loss": 0.9017,
      "step": 7961
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.990947961807251,
      "learning_rate": 1.5532057630259082e-05,
      "loss": 0.7585,
      "step": 7962
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7914471626281738,
      "learning_rate": 1.5531418713861293e-05,
      "loss": 0.8239,
      "step": 7963
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8045765161514282,
      "learning_rate": 1.55307797974635e-05,
      "loss": 0.9307,
      "step": 7964
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9162153005599976,
      "learning_rate": 1.5530140881065713e-05,
      "loss": 0.9334,
      "step": 7965
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8236751556396484,
      "learning_rate": 1.5529501964667925e-05,
      "loss": 0.4658,
      "step": 7966
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9474176168441772,
      "learning_rate": 1.5528863048270136e-05,
      "loss": 0.934,
      "step": 7967
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.5623676776885986,
      "learning_rate": 1.5528224131872344e-05,
      "loss": 0.9715,
      "step": 7968
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.649514675140381,
      "learning_rate": 1.5527585215474556e-05,
      "loss": 1.3899,
      "step": 7969
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6433109045028687,
      "learning_rate": 1.5526946299076767e-05,
      "loss": 0.9065,
      "step": 7970
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.039040565490723,
      "learning_rate": 1.552630738267898e-05,
      "loss": 1.7442,
      "step": 7971
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.133697271347046,
      "learning_rate": 1.5525668466281187e-05,
      "loss": 1.0042,
      "step": 7972
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9219155311584473,
      "learning_rate": 1.55250295498834e-05,
      "loss": 0.9695,
      "step": 7973
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8527967929840088,
      "learning_rate": 1.552439063348561e-05,
      "loss": 0.9493,
      "step": 7974
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.707735538482666,
      "learning_rate": 1.552375171708782e-05,
      "loss": 0.8957,
      "step": 7975
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3302505016326904,
      "learning_rate": 1.552311280069003e-05,
      "loss": 1.0585,
      "step": 7976
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0327155590057373,
      "learning_rate": 1.552247388429224e-05,
      "loss": 0.9275,
      "step": 7977
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0477142333984375,
      "learning_rate": 1.5521834967894453e-05,
      "loss": 0.9641,
      "step": 7978
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8676626682281494,
      "learning_rate": 1.5521196051496664e-05,
      "loss": 0.8604,
      "step": 7979
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1485824584960938,
      "learning_rate": 1.5520557135098873e-05,
      "loss": 0.9403,
      "step": 7980
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5946333408355713,
      "learning_rate": 1.5519918218701084e-05,
      "loss": 0.8655,
      "step": 7981
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8440444469451904,
      "learning_rate": 1.5519279302303292e-05,
      "loss": 1.0533,
      "step": 7982
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7085511684417725,
      "learning_rate": 1.5518640385905507e-05,
      "loss": 0.9177,
      "step": 7983
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0861470699310303,
      "learning_rate": 1.5518001469507715e-05,
      "loss": 1.0368,
      "step": 7984
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2923827171325684,
      "learning_rate": 1.5517362553109927e-05,
      "loss": 0.9111,
      "step": 7985
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.572462320327759,
      "learning_rate": 1.5516723636712135e-05,
      "loss": 1.1066,
      "step": 7986
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.313268184661865,
      "learning_rate": 1.5516084720314347e-05,
      "loss": 1.0947,
      "step": 7987
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6818245649337769,
      "learning_rate": 1.5515445803916558e-05,
      "loss": 0.9089,
      "step": 7988
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8023183345794678,
      "learning_rate": 1.551480688751877e-05,
      "loss": 0.9723,
      "step": 7989
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9774724245071411,
      "learning_rate": 1.5514167971120978e-05,
      "loss": 1.1013,
      "step": 7990
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.019984722137451,
      "learning_rate": 1.551352905472319e-05,
      "loss": 1.2209,
      "step": 7991
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.994284987449646,
      "learning_rate": 1.55128901383254e-05,
      "loss": 0.9035,
      "step": 7992
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6687675714492798,
      "learning_rate": 1.5512251221927613e-05,
      "loss": 0.7393,
      "step": 7993
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.462770462036133,
      "learning_rate": 1.551161230552982e-05,
      "loss": 0.8919,
      "step": 7994
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.204165935516357,
      "learning_rate": 1.5510973389132032e-05,
      "loss": 1.0183,
      "step": 7995
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7754461765289307,
      "learning_rate": 1.5510334472734244e-05,
      "loss": 0.8819,
      "step": 7996
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7992948293685913,
      "learning_rate": 1.5509695556336455e-05,
      "loss": 1.007,
      "step": 7997
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0351955890655518,
      "learning_rate": 1.5509056639938664e-05,
      "loss": 0.9997,
      "step": 7998
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.087523460388184,
      "learning_rate": 1.5508417723540875e-05,
      "loss": 1.6603,
      "step": 7999
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7432571649551392,
      "learning_rate": 1.5507778807143087e-05,
      "loss": 1.16,
      "step": 8000
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.701648712158203,
      "learning_rate": 1.5507139890745298e-05,
      "loss": 1.1704,
      "step": 8001
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.897793173789978,
      "learning_rate": 1.5506500974347506e-05,
      "loss": 0.9312,
      "step": 8002
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.50283944606781,
      "learning_rate": 1.5505862057949718e-05,
      "loss": 0.3787,
      "step": 8003
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.907128095626831,
      "learning_rate": 1.550522314155193e-05,
      "loss": 1.0554,
      "step": 8004
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6914708614349365,
      "learning_rate": 1.550458422515414e-05,
      "loss": 0.8811,
      "step": 8005
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.688064455986023,
      "learning_rate": 1.550394530875635e-05,
      "loss": 0.7755,
      "step": 8006
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.584956169128418,
      "learning_rate": 1.550330639235856e-05,
      "loss": 0.6365,
      "step": 8007
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.479506254196167,
      "learning_rate": 1.5502667475960772e-05,
      "loss": 1.0608,
      "step": 8008
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9664101600646973,
      "learning_rate": 1.5502028559562984e-05,
      "loss": 0.933,
      "step": 8009
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.998053789138794,
      "learning_rate": 1.5501389643165192e-05,
      "loss": 1.0858,
      "step": 8010
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.483232021331787,
      "learning_rate": 1.5500750726767403e-05,
      "loss": 1.4197,
      "step": 8011
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1431567668914795,
      "learning_rate": 1.5500111810369615e-05,
      "loss": 0.9351,
      "step": 8012
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3288087844848633,
      "learning_rate": 1.5499472893971827e-05,
      "loss": 1.0548,
      "step": 8013
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0179131031036377,
      "learning_rate": 1.5498833977574035e-05,
      "loss": 0.8366,
      "step": 8014
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.80195951461792,
      "learning_rate": 1.5498195061176246e-05,
      "loss": 0.4104,
      "step": 8015
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.32822847366333,
      "learning_rate": 1.5497556144778458e-05,
      "loss": 1.7749,
      "step": 8016
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.58193838596344,
      "learning_rate": 1.549691722838067e-05,
      "loss": 0.3785,
      "step": 8017
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2721848487854004,
      "learning_rate": 1.5496278311982878e-05,
      "loss": 0.9711,
      "step": 8018
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.066899061203003,
      "learning_rate": 1.549563939558509e-05,
      "loss": 0.8201,
      "step": 8019
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.819210410118103,
      "learning_rate": 1.5495000479187297e-05,
      "loss": 0.9034,
      "step": 8020
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2615966796875,
      "learning_rate": 1.5494361562789512e-05,
      "loss": 1.3165,
      "step": 8021
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4220874309539795,
      "learning_rate": 1.549372264639172e-05,
      "loss": 0.9614,
      "step": 8022
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.757659912109375,
      "learning_rate": 1.5493083729993932e-05,
      "loss": 1.0034,
      "step": 8023
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9227933883666992,
      "learning_rate": 1.549244481359614e-05,
      "loss": 0.9179,
      "step": 8024
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9671456813812256,
      "learning_rate": 1.549180589719835e-05,
      "loss": 0.9565,
      "step": 8025
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.482324123382568,
      "learning_rate": 1.5491166980800563e-05,
      "loss": 1.3109,
      "step": 8026
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.06018590927124,
      "learning_rate": 1.5490528064402775e-05,
      "loss": 1.281,
      "step": 8027
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8695014715194702,
      "learning_rate": 1.5489889148004983e-05,
      "loss": 0.7637,
      "step": 8028
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5699796676635742,
      "learning_rate": 1.5489250231607194e-05,
      "loss": 0.3978,
      "step": 8029
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.609617829322815,
      "learning_rate": 1.5488611315209406e-05,
      "loss": 0.8475,
      "step": 8030
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4424006938934326,
      "learning_rate": 1.5487972398811617e-05,
      "loss": 1.1822,
      "step": 8031
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.9641432762145996,
      "learning_rate": 1.5487333482413826e-05,
      "loss": 1.1369,
      "step": 8032
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3431687355041504,
      "learning_rate": 1.5486694566016037e-05,
      "loss": 0.881,
      "step": 8033
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4641830921173096,
      "learning_rate": 1.548605564961825e-05,
      "loss": 1.0401,
      "step": 8034
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.092965602874756,
      "learning_rate": 1.548541673322046e-05,
      "loss": 0.6789,
      "step": 8035
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.985068440437317,
      "learning_rate": 1.548477781682267e-05,
      "loss": 1.2606,
      "step": 8036
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9902653694152832,
      "learning_rate": 1.548413890042488e-05,
      "loss": 0.8199,
      "step": 8037
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.588808059692383,
      "learning_rate": 1.548349998402709e-05,
      "loss": 1.1422,
      "step": 8038
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.7915070056915283,
      "learning_rate": 1.5482861067629303e-05,
      "loss": 1.1788,
      "step": 8039
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4603317975997925,
      "learning_rate": 1.548222215123151e-05,
      "loss": 0.4289,
      "step": 8040
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0121543407440186,
      "learning_rate": 1.5481583234833723e-05,
      "loss": 0.905,
      "step": 8041
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.899935245513916,
      "learning_rate": 1.5480944318435934e-05,
      "loss": 0.9104,
      "step": 8042
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6663243770599365,
      "learning_rate": 1.5480305402038146e-05,
      "loss": 0.9806,
      "step": 8043
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.381439208984375,
      "learning_rate": 1.5479666485640354e-05,
      "loss": 0.9019,
      "step": 8044
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.791205883026123,
      "learning_rate": 1.5479027569242566e-05,
      "loss": 0.9149,
      "step": 8045
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7728700637817383,
      "learning_rate": 1.5478388652844777e-05,
      "loss": 0.3375,
      "step": 8046
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.915955662727356,
      "learning_rate": 1.547774973644699e-05,
      "loss": 0.8599,
      "step": 8047
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7380621433258057,
      "learning_rate": 1.5477110820049197e-05,
      "loss": 0.9463,
      "step": 8048
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6513066291809082,
      "learning_rate": 1.547647190365141e-05,
      "loss": 0.971,
      "step": 8049
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.312840700149536,
      "learning_rate": 1.547583298725362e-05,
      "loss": 0.9702,
      "step": 8050
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2143774032592773,
      "learning_rate": 1.547519407085583e-05,
      "loss": 0.9594,
      "step": 8051
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.170008659362793,
      "learning_rate": 1.547455515445804e-05,
      "loss": 0.9976,
      "step": 8052
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.235567569732666,
      "learning_rate": 1.547391623806025e-05,
      "loss": 1.1504,
      "step": 8053
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.5402774810791016,
      "learning_rate": 1.547327732166246e-05,
      "loss": 1.0475,
      "step": 8054
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3177239894866943,
      "learning_rate": 1.5472638405264674e-05,
      "loss": 1.2653,
      "step": 8055
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5903302431106567,
      "learning_rate": 1.5471999488866882e-05,
      "loss": 0.9474,
      "step": 8056
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.6350972652435303,
      "learning_rate": 1.5471360572469094e-05,
      "loss": 1.1735,
      "step": 8057
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8526315689086914,
      "learning_rate": 1.5470721656071302e-05,
      "loss": 0.8986,
      "step": 8058
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.721670389175415,
      "learning_rate": 1.5470082739673517e-05,
      "loss": 0.9043,
      "step": 8059
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.422451138496399,
      "learning_rate": 1.5469443823275725e-05,
      "loss": 0.4394,
      "step": 8060
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7735910415649414,
      "learning_rate": 1.5468804906877937e-05,
      "loss": 0.9249,
      "step": 8061
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4135656356811523,
      "learning_rate": 1.5468165990480145e-05,
      "loss": 1.0616,
      "step": 8062
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8793408870697021,
      "learning_rate": 1.5467527074082356e-05,
      "loss": 0.8915,
      "step": 8063
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7552027702331543,
      "learning_rate": 1.5466888157684568e-05,
      "loss": 0.7914,
      "step": 8064
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.646676778793335,
      "learning_rate": 1.546624924128678e-05,
      "loss": 0.8716,
      "step": 8065
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1561741828918457,
      "learning_rate": 1.5465610324888988e-05,
      "loss": 0.9798,
      "step": 8066
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9496511220932007,
      "learning_rate": 1.54649714084912e-05,
      "loss": 0.9579,
      "step": 8067
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7398735284805298,
      "learning_rate": 1.546433249209341e-05,
      "loss": 0.9934,
      "step": 8068
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.295497179031372,
      "learning_rate": 1.5463693575695622e-05,
      "loss": 1.019,
      "step": 8069
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.675859212875366,
      "learning_rate": 1.546305465929783e-05,
      "loss": 1.1844,
      "step": 8070
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.200653076171875,
      "learning_rate": 1.5462415742900042e-05,
      "loss": 0.9251,
      "step": 8071
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.1908652782440186,
      "learning_rate": 1.5461776826502254e-05,
      "loss": 0.6618,
      "step": 8072
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3245177268981934,
      "learning_rate": 1.5461137910104465e-05,
      "loss": 1.2099,
      "step": 8073
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1701860427856445,
      "learning_rate": 1.5460498993706673e-05,
      "loss": 0.9794,
      "step": 8074
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.83975887298584,
      "learning_rate": 1.5459860077308885e-05,
      "loss": 1.8159,
      "step": 8075
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9791911840438843,
      "learning_rate": 1.5459221160911096e-05,
      "loss": 1.1479,
      "step": 8076
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6521823406219482,
      "learning_rate": 1.5458582244513308e-05,
      "loss": 1.0044,
      "step": 8077
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7315505743026733,
      "learning_rate": 1.5457943328115516e-05,
      "loss": 0.9475,
      "step": 8078
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6980994939804077,
      "learning_rate": 1.5457304411717728e-05,
      "loss": 0.4779,
      "step": 8079
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.6091763973236084,
      "learning_rate": 1.545666549531994e-05,
      "loss": 0.8101,
      "step": 8080
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9576420783996582,
      "learning_rate": 1.545602657892215e-05,
      "loss": 0.9629,
      "step": 8081
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1059634685516357,
      "learning_rate": 1.545538766252436e-05,
      "loss": 0.893,
      "step": 8082
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5955240726470947,
      "learning_rate": 1.545474874612657e-05,
      "loss": 0.9838,
      "step": 8083
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5825284719467163,
      "learning_rate": 1.5454109829728782e-05,
      "loss": 0.7866,
      "step": 8084
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.366917371749878,
      "learning_rate": 1.5453470913330994e-05,
      "loss": 0.8489,
      "step": 8085
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.19018292427063,
      "learning_rate": 1.5452831996933202e-05,
      "loss": 0.9085,
      "step": 8086
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4358898401260376,
      "learning_rate": 1.5452193080535413e-05,
      "loss": 0.424,
      "step": 8087
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9440901279449463,
      "learning_rate": 1.5451554164137625e-05,
      "loss": 0.8913,
      "step": 8088
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1846513748168945,
      "learning_rate": 1.5450915247739836e-05,
      "loss": 1.0242,
      "step": 8089
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6596609354019165,
      "learning_rate": 1.5450276331342045e-05,
      "loss": 0.9117,
      "step": 8090
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1204590797424316,
      "learning_rate": 1.5449637414944256e-05,
      "loss": 0.9055,
      "step": 8091
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0519769191741943,
      "learning_rate": 1.5448998498546464e-05,
      "loss": 0.9005,
      "step": 8092
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.619889497756958,
      "learning_rate": 1.544835958214868e-05,
      "loss": 0.9457,
      "step": 8093
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6749383211135864,
      "learning_rate": 1.5447720665750887e-05,
      "loss": 0.8821,
      "step": 8094
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.105954647064209,
      "learning_rate": 1.54470817493531e-05,
      "loss": 1.2362,
      "step": 8095
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0731801986694336,
      "learning_rate": 1.5446442832955307e-05,
      "loss": 0.909,
      "step": 8096
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.8268933296203613,
      "learning_rate": 1.544580391655752e-05,
      "loss": 1.0385,
      "step": 8097
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.943190574645996,
      "learning_rate": 1.544516500015973e-05,
      "loss": 1.0772,
      "step": 8098
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.665856957435608,
      "learning_rate": 1.544452608376194e-05,
      "loss": 0.8316,
      "step": 8099
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7496356964111328,
      "learning_rate": 1.544388716736415e-05,
      "loss": 1.0264,
      "step": 8100
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.501125454902649,
      "learning_rate": 1.544324825096636e-05,
      "loss": 0.4525,
      "step": 8101
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.041597366333008,
      "learning_rate": 1.5442609334568573e-05,
      "loss": 0.901,
      "step": 8102
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.53564190864563,
      "learning_rate": 1.5441970418170784e-05,
      "loss": 1.0157,
      "step": 8103
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.808822512626648,
      "learning_rate": 1.5441331501772993e-05,
      "loss": 0.9124,
      "step": 8104
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0470333099365234,
      "learning_rate": 1.5440692585375204e-05,
      "loss": 0.986,
      "step": 8105
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.70096755027771,
      "learning_rate": 1.5440053668977416e-05,
      "loss": 0.9704,
      "step": 8106
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8435916900634766,
      "learning_rate": 1.5439414752579627e-05,
      "loss": 0.9511,
      "step": 8107
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.033813953399658,
      "learning_rate": 1.5438775836181835e-05,
      "loss": 0.7781,
      "step": 8108
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.06498122215271,
      "learning_rate": 1.5438136919784047e-05,
      "loss": 0.9828,
      "step": 8109
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6934149265289307,
      "learning_rate": 1.543749800338626e-05,
      "loss": 0.9285,
      "step": 8110
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0459845066070557,
      "learning_rate": 1.543685908698847e-05,
      "loss": 1.2428,
      "step": 8111
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.3377532958984375,
      "learning_rate": 1.5436220170590678e-05,
      "loss": 1.0541,
      "step": 8112
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.745598554611206,
      "learning_rate": 1.543558125419289e-05,
      "loss": 1.0323,
      "step": 8113
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6889739036560059,
      "learning_rate": 1.54349423377951e-05,
      "loss": 0.9335,
      "step": 8114
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3665236234664917,
      "learning_rate": 1.5434303421397313e-05,
      "loss": 0.9505,
      "step": 8115
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6362899541854858,
      "learning_rate": 1.543366450499952e-05,
      "loss": 0.9033,
      "step": 8116
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7525343894958496,
      "learning_rate": 1.5433025588601733e-05,
      "loss": 0.8887,
      "step": 8117
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6966243982315063,
      "learning_rate": 1.5432386672203944e-05,
      "loss": 1.0072,
      "step": 8118
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5523159503936768,
      "learning_rate": 1.5431747755806156e-05,
      "loss": 0.9635,
      "step": 8119
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.9245758056640625,
      "learning_rate": 1.5431108839408364e-05,
      "loss": 1.299,
      "step": 8120
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.455066204071045,
      "learning_rate": 1.5430469923010575e-05,
      "loss": 0.7542,
      "step": 8121
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.040318727493286,
      "learning_rate": 1.5429831006612787e-05,
      "loss": 1.0054,
      "step": 8122
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8217575550079346,
      "learning_rate": 1.5429192090215e-05,
      "loss": 0.9527,
      "step": 8123
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9636505842208862,
      "learning_rate": 1.5428553173817207e-05,
      "loss": 0.8843,
      "step": 8124
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.051687717437744,
      "learning_rate": 1.5427914257419418e-05,
      "loss": 1.0949,
      "step": 8125
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8047990798950195,
      "learning_rate": 1.542727534102163e-05,
      "loss": 0.9045,
      "step": 8126
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.825122117996216,
      "learning_rate": 1.542663642462384e-05,
      "loss": 0.9298,
      "step": 8127
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.8619344234466553,
      "learning_rate": 1.542599750822605e-05,
      "loss": 0.659,
      "step": 8128
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5667062997817993,
      "learning_rate": 1.542535859182826e-05,
      "loss": 0.8313,
      "step": 8129
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.685000419616699,
      "learning_rate": 1.542471967543047e-05,
      "loss": 1.0622,
      "step": 8130
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.093740224838257,
      "learning_rate": 1.5424080759032684e-05,
      "loss": 1.1432,
      "step": 8131
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.802931308746338,
      "learning_rate": 1.5423441842634892e-05,
      "loss": 0.9739,
      "step": 8132
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.66110098361969,
      "learning_rate": 1.5422802926237104e-05,
      "loss": 0.8965,
      "step": 8133
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.133843183517456,
      "learning_rate": 1.5422164009839312e-05,
      "loss": 1.0046,
      "step": 8134
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.17688512802124,
      "learning_rate": 1.5421525093441523e-05,
      "loss": 1.3063,
      "step": 8135
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.6916956901550293,
      "learning_rate": 1.5420886177043735e-05,
      "loss": 0.8685,
      "step": 8136
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.1590631008148193,
      "learning_rate": 1.5420247260645947e-05,
      "loss": 0.9206,
      "step": 8137
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.179116249084473,
      "learning_rate": 1.5419608344248155e-05,
      "loss": 1.3292,
      "step": 8138
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9480332136154175,
      "learning_rate": 1.5418969427850366e-05,
      "loss": 0.907,
      "step": 8139
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9779824018478394,
      "learning_rate": 1.5418330511452578e-05,
      "loss": 0.9584,
      "step": 8140
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.256434440612793,
      "learning_rate": 1.541769159505479e-05,
      "loss": 0.8598,
      "step": 8141
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1117098331451416,
      "learning_rate": 1.5417052678656998e-05,
      "loss": 1.0905,
      "step": 8142
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.0223388671875,
      "learning_rate": 1.541641376225921e-05,
      "loss": 0.7438,
      "step": 8143
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.30144739151001,
      "learning_rate": 1.541577484586142e-05,
      "loss": 1.0781,
      "step": 8144
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7229197025299072,
      "learning_rate": 1.5415135929463632e-05,
      "loss": 0.566,
      "step": 8145
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.022034645080566,
      "learning_rate": 1.541449701306584e-05,
      "loss": 1.2952,
      "step": 8146
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6266783475875854,
      "learning_rate": 1.5413858096668052e-05,
      "loss": 0.8834,
      "step": 8147
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5237624645233154,
      "learning_rate": 1.5413219180270263e-05,
      "loss": 0.9259,
      "step": 8148
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7496916055679321,
      "learning_rate": 1.5412580263872475e-05,
      "loss": 0.932,
      "step": 8149
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3109564781188965,
      "learning_rate": 1.5411941347474683e-05,
      "loss": 1.0667,
      "step": 8150
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.617945432662964,
      "learning_rate": 1.5411302431076895e-05,
      "loss": 1.0089,
      "step": 8151
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.417633056640625,
      "learning_rate": 1.5410663514679106e-05,
      "loss": 1.2949,
      "step": 8152
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9741123914718628,
      "learning_rate": 1.5410024598281318e-05,
      "loss": 0.8582,
      "step": 8153
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.623274087905884,
      "learning_rate": 1.5409385681883526e-05,
      "loss": 1.0054,
      "step": 8154
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.232418060302734,
      "learning_rate": 1.5408746765485737e-05,
      "loss": 0.7705,
      "step": 8155
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9020116329193115,
      "learning_rate": 1.540810784908795e-05,
      "loss": 0.8739,
      "step": 8156
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.050828695297241,
      "learning_rate": 1.540746893269016e-05,
      "loss": 0.9515,
      "step": 8157
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.93721604347229,
      "learning_rate": 1.540683001629237e-05,
      "loss": 0.8934,
      "step": 8158
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6674245595932007,
      "learning_rate": 1.540619109989458e-05,
      "loss": 0.8475,
      "step": 8159
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.029346942901611,
      "learning_rate": 1.5405552183496792e-05,
      "loss": 1.2138,
      "step": 8160
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2597856521606445,
      "learning_rate": 1.5404913267099003e-05,
      "loss": 0.9864,
      "step": 8161
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.493954658508301,
      "learning_rate": 1.540427435070121e-05,
      "loss": 1.5638,
      "step": 8162
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5211024284362793,
      "learning_rate": 1.5403635434303423e-05,
      "loss": 0.3953,
      "step": 8163
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2282631397247314,
      "learning_rate": 1.540299651790563e-05,
      "loss": 1.224,
      "step": 8164
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7133771181106567,
      "learning_rate": 1.5402357601507846e-05,
      "loss": 0.8877,
      "step": 8165
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5701677799224854,
      "learning_rate": 1.5401718685110054e-05,
      "loss": 0.8884,
      "step": 8166
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.990927815437317,
      "learning_rate": 1.5401079768712266e-05,
      "loss": 1.2171,
      "step": 8167
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.987850308418274,
      "learning_rate": 1.5400440852314474e-05,
      "loss": 1.0392,
      "step": 8168
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4161081314086914,
      "learning_rate": 1.5399801935916686e-05,
      "loss": 0.9608,
      "step": 8169
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.233316659927368,
      "learning_rate": 1.5399163019518897e-05,
      "loss": 1.1557,
      "step": 8170
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.8752787113189697,
      "learning_rate": 1.539852410312111e-05,
      "loss": 1.0198,
      "step": 8171
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0684731006622314,
      "learning_rate": 1.5397885186723317e-05,
      "loss": 0.9098,
      "step": 8172
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.006585121154785,
      "learning_rate": 1.539724627032553e-05,
      "loss": 1.1137,
      "step": 8173
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.20314359664917,
      "learning_rate": 1.539660735392774e-05,
      "loss": 1.0601,
      "step": 8174
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6548326015472412,
      "learning_rate": 1.539596843752995e-05,
      "loss": 0.9449,
      "step": 8175
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6849911212921143,
      "learning_rate": 1.539532952113216e-05,
      "loss": 0.3286,
      "step": 8176
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6308740377426147,
      "learning_rate": 1.539469060473437e-05,
      "loss": 0.4373,
      "step": 8177
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5833535194396973,
      "learning_rate": 1.5394051688336583e-05,
      "loss": 0.9793,
      "step": 8178
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.928453803062439,
      "learning_rate": 1.5393412771938794e-05,
      "loss": 1.017,
      "step": 8179
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.664014458656311,
      "learning_rate": 1.5392773855541002e-05,
      "loss": 0.8941,
      "step": 8180
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0550649166107178,
      "learning_rate": 1.5392134939143214e-05,
      "loss": 0.9237,
      "step": 8181
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7395306825637817,
      "learning_rate": 1.5391496022745426e-05,
      "loss": 0.9683,
      "step": 8182
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.729466199874878,
      "learning_rate": 1.5390857106347637e-05,
      "loss": 0.9718,
      "step": 8183
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.234696626663208,
      "learning_rate": 1.5390218189949845e-05,
      "loss": 1.0496,
      "step": 8184
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.9420523643493652,
      "learning_rate": 1.5389579273552057e-05,
      "loss": 1.6171,
      "step": 8185
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7297836542129517,
      "learning_rate": 1.538894035715427e-05,
      "loss": 1.0023,
      "step": 8186
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.902902841567993,
      "learning_rate": 1.538830144075648e-05,
      "loss": 0.9446,
      "step": 8187
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.205273389816284,
      "learning_rate": 1.5387662524358688e-05,
      "loss": 0.9839,
      "step": 8188
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.876833200454712,
      "learning_rate": 1.53870236079609e-05,
      "loss": 0.976,
      "step": 8189
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.828428030014038,
      "learning_rate": 1.538638469156311e-05,
      "loss": 0.9884,
      "step": 8190
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6548712253570557,
      "learning_rate": 1.5385745775165323e-05,
      "loss": 0.9785,
      "step": 8191
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.640496850013733,
      "learning_rate": 1.538510685876753e-05,
      "loss": 0.4997,
      "step": 8192
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9865106344223022,
      "learning_rate": 1.5384467942369742e-05,
      "loss": 0.9615,
      "step": 8193
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1296653747558594,
      "learning_rate": 1.5383829025971954e-05,
      "loss": 0.9052,
      "step": 8194
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.947121262550354,
      "learning_rate": 1.5383190109574165e-05,
      "loss": 0.9589,
      "step": 8195
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7378472089767456,
      "learning_rate": 1.5382551193176374e-05,
      "loss": 1.2559,
      "step": 8196
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.399713516235352,
      "learning_rate": 1.5381912276778585e-05,
      "loss": 1.3056,
      "step": 8197
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.105910301208496,
      "learning_rate": 1.5381273360380797e-05,
      "loss": 0.8796,
      "step": 8198
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1489028930664062,
      "learning_rate": 1.5380634443983008e-05,
      "loss": 0.8364,
      "step": 8199
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1851820945739746,
      "learning_rate": 1.5379995527585216e-05,
      "loss": 1.2321,
      "step": 8200
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2694902420043945,
      "learning_rate": 1.5379356611187428e-05,
      "loss": 1.0905,
      "step": 8201
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.1542537212371826,
      "learning_rate": 1.5378717694789636e-05,
      "loss": 0.994,
      "step": 8202
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4515862464904785,
      "learning_rate": 1.537807877839185e-05,
      "loss": 0.876,
      "step": 8203
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7303543090820312,
      "learning_rate": 1.537743986199406e-05,
      "loss": 1.0101,
      "step": 8204
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9432868957519531,
      "learning_rate": 1.537680094559627e-05,
      "loss": 0.9893,
      "step": 8205
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.5154595375061035,
      "learning_rate": 1.537616202919848e-05,
      "loss": 1.1675,
      "step": 8206
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.937152147293091,
      "learning_rate": 1.537552311280069e-05,
      "loss": 1.4106,
      "step": 8207
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7530932426452637,
      "learning_rate": 1.5374884196402902e-05,
      "loss": 0.939,
      "step": 8208
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.594851493835449,
      "learning_rate": 1.5374245280005114e-05,
      "loss": 1.3494,
      "step": 8209
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9860022068023682,
      "learning_rate": 1.5373606363607322e-05,
      "loss": 0.9554,
      "step": 8210
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1976733207702637,
      "learning_rate": 1.5372967447209533e-05,
      "loss": 0.9813,
      "step": 8211
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0231099128723145,
      "learning_rate": 1.5372328530811745e-05,
      "loss": 1.3224,
      "step": 8212
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0153448581695557,
      "learning_rate": 1.5371689614413956e-05,
      "loss": 1.0379,
      "step": 8213
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9153988361358643,
      "learning_rate": 1.5371050698016165e-05,
      "loss": 0.9759,
      "step": 8214
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6652171611785889,
      "learning_rate": 1.5370411781618376e-05,
      "loss": 0.9076,
      "step": 8215
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7331798076629639,
      "learning_rate": 1.5369772865220588e-05,
      "loss": 0.9154,
      "step": 8216
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0884947776794434,
      "learning_rate": 1.53691339488228e-05,
      "loss": 0.9176,
      "step": 8217
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9913556575775146,
      "learning_rate": 1.5368495032425007e-05,
      "loss": 1.2654,
      "step": 8218
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.5838828086853027,
      "learning_rate": 1.536785611602722e-05,
      "loss": 0.9742,
      "step": 8219
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6305104494094849,
      "learning_rate": 1.536721719962943e-05,
      "loss": 0.8847,
      "step": 8220
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1120212078094482,
      "learning_rate": 1.5366578283231642e-05,
      "loss": 0.9591,
      "step": 8221
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6423814296722412,
      "learning_rate": 1.536593936683385e-05,
      "loss": 0.8863,
      "step": 8222
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3338255882263184,
      "learning_rate": 1.536530045043606e-05,
      "loss": 0.7813,
      "step": 8223
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.790636420249939,
      "learning_rate": 1.5364661534038273e-05,
      "loss": 0.9269,
      "step": 8224
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.632249116897583,
      "learning_rate": 1.5364022617640485e-05,
      "loss": 0.8482,
      "step": 8225
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.2255001068115234,
      "learning_rate": 1.5363383701242693e-05,
      "loss": 0.9585,
      "step": 8226
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5894581079483032,
      "learning_rate": 1.5362744784844904e-05,
      "loss": 0.7632,
      "step": 8227
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7373005151748657,
      "learning_rate": 1.5362105868447116e-05,
      "loss": 0.9885,
      "step": 8228
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9398771524429321,
      "learning_rate": 1.5361466952049328e-05,
      "loss": 1.3065,
      "step": 8229
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6475223302841187,
      "learning_rate": 1.5360828035651536e-05,
      "loss": 0.8543,
      "step": 8230
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.102160692214966,
      "learning_rate": 1.5360189119253747e-05,
      "loss": 1.0226,
      "step": 8231
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6319044828414917,
      "learning_rate": 1.535955020285596e-05,
      "loss": 0.9403,
      "step": 8232
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0828428268432617,
      "learning_rate": 1.535891128645817e-05,
      "loss": 0.9182,
      "step": 8233
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3110246658325195,
      "learning_rate": 1.535827237006038e-05,
      "loss": 0.758,
      "step": 8234
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.115760803222656,
      "learning_rate": 1.535763345366259e-05,
      "loss": 1.5937,
      "step": 8235
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8711941242218018,
      "learning_rate": 1.5356994537264798e-05,
      "loss": 0.9079,
      "step": 8236
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.142958879470825,
      "learning_rate": 1.5356355620867013e-05,
      "loss": 0.9983,
      "step": 8237
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.897828221321106,
      "learning_rate": 1.535571670446922e-05,
      "loss": 0.9313,
      "step": 8238
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5999559164047241,
      "learning_rate": 1.5355077788071433e-05,
      "loss": 0.8844,
      "step": 8239
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.922366738319397,
      "learning_rate": 1.535443887167364e-05,
      "loss": 0.9421,
      "step": 8240
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0735504627227783,
      "learning_rate": 1.5353799955275856e-05,
      "loss": 1.0208,
      "step": 8241
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0056726932525635,
      "learning_rate": 1.5353161038878064e-05,
      "loss": 0.5324,
      "step": 8242
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7322008609771729,
      "learning_rate": 1.5352522122480276e-05,
      "loss": 0.4382,
      "step": 8243
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.07609486579895,
      "learning_rate": 1.5351883206082484e-05,
      "loss": 0.9789,
      "step": 8244
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7934490442276,
      "learning_rate": 1.5351244289684695e-05,
      "loss": 0.9662,
      "step": 8245
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.323948860168457,
      "learning_rate": 1.5350605373286907e-05,
      "loss": 1.2349,
      "step": 8246
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8277578353881836,
      "learning_rate": 1.534996645688912e-05,
      "loss": 0.8751,
      "step": 8247
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3932185173034668,
      "learning_rate": 1.5349327540491327e-05,
      "loss": 0.4031,
      "step": 8248
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1062324047088623,
      "learning_rate": 1.5348688624093538e-05,
      "loss": 1.0306,
      "step": 8249
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7804062366485596,
      "learning_rate": 1.534804970769575e-05,
      "loss": 0.878,
      "step": 8250
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.798364281654358,
      "learning_rate": 1.534741079129796e-05,
      "loss": 0.8795,
      "step": 8251
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.422344207763672,
      "learning_rate": 1.534677187490017e-05,
      "loss": 1.1004,
      "step": 8252
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.61684250831604,
      "learning_rate": 1.534613295850238e-05,
      "loss": 1.0101,
      "step": 8253
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.7021114826202393,
      "learning_rate": 1.5345494042104593e-05,
      "loss": 0.9931,
      "step": 8254
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.651466965675354,
      "learning_rate": 1.5344855125706804e-05,
      "loss": 0.8967,
      "step": 8255
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8270862102508545,
      "learning_rate": 1.5344216209309012e-05,
      "loss": 0.9281,
      "step": 8256
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.0907139778137207,
      "learning_rate": 1.5343577292911224e-05,
      "loss": 0.4176,
      "step": 8257
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8387213945388794,
      "learning_rate": 1.5342938376513435e-05,
      "loss": 0.8929,
      "step": 8258
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.6296494007110596,
      "learning_rate": 1.5342299460115647e-05,
      "loss": 0.8533,
      "step": 8259
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.575503945350647,
      "learning_rate": 1.5341660543717855e-05,
      "loss": 0.7996,
      "step": 8260
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.008804798126221,
      "learning_rate": 1.5341021627320067e-05,
      "loss": 1.4418,
      "step": 8261
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.177762985229492,
      "learning_rate": 1.5340382710922278e-05,
      "loss": 1.0157,
      "step": 8262
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.7375168800354,
      "learning_rate": 1.533974379452449e-05,
      "loss": 1.0167,
      "step": 8263
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.851826786994934,
      "learning_rate": 1.5339104878126698e-05,
      "loss": 0.8759,
      "step": 8264
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.303781509399414,
      "learning_rate": 1.533846596172891e-05,
      "loss": 1.3869,
      "step": 8265
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.615022897720337,
      "learning_rate": 1.533782704533112e-05,
      "loss": 0.884,
      "step": 8266
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6149113178253174,
      "learning_rate": 1.5337188128933332e-05,
      "loss": 0.3775,
      "step": 8267
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4367258548736572,
      "learning_rate": 1.533654921253554e-05,
      "loss": 0.9727,
      "step": 8268
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3835092782974243,
      "learning_rate": 1.5335910296137752e-05,
      "loss": 0.3887,
      "step": 8269
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6222774982452393,
      "learning_rate": 1.5335271379739964e-05,
      "loss": 0.8501,
      "step": 8270
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.541490077972412,
      "learning_rate": 1.5334632463342175e-05,
      "loss": 1.3045,
      "step": 8271
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.649834632873535,
      "learning_rate": 1.5333993546944383e-05,
      "loss": 1.236,
      "step": 8272
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7190853357315063,
      "learning_rate": 1.5333354630546595e-05,
      "loss": 0.9452,
      "step": 8273
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7964999675750732,
      "learning_rate": 1.5332715714148803e-05,
      "loss": 1.1369,
      "step": 8274
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.025642156600952,
      "learning_rate": 1.5332076797751018e-05,
      "loss": 0.9547,
      "step": 8275
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.398430347442627,
      "learning_rate": 1.5331437881353226e-05,
      "loss": 1.3848,
      "step": 8276
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.996023178100586,
      "learning_rate": 1.5330798964955438e-05,
      "loss": 1.099,
      "step": 8277
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5844392776489258,
      "learning_rate": 1.5330160048557646e-05,
      "loss": 0.8945,
      "step": 8278
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8524181842803955,
      "learning_rate": 1.5329521132159857e-05,
      "loss": 0.9914,
      "step": 8279
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6899330615997314,
      "learning_rate": 1.532888221576207e-05,
      "loss": 0.9943,
      "step": 8280
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.932880163192749,
      "learning_rate": 1.532824329936428e-05,
      "loss": 1.0699,
      "step": 8281
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9999438524246216,
      "learning_rate": 1.532760438296649e-05,
      "loss": 0.9894,
      "step": 8282
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8419837951660156,
      "learning_rate": 1.53269654665687e-05,
      "loss": 0.9767,
      "step": 8283
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.931186676025391,
      "learning_rate": 1.5326326550170912e-05,
      "loss": 1.429,
      "step": 8284
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5092562437057495,
      "learning_rate": 1.5325687633773123e-05,
      "loss": 0.8559,
      "step": 8285
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3878748416900635,
      "learning_rate": 1.532504871737533e-05,
      "loss": 0.8498,
      "step": 8286
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3475422859191895,
      "learning_rate": 1.5324409800977543e-05,
      "loss": 0.9071,
      "step": 8287
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.947218418121338,
      "learning_rate": 1.5323770884579755e-05,
      "loss": 1.0851,
      "step": 8288
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8742104768753052,
      "learning_rate": 1.5323131968181966e-05,
      "loss": 1.1071,
      "step": 8289
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1156563758850098,
      "learning_rate": 1.5322493051784174e-05,
      "loss": 0.8792,
      "step": 8290
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6244131326675415,
      "learning_rate": 1.5321854135386386e-05,
      "loss": 0.4472,
      "step": 8291
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2426583766937256,
      "learning_rate": 1.5321215218988597e-05,
      "loss": 0.8324,
      "step": 8292
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4392260313034058,
      "learning_rate": 1.532057630259081e-05,
      "loss": 0.3693,
      "step": 8293
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6601356267929077,
      "learning_rate": 1.5319937386193017e-05,
      "loss": 1.032,
      "step": 8294
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.572240948677063,
      "learning_rate": 1.531929846979523e-05,
      "loss": 1.0139,
      "step": 8295
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.601339101791382,
      "learning_rate": 1.531865955339744e-05,
      "loss": 0.9603,
      "step": 8296
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8142526149749756,
      "learning_rate": 1.5318020636999652e-05,
      "loss": 0.855,
      "step": 8297
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3915886878967285,
      "learning_rate": 1.531738172060186e-05,
      "loss": 1.3908,
      "step": 8298
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7269182205200195,
      "learning_rate": 1.531674280420407e-05,
      "loss": 0.8372,
      "step": 8299
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.167865753173828,
      "learning_rate": 1.5316103887806283e-05,
      "loss": 1.0444,
      "step": 8300
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.524202346801758,
      "learning_rate": 1.5315464971408495e-05,
      "loss": 0.7909,
      "step": 8301
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6621086597442627,
      "learning_rate": 1.5314826055010703e-05,
      "loss": 0.8942,
      "step": 8302
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0867340564727783,
      "learning_rate": 1.5314187138612914e-05,
      "loss": 1.3344,
      "step": 8303
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0985071659088135,
      "learning_rate": 1.5313548222215126e-05,
      "loss": 1.0092,
      "step": 8304
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2668970823287964,
      "learning_rate": 1.5312909305817337e-05,
      "loss": 0.8113,
      "step": 8305
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.438187599182129,
      "learning_rate": 1.5312270389419546e-05,
      "loss": 0.9923,
      "step": 8306
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8918973207473755,
      "learning_rate": 1.5311631473021757e-05,
      "loss": 0.923,
      "step": 8307
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1523280143737793,
      "learning_rate": 1.531099255662397e-05,
      "loss": 0.4392,
      "step": 8308
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.54693603515625,
      "learning_rate": 1.531035364022618e-05,
      "loss": 1.1463,
      "step": 8309
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0437443256378174,
      "learning_rate": 1.530971472382839e-05,
      "loss": 0.9493,
      "step": 8310
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.868489980697632,
      "learning_rate": 1.53090758074306e-05,
      "loss": 0.9036,
      "step": 8311
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.303961992263794,
      "learning_rate": 1.5308436891032808e-05,
      "loss": 1.0433,
      "step": 8312
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.841385841369629,
      "learning_rate": 1.530779797463502e-05,
      "loss": 1.0218,
      "step": 8313
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5603702068328857,
      "learning_rate": 1.530715905823723e-05,
      "loss": 0.82,
      "step": 8314
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.6127777099609375,
      "learning_rate": 1.530652014183944e-05,
      "loss": 0.7845,
      "step": 8315
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2235465049743652,
      "learning_rate": 1.530588122544165e-05,
      "loss": 1.3127,
      "step": 8316
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.692790508270264,
      "learning_rate": 1.5305242309043862e-05,
      "loss": 1.1335,
      "step": 8317
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8359277248382568,
      "learning_rate": 1.5304603392646074e-05,
      "loss": 0.8523,
      "step": 8318
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.581315040588379,
      "learning_rate": 1.5303964476248282e-05,
      "loss": 0.694,
      "step": 8319
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.867174744606018,
      "learning_rate": 1.5303325559850494e-05,
      "loss": 0.9039,
      "step": 8320
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7108650207519531,
      "learning_rate": 1.5302686643452705e-05,
      "loss": 0.9495,
      "step": 8321
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8537412881851196,
      "learning_rate": 1.5302047727054917e-05,
      "loss": 0.8653,
      "step": 8322
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.452425718307495,
      "learning_rate": 1.5301408810657125e-05,
      "loss": 0.9415,
      "step": 8323
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.138200521469116,
      "learning_rate": 1.5300769894259336e-05,
      "loss": 0.9242,
      "step": 8324
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6923320293426514,
      "learning_rate": 1.5300130977861548e-05,
      "loss": 0.9877,
      "step": 8325
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.03352427482605,
      "learning_rate": 1.529949206146376e-05,
      "loss": 0.9678,
      "step": 8326
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.223578453063965,
      "learning_rate": 1.5298853145065968e-05,
      "loss": 1.4536,
      "step": 8327
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5213148593902588,
      "learning_rate": 1.529821422866818e-05,
      "loss": 0.8432,
      "step": 8328
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8535456657409668,
      "learning_rate": 1.529757531227039e-05,
      "loss": 0.89,
      "step": 8329
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6447793245315552,
      "learning_rate": 1.5296936395872602e-05,
      "loss": 0.8939,
      "step": 8330
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.534874439239502,
      "learning_rate": 1.529629747947481e-05,
      "loss": 0.6708,
      "step": 8331
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8767023086547852,
      "learning_rate": 1.5295658563077022e-05,
      "loss": 0.9214,
      "step": 8332
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0974159240722656,
      "learning_rate": 1.5295019646679234e-05,
      "loss": 0.9907,
      "step": 8333
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.470755934715271,
      "learning_rate": 1.5294380730281445e-05,
      "loss": 0.7715,
      "step": 8334
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.371425151824951,
      "learning_rate": 1.5293741813883653e-05,
      "loss": 0.9546,
      "step": 8335
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.694096326828003,
      "learning_rate": 1.5293102897485865e-05,
      "loss": 0.9218,
      "step": 8336
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6987890005111694,
      "learning_rate": 1.5292463981088076e-05,
      "loss": 0.834,
      "step": 8337
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.6834728717803955,
      "learning_rate": 1.5291825064690288e-05,
      "loss": 1.1749,
      "step": 8338
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4590626955032349,
      "learning_rate": 1.5291186148292496e-05,
      "loss": 0.8888,
      "step": 8339
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4351691007614136,
      "learning_rate": 1.5290547231894708e-05,
      "loss": 0.7744,
      "step": 8340
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5651497840881348,
      "learning_rate": 1.5289908315496916e-05,
      "loss": 0.9285,
      "step": 8341
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.940679669380188,
      "learning_rate": 1.528926939909913e-05,
      "loss": 0.8972,
      "step": 8342
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0061471462249756,
      "learning_rate": 1.528863048270134e-05,
      "loss": 0.9132,
      "step": 8343
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6834759712219238,
      "learning_rate": 1.528799156630355e-05,
      "loss": 0.8431,
      "step": 8344
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.18619704246521,
      "learning_rate": 1.528735264990576e-05,
      "loss": 0.9539,
      "step": 8345
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1973276138305664,
      "learning_rate": 1.528671373350797e-05,
      "loss": 0.9918,
      "step": 8346
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.435310363769531,
      "learning_rate": 1.5286074817110182e-05,
      "loss": 1.9119,
      "step": 8347
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.99844491481781,
      "learning_rate": 1.5285435900712393e-05,
      "loss": 0.8605,
      "step": 8348
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9318724870681763,
      "learning_rate": 1.52847969843146e-05,
      "loss": 0.9126,
      "step": 8349
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6845496892929077,
      "learning_rate": 1.5284158067916813e-05,
      "loss": 0.8938,
      "step": 8350
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4977686405181885,
      "learning_rate": 1.5283519151519025e-05,
      "loss": 1.0477,
      "step": 8351
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9459519386291504,
      "learning_rate": 1.5282880235121236e-05,
      "loss": 0.9366,
      "step": 8352
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8213601112365723,
      "learning_rate": 1.5282241318723444e-05,
      "loss": 1.2386,
      "step": 8353
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.4878926277160645,
      "learning_rate": 1.5281602402325656e-05,
      "loss": 1.3239,
      "step": 8354
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9966998100280762,
      "learning_rate": 1.5280963485927867e-05,
      "loss": 1.0338,
      "step": 8355
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8318824768066406,
      "learning_rate": 1.528032456953008e-05,
      "loss": 0.9292,
      "step": 8356
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.516356945037842,
      "learning_rate": 1.5279685653132287e-05,
      "loss": 0.8689,
      "step": 8357
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.44113826751709,
      "learning_rate": 1.52790467367345e-05,
      "loss": 1.0878,
      "step": 8358
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.544863700866699,
      "learning_rate": 1.527840782033671e-05,
      "loss": 0.8629,
      "step": 8359
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8110325336456299,
      "learning_rate": 1.527776890393892e-05,
      "loss": 0.944,
      "step": 8360
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.939523696899414,
      "learning_rate": 1.527712998754113e-05,
      "loss": 0.9195,
      "step": 8361
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7315645217895508,
      "learning_rate": 1.527649107114334e-05,
      "loss": 0.9075,
      "step": 8362
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8597283363342285,
      "learning_rate": 1.5275852154745553e-05,
      "loss": 0.9182,
      "step": 8363
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0224692821502686,
      "learning_rate": 1.5275213238347764e-05,
      "loss": 0.5884,
      "step": 8364
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.243534564971924,
      "learning_rate": 1.5274574321949973e-05,
      "loss": 0.9532,
      "step": 8365
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3213367462158203,
      "learning_rate": 1.5273935405552184e-05,
      "loss": 0.7878,
      "step": 8366
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.1137118339538574,
      "learning_rate": 1.5273296489154396e-05,
      "loss": 0.9817,
      "step": 8367
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.871512532234192,
      "learning_rate": 1.5272657572756607e-05,
      "loss": 0.9075,
      "step": 8368
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4980783462524414,
      "learning_rate": 1.5272018656358815e-05,
      "loss": 0.8457,
      "step": 8369
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2151598930358887,
      "learning_rate": 1.5271379739961027e-05,
      "loss": 0.8633,
      "step": 8370
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0368399620056152,
      "learning_rate": 1.527074082356324e-05,
      "loss": 0.9384,
      "step": 8371
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5781195163726807,
      "learning_rate": 1.527010190716545e-05,
      "loss": 0.8841,
      "step": 8372
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.03161096572876,
      "learning_rate": 1.5269462990767658e-05,
      "loss": 1.5415,
      "step": 8373
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1369080543518066,
      "learning_rate": 1.526882407436987e-05,
      "loss": 0.8646,
      "step": 8374
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4381941556930542,
      "learning_rate": 1.526818515797208e-05,
      "loss": 0.8605,
      "step": 8375
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9867576360702515,
      "learning_rate": 1.5267546241574293e-05,
      "loss": 0.8079,
      "step": 8376
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8220982551574707,
      "learning_rate": 1.52669073251765e-05,
      "loss": 0.9242,
      "step": 8377
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.799310326576233,
      "learning_rate": 1.5266268408778713e-05,
      "loss": 0.4039,
      "step": 8378
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5514463186264038,
      "learning_rate": 1.526562949238092e-05,
      "loss": 0.8486,
      "step": 8379
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.481853723526001,
      "learning_rate": 1.5264990575983136e-05,
      "loss": 1.0436,
      "step": 8380
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8717619180679321,
      "learning_rate": 1.5264351659585344e-05,
      "loss": 0.5041,
      "step": 8381
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.022435665130615,
      "learning_rate": 1.5263712743187555e-05,
      "loss": 1.2376,
      "step": 8382
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9093035459518433,
      "learning_rate": 1.5263073826789764e-05,
      "loss": 0.8805,
      "step": 8383
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.584310054779053,
      "learning_rate": 1.5262434910391975e-05,
      "loss": 0.928,
      "step": 8384
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.576912522315979,
      "learning_rate": 1.5261795993994187e-05,
      "loss": 0.8738,
      "step": 8385
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.7770416736602783,
      "learning_rate": 1.5261157077596398e-05,
      "loss": 0.8976,
      "step": 8386
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.735944390296936,
      "learning_rate": 1.5260518161198606e-05,
      "loss": 0.9224,
      "step": 8387
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1654109954833984,
      "learning_rate": 1.5259879244800818e-05,
      "loss": 0.9147,
      "step": 8388
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.670948505401611,
      "learning_rate": 1.525924032840303e-05,
      "loss": 1.8422,
      "step": 8389
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8423750400543213,
      "learning_rate": 1.5258601412005241e-05,
      "loss": 1.2171,
      "step": 8390
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9722548723220825,
      "learning_rate": 1.525796249560745e-05,
      "loss": 0.8511,
      "step": 8391
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.815767288208008,
      "learning_rate": 1.5257323579209662e-05,
      "loss": 1.0543,
      "step": 8392
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.803058385848999,
      "learning_rate": 1.5256684662811872e-05,
      "loss": 0.9493,
      "step": 8393
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.396974563598633,
      "learning_rate": 1.5256045746414084e-05,
      "loss": 1.3362,
      "step": 8394
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9263454675674438,
      "learning_rate": 1.5255406830016292e-05,
      "loss": 0.9122,
      "step": 8395
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.730982780456543,
      "learning_rate": 1.5254767913618505e-05,
      "loss": 0.9598,
      "step": 8396
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9802093505859375,
      "learning_rate": 1.5254128997220713e-05,
      "loss": 0.9623,
      "step": 8397
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7430001497268677,
      "learning_rate": 1.5253490080822927e-05,
      "loss": 0.8914,
      "step": 8398
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.122480630874634,
      "learning_rate": 1.5252851164425135e-05,
      "loss": 1.1419,
      "step": 8399
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.9410102367401123,
      "learning_rate": 1.5252212248027348e-05,
      "loss": 1.0121,
      "step": 8400
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.848159074783325,
      "learning_rate": 1.5251573331629556e-05,
      "loss": 0.9955,
      "step": 8401
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.178905725479126,
      "learning_rate": 1.5250934415231768e-05,
      "loss": 1.0855,
      "step": 8402
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.38600754737854,
      "learning_rate": 1.5250295498833978e-05,
      "loss": 0.4038,
      "step": 8403
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8753418922424316,
      "learning_rate": 1.5249656582436189e-05,
      "loss": 1.1915,
      "step": 8404
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2446272373199463,
      "learning_rate": 1.5249017666038399e-05,
      "loss": 1.011,
      "step": 8405
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9763513803482056,
      "learning_rate": 1.524837874964061e-05,
      "loss": 1.2087,
      "step": 8406
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9880763292312622,
      "learning_rate": 1.524773983324282e-05,
      "loss": 0.938,
      "step": 8407
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6575850248336792,
      "learning_rate": 1.5247100916845032e-05,
      "loss": 1.0027,
      "step": 8408
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8623316287994385,
      "learning_rate": 1.5246462000447242e-05,
      "loss": 0.9749,
      "step": 8409
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7743315696716309,
      "learning_rate": 1.5245823084049453e-05,
      "loss": 0.8962,
      "step": 8410
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.0848894119262695,
      "learning_rate": 1.5245184167651663e-05,
      "loss": 1.1415,
      "step": 8411
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0309722423553467,
      "learning_rate": 1.5244545251253875e-05,
      "loss": 0.989,
      "step": 8412
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.6917619705200195,
      "learning_rate": 1.5243906334856085e-05,
      "loss": 0.969,
      "step": 8413
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.267223596572876,
      "learning_rate": 1.5243267418458296e-05,
      "loss": 0.8356,
      "step": 8414
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.951117992401123,
      "learning_rate": 1.5242628502060506e-05,
      "loss": 1.0282,
      "step": 8415
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8397177457809448,
      "learning_rate": 1.5241989585662717e-05,
      "loss": 0.9436,
      "step": 8416
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.618382215499878,
      "learning_rate": 1.5241350669264927e-05,
      "loss": 0.8722,
      "step": 8417
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.653767466545105,
      "learning_rate": 1.5240711752867139e-05,
      "loss": 0.8899,
      "step": 8418
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1189489364624023,
      "learning_rate": 1.5240072836469349e-05,
      "loss": 0.947,
      "step": 8419
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.797107219696045,
      "learning_rate": 1.523943392007156e-05,
      "loss": 0.9648,
      "step": 8420
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9679617881774902,
      "learning_rate": 1.523879500367377e-05,
      "loss": 0.9627,
      "step": 8421
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.107358932495117,
      "learning_rate": 1.5238156087275982e-05,
      "loss": 0.7814,
      "step": 8422
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1838884353637695,
      "learning_rate": 1.5237517170878192e-05,
      "loss": 0.9969,
      "step": 8423
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.743659496307373,
      "learning_rate": 1.5236878254480403e-05,
      "loss": 0.9452,
      "step": 8424
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9945764541625977,
      "learning_rate": 1.5236239338082613e-05,
      "loss": 0.8928,
      "step": 8425
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.154137134552002,
      "learning_rate": 1.5235600421684824e-05,
      "loss": 1.0205,
      "step": 8426
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.6160221099853516,
      "learning_rate": 1.5234961505287034e-05,
      "loss": 0.8594,
      "step": 8427
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.607285737991333,
      "learning_rate": 1.5234322588889246e-05,
      "loss": 0.9748,
      "step": 8428
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5953422784805298,
      "learning_rate": 1.5233683672491456e-05,
      "loss": 0.4098,
      "step": 8429
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.532499074935913,
      "learning_rate": 1.5233044756093667e-05,
      "loss": 1.0666,
      "step": 8430
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.3455119132995605,
      "learning_rate": 1.5232405839695877e-05,
      "loss": 0.9891,
      "step": 8431
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.975225567817688,
      "learning_rate": 1.5231766923298089e-05,
      "loss": 0.9219,
      "step": 8432
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.552301287651062,
      "learning_rate": 1.5231128006900297e-05,
      "loss": 0.8994,
      "step": 8433
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.291266918182373,
      "learning_rate": 1.523048909050251e-05,
      "loss": 0.9584,
      "step": 8434
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4543378353118896,
      "learning_rate": 1.5229850174104718e-05,
      "loss": 0.3894,
      "step": 8435
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.303563117980957,
      "learning_rate": 1.5229211257706931e-05,
      "loss": 1.2109,
      "step": 8436
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.659743309020996,
      "learning_rate": 1.522857234130914e-05,
      "loss": 1.3777,
      "step": 8437
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7312822341918945,
      "learning_rate": 1.5227933424911351e-05,
      "loss": 0.4552,
      "step": 8438
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.764523983001709,
      "learning_rate": 1.5227294508513561e-05,
      "loss": 0.9118,
      "step": 8439
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7110369205474854,
      "learning_rate": 1.5226655592115773e-05,
      "loss": 0.8771,
      "step": 8440
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6713416576385498,
      "learning_rate": 1.5226016675717982e-05,
      "loss": 0.9227,
      "step": 8441
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.796365976333618,
      "learning_rate": 1.5225377759320194e-05,
      "loss": 0.9445,
      "step": 8442
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.972191333770752,
      "learning_rate": 1.5224738842922404e-05,
      "loss": 0.947,
      "step": 8443
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.306885004043579,
      "learning_rate": 1.5224099926524615e-05,
      "loss": 0.7884,
      "step": 8444
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.34487247467041,
      "learning_rate": 1.5223461010126825e-05,
      "loss": 1.1765,
      "step": 8445
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.560532569885254,
      "learning_rate": 1.5222822093729037e-05,
      "loss": 1.0742,
      "step": 8446
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.052281856536865,
      "learning_rate": 1.5222183177331247e-05,
      "loss": 1.7975,
      "step": 8447
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.7616500854492188,
      "learning_rate": 1.5221544260933458e-05,
      "loss": 1.1386,
      "step": 8448
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0109617710113525,
      "learning_rate": 1.5220905344535668e-05,
      "loss": 1.1779,
      "step": 8449
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5298399925231934,
      "learning_rate": 1.522026642813788e-05,
      "loss": 0.8899,
      "step": 8450
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3863253593444824,
      "learning_rate": 1.521962751174009e-05,
      "loss": 0.7186,
      "step": 8451
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7213211059570312,
      "learning_rate": 1.5218988595342301e-05,
      "loss": 0.9716,
      "step": 8452
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.759957790374756,
      "learning_rate": 1.521834967894451e-05,
      "loss": 1.402,
      "step": 8453
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3243467807769775,
      "learning_rate": 1.5217710762546722e-05,
      "loss": 1.0102,
      "step": 8454
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.527271032333374,
      "learning_rate": 1.5217071846148932e-05,
      "loss": 1.0043,
      "step": 8455
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.998166799545288,
      "learning_rate": 1.5216432929751144e-05,
      "loss": 1.2229,
      "step": 8456
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5131583213806152,
      "learning_rate": 1.5215794013353354e-05,
      "loss": 0.594,
      "step": 8457
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8753403425216675,
      "learning_rate": 1.5215155096955565e-05,
      "loss": 0.9385,
      "step": 8458
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.320533275604248,
      "learning_rate": 1.5214516180557775e-05,
      "loss": 1.4212,
      "step": 8459
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2859578132629395,
      "learning_rate": 1.5213877264159987e-05,
      "loss": 0.9381,
      "step": 8460
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5520657300949097,
      "learning_rate": 1.5213238347762196e-05,
      "loss": 0.917,
      "step": 8461
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.170341491699219,
      "learning_rate": 1.5212599431364408e-05,
      "loss": 0.997,
      "step": 8462
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.296748161315918,
      "learning_rate": 1.5211960514966618e-05,
      "loss": 1.0728,
      "step": 8463
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9602397680282593,
      "learning_rate": 1.521132159856883e-05,
      "loss": 0.9251,
      "step": 8464
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.548365831375122,
      "learning_rate": 1.521068268217104e-05,
      "loss": 1.0456,
      "step": 8465
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3172638416290283,
      "learning_rate": 1.521004376577325e-05,
      "loss": 0.3957,
      "step": 8466
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.672863006591797,
      "learning_rate": 1.520940484937546e-05,
      "loss": 1.2704,
      "step": 8467
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.6957015991210938,
      "learning_rate": 1.5208765932977672e-05,
      "loss": 0.6609,
      "step": 8468
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0741541385650635,
      "learning_rate": 1.520812701657988e-05,
      "loss": 0.8668,
      "step": 8469
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.0425500869750977,
      "learning_rate": 1.5207488100182094e-05,
      "loss": 0.833,
      "step": 8470
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8087154626846313,
      "learning_rate": 1.5206849183784302e-05,
      "loss": 0.9472,
      "step": 8471
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3930771350860596,
      "learning_rate": 1.5206210267386515e-05,
      "loss": 1.0323,
      "step": 8472
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1330180168151855,
      "learning_rate": 1.5205571350988723e-05,
      "loss": 1.1849,
      "step": 8473
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4336814880371094,
      "learning_rate": 1.5204932434590935e-05,
      "loss": 0.9086,
      "step": 8474
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.53810977935791,
      "learning_rate": 1.5204293518193145e-05,
      "loss": 1.2735,
      "step": 8475
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7316956520080566,
      "learning_rate": 1.5203654601795356e-05,
      "loss": 0.9913,
      "step": 8476
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.10318660736084,
      "learning_rate": 1.5203015685397566e-05,
      "loss": 0.9933,
      "step": 8477
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.706030011177063,
      "learning_rate": 1.5202376768999777e-05,
      "loss": 0.8929,
      "step": 8478
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6210006475448608,
      "learning_rate": 1.5201737852601987e-05,
      "loss": 0.8992,
      "step": 8479
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.165131092071533,
      "learning_rate": 1.5201098936204199e-05,
      "loss": 0.9803,
      "step": 8480
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8624064922332764,
      "learning_rate": 1.5200460019806409e-05,
      "loss": 1.0261,
      "step": 8481
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6878342628479004,
      "learning_rate": 1.519982110340862e-05,
      "loss": 0.4536,
      "step": 8482
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.9034688472747803,
      "learning_rate": 1.519918218701083e-05,
      "loss": 0.8957,
      "step": 8483
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5523722171783447,
      "learning_rate": 1.5198543270613042e-05,
      "loss": 0.8731,
      "step": 8484
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5158047676086426,
      "learning_rate": 1.5197904354215252e-05,
      "loss": 0.9662,
      "step": 8485
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8123607635498047,
      "learning_rate": 1.5197265437817463e-05,
      "loss": 0.9147,
      "step": 8486
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.166593074798584,
      "learning_rate": 1.5196626521419673e-05,
      "loss": 1.1099,
      "step": 8487
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0201077461242676,
      "learning_rate": 1.5195987605021884e-05,
      "loss": 0.6819,
      "step": 8488
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9880282878875732,
      "learning_rate": 1.5195348688624094e-05,
      "loss": 0.8793,
      "step": 8489
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0746827125549316,
      "learning_rate": 1.5194709772226306e-05,
      "loss": 1.0165,
      "step": 8490
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.610628604888916,
      "learning_rate": 1.5194070855828516e-05,
      "loss": 0.8905,
      "step": 8491
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1515142917633057,
      "learning_rate": 1.5193431939430727e-05,
      "loss": 1.1111,
      "step": 8492
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5511771440505981,
      "learning_rate": 1.5192793023032937e-05,
      "loss": 0.8525,
      "step": 8493
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9657186269760132,
      "learning_rate": 1.5192154106635149e-05,
      "loss": 1.0164,
      "step": 8494
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7337485551834106,
      "learning_rate": 1.5191515190237359e-05,
      "loss": 0.9042,
      "step": 8495
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8815600872039795,
      "learning_rate": 1.519087627383957e-05,
      "loss": 1.0239,
      "step": 8496
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9515475034713745,
      "learning_rate": 1.519023735744178e-05,
      "loss": 0.8367,
      "step": 8497
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6474659442901611,
      "learning_rate": 1.5189598441043991e-05,
      "loss": 0.8659,
      "step": 8498
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.642411470413208,
      "learning_rate": 1.5188959524646201e-05,
      "loss": 0.9341,
      "step": 8499
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.648087978363037,
      "learning_rate": 1.5188320608248413e-05,
      "loss": 0.3861,
      "step": 8500
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.032766819000244,
      "learning_rate": 1.5187681691850623e-05,
      "loss": 0.8651,
      "step": 8501
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.62188982963562,
      "learning_rate": 1.5187042775452834e-05,
      "loss": 1.1507,
      "step": 8502
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9069987535476685,
      "learning_rate": 1.5186403859055044e-05,
      "loss": 0.9201,
      "step": 8503
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6138968467712402,
      "learning_rate": 1.5185764942657256e-05,
      "loss": 0.471,
      "step": 8504
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8170157670974731,
      "learning_rate": 1.5185126026259464e-05,
      "loss": 0.7259,
      "step": 8505
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5893912315368652,
      "learning_rate": 1.5184487109861677e-05,
      "loss": 1.1151,
      "step": 8506
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6396015882492065,
      "learning_rate": 1.5183848193463885e-05,
      "loss": 0.8402,
      "step": 8507
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6120036840438843,
      "learning_rate": 1.5183209277066098e-05,
      "loss": 0.8565,
      "step": 8508
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.640594482421875,
      "learning_rate": 1.5182570360668307e-05,
      "loss": 0.7572,
      "step": 8509
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5713778734207153,
      "learning_rate": 1.5181931444270518e-05,
      "loss": 0.9584,
      "step": 8510
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6315628290176392,
      "learning_rate": 1.5181292527872728e-05,
      "loss": 1.0846,
      "step": 8511
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.027867078781128,
      "learning_rate": 1.518065361147494e-05,
      "loss": 0.9844,
      "step": 8512
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9518793821334839,
      "learning_rate": 1.518001469507715e-05,
      "loss": 0.9434,
      "step": 8513
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.9471373558044434,
      "learning_rate": 1.5179375778679361e-05,
      "loss": 0.7941,
      "step": 8514
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.083741664886475,
      "learning_rate": 1.517873686228157e-05,
      "loss": 0.9489,
      "step": 8515
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9278370141983032,
      "learning_rate": 1.5178097945883782e-05,
      "loss": 0.8912,
      "step": 8516
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.665419340133667,
      "learning_rate": 1.5177459029485992e-05,
      "loss": 1.0222,
      "step": 8517
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0079519748687744,
      "learning_rate": 1.5176820113088204e-05,
      "loss": 0.9338,
      "step": 8518
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0880885124206543,
      "learning_rate": 1.5176181196690414e-05,
      "loss": 0.8961,
      "step": 8519
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.745301365852356,
      "learning_rate": 1.5175542280292625e-05,
      "loss": 0.9485,
      "step": 8520
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.4429450035095215,
      "learning_rate": 1.5174903363894835e-05,
      "loss": 1.2613,
      "step": 8521
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.3801305294036865,
      "learning_rate": 1.5174264447497047e-05,
      "loss": 1.3028,
      "step": 8522
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.691706895828247,
      "learning_rate": 1.5173625531099256e-05,
      "loss": 0.9525,
      "step": 8523
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8622241020202637,
      "learning_rate": 1.5172986614701468e-05,
      "loss": 0.7727,
      "step": 8524
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.611682176589966,
      "learning_rate": 1.5172347698303678e-05,
      "loss": 0.7193,
      "step": 8525
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.2040040493011475,
      "learning_rate": 1.517170878190589e-05,
      "loss": 0.8593,
      "step": 8526
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.455953359603882,
      "learning_rate": 1.51710698655081e-05,
      "loss": 1.0002,
      "step": 8527
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.401907444000244,
      "learning_rate": 1.517043094911031e-05,
      "loss": 0.927,
      "step": 8528
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5175756216049194,
      "learning_rate": 1.516979203271252e-05,
      "loss": 0.3995,
      "step": 8529
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8589369058609009,
      "learning_rate": 1.5169153116314732e-05,
      "loss": 0.9413,
      "step": 8530
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.816879391670227,
      "learning_rate": 1.5168514199916942e-05,
      "loss": 0.86,
      "step": 8531
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.394282341003418,
      "learning_rate": 1.5167875283519154e-05,
      "loss": 0.9778,
      "step": 8532
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.9821979999542236,
      "learning_rate": 1.5167236367121363e-05,
      "loss": 0.7453,
      "step": 8533
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9647026062011719,
      "learning_rate": 1.5166597450723575e-05,
      "loss": 0.9032,
      "step": 8534
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.722780704498291,
      "learning_rate": 1.5165958534325785e-05,
      "loss": 0.8461,
      "step": 8535
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9234172105789185,
      "learning_rate": 1.5165319617927996e-05,
      "loss": 0.998,
      "step": 8536
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0316736698150635,
      "learning_rate": 1.5164680701530206e-05,
      "loss": 1.2953,
      "step": 8537
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.705109119415283,
      "learning_rate": 1.5164041785132418e-05,
      "loss": 0.9493,
      "step": 8538
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9719061851501465,
      "learning_rate": 1.5163402868734628e-05,
      "loss": 0.9596,
      "step": 8539
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7557423114776611,
      "learning_rate": 1.516276395233684e-05,
      "loss": 0.7845,
      "step": 8540
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.270052909851074,
      "learning_rate": 1.5162125035939047e-05,
      "loss": 1.1176,
      "step": 8541
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5461753606796265,
      "learning_rate": 1.516148611954126e-05,
      "loss": 0.9439,
      "step": 8542
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7600427865982056,
      "learning_rate": 1.5160847203143469e-05,
      "loss": 0.9746,
      "step": 8543
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.642326593399048,
      "learning_rate": 1.5160208286745682e-05,
      "loss": 0.555,
      "step": 8544
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7171255350112915,
      "learning_rate": 1.515956937034789e-05,
      "loss": 0.8732,
      "step": 8545
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5999982357025146,
      "learning_rate": 1.5158930453950103e-05,
      "loss": 0.7946,
      "step": 8546
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4805229902267456,
      "learning_rate": 1.5158291537552312e-05,
      "loss": 0.3916,
      "step": 8547
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.006831169128418,
      "learning_rate": 1.5157652621154523e-05,
      "loss": 0.9722,
      "step": 8548
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1519389152526855,
      "learning_rate": 1.5157013704756733e-05,
      "loss": 0.9916,
      "step": 8549
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5763808488845825,
      "learning_rate": 1.5156374788358944e-05,
      "loss": 0.9148,
      "step": 8550
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7734640836715698,
      "learning_rate": 1.5155735871961154e-05,
      "loss": 1.0564,
      "step": 8551
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.879523515701294,
      "learning_rate": 1.5155096955563366e-05,
      "loss": 1.0056,
      "step": 8552
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5090980529785156,
      "learning_rate": 1.5154458039165576e-05,
      "loss": 0.8293,
      "step": 8553
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.48544692993164,
      "learning_rate": 1.5153819122767787e-05,
      "loss": 1.6663,
      "step": 8554
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7133934497833252,
      "learning_rate": 1.5153180206369997e-05,
      "loss": 0.8648,
      "step": 8555
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.604040265083313,
      "learning_rate": 1.5152541289972209e-05,
      "loss": 0.9485,
      "step": 8556
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.064063549041748,
      "learning_rate": 1.5151902373574419e-05,
      "loss": 1.5863,
      "step": 8557
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.958587646484375,
      "learning_rate": 1.515126345717663e-05,
      "loss": 1.0434,
      "step": 8558
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1014373302459717,
      "learning_rate": 1.515062454077884e-05,
      "loss": 0.9461,
      "step": 8559
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.260009288787842,
      "learning_rate": 1.5149985624381051e-05,
      "loss": 0.8006,
      "step": 8560
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9217149019241333,
      "learning_rate": 1.5149346707983261e-05,
      "loss": 0.9594,
      "step": 8561
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4423654079437256,
      "learning_rate": 1.5148707791585473e-05,
      "loss": 1.2097,
      "step": 8562
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.128730535507202,
      "learning_rate": 1.5148068875187683e-05,
      "loss": 1.1576,
      "step": 8563
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.574855327606201,
      "learning_rate": 1.5147429958789894e-05,
      "loss": 1.6936,
      "step": 8564
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.988785743713379,
      "learning_rate": 1.5146791042392104e-05,
      "loss": 0.9266,
      "step": 8565
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1767711639404297,
      "learning_rate": 1.5146152125994316e-05,
      "loss": 1.2996,
      "step": 8566
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7418936491012573,
      "learning_rate": 1.5145513209596526e-05,
      "loss": 0.9339,
      "step": 8567
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4740307331085205,
      "learning_rate": 1.5144874293198737e-05,
      "loss": 0.8911,
      "step": 8568
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1135828495025635,
      "learning_rate": 1.5144235376800947e-05,
      "loss": 0.9777,
      "step": 8569
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8087091445922852,
      "learning_rate": 1.5143596460403158e-05,
      "loss": 0.9547,
      "step": 8570
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7919350862503052,
      "learning_rate": 1.5142957544005368e-05,
      "loss": 0.9831,
      "step": 8571
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.911423921585083,
      "learning_rate": 1.514231862760758e-05,
      "loss": 1.2143,
      "step": 8572
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.272071361541748,
      "learning_rate": 1.514167971120979e-05,
      "loss": 1.0104,
      "step": 8573
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.231391191482544,
      "learning_rate": 1.5141040794812001e-05,
      "loss": 0.8987,
      "step": 8574
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4425569772720337,
      "learning_rate": 1.5140401878414211e-05,
      "loss": 0.8775,
      "step": 8575
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1203665733337402,
      "learning_rate": 1.5139762962016423e-05,
      "loss": 0.9774,
      "step": 8576
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9536194801330566,
      "learning_rate": 1.5139124045618633e-05,
      "loss": 0.902,
      "step": 8577
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8918216228485107,
      "learning_rate": 1.5138485129220844e-05,
      "loss": 1.2024,
      "step": 8578
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.446309804916382,
      "learning_rate": 1.5137846212823052e-05,
      "loss": 1.0718,
      "step": 8579
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.06710147857666,
      "learning_rate": 1.5137207296425265e-05,
      "loss": 1.3173,
      "step": 8580
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7951878309249878,
      "learning_rate": 1.5136568380027474e-05,
      "loss": 0.9759,
      "step": 8581
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.708522915840149,
      "learning_rate": 1.5135929463629687e-05,
      "loss": 0.9398,
      "step": 8582
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4290013313293457,
      "learning_rate": 1.5135290547231895e-05,
      "loss": 0.935,
      "step": 8583
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6261382102966309,
      "learning_rate": 1.5134651630834107e-05,
      "loss": 0.9351,
      "step": 8584
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1129941940307617,
      "learning_rate": 1.5134012714436316e-05,
      "loss": 1.0007,
      "step": 8585
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8448439836502075,
      "learning_rate": 1.5133373798038528e-05,
      "loss": 1.0454,
      "step": 8586
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4890739917755127,
      "learning_rate": 1.5132734881640738e-05,
      "loss": 1.2766,
      "step": 8587
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3101205825805664,
      "learning_rate": 1.513209596524295e-05,
      "loss": 0.8971,
      "step": 8588
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.13893985748291,
      "learning_rate": 1.513145704884516e-05,
      "loss": 1.3441,
      "step": 8589
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.282644748687744,
      "learning_rate": 1.513081813244737e-05,
      "loss": 1.1049,
      "step": 8590
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7778862714767456,
      "learning_rate": 1.513017921604958e-05,
      "loss": 1.0253,
      "step": 8591
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.884793758392334,
      "learning_rate": 1.5129540299651792e-05,
      "loss": 1.3391,
      "step": 8592
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.029083490371704,
      "learning_rate": 1.5128901383254002e-05,
      "loss": 0.8941,
      "step": 8593
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6072629690170288,
      "learning_rate": 1.5128262466856214e-05,
      "loss": 0.8132,
      "step": 8594
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.757919192314148,
      "learning_rate": 1.5127623550458423e-05,
      "loss": 0.9165,
      "step": 8595
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.291271686553955,
      "learning_rate": 1.5126984634060635e-05,
      "loss": 1.3998,
      "step": 8596
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.08652663230896,
      "learning_rate": 1.5126345717662845e-05,
      "loss": 0.9726,
      "step": 8597
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8741416931152344,
      "learning_rate": 1.5125706801265056e-05,
      "loss": 0.9472,
      "step": 8598
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.567277431488037,
      "learning_rate": 1.5125067884867266e-05,
      "loss": 0.7733,
      "step": 8599
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.721247911453247,
      "learning_rate": 1.5124428968469478e-05,
      "loss": 1.2176,
      "step": 8600
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.544548988342285,
      "learning_rate": 1.5123790052071688e-05,
      "loss": 1.3967,
      "step": 8601
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.547370195388794,
      "learning_rate": 1.51231511356739e-05,
      "loss": 0.8348,
      "step": 8602
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1701560020446777,
      "learning_rate": 1.5122512219276109e-05,
      "loss": 1.4496,
      "step": 8603
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.826756238937378,
      "learning_rate": 1.512187330287832e-05,
      "loss": 1.0332,
      "step": 8604
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9600013494491577,
      "learning_rate": 1.512123438648053e-05,
      "loss": 0.9879,
      "step": 8605
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4504706859588623,
      "learning_rate": 1.5120595470082742e-05,
      "loss": 0.9317,
      "step": 8606
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7354660034179688,
      "learning_rate": 1.5119956553684952e-05,
      "loss": 0.9554,
      "step": 8607
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9121794700622559,
      "learning_rate": 1.5119317637287163e-05,
      "loss": 1.0043,
      "step": 8608
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.033273696899414,
      "learning_rate": 1.5118678720889373e-05,
      "loss": 1.0892,
      "step": 8609
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8104323148727417,
      "learning_rate": 1.5118039804491585e-05,
      "loss": 0.9611,
      "step": 8610
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.232496738433838,
      "learning_rate": 1.5117400888093795e-05,
      "loss": 0.992,
      "step": 8611
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.067084550857544,
      "learning_rate": 1.5116761971696006e-05,
      "loss": 0.9104,
      "step": 8612
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.562777042388916,
      "learning_rate": 1.5116123055298216e-05,
      "loss": 1.1312,
      "step": 8613
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.05462646484375,
      "learning_rate": 1.5115484138900428e-05,
      "loss": 1.2097,
      "step": 8614
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6299370527267456,
      "learning_rate": 1.5114845222502636e-05,
      "loss": 0.8852,
      "step": 8615
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7772059440612793,
      "learning_rate": 1.5114206306104849e-05,
      "loss": 0.9355,
      "step": 8616
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.9100704193115234,
      "learning_rate": 1.5113567389707057e-05,
      "loss": 0.9034,
      "step": 8617
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.137439250946045,
      "learning_rate": 1.511292847330927e-05,
      "loss": 1.2291,
      "step": 8618
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.97323739528656,
      "learning_rate": 1.5112289556911479e-05,
      "loss": 0.4357,
      "step": 8619
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.036487102508545,
      "learning_rate": 1.511165064051369e-05,
      "loss": 1.4387,
      "step": 8620
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6966116428375244,
      "learning_rate": 1.51110117241159e-05,
      "loss": 0.8113,
      "step": 8621
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.343246459960938,
      "learning_rate": 1.5110372807718111e-05,
      "loss": 1.5962,
      "step": 8622
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1566286087036133,
      "learning_rate": 1.5109733891320321e-05,
      "loss": 1.0328,
      "step": 8623
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9783835411071777,
      "learning_rate": 1.5109094974922533e-05,
      "loss": 0.8612,
      "step": 8624
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.653366804122925,
      "learning_rate": 1.5108456058524743e-05,
      "loss": 0.8725,
      "step": 8625
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.56300163269043,
      "learning_rate": 1.5107817142126954e-05,
      "loss": 0.9077,
      "step": 8626
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2027769088745117,
      "learning_rate": 1.5107178225729164e-05,
      "loss": 0.7771,
      "step": 8627
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.261096954345703,
      "learning_rate": 1.5106539309331376e-05,
      "loss": 1.2429,
      "step": 8628
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8756256103515625,
      "learning_rate": 1.5105900392933586e-05,
      "loss": 0.7647,
      "step": 8629
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7245601415634155,
      "learning_rate": 1.5105261476535797e-05,
      "loss": 0.4271,
      "step": 8630
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.8930931091308594,
      "learning_rate": 1.5104622560138007e-05,
      "loss": 1.0687,
      "step": 8631
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.699116826057434,
      "learning_rate": 1.5103983643740218e-05,
      "loss": 0.827,
      "step": 8632
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.54398775100708,
      "learning_rate": 1.5103344727342428e-05,
      "loss": 0.9556,
      "step": 8633
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9334591627120972,
      "learning_rate": 1.510270581094464e-05,
      "loss": 0.8933,
      "step": 8634
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.112192153930664,
      "learning_rate": 1.510206689454685e-05,
      "loss": 1.108,
      "step": 8635
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.73092782497406,
      "learning_rate": 1.5101427978149061e-05,
      "loss": 0.9575,
      "step": 8636
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6977139711380005,
      "learning_rate": 1.5100789061751271e-05,
      "loss": 0.9021,
      "step": 8637
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.313129186630249,
      "learning_rate": 1.5100150145353483e-05,
      "loss": 1.3139,
      "step": 8638
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.2632088661193848,
      "learning_rate": 1.5099511228955693e-05,
      "loss": 1.1033,
      "step": 8639
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7079246044158936,
      "learning_rate": 1.5098872312557904e-05,
      "loss": 0.921,
      "step": 8640
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.483520269393921,
      "learning_rate": 1.5098233396160114e-05,
      "loss": 1.2339,
      "step": 8641
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7326315641403198,
      "learning_rate": 1.5097594479762325e-05,
      "loss": 1.0067,
      "step": 8642
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.7679481506347656,
      "learning_rate": 1.5096955563364535e-05,
      "loss": 0.9128,
      "step": 8643
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.501756191253662,
      "learning_rate": 1.5096316646966747e-05,
      "loss": 1.3255,
      "step": 8644
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8560593128204346,
      "learning_rate": 1.5095677730568957e-05,
      "loss": 1.1472,
      "step": 8645
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9300941228866577,
      "learning_rate": 1.5095038814171168e-05,
      "loss": 0.9106,
      "step": 8646
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8642598390579224,
      "learning_rate": 1.5094399897773378e-05,
      "loss": 0.7978,
      "step": 8647
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7568297386169434,
      "learning_rate": 1.509376098137559e-05,
      "loss": 0.7852,
      "step": 8648
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6069093942642212,
      "learning_rate": 1.50931220649778e-05,
      "loss": 0.8577,
      "step": 8649
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.023533344268799,
      "learning_rate": 1.5092483148580011e-05,
      "loss": 1.0261,
      "step": 8650
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.555600881576538,
      "learning_rate": 1.509184423218222e-05,
      "loss": 1.0106,
      "step": 8651
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3838731050491333,
      "learning_rate": 1.5091205315784432e-05,
      "loss": 0.8944,
      "step": 8652
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.198528289794922,
      "learning_rate": 1.509056639938664e-05,
      "loss": 1.1656,
      "step": 8653
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8582897186279297,
      "learning_rate": 1.5089927482988854e-05,
      "loss": 0.8999,
      "step": 8654
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.718292236328125,
      "learning_rate": 1.5089288566591062e-05,
      "loss": 0.8676,
      "step": 8655
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.388787269592285,
      "learning_rate": 1.5088649650193274e-05,
      "loss": 1.0423,
      "step": 8656
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3427209854125977,
      "learning_rate": 1.5088010733795483e-05,
      "loss": 0.8916,
      "step": 8657
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5975404977798462,
      "learning_rate": 1.5087371817397695e-05,
      "loss": 0.4111,
      "step": 8658
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8622753620147705,
      "learning_rate": 1.5086732900999905e-05,
      "loss": 0.9567,
      "step": 8659
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.45039701461792,
      "learning_rate": 1.5086093984602116e-05,
      "loss": 1.4648,
      "step": 8660
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.798084259033203,
      "learning_rate": 1.5085455068204326e-05,
      "loss": 1.0857,
      "step": 8661
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6825506687164307,
      "learning_rate": 1.5084816151806538e-05,
      "loss": 0.9443,
      "step": 8662
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0719308853149414,
      "learning_rate": 1.5084177235408748e-05,
      "loss": 0.9435,
      "step": 8663
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6319016218185425,
      "learning_rate": 1.508353831901096e-05,
      "loss": 0.8454,
      "step": 8664
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8034093379974365,
      "learning_rate": 1.5082899402613169e-05,
      "loss": 0.9766,
      "step": 8665
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8665083646774292,
      "learning_rate": 1.508226048621538e-05,
      "loss": 0.7788,
      "step": 8666
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9708327054977417,
      "learning_rate": 1.508162156981759e-05,
      "loss": 1.0403,
      "step": 8667
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.011818885803223,
      "learning_rate": 1.5080982653419802e-05,
      "loss": 1.6815,
      "step": 8668
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0131192207336426,
      "learning_rate": 1.5080343737022012e-05,
      "loss": 0.9088,
      "step": 8669
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.5346240997314453,
      "learning_rate": 1.5079704820624223e-05,
      "loss": 0.9182,
      "step": 8670
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.858011484146118,
      "learning_rate": 1.5079065904226433e-05,
      "loss": 1.1338,
      "step": 8671
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6425782442092896,
      "learning_rate": 1.5078426987828645e-05,
      "loss": 0.9447,
      "step": 8672
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8312469720840454,
      "learning_rate": 1.5077788071430855e-05,
      "loss": 0.9633,
      "step": 8673
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8492431640625,
      "learning_rate": 1.5077149155033066e-05,
      "loss": 0.9186,
      "step": 8674
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4088003635406494,
      "learning_rate": 1.5076510238635276e-05,
      "loss": 0.9681,
      "step": 8675
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.647702932357788,
      "learning_rate": 1.5075871322237488e-05,
      "loss": 0.4826,
      "step": 8676
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1500794887542725,
      "learning_rate": 1.5075232405839697e-05,
      "loss": 0.9209,
      "step": 8677
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8320297002792358,
      "learning_rate": 1.5074593489441909e-05,
      "loss": 0.9217,
      "step": 8678
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.212347030639648,
      "learning_rate": 1.5073954573044119e-05,
      "loss": 1.0597,
      "step": 8679
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.900348424911499,
      "learning_rate": 1.507331565664633e-05,
      "loss": 0.9332,
      "step": 8680
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5441645383834839,
      "learning_rate": 1.507267674024854e-05,
      "loss": 0.4311,
      "step": 8681
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.617372751235962,
      "learning_rate": 1.5072037823850752e-05,
      "loss": 0.884,
      "step": 8682
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.322770833969116,
      "learning_rate": 1.5071398907452962e-05,
      "loss": 0.7421,
      "step": 8683
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8275587558746338,
      "learning_rate": 1.5070759991055173e-05,
      "loss": 0.9633,
      "step": 8684
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.972946286201477,
      "learning_rate": 1.5070121074657383e-05,
      "loss": 0.8174,
      "step": 8685
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8096572160720825,
      "learning_rate": 1.5069482158259595e-05,
      "loss": 0.9559,
      "step": 8686
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6688663959503174,
      "learning_rate": 1.5068843241861803e-05,
      "loss": 0.8347,
      "step": 8687
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6921433210372925,
      "learning_rate": 1.5068204325464016e-05,
      "loss": 0.9337,
      "step": 8688
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2545392513275146,
      "learning_rate": 1.5067565409066224e-05,
      "loss": 1.1719,
      "step": 8689
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5613456964492798,
      "learning_rate": 1.5066926492668437e-05,
      "loss": 0.9004,
      "step": 8690
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.996687650680542,
      "learning_rate": 1.5066287576270646e-05,
      "loss": 0.8388,
      "step": 8691
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.608724594116211,
      "learning_rate": 1.5065648659872859e-05,
      "loss": 1.0124,
      "step": 8692
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5165157318115234,
      "learning_rate": 1.5065009743475067e-05,
      "loss": 0.8365,
      "step": 8693
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1070199012756348,
      "learning_rate": 1.5064370827077279e-05,
      "loss": 0.8856,
      "step": 8694
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.255110502243042,
      "learning_rate": 1.5063731910679488e-05,
      "loss": 0.9264,
      "step": 8695
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2884323596954346,
      "learning_rate": 1.50630929942817e-05,
      "loss": 1.1344,
      "step": 8696
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.062244176864624,
      "learning_rate": 1.506245407788391e-05,
      "loss": 0.9803,
      "step": 8697
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5274901390075684,
      "learning_rate": 1.5061815161486121e-05,
      "loss": 0.4759,
      "step": 8698
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.5170140266418457,
      "learning_rate": 1.5061176245088331e-05,
      "loss": 1.1053,
      "step": 8699
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.151041030883789,
      "learning_rate": 1.5060537328690541e-05,
      "loss": 1.2779,
      "step": 8700
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.1455893516540527,
      "learning_rate": 1.5059898412292753e-05,
      "loss": 0.9267,
      "step": 8701
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.733426809310913,
      "learning_rate": 1.5059259495894962e-05,
      "loss": 0.983,
      "step": 8702
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.528181076049805,
      "learning_rate": 1.5058620579497174e-05,
      "loss": 1.174,
      "step": 8703
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9766478538513184,
      "learning_rate": 1.5057981663099384e-05,
      "loss": 0.8401,
      "step": 8704
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9362208843231201,
      "learning_rate": 1.5057342746701595e-05,
      "loss": 1.0541,
      "step": 8705
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8647791147232056,
      "learning_rate": 1.5056703830303805e-05,
      "loss": 0.9182,
      "step": 8706
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.730353832244873,
      "learning_rate": 1.5056064913906017e-05,
      "loss": 0.8811,
      "step": 8707
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5778576135635376,
      "learning_rate": 1.5055425997508227e-05,
      "loss": 0.8487,
      "step": 8708
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.288379192352295,
      "learning_rate": 1.5054787081110438e-05,
      "loss": 1.2272,
      "step": 8709
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5058811902999878,
      "learning_rate": 1.5054148164712648e-05,
      "loss": 0.7922,
      "step": 8710
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.197730779647827,
      "learning_rate": 1.505350924831486e-05,
      "loss": 1.0611,
      "step": 8711
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.253694534301758,
      "learning_rate": 1.505287033191707e-05,
      "loss": 1.2889,
      "step": 8712
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.4461541175842285,
      "learning_rate": 1.5052231415519281e-05,
      "loss": 1.4721,
      "step": 8713
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8072344064712524,
      "learning_rate": 1.505159249912149e-05,
      "loss": 0.7975,
      "step": 8714
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.4536120891571045,
      "learning_rate": 1.5050953582723702e-05,
      "loss": 1.0438,
      "step": 8715
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0489742755889893,
      "learning_rate": 1.5050314666325912e-05,
      "loss": 1.0444,
      "step": 8716
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7427445650100708,
      "learning_rate": 1.5049675749928124e-05,
      "loss": 0.8569,
      "step": 8717
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2539336681365967,
      "learning_rate": 1.5049036833530332e-05,
      "loss": 1.2489,
      "step": 8718
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9163992404937744,
      "learning_rate": 1.5048397917132545e-05,
      "loss": 0.7086,
      "step": 8719
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.6812145709991455,
      "learning_rate": 1.5047759000734753e-05,
      "loss": 1.429,
      "step": 8720
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6250145435333252,
      "learning_rate": 1.5047120084336967e-05,
      "loss": 0.8676,
      "step": 8721
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.776179313659668,
      "learning_rate": 1.5046481167939175e-05,
      "loss": 1.2467,
      "step": 8722
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.036027431488037,
      "learning_rate": 1.5045842251541388e-05,
      "loss": 1.0109,
      "step": 8723
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.402366876602173,
      "learning_rate": 1.5045203335143596e-05,
      "loss": 1.0567,
      "step": 8724
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.431744575500488,
      "learning_rate": 1.5044564418745808e-05,
      "loss": 0.9312,
      "step": 8725
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1239709854125977,
      "learning_rate": 1.5043925502348018e-05,
      "loss": 1.0454,
      "step": 8726
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.005388259887695,
      "learning_rate": 1.5043286585950229e-05,
      "loss": 0.8878,
      "step": 8727
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.07020902633667,
      "learning_rate": 1.5042647669552439e-05,
      "loss": 1.0276,
      "step": 8728
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.326686143875122,
      "learning_rate": 1.504200875315465e-05,
      "loss": 0.9368,
      "step": 8729
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.465019702911377,
      "learning_rate": 1.504136983675686e-05,
      "loss": 1.4711,
      "step": 8730
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7684249877929688,
      "learning_rate": 1.5040730920359072e-05,
      "loss": 1.1149,
      "step": 8731
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.520101308822632,
      "learning_rate": 1.5040092003961282e-05,
      "loss": 0.9159,
      "step": 8732
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0011956691741943,
      "learning_rate": 1.5039453087563493e-05,
      "loss": 0.9229,
      "step": 8733
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9004677534103394,
      "learning_rate": 1.5038814171165703e-05,
      "loss": 0.8777,
      "step": 8734
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5663304328918457,
      "learning_rate": 1.5038175254767915e-05,
      "loss": 0.8538,
      "step": 8735
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.912801742553711,
      "learning_rate": 1.5037536338370125e-05,
      "loss": 0.5291,
      "step": 8736
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.363292932510376,
      "learning_rate": 1.5036897421972336e-05,
      "loss": 0.9796,
      "step": 8737
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6434605121612549,
      "learning_rate": 1.5036258505574546e-05,
      "loss": 0.9458,
      "step": 8738
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4830918312072754,
      "learning_rate": 1.5035619589176757e-05,
      "loss": 0.856,
      "step": 8739
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.700694561004639,
      "learning_rate": 1.5034980672778967e-05,
      "loss": 1.6629,
      "step": 8740
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0112617015838623,
      "learning_rate": 1.5034341756381179e-05,
      "loss": 0.9402,
      "step": 8741
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.510061502456665,
      "learning_rate": 1.5033702839983389e-05,
      "loss": 1.0316,
      "step": 8742
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.272035598754883,
      "learning_rate": 1.50330639235856e-05,
      "loss": 1.1659,
      "step": 8743
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.81612491607666,
      "learning_rate": 1.503242500718781e-05,
      "loss": 0.9573,
      "step": 8744
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.513700008392334,
      "learning_rate": 1.5031786090790022e-05,
      "loss": 1.011,
      "step": 8745
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0559394359588623,
      "learning_rate": 1.5031147174392232e-05,
      "loss": 0.9113,
      "step": 8746
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.2294387817382812,
      "learning_rate": 1.5030508257994443e-05,
      "loss": 0.8718,
      "step": 8747
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8216761350631714,
      "learning_rate": 1.5029869341596653e-05,
      "loss": 0.9556,
      "step": 8748
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.793783187866211,
      "learning_rate": 1.5029230425198864e-05,
      "loss": 0.9656,
      "step": 8749
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7461973428726196,
      "learning_rate": 1.5028591508801074e-05,
      "loss": 0.9688,
      "step": 8750
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8048640489578247,
      "learning_rate": 1.5027952592403286e-05,
      "loss": 1.0204,
      "step": 8751
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.639488935470581,
      "learning_rate": 1.5027313676005496e-05,
      "loss": 0.8171,
      "step": 8752
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7006964683532715,
      "learning_rate": 1.5026674759607707e-05,
      "loss": 0.9636,
      "step": 8753
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.21193265914917,
      "learning_rate": 1.5026035843209915e-05,
      "loss": 0.9666,
      "step": 8754
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.522812843322754,
      "learning_rate": 1.5025396926812129e-05,
      "loss": 2.3717,
      "step": 8755
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8775901794433594,
      "learning_rate": 1.5024758010414337e-05,
      "loss": 0.993,
      "step": 8756
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0366694927215576,
      "learning_rate": 1.502411909401655e-05,
      "loss": 0.5794,
      "step": 8757
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5821255445480347,
      "learning_rate": 1.5023480177618758e-05,
      "loss": 0.9304,
      "step": 8758
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6520674228668213,
      "learning_rate": 1.5022841261220971e-05,
      "loss": 0.7535,
      "step": 8759
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.639693021774292,
      "learning_rate": 1.502220234482318e-05,
      "loss": 0.9527,
      "step": 8760
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7245910167694092,
      "learning_rate": 1.5021563428425391e-05,
      "loss": 0.9,
      "step": 8761
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.4805331230163574,
      "learning_rate": 1.5020924512027601e-05,
      "loss": 1.0941,
      "step": 8762
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.725245714187622,
      "learning_rate": 1.5020285595629813e-05,
      "loss": 0.9093,
      "step": 8763
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6588712930679321,
      "learning_rate": 1.5019646679232022e-05,
      "loss": 0.9637,
      "step": 8764
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8735909461975098,
      "learning_rate": 1.5019007762834234e-05,
      "loss": 0.9571,
      "step": 8765
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.051440715789795,
      "learning_rate": 1.5018368846436444e-05,
      "loss": 0.8471,
      "step": 8766
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.005789279937744,
      "learning_rate": 1.5017729930038655e-05,
      "loss": 0.8076,
      "step": 8767
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7126963138580322,
      "learning_rate": 1.5017091013640865e-05,
      "loss": 0.9352,
      "step": 8768
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7292745113372803,
      "learning_rate": 1.5016452097243077e-05,
      "loss": 0.9335,
      "step": 8769
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9052835702896118,
      "learning_rate": 1.5015813180845287e-05,
      "loss": 0.7887,
      "step": 8770
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.245521545410156,
      "learning_rate": 1.5015174264447498e-05,
      "loss": 0.6106,
      "step": 8771
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5702316761016846,
      "learning_rate": 1.5014535348049708e-05,
      "loss": 0.8857,
      "step": 8772
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5007786750793457,
      "learning_rate": 1.501389643165192e-05,
      "loss": 0.4335,
      "step": 8773
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7610331773757935,
      "learning_rate": 1.501325751525413e-05,
      "loss": 0.9769,
      "step": 8774
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7924424409866333,
      "learning_rate": 1.5012618598856341e-05,
      "loss": 0.8956,
      "step": 8775
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2984023094177246,
      "learning_rate": 1.501197968245855e-05,
      "loss": 1.0841,
      "step": 8776
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8769997358322144,
      "learning_rate": 1.5011340766060762e-05,
      "loss": 0.8509,
      "step": 8777
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.835574746131897,
      "learning_rate": 1.5010701849662972e-05,
      "loss": 0.8945,
      "step": 8778
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6948140859603882,
      "learning_rate": 1.5010062933265184e-05,
      "loss": 1.2068,
      "step": 8779
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7334134578704834,
      "learning_rate": 1.5009424016867394e-05,
      "loss": 0.9697,
      "step": 8780
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.858640432357788,
      "learning_rate": 1.5008785100469605e-05,
      "loss": 0.9808,
      "step": 8781
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.1252613067626953,
      "learning_rate": 1.5008146184071815e-05,
      "loss": 1.1216,
      "step": 8782
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.3515214920043945,
      "learning_rate": 1.5007507267674027e-05,
      "loss": 0.9903,
      "step": 8783
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6940679550170898,
      "learning_rate": 1.5006868351276236e-05,
      "loss": 0.7969,
      "step": 8784
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7663155794143677,
      "learning_rate": 1.5006229434878448e-05,
      "loss": 0.8498,
      "step": 8785
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.5226519107818604,
      "learning_rate": 1.5005590518480658e-05,
      "loss": 1.0547,
      "step": 8786
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6782207489013672,
      "learning_rate": 1.500495160208287e-05,
      "loss": 0.9239,
      "step": 8787
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.725355625152588,
      "learning_rate": 1.500431268568508e-05,
      "loss": 0.8741,
      "step": 8788
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.167546033859253,
      "learning_rate": 1.500367376928729e-05,
      "loss": 0.988,
      "step": 8789
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4738006591796875,
      "learning_rate": 1.50030348528895e-05,
      "loss": 1.1216,
      "step": 8790
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.223547458648682,
      "learning_rate": 1.5002395936491712e-05,
      "loss": 1.3338,
      "step": 8791
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9200005531311035,
      "learning_rate": 1.500175702009392e-05,
      "loss": 0.925,
      "step": 8792
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9497019052505493,
      "learning_rate": 1.5001118103696134e-05,
      "loss": 0.8186,
      "step": 8793
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.5733234882354736,
      "learning_rate": 1.5000479187298342e-05,
      "loss": 0.9558,
      "step": 8794
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.488155484199524,
      "learning_rate": 1.4999840270900555e-05,
      "loss": 0.4027,
      "step": 8795
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9926050901412964,
      "learning_rate": 1.4999201354502763e-05,
      "loss": 1.0536,
      "step": 8796
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1826589107513428,
      "learning_rate": 1.4998562438104975e-05,
      "loss": 0.9817,
      "step": 8797
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7360378503799438,
      "learning_rate": 1.4997923521707185e-05,
      "loss": 0.8708,
      "step": 8798
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.33286714553833,
      "learning_rate": 1.4997284605309396e-05,
      "loss": 0.8194,
      "step": 8799
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.0422475337982178,
      "learning_rate": 1.4996645688911606e-05,
      "loss": 0.8948,
      "step": 8800
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0210840702056885,
      "learning_rate": 1.4996006772513817e-05,
      "loss": 1.0083,
      "step": 8801
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.756880521774292,
      "learning_rate": 1.4995367856116027e-05,
      "loss": 1.0241,
      "step": 8802
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.423695087432861,
      "learning_rate": 1.4994728939718239e-05,
      "loss": 1.4664,
      "step": 8803
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8249071836471558,
      "learning_rate": 1.4994090023320449e-05,
      "loss": 1.0308,
      "step": 8804
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6461334228515625,
      "learning_rate": 1.499345110692266e-05,
      "loss": 0.8688,
      "step": 8805
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7784478664398193,
      "learning_rate": 1.499281219052487e-05,
      "loss": 0.4402,
      "step": 8806
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2606499195098877,
      "learning_rate": 1.4992173274127082e-05,
      "loss": 0.9032,
      "step": 8807
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.427285194396973,
      "learning_rate": 1.4991534357729292e-05,
      "loss": 1.2117,
      "step": 8808
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.615037441253662,
      "learning_rate": 1.4990895441331503e-05,
      "loss": 0.4528,
      "step": 8809
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.9206883907318115,
      "learning_rate": 1.4990256524933713e-05,
      "loss": 1.1428,
      "step": 8810
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0475165843963623,
      "learning_rate": 1.4989617608535924e-05,
      "loss": 1.0411,
      "step": 8811
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8032718896865845,
      "learning_rate": 1.4988978692138134e-05,
      "loss": 0.9307,
      "step": 8812
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.940798044204712,
      "learning_rate": 1.4988339775740346e-05,
      "loss": 0.9513,
      "step": 8813
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9155921936035156,
      "learning_rate": 1.4987700859342556e-05,
      "loss": 0.9991,
      "step": 8814
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.837091326713562,
      "learning_rate": 1.4987061942944767e-05,
      "loss": 1.0179,
      "step": 8815
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0046255588531494,
      "learning_rate": 1.4986423026546977e-05,
      "loss": 1.1218,
      "step": 8816
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.3818883895874023,
      "learning_rate": 1.4985784110149189e-05,
      "loss": 0.6514,
      "step": 8817
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.704209327697754,
      "learning_rate": 1.4985145193751399e-05,
      "loss": 0.4399,
      "step": 8818
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1663730144500732,
      "learning_rate": 1.498450627735361e-05,
      "loss": 1.1041,
      "step": 8819
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0558693408966064,
      "learning_rate": 1.498386736095582e-05,
      "loss": 0.8394,
      "step": 8820
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7061545848846436,
      "learning_rate": 1.4983228444558031e-05,
      "loss": 0.9303,
      "step": 8821
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7289384603500366,
      "learning_rate": 1.4982589528160241e-05,
      "loss": 1.2286,
      "step": 8822
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7392609119415283,
      "learning_rate": 1.4981950611762453e-05,
      "loss": 0.8054,
      "step": 8823
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8785182237625122,
      "learning_rate": 1.4981311695364663e-05,
      "loss": 0.8519,
      "step": 8824
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.1306569576263428,
      "learning_rate": 1.4980672778966874e-05,
      "loss": 1.0186,
      "step": 8825
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.833541750907898,
      "learning_rate": 1.4980033862569084e-05,
      "loss": 0.8643,
      "step": 8826
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7903952598571777,
      "learning_rate": 1.4979394946171296e-05,
      "loss": 0.9114,
      "step": 8827
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0798957347869873,
      "learning_rate": 1.4978756029773504e-05,
      "loss": 0.892,
      "step": 8828
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6930925846099854,
      "learning_rate": 1.4978117113375717e-05,
      "loss": 0.9302,
      "step": 8829
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.617551565170288,
      "learning_rate": 1.4977478196977925e-05,
      "loss": 1.0173,
      "step": 8830
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6732062101364136,
      "learning_rate": 1.4976839280580138e-05,
      "loss": 1.0101,
      "step": 8831
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9836150407791138,
      "learning_rate": 1.4976200364182347e-05,
      "loss": 0.9611,
      "step": 8832
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7654976844787598,
      "learning_rate": 1.4975561447784558e-05,
      "loss": 0.9615,
      "step": 8833
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.867405652999878,
      "learning_rate": 1.4974922531386768e-05,
      "loss": 0.9435,
      "step": 8834
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0985894203186035,
      "learning_rate": 1.497428361498898e-05,
      "loss": 0.8467,
      "step": 8835
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0907750129699707,
      "learning_rate": 1.497364469859119e-05,
      "loss": 0.94,
      "step": 8836
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.890707492828369,
      "learning_rate": 1.4973005782193401e-05,
      "loss": 1.3544,
      "step": 8837
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8538811206817627,
      "learning_rate": 1.497236686579561e-05,
      "loss": 1.2441,
      "step": 8838
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6415072679519653,
      "learning_rate": 1.4971727949397822e-05,
      "loss": 0.932,
      "step": 8839
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5509703159332275,
      "learning_rate": 1.4971089033000032e-05,
      "loss": 0.8498,
      "step": 8840
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.003537893295288,
      "learning_rate": 1.4970450116602244e-05,
      "loss": 0.9556,
      "step": 8841
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5114550590515137,
      "learning_rate": 1.4969811200204454e-05,
      "loss": 0.4225,
      "step": 8842
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5635775327682495,
      "learning_rate": 1.4969172283806665e-05,
      "loss": 0.8612,
      "step": 8843
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.771679162979126,
      "learning_rate": 1.4968533367408875e-05,
      "loss": 0.9162,
      "step": 8844
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5877937078475952,
      "learning_rate": 1.4967894451011087e-05,
      "loss": 0.902,
      "step": 8845
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9585702419281006,
      "learning_rate": 1.4967255534613296e-05,
      "loss": 0.9413,
      "step": 8846
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.679876685142517,
      "learning_rate": 1.4966616618215508e-05,
      "loss": 0.8506,
      "step": 8847
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6347153186798096,
      "learning_rate": 1.4965977701817718e-05,
      "loss": 0.9,
      "step": 8848
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4459269046783447,
      "learning_rate": 1.496533878541993e-05,
      "loss": 1.1067,
      "step": 8849
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.805104970932007,
      "learning_rate": 1.496469986902214e-05,
      "loss": 1.2023,
      "step": 8850
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9026464223861694,
      "learning_rate": 1.496406095262435e-05,
      "loss": 0.9551,
      "step": 8851
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.1409623622894287,
      "learning_rate": 1.496342203622656e-05,
      "loss": 0.6881,
      "step": 8852
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7417300939559937,
      "learning_rate": 1.4962783119828772e-05,
      "loss": 0.9784,
      "step": 8853
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9346264600753784,
      "learning_rate": 1.4962144203430982e-05,
      "loss": 0.9931,
      "step": 8854
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4119956493377686,
      "learning_rate": 1.4961505287033194e-05,
      "loss": 0.7746,
      "step": 8855
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1065735816955566,
      "learning_rate": 1.4960866370635403e-05,
      "loss": 0.915,
      "step": 8856
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.499668836593628,
      "learning_rate": 1.4960227454237615e-05,
      "loss": 0.7794,
      "step": 8857
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8789196014404297,
      "learning_rate": 1.4959588537839825e-05,
      "loss": 1.038,
      "step": 8858
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.090175151824951,
      "learning_rate": 1.4958949621442036e-05,
      "loss": 0.9208,
      "step": 8859
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0007407665252686,
      "learning_rate": 1.4958310705044246e-05,
      "loss": 1.0838,
      "step": 8860
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7581712007522583,
      "learning_rate": 1.4957671788646458e-05,
      "loss": 0.9118,
      "step": 8861
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.851960301399231,
      "learning_rate": 1.4957032872248668e-05,
      "loss": 0.98,
      "step": 8862
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7132576704025269,
      "learning_rate": 1.495639395585088e-05,
      "loss": 0.9617,
      "step": 8863
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0040788650512695,
      "learning_rate": 1.4955755039453087e-05,
      "loss": 0.8769,
      "step": 8864
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.430856466293335,
      "learning_rate": 1.49551161230553e-05,
      "loss": 0.8339,
      "step": 8865
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.09305477142334,
      "learning_rate": 1.4954477206657509e-05,
      "loss": 0.9377,
      "step": 8866
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6120471954345703,
      "learning_rate": 1.4953838290259722e-05,
      "loss": 0.934,
      "step": 8867
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.065366744995117,
      "learning_rate": 1.495319937386193e-05,
      "loss": 1.4858,
      "step": 8868
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7819533348083496,
      "learning_rate": 1.4952560457464143e-05,
      "loss": 0.9407,
      "step": 8869
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.861769437789917,
      "learning_rate": 1.4951921541066352e-05,
      "loss": 0.6464,
      "step": 8870
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.393465518951416,
      "learning_rate": 1.4951282624668563e-05,
      "loss": 1.025,
      "step": 8871
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.053281545639038,
      "learning_rate": 1.4950643708270773e-05,
      "loss": 0.9977,
      "step": 8872
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2610340118408203,
      "learning_rate": 1.4950004791872984e-05,
      "loss": 1.0907,
      "step": 8873
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9159235954284668,
      "learning_rate": 1.4949365875475194e-05,
      "loss": 0.98,
      "step": 8874
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.665591239929199,
      "learning_rate": 1.4948726959077406e-05,
      "loss": 1.1363,
      "step": 8875
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0099871158599854,
      "learning_rate": 1.4948088042679616e-05,
      "loss": 1.1855,
      "step": 8876
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5734708309173584,
      "learning_rate": 1.4947449126281827e-05,
      "loss": 0.8734,
      "step": 8877
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4465088844299316,
      "learning_rate": 1.4946810209884037e-05,
      "loss": 0.8757,
      "step": 8878
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0571508407592773,
      "learning_rate": 1.4946171293486249e-05,
      "loss": 0.8918,
      "step": 8879
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.714842677116394,
      "learning_rate": 1.4945532377088459e-05,
      "loss": 0.8913,
      "step": 8880
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5908820629119873,
      "learning_rate": 1.494489346069067e-05,
      "loss": 0.8318,
      "step": 8881
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.490553379058838,
      "learning_rate": 1.494425454429288e-05,
      "loss": 0.8493,
      "step": 8882
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.715990424156189,
      "learning_rate": 1.4943615627895091e-05,
      "loss": 0.9763,
      "step": 8883
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2068612575531006,
      "learning_rate": 1.4942976711497301e-05,
      "loss": 0.8986,
      "step": 8884
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1908669471740723,
      "learning_rate": 1.4942337795099513e-05,
      "loss": 1.0641,
      "step": 8885
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3963937759399414,
      "learning_rate": 1.4941698878701723e-05,
      "loss": 0.9278,
      "step": 8886
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8048228025436401,
      "learning_rate": 1.4941059962303934e-05,
      "loss": 1.2747,
      "step": 8887
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.06880259513855,
      "learning_rate": 1.4940421045906144e-05,
      "loss": 0.9091,
      "step": 8888
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.526808500289917,
      "learning_rate": 1.4939782129508356e-05,
      "loss": 0.8581,
      "step": 8889
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4928274154663086,
      "learning_rate": 1.4939143213110566e-05,
      "loss": 0.7163,
      "step": 8890
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.465343713760376,
      "learning_rate": 1.4938504296712777e-05,
      "loss": 0.5387,
      "step": 8891
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.030648946762085,
      "learning_rate": 1.4937865380314987e-05,
      "loss": 1.2585,
      "step": 8892
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4518635272979736,
      "learning_rate": 1.4937226463917198e-05,
      "loss": 0.6115,
      "step": 8893
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.366668939590454,
      "learning_rate": 1.4936587547519408e-05,
      "loss": 0.7251,
      "step": 8894
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.3618693351745605,
      "learning_rate": 1.493594863112162e-05,
      "loss": 1.2275,
      "step": 8895
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9123494625091553,
      "learning_rate": 1.493530971472383e-05,
      "loss": 0.9601,
      "step": 8896
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9008309841156006,
      "learning_rate": 1.4934670798326041e-05,
      "loss": 1.0677,
      "step": 8897
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9499220848083496,
      "learning_rate": 1.4934031881928251e-05,
      "loss": 0.7502,
      "step": 8898
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7038694620132446,
      "learning_rate": 1.4933392965530463e-05,
      "loss": 0.9542,
      "step": 8899
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.157074451446533,
      "learning_rate": 1.4932754049132673e-05,
      "loss": 0.8492,
      "step": 8900
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.597400665283203,
      "learning_rate": 1.4932115132734884e-05,
      "loss": 1.5794,
      "step": 8901
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4673819541931152,
      "learning_rate": 1.4931476216337092e-05,
      "loss": 0.429,
      "step": 8902
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.609931468963623,
      "learning_rate": 1.4930837299939305e-05,
      "loss": 1.2788,
      "step": 8903
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.7056894302368164,
      "learning_rate": 1.4930198383541514e-05,
      "loss": 1.1012,
      "step": 8904
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2510650157928467,
      "learning_rate": 1.4929559467143727e-05,
      "loss": 0.9272,
      "step": 8905
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9528889656066895,
      "learning_rate": 1.4928920550745935e-05,
      "loss": 1.0153,
      "step": 8906
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.132279634475708,
      "learning_rate": 1.4928281634348147e-05,
      "loss": 0.9963,
      "step": 8907
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5234930515289307,
      "learning_rate": 1.4927642717950356e-05,
      "loss": 0.9136,
      "step": 8908
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.691391110420227,
      "learning_rate": 1.4927003801552568e-05,
      "loss": 1.0676,
      "step": 8909
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.174858808517456,
      "learning_rate": 1.4926364885154778e-05,
      "loss": 0.7846,
      "step": 8910
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3867918252944946,
      "learning_rate": 1.492572596875699e-05,
      "loss": 0.8553,
      "step": 8911
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5885354280471802,
      "learning_rate": 1.49250870523592e-05,
      "loss": 0.413,
      "step": 8912
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2535059452056885,
      "learning_rate": 1.492444813596141e-05,
      "loss": 0.9433,
      "step": 8913
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0875566005706787,
      "learning_rate": 1.492380921956362e-05,
      "loss": 0.9118,
      "step": 8914
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.9743576049804688,
      "learning_rate": 1.4923170303165832e-05,
      "loss": 1.2097,
      "step": 8915
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.931382894515991,
      "learning_rate": 1.4922531386768042e-05,
      "loss": 1.087,
      "step": 8916
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.102508544921875,
      "learning_rate": 1.4921892470370254e-05,
      "loss": 0.9501,
      "step": 8917
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0542657375335693,
      "learning_rate": 1.4921253553972463e-05,
      "loss": 1.0413,
      "step": 8918
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2254793643951416,
      "learning_rate": 1.4920614637574675e-05,
      "loss": 0.9649,
      "step": 8919
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.227015018463135,
      "learning_rate": 1.4919975721176885e-05,
      "loss": 1.1844,
      "step": 8920
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.562596559524536,
      "learning_rate": 1.4919336804779096e-05,
      "loss": 1.0557,
      "step": 8921
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4140396118164062,
      "learning_rate": 1.4918697888381306e-05,
      "loss": 0.9434,
      "step": 8922
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0670952796936035,
      "learning_rate": 1.4918058971983518e-05,
      "loss": 1.0648,
      "step": 8923
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.619882345199585,
      "learning_rate": 1.4917420055585728e-05,
      "loss": 0.8876,
      "step": 8924
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9252876043319702,
      "learning_rate": 1.491678113918794e-05,
      "loss": 1.2402,
      "step": 8925
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8184484243392944,
      "learning_rate": 1.4916142222790149e-05,
      "loss": 0.9245,
      "step": 8926
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.7252416610717773,
      "learning_rate": 1.491550330639236e-05,
      "loss": 1.0322,
      "step": 8927
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8323371410369873,
      "learning_rate": 1.491486438999457e-05,
      "loss": 0.9709,
      "step": 8928
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6559388637542725,
      "learning_rate": 1.4914225473596782e-05,
      "loss": 1.1226,
      "step": 8929
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.371798515319824,
      "learning_rate": 1.4913586557198992e-05,
      "loss": 1.1693,
      "step": 8930
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6665974855422974,
      "learning_rate": 1.4912947640801203e-05,
      "loss": 1.0152,
      "step": 8931
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6332197189331055,
      "learning_rate": 1.4912308724403413e-05,
      "loss": 0.8716,
      "step": 8932
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0768861770629883,
      "learning_rate": 1.4911669808005625e-05,
      "loss": 0.9481,
      "step": 8933
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8276515007019043,
      "learning_rate": 1.4911030891607835e-05,
      "loss": 1.0419,
      "step": 8934
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9618526697158813,
      "learning_rate": 1.4910391975210046e-05,
      "loss": 1.0286,
      "step": 8935
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6624265909194946,
      "learning_rate": 1.4909753058812256e-05,
      "loss": 0.9058,
      "step": 8936
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4708064794540405,
      "learning_rate": 1.4909114142414468e-05,
      "loss": 0.3548,
      "step": 8937
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.040693759918213,
      "learning_rate": 1.4908475226016676e-05,
      "loss": 0.8947,
      "step": 8938
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.8064048290252686,
      "learning_rate": 1.4907836309618889e-05,
      "loss": 1.0816,
      "step": 8939
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6866899728775024,
      "learning_rate": 1.4907197393221097e-05,
      "loss": 0.9937,
      "step": 8940
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.504680871963501,
      "learning_rate": 1.490655847682331e-05,
      "loss": 0.9147,
      "step": 8941
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5967209339141846,
      "learning_rate": 1.4905919560425519e-05,
      "loss": 0.8667,
      "step": 8942
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.741106629371643,
      "learning_rate": 1.490528064402773e-05,
      "loss": 0.9565,
      "step": 8943
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5265040397644043,
      "learning_rate": 1.490464172762994e-05,
      "loss": 0.8929,
      "step": 8944
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6598387956619263,
      "learning_rate": 1.4904002811232151e-05,
      "loss": 0.9476,
      "step": 8945
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.9864604473114014,
      "learning_rate": 1.4903363894834361e-05,
      "loss": 1.0841,
      "step": 8946
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0815329551696777,
      "learning_rate": 1.4902724978436573e-05,
      "loss": 1.0119,
      "step": 8947
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.389984607696533,
      "learning_rate": 1.4902086062038783e-05,
      "loss": 0.8974,
      "step": 8948
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7070810794830322,
      "learning_rate": 1.4901447145640994e-05,
      "loss": 0.8339,
      "step": 8949
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.604702949523926,
      "learning_rate": 1.4900808229243204e-05,
      "loss": 0.9967,
      "step": 8950
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.4902923107147217,
      "learning_rate": 1.4900169312845416e-05,
      "loss": 0.8266,
      "step": 8951
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2933106422424316,
      "learning_rate": 1.4899530396447626e-05,
      "loss": 0.5677,
      "step": 8952
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.645169734954834,
      "learning_rate": 1.4898891480049837e-05,
      "loss": 1.109,
      "step": 8953
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7038061618804932,
      "learning_rate": 1.4898252563652047e-05,
      "loss": 0.8797,
      "step": 8954
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.8444478511810303,
      "learning_rate": 1.4897613647254258e-05,
      "loss": 0.8197,
      "step": 8955
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.3012921810150146,
      "learning_rate": 1.4896974730856468e-05,
      "loss": 1.3073,
      "step": 8956
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.599669933319092,
      "learning_rate": 1.489633581445868e-05,
      "loss": 1.7481,
      "step": 8957
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.317673683166504,
      "learning_rate": 1.489569689806089e-05,
      "loss": 1.0874,
      "step": 8958
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5671836137771606,
      "learning_rate": 1.4895057981663101e-05,
      "loss": 0.8655,
      "step": 8959
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1864817142486572,
      "learning_rate": 1.4894419065265311e-05,
      "loss": 1.15,
      "step": 8960
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.573375701904297,
      "learning_rate": 1.4893780148867523e-05,
      "loss": 1.2817,
      "step": 8961
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8727511167526245,
      "learning_rate": 1.4893141232469733e-05,
      "loss": 0.9574,
      "step": 8962
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.154695510864258,
      "learning_rate": 1.4892502316071944e-05,
      "loss": 1.1959,
      "step": 8963
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2616114616394043,
      "learning_rate": 1.4891863399674154e-05,
      "loss": 1.0342,
      "step": 8964
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9296252727508545,
      "learning_rate": 1.4891224483276365e-05,
      "loss": 0.9936,
      "step": 8965
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2869369983673096,
      "learning_rate": 1.4890585566878575e-05,
      "loss": 1.2114,
      "step": 8966
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.893315076828003,
      "learning_rate": 1.4889946650480787e-05,
      "loss": 0.8365,
      "step": 8967
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.045605659484863,
      "learning_rate": 1.4889307734082997e-05,
      "loss": 1.394,
      "step": 8968
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.005581855773926,
      "learning_rate": 1.4888668817685208e-05,
      "loss": 0.8704,
      "step": 8969
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8164513111114502,
      "learning_rate": 1.4888029901287418e-05,
      "loss": 0.933,
      "step": 8970
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.287848711013794,
      "learning_rate": 1.488739098488963e-05,
      "loss": 0.9257,
      "step": 8971
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.601067066192627,
      "learning_rate": 1.488675206849184e-05,
      "loss": 1.1457,
      "step": 8972
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.827864170074463,
      "learning_rate": 1.4886113152094051e-05,
      "loss": 0.9844,
      "step": 8973
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6878142356872559,
      "learning_rate": 1.488547423569626e-05,
      "loss": 0.92,
      "step": 8974
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.659471869468689,
      "learning_rate": 1.4884835319298472e-05,
      "loss": 0.8768,
      "step": 8975
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.558620572090149,
      "learning_rate": 1.488419640290068e-05,
      "loss": 0.4414,
      "step": 8976
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.296529769897461,
      "learning_rate": 1.4883557486502894e-05,
      "loss": 0.8419,
      "step": 8977
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6066256761550903,
      "learning_rate": 1.4882918570105102e-05,
      "loss": 0.9169,
      "step": 8978
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7073721885681152,
      "learning_rate": 1.4882279653707314e-05,
      "loss": 1.181,
      "step": 8979
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8993338346481323,
      "learning_rate": 1.4881640737309523e-05,
      "loss": 1.0236,
      "step": 8980
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.365979194641113,
      "learning_rate": 1.4881001820911735e-05,
      "loss": 1.1615,
      "step": 8981
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6497952938079834,
      "learning_rate": 1.4880362904513945e-05,
      "loss": 0.8741,
      "step": 8982
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.2131266593933105,
      "learning_rate": 1.4879723988116156e-05,
      "loss": 1.3455,
      "step": 8983
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1765475273132324,
      "learning_rate": 1.4879085071718366e-05,
      "loss": 0.9806,
      "step": 8984
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1972134113311768,
      "learning_rate": 1.4878446155320578e-05,
      "loss": 0.9427,
      "step": 8985
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.152205467224121,
      "learning_rate": 1.4877807238922788e-05,
      "loss": 1.1655,
      "step": 8986
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9090888500213623,
      "learning_rate": 1.4877168322525e-05,
      "loss": 0.9979,
      "step": 8987
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8781592845916748,
      "learning_rate": 1.4876529406127209e-05,
      "loss": 1.0012,
      "step": 8988
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4384851455688477,
      "learning_rate": 1.487589048972942e-05,
      "loss": 1.062,
      "step": 8989
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.711676836013794,
      "learning_rate": 1.487525157333163e-05,
      "loss": 0.9632,
      "step": 8990
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8704153299331665,
      "learning_rate": 1.4874612656933842e-05,
      "loss": 0.9167,
      "step": 8991
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1105926036834717,
      "learning_rate": 1.4873973740536052e-05,
      "loss": 0.9038,
      "step": 8992
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.965555191040039,
      "learning_rate": 1.4873334824138263e-05,
      "loss": 1.1063,
      "step": 8993
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5534474849700928,
      "learning_rate": 1.4872695907740473e-05,
      "loss": 0.8353,
      "step": 8994
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0899999141693115,
      "learning_rate": 1.4872056991342685e-05,
      "loss": 0.8355,
      "step": 8995
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9571012258529663,
      "learning_rate": 1.4871418074944895e-05,
      "loss": 0.9469,
      "step": 8996
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4944911003112793,
      "learning_rate": 1.4870779158547106e-05,
      "loss": 0.3875,
      "step": 8997
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4791359901428223,
      "learning_rate": 1.4870140242149316e-05,
      "loss": 1.0548,
      "step": 8998
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.568559169769287,
      "learning_rate": 1.4869501325751528e-05,
      "loss": 1.0235,
      "step": 8999
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.885514736175537,
      "learning_rate": 1.4868862409353737e-05,
      "loss": 0.8887,
      "step": 9000
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4643502235412598,
      "learning_rate": 1.4868223492955949e-05,
      "loss": 0.9185,
      "step": 9001
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7178535461425781,
      "learning_rate": 1.4867584576558159e-05,
      "loss": 0.8011,
      "step": 9002
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.108309745788574,
      "learning_rate": 1.486694566016037e-05,
      "loss": 0.9776,
      "step": 9003
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.898200273513794,
      "learning_rate": 1.486630674376258e-05,
      "loss": 0.9421,
      "step": 9004
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1763365268707275,
      "learning_rate": 1.4865667827364792e-05,
      "loss": 0.9598,
      "step": 9005
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5804953575134277,
      "learning_rate": 1.4865028910967002e-05,
      "loss": 0.8132,
      "step": 9006
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.746803283691406,
      "learning_rate": 1.4864389994569213e-05,
      "loss": 1.0018,
      "step": 9007
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.865322470664978,
      "learning_rate": 1.4863751078171423e-05,
      "loss": 0.8956,
      "step": 9008
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.357924222946167,
      "learning_rate": 1.4863112161773635e-05,
      "loss": 0.575,
      "step": 9009
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2503175735473633,
      "learning_rate": 1.4862473245375843e-05,
      "loss": 0.8724,
      "step": 9010
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3502588272094727,
      "learning_rate": 1.4861834328978056e-05,
      "loss": 1.2078,
      "step": 9011
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5880666971206665,
      "learning_rate": 1.4861195412580264e-05,
      "loss": 0.9902,
      "step": 9012
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5691790580749512,
      "learning_rate": 1.4860556496182477e-05,
      "loss": 0.3828,
      "step": 9013
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9307749271392822,
      "learning_rate": 1.4859917579784686e-05,
      "loss": 0.9351,
      "step": 9014
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7814702987670898,
      "learning_rate": 1.4859278663386899e-05,
      "loss": 0.4604,
      "step": 9015
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0933339595794678,
      "learning_rate": 1.4858639746989107e-05,
      "loss": 1.2812,
      "step": 9016
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8751683235168457,
      "learning_rate": 1.4858000830591318e-05,
      "loss": 0.8742,
      "step": 9017
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9504696130752563,
      "learning_rate": 1.4857361914193528e-05,
      "loss": 1.0278,
      "step": 9018
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.475036144256592,
      "learning_rate": 1.485672299779574e-05,
      "loss": 1.1295,
      "step": 9019
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.817326784133911,
      "learning_rate": 1.485608408139795e-05,
      "loss": 0.9774,
      "step": 9020
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.705345869064331,
      "learning_rate": 1.4855445165000161e-05,
      "loss": 0.8545,
      "step": 9021
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.65962553024292,
      "learning_rate": 1.4854806248602371e-05,
      "loss": 0.9679,
      "step": 9022
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7615169286727905,
      "learning_rate": 1.4854167332204583e-05,
      "loss": 1.2152,
      "step": 9023
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8247181177139282,
      "learning_rate": 1.4853528415806793e-05,
      "loss": 0.9444,
      "step": 9024
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6976852416992188,
      "learning_rate": 1.4852889499409004e-05,
      "loss": 0.6324,
      "step": 9025
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.325242757797241,
      "learning_rate": 1.4852250583011214e-05,
      "loss": 0.9587,
      "step": 9026
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.989389181137085,
      "learning_rate": 1.4851611666613425e-05,
      "loss": 1.0997,
      "step": 9027
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.852639198303223,
      "learning_rate": 1.4850972750215635e-05,
      "loss": 1.2349,
      "step": 9028
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6983901262283325,
      "learning_rate": 1.4850333833817847e-05,
      "loss": 1.2995,
      "step": 9029
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.779354214668274,
      "learning_rate": 1.4849694917420057e-05,
      "loss": 1.0908,
      "step": 9030
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.548211574554443,
      "learning_rate": 1.4849056001022268e-05,
      "loss": 1.1489,
      "step": 9031
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5973037481307983,
      "learning_rate": 1.4848417084624478e-05,
      "loss": 0.5066,
      "step": 9032
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.912644386291504,
      "learning_rate": 1.484777816822669e-05,
      "loss": 1.2364,
      "step": 9033
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2876880168914795,
      "learning_rate": 1.48471392518289e-05,
      "loss": 0.6213,
      "step": 9034
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.786968469619751,
      "learning_rate": 1.4846500335431111e-05,
      "loss": 1.3107,
      "step": 9035
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.9728310108184814,
      "learning_rate": 1.4845861419033321e-05,
      "loss": 1.3238,
      "step": 9036
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1064000129699707,
      "learning_rate": 1.4845222502635532e-05,
      "loss": 1.2549,
      "step": 9037
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0074408054351807,
      "learning_rate": 1.4844583586237742e-05,
      "loss": 0.8102,
      "step": 9038
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5714291334152222,
      "learning_rate": 1.4843944669839954e-05,
      "loss": 0.8597,
      "step": 9039
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9808803796768188,
      "learning_rate": 1.4843305753442164e-05,
      "loss": 0.8637,
      "step": 9040
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3885276317596436,
      "learning_rate": 1.4842666837044375e-05,
      "loss": 0.7411,
      "step": 9041
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.050677537918091,
      "learning_rate": 1.4842027920646585e-05,
      "loss": 0.8379,
      "step": 9042
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.972888231277466,
      "learning_rate": 1.4841389004248797e-05,
      "loss": 0.8758,
      "step": 9043
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0938732624053955,
      "learning_rate": 1.4840750087851007e-05,
      "loss": 0.7857,
      "step": 9044
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7653573751449585,
      "learning_rate": 1.4840111171453218e-05,
      "loss": 0.9379,
      "step": 9045
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8700019121170044,
      "learning_rate": 1.4839472255055428e-05,
      "loss": 0.8788,
      "step": 9046
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7405399084091187,
      "learning_rate": 1.483883333865764e-05,
      "loss": 0.9528,
      "step": 9047
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0403904914855957,
      "learning_rate": 1.4838194422259848e-05,
      "loss": 0.9771,
      "step": 9048
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.692114233970642,
      "learning_rate": 1.4837555505862061e-05,
      "loss": 0.8884,
      "step": 9049
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5430253744125366,
      "learning_rate": 1.4836916589464269e-05,
      "loss": 0.949,
      "step": 9050
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2983391284942627,
      "learning_rate": 1.4836277673066482e-05,
      "loss": 1.0318,
      "step": 9051
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9485751390457153,
      "learning_rate": 1.483563875666869e-05,
      "loss": 0.9482,
      "step": 9052
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.392993927001953,
      "learning_rate": 1.4834999840270902e-05,
      "loss": 1.0542,
      "step": 9053
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.120358943939209,
      "learning_rate": 1.4834360923873112e-05,
      "loss": 1.6294,
      "step": 9054
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5011128187179565,
      "learning_rate": 1.4833722007475323e-05,
      "loss": 0.89,
      "step": 9055
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9349019527435303,
      "learning_rate": 1.4833083091077533e-05,
      "loss": 0.9313,
      "step": 9056
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.352173089981079,
      "learning_rate": 1.4832444174679745e-05,
      "loss": 0.7167,
      "step": 9057
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0342187881469727,
      "learning_rate": 1.4831805258281955e-05,
      "loss": 0.4733,
      "step": 9058
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2113003730773926,
      "learning_rate": 1.4831166341884166e-05,
      "loss": 0.9626,
      "step": 9059
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9413058757781982,
      "learning_rate": 1.4830527425486376e-05,
      "loss": 0.6083,
      "step": 9060
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.030700445175171,
      "learning_rate": 1.4829888509088588e-05,
      "loss": 0.9152,
      "step": 9061
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.604321002960205,
      "learning_rate": 1.4829249592690797e-05,
      "loss": 1.0169,
      "step": 9062
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.978726387023926,
      "learning_rate": 1.4828610676293009e-05,
      "loss": 1.627,
      "step": 9063
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6183667182922363,
      "learning_rate": 1.4827971759895219e-05,
      "loss": 0.9097,
      "step": 9064
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6422629356384277,
      "learning_rate": 1.482733284349743e-05,
      "loss": 0.6569,
      "step": 9065
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0674400329589844,
      "learning_rate": 1.482669392709964e-05,
      "loss": 0.723,
      "step": 9066
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.588965892791748,
      "learning_rate": 1.4826055010701852e-05,
      "loss": 1.0258,
      "step": 9067
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.12515926361084,
      "learning_rate": 1.4825416094304062e-05,
      "loss": 0.8919,
      "step": 9068
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9711081981658936,
      "learning_rate": 1.4824777177906273e-05,
      "loss": 1.3627,
      "step": 9069
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0524754524230957,
      "learning_rate": 1.4824138261508483e-05,
      "loss": 1.0392,
      "step": 9070
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1492724418640137,
      "learning_rate": 1.4823499345110695e-05,
      "loss": 1.2378,
      "step": 9071
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8876092433929443,
      "learning_rate": 1.4822860428712904e-05,
      "loss": 0.9349,
      "step": 9072
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8838624954223633,
      "learning_rate": 1.4822221512315116e-05,
      "loss": 1.1341,
      "step": 9073
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4870855808258057,
      "learning_rate": 1.4821582595917326e-05,
      "loss": 1.1639,
      "step": 9074
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.706836223602295,
      "learning_rate": 1.4820943679519537e-05,
      "loss": 0.9262,
      "step": 9075
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8622902631759644,
      "learning_rate": 1.4820304763121747e-05,
      "loss": 0.8872,
      "step": 9076
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6392836570739746,
      "learning_rate": 1.4819665846723959e-05,
      "loss": 0.869,
      "step": 9077
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1070022583007812,
      "learning_rate": 1.4819026930326169e-05,
      "loss": 1.3396,
      "step": 9078
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.435824394226074,
      "learning_rate": 1.481838801392838e-05,
      "loss": 0.9182,
      "step": 9079
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7684067487716675,
      "learning_rate": 1.481774909753059e-05,
      "loss": 0.5196,
      "step": 9080
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3095431327819824,
      "learning_rate": 1.4817110181132802e-05,
      "loss": 0.8031,
      "step": 9081
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6832094192504883,
      "learning_rate": 1.4816471264735011e-05,
      "loss": 0.9059,
      "step": 9082
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9929029941558838,
      "learning_rate": 1.4815832348337223e-05,
      "loss": 1.0651,
      "step": 9083
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.111217498779297,
      "learning_rate": 1.4815193431939431e-05,
      "loss": 0.9595,
      "step": 9084
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.045921564102173,
      "learning_rate": 1.4814554515541641e-05,
      "loss": 0.9692,
      "step": 9085
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5418508052825928,
      "learning_rate": 1.4813915599143853e-05,
      "loss": 1.0543,
      "step": 9086
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5866193771362305,
      "learning_rate": 1.4813276682746062e-05,
      "loss": 0.5689,
      "step": 9087
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4603749513626099,
      "learning_rate": 1.4812637766348274e-05,
      "loss": 0.7912,
      "step": 9088
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8626612424850464,
      "learning_rate": 1.4811998849950484e-05,
      "loss": 0.9088,
      "step": 9089
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3434122800827026,
      "learning_rate": 1.4811359933552695e-05,
      "loss": 0.8407,
      "step": 9090
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1209421157836914,
      "learning_rate": 1.4810721017154905e-05,
      "loss": 1.1782,
      "step": 9091
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7849335670471191,
      "learning_rate": 1.4810082100757117e-05,
      "loss": 0.9083,
      "step": 9092
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8957558870315552,
      "learning_rate": 1.4809443184359327e-05,
      "loss": 1.0116,
      "step": 9093
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9198886156082153,
      "learning_rate": 1.4808804267961538e-05,
      "loss": 0.9593,
      "step": 9094
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6087491512298584,
      "learning_rate": 1.4808165351563748e-05,
      "loss": 0.8318,
      "step": 9095
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2577598094940186,
      "learning_rate": 1.480752643516596e-05,
      "loss": 0.9525,
      "step": 9096
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.8321800231933594,
      "learning_rate": 1.480688751876817e-05,
      "loss": 0.9397,
      "step": 9097
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.75475013256073,
      "learning_rate": 1.4806248602370381e-05,
      "loss": 1.224,
      "step": 9098
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.8741230964660645,
      "learning_rate": 1.480560968597259e-05,
      "loss": 1.4059,
      "step": 9099
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.6368303298950195,
      "learning_rate": 1.4804970769574802e-05,
      "loss": 1.1166,
      "step": 9100
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4160616397857666,
      "learning_rate": 1.4804331853177012e-05,
      "loss": 0.848,
      "step": 9101
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8474152088165283,
      "learning_rate": 1.4803692936779224e-05,
      "loss": 0.869,
      "step": 9102
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8545541763305664,
      "learning_rate": 1.4803054020381434e-05,
      "loss": 0.9883,
      "step": 9103
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.828737258911133,
      "learning_rate": 1.4802415103983645e-05,
      "loss": 1.0097,
      "step": 9104
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2464518547058105,
      "learning_rate": 1.4801776187585855e-05,
      "loss": 0.7164,
      "step": 9105
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5093791484832764,
      "learning_rate": 1.4801137271188067e-05,
      "loss": 1.0135,
      "step": 9106
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7018836736679077,
      "learning_rate": 1.4800498354790276e-05,
      "loss": 0.7502,
      "step": 9107
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5645631551742554,
      "learning_rate": 1.4799859438392488e-05,
      "loss": 0.9018,
      "step": 9108
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8981877565383911,
      "learning_rate": 1.4799220521994698e-05,
      "loss": 1.0822,
      "step": 9109
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8378503322601318,
      "learning_rate": 1.479858160559691e-05,
      "loss": 0.8436,
      "step": 9110
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0478930473327637,
      "learning_rate": 1.479794268919912e-05,
      "loss": 0.9054,
      "step": 9111
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.541426420211792,
      "learning_rate": 1.479730377280133e-05,
      "loss": 0.7689,
      "step": 9112
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0690460205078125,
      "learning_rate": 1.479666485640354e-05,
      "loss": 0.7582,
      "step": 9113
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1796581745147705,
      "learning_rate": 1.4796025940005752e-05,
      "loss": 0.9805,
      "step": 9114
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1462783813476562,
      "learning_rate": 1.479538702360796e-05,
      "loss": 1.0361,
      "step": 9115
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.448864221572876,
      "learning_rate": 1.4794748107210174e-05,
      "loss": 1.0322,
      "step": 9116
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.969987154006958,
      "learning_rate": 1.4794109190812382e-05,
      "loss": 0.7377,
      "step": 9117
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.398995876312256,
      "learning_rate": 1.4793470274414595e-05,
      "loss": 0.7057,
      "step": 9118
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5453404188156128,
      "learning_rate": 1.4792831358016803e-05,
      "loss": 0.4947,
      "step": 9119
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.451947212219238,
      "learning_rate": 1.4792192441619015e-05,
      "loss": 1.0368,
      "step": 9120
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.024282217025757,
      "learning_rate": 1.4791553525221225e-05,
      "loss": 0.914,
      "step": 9121
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7617791891098022,
      "learning_rate": 1.4790914608823436e-05,
      "loss": 0.8873,
      "step": 9122
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.282385349273682,
      "learning_rate": 1.4790275692425646e-05,
      "loss": 1.2325,
      "step": 9123
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6495530605316162,
      "learning_rate": 1.4789636776027857e-05,
      "loss": 0.8851,
      "step": 9124
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8224295377731323,
      "learning_rate": 1.4788997859630067e-05,
      "loss": 0.8487,
      "step": 9125
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4308087825775146,
      "learning_rate": 1.4788358943232279e-05,
      "loss": 1.2224,
      "step": 9126
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9811935424804688,
      "learning_rate": 1.4787720026834489e-05,
      "loss": 0.9715,
      "step": 9127
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.350533962249756,
      "learning_rate": 1.47870811104367e-05,
      "loss": 0.9896,
      "step": 9128
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.657576560974121,
      "learning_rate": 1.478644219403891e-05,
      "loss": 0.9149,
      "step": 9129
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7897264957427979,
      "learning_rate": 1.4785803277641122e-05,
      "loss": 1.0287,
      "step": 9130
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8492906093597412,
      "learning_rate": 1.4785164361243332e-05,
      "loss": 1.1813,
      "step": 9131
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7018846273422241,
      "learning_rate": 1.4784525444845543e-05,
      "loss": 0.9268,
      "step": 9132
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6301549673080444,
      "learning_rate": 1.4783886528447753e-05,
      "loss": 0.8471,
      "step": 9133
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9331862926483154,
      "learning_rate": 1.4783247612049964e-05,
      "loss": 1.0747,
      "step": 9134
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.855809211730957,
      "learning_rate": 1.4782608695652174e-05,
      "loss": 1.092,
      "step": 9135
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0386807918548584,
      "learning_rate": 1.4781969779254386e-05,
      "loss": 0.8774,
      "step": 9136
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.387057304382324,
      "learning_rate": 1.4781330862856596e-05,
      "loss": 0.7112,
      "step": 9137
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.4956183433532715,
      "learning_rate": 1.4780691946458807e-05,
      "loss": 0.8225,
      "step": 9138
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0903711318969727,
      "learning_rate": 1.4780053030061017e-05,
      "loss": 0.9503,
      "step": 9139
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.817164182662964,
      "learning_rate": 1.4779414113663229e-05,
      "loss": 0.82,
      "step": 9140
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.080787420272827,
      "learning_rate": 1.4778775197265439e-05,
      "loss": 1.1234,
      "step": 9141
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5670002698898315,
      "learning_rate": 1.477813628086765e-05,
      "loss": 0.7699,
      "step": 9142
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5372684001922607,
      "learning_rate": 1.477749736446986e-05,
      "loss": 1.0364,
      "step": 9143
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.94190514087677,
      "learning_rate": 1.4776858448072071e-05,
      "loss": 0.8872,
      "step": 9144
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7275664806365967,
      "learning_rate": 1.4776219531674281e-05,
      "loss": 0.8224,
      "step": 9145
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3569467067718506,
      "learning_rate": 1.4775580615276493e-05,
      "loss": 0.8175,
      "step": 9146
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6260995864868164,
      "learning_rate": 1.4774941698878703e-05,
      "loss": 0.8558,
      "step": 9147
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5112261772155762,
      "learning_rate": 1.4774302782480914e-05,
      "loss": 0.3523,
      "step": 9148
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.971585988998413,
      "learning_rate": 1.4773663866083124e-05,
      "loss": 1.0442,
      "step": 9149
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.878108263015747,
      "learning_rate": 1.4773024949685336e-05,
      "loss": 0.864,
      "step": 9150
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8148761987686157,
      "learning_rate": 1.4772386033287544e-05,
      "loss": 0.933,
      "step": 9151
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2869913578033447,
      "learning_rate": 1.4771747116889757e-05,
      "loss": 0.9202,
      "step": 9152
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.7964258193969727,
      "learning_rate": 1.4771108200491965e-05,
      "loss": 0.9667,
      "step": 9153
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.031139850616455,
      "learning_rate": 1.4770469284094178e-05,
      "loss": 0.9964,
      "step": 9154
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9270119667053223,
      "learning_rate": 1.4769830367696387e-05,
      "loss": 1.2099,
      "step": 9155
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.894694447517395,
      "learning_rate": 1.4769191451298598e-05,
      "loss": 1.0121,
      "step": 9156
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.932724118232727,
      "learning_rate": 1.4768552534900808e-05,
      "loss": 0.929,
      "step": 9157
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.594077706336975,
      "learning_rate": 1.476791361850302e-05,
      "loss": 0.8511,
      "step": 9158
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.059445381164551,
      "learning_rate": 1.476727470210523e-05,
      "loss": 1.0464,
      "step": 9159
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5706901550292969,
      "learning_rate": 1.4766635785707441e-05,
      "loss": 0.9366,
      "step": 9160
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7760953903198242,
      "learning_rate": 1.476599686930965e-05,
      "loss": 0.9327,
      "step": 9161
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7988178730010986,
      "learning_rate": 1.4765357952911862e-05,
      "loss": 0.9875,
      "step": 9162
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8575623035430908,
      "learning_rate": 1.4764719036514072e-05,
      "loss": 1.2519,
      "step": 9163
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8587324619293213,
      "learning_rate": 1.4764080120116284e-05,
      "loss": 0.8748,
      "step": 9164
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.66991925239563,
      "learning_rate": 1.4763441203718494e-05,
      "loss": 0.6605,
      "step": 9165
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0604453086853027,
      "learning_rate": 1.4762802287320705e-05,
      "loss": 0.9759,
      "step": 9166
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.034013032913208,
      "learning_rate": 1.4762163370922915e-05,
      "loss": 1.0374,
      "step": 9167
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9616118669509888,
      "learning_rate": 1.4761524454525127e-05,
      "loss": 0.831,
      "step": 9168
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.764406681060791,
      "learning_rate": 1.4760885538127336e-05,
      "loss": 1.0908,
      "step": 9169
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6032445430755615,
      "learning_rate": 1.4760246621729548e-05,
      "loss": 0.9335,
      "step": 9170
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9981186389923096,
      "learning_rate": 1.4759607705331758e-05,
      "loss": 1.0685,
      "step": 9171
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9929343461990356,
      "learning_rate": 1.475896878893397e-05,
      "loss": 0.9328,
      "step": 9172
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.260904550552368,
      "learning_rate": 1.475832987253618e-05,
      "loss": 0.8526,
      "step": 9173
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.559492826461792,
      "learning_rate": 1.475769095613839e-05,
      "loss": 0.9681,
      "step": 9174
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.857625722885132,
      "learning_rate": 1.47570520397406e-05,
      "loss": 1.0932,
      "step": 9175
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1150872707366943,
      "learning_rate": 1.4756413123342812e-05,
      "loss": 0.9502,
      "step": 9176
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.233748435974121,
      "learning_rate": 1.4755774206945022e-05,
      "loss": 0.8166,
      "step": 9177
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8183304071426392,
      "learning_rate": 1.4755135290547234e-05,
      "loss": 0.9152,
      "step": 9178
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9764281511306763,
      "learning_rate": 1.4754496374149443e-05,
      "loss": 0.8718,
      "step": 9179
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1175310611724854,
      "learning_rate": 1.4753857457751655e-05,
      "loss": 0.936,
      "step": 9180
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.0350303649902344,
      "learning_rate": 1.4753218541353865e-05,
      "loss": 1.23,
      "step": 9181
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.741235375404358,
      "learning_rate": 1.4752579624956076e-05,
      "loss": 0.9909,
      "step": 9182
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.993730902671814,
      "learning_rate": 1.4751940708558286e-05,
      "loss": 0.9543,
      "step": 9183
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.681586265563965,
      "learning_rate": 1.4751301792160498e-05,
      "loss": 1.3789,
      "step": 9184
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9776256084442139,
      "learning_rate": 1.4750662875762708e-05,
      "loss": 0.9975,
      "step": 9185
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.764664888381958,
      "learning_rate": 1.475002395936492e-05,
      "loss": 1.0889,
      "step": 9186
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8684461116790771,
      "learning_rate": 1.4749385042967127e-05,
      "loss": 0.9719,
      "step": 9187
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.005235195159912,
      "learning_rate": 1.474874612656934e-05,
      "loss": 0.9674,
      "step": 9188
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.061586380004883,
      "learning_rate": 1.4748107210171549e-05,
      "loss": 1.154,
      "step": 9189
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7670578956604004,
      "learning_rate": 1.4747468293773762e-05,
      "loss": 0.8848,
      "step": 9190
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5745302438735962,
      "learning_rate": 1.474682937737597e-05,
      "loss": 0.8213,
      "step": 9191
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5738753080368042,
      "learning_rate": 1.4746190460978183e-05,
      "loss": 0.9744,
      "step": 9192
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5114867687225342,
      "learning_rate": 1.4745551544580392e-05,
      "loss": 0.8987,
      "step": 9193
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.767512559890747,
      "learning_rate": 1.4744912628182603e-05,
      "loss": 0.7667,
      "step": 9194
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6688369512557983,
      "learning_rate": 1.4744273711784813e-05,
      "loss": 0.9217,
      "step": 9195
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6905367374420166,
      "learning_rate": 1.4743634795387024e-05,
      "loss": 0.8768,
      "step": 9196
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0474720001220703,
      "learning_rate": 1.4742995878989234e-05,
      "loss": 0.8258,
      "step": 9197
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4401490688323975,
      "learning_rate": 1.4742356962591446e-05,
      "loss": 0.9615,
      "step": 9198
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9521222114562988,
      "learning_rate": 1.4741718046193656e-05,
      "loss": 0.907,
      "step": 9199
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9319642782211304,
      "learning_rate": 1.4741079129795867e-05,
      "loss": 1.1366,
      "step": 9200
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1841588020324707,
      "learning_rate": 1.4740440213398077e-05,
      "loss": 0.995,
      "step": 9201
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.057608127593994,
      "learning_rate": 1.4739801297000289e-05,
      "loss": 0.932,
      "step": 9202
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.4962239265441895,
      "learning_rate": 1.4739162380602499e-05,
      "loss": 0.9418,
      "step": 9203
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6915998458862305,
      "learning_rate": 1.473852346420471e-05,
      "loss": 0.9507,
      "step": 9204
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7743916511535645,
      "learning_rate": 1.473788454780692e-05,
      "loss": 0.8942,
      "step": 9205
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9480164051055908,
      "learning_rate": 1.4737245631409131e-05,
      "loss": 0.9292,
      "step": 9206
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9663752317428589,
      "learning_rate": 1.4736606715011341e-05,
      "loss": 0.9243,
      "step": 9207
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.212213039398193,
      "learning_rate": 1.4735967798613553e-05,
      "loss": 1.4493,
      "step": 9208
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7215912342071533,
      "learning_rate": 1.4735328882215763e-05,
      "loss": 0.8885,
      "step": 9209
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1149377822875977,
      "learning_rate": 1.4734689965817974e-05,
      "loss": 1.0786,
      "step": 9210
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1875357627868652,
      "learning_rate": 1.4734051049420184e-05,
      "loss": 0.8946,
      "step": 9211
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.431417942047119,
      "learning_rate": 1.4733412133022396e-05,
      "loss": 1.0065,
      "step": 9212
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9081182479858398,
      "learning_rate": 1.4732773216624606e-05,
      "loss": 0.8838,
      "step": 9213
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.756270408630371,
      "learning_rate": 1.4732134300226817e-05,
      "loss": 1.0086,
      "step": 9214
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.119872570037842,
      "learning_rate": 1.4731495383829027e-05,
      "loss": 0.8458,
      "step": 9215
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.4168503284454346,
      "learning_rate": 1.4730856467431238e-05,
      "loss": 0.8917,
      "step": 9216
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5539675951004028,
      "learning_rate": 1.4730217551033448e-05,
      "loss": 0.9711,
      "step": 9217
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5814294815063477,
      "learning_rate": 1.472957863463566e-05,
      "loss": 0.6243,
      "step": 9218
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9163097143173218,
      "learning_rate": 1.472893971823787e-05,
      "loss": 0.9538,
      "step": 9219
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2820489406585693,
      "learning_rate": 1.4728300801840081e-05,
      "loss": 1.0697,
      "step": 9220
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2056188583374023,
      "learning_rate": 1.4727661885442291e-05,
      "loss": 0.8197,
      "step": 9221
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.760627269744873,
      "learning_rate": 1.4727022969044503e-05,
      "loss": 0.8795,
      "step": 9222
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9354242086410522,
      "learning_rate": 1.472638405264671e-05,
      "loss": 0.9616,
      "step": 9223
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.348602056503296,
      "learning_rate": 1.4725745136248924e-05,
      "loss": 0.89,
      "step": 9224
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.955352783203125,
      "learning_rate": 1.4725106219851132e-05,
      "loss": 0.8926,
      "step": 9225
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1637837886810303,
      "learning_rate": 1.4724467303453345e-05,
      "loss": 0.9536,
      "step": 9226
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9417246580123901,
      "learning_rate": 1.4723828387055554e-05,
      "loss": 1.2773,
      "step": 9227
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.915114164352417,
      "learning_rate": 1.4723189470657767e-05,
      "loss": 1.0405,
      "step": 9228
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2184371948242188,
      "learning_rate": 1.4722550554259975e-05,
      "loss": 1.0063,
      "step": 9229
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2905447483062744,
      "learning_rate": 1.4721911637862187e-05,
      "loss": 1.0807,
      "step": 9230
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.160799264907837,
      "learning_rate": 1.4721272721464396e-05,
      "loss": 0.5936,
      "step": 9231
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.977428674697876,
      "learning_rate": 1.4720633805066608e-05,
      "loss": 0.857,
      "step": 9232
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.493258953094482,
      "learning_rate": 1.4719994888668818e-05,
      "loss": 1.1969,
      "step": 9233
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.634106397628784,
      "learning_rate": 1.471935597227103e-05,
      "loss": 1.1974,
      "step": 9234
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5120654106140137,
      "learning_rate": 1.471871705587324e-05,
      "loss": 1.0696,
      "step": 9235
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9372142553329468,
      "learning_rate": 1.471807813947545e-05,
      "loss": 0.9395,
      "step": 9236
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9160311222076416,
      "learning_rate": 1.471743922307766e-05,
      "loss": 0.9216,
      "step": 9237
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0112314224243164,
      "learning_rate": 1.4716800306679872e-05,
      "loss": 0.9749,
      "step": 9238
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8948795795440674,
      "learning_rate": 1.4716161390282082e-05,
      "loss": 1.0342,
      "step": 9239
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4646458625793457,
      "learning_rate": 1.4715522473884294e-05,
      "loss": 0.9574,
      "step": 9240
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.982753753662109,
      "learning_rate": 1.4714883557486503e-05,
      "loss": 1.8313,
      "step": 9241
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.428230047225952,
      "learning_rate": 1.4714244641088715e-05,
      "loss": 0.7243,
      "step": 9242
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0692224502563477,
      "learning_rate": 1.4713605724690925e-05,
      "loss": 0.9319,
      "step": 9243
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0649759769439697,
      "learning_rate": 1.4712966808293136e-05,
      "loss": 0.9006,
      "step": 9244
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7292100191116333,
      "learning_rate": 1.4712327891895346e-05,
      "loss": 0.939,
      "step": 9245
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7721272706985474,
      "learning_rate": 1.4711688975497558e-05,
      "loss": 0.8165,
      "step": 9246
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7556488513946533,
      "learning_rate": 1.4711050059099768e-05,
      "loss": 0.9686,
      "step": 9247
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7853503227233887,
      "learning_rate": 1.471041114270198e-05,
      "loss": 1.0331,
      "step": 9248
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8263673782348633,
      "learning_rate": 1.4709772226304189e-05,
      "loss": 0.7975,
      "step": 9249
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5190730094909668,
      "learning_rate": 1.47091333099064e-05,
      "loss": 0.819,
      "step": 9250
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6125718355178833,
      "learning_rate": 1.470849439350861e-05,
      "loss": 0.8756,
      "step": 9251
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.159393548965454,
      "learning_rate": 1.4707855477110822e-05,
      "loss": 1.2187,
      "step": 9252
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.887401580810547,
      "learning_rate": 1.4707216560713032e-05,
      "loss": 1.4347,
      "step": 9253
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8884830474853516,
      "learning_rate": 1.4706577644315243e-05,
      "loss": 0.872,
      "step": 9254
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7466371059417725,
      "learning_rate": 1.4705938727917453e-05,
      "loss": 1.2185,
      "step": 9255
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7713069915771484,
      "learning_rate": 1.4705299811519665e-05,
      "loss": 0.9876,
      "step": 9256
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.831048846244812,
      "learning_rate": 1.4704660895121875e-05,
      "loss": 0.9545,
      "step": 9257
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.050793170928955,
      "learning_rate": 1.4704021978724086e-05,
      "loss": 0.9807,
      "step": 9258
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9955861568450928,
      "learning_rate": 1.4703383062326296e-05,
      "loss": 0.8499,
      "step": 9259
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.84507155418396,
      "learning_rate": 1.4702744145928508e-05,
      "loss": 1.1257,
      "step": 9260
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.09537672996521,
      "learning_rate": 1.4702105229530716e-05,
      "loss": 1.0327,
      "step": 9261
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9167314767837524,
      "learning_rate": 1.4701466313132929e-05,
      "loss": 0.9348,
      "step": 9262
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7368675470352173,
      "learning_rate": 1.4700827396735137e-05,
      "loss": 0.9606,
      "step": 9263
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9481658935546875,
      "learning_rate": 1.470018848033735e-05,
      "loss": 0.9387,
      "step": 9264
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.074704647064209,
      "learning_rate": 1.4699549563939559e-05,
      "loss": 1.063,
      "step": 9265
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.175723075866699,
      "learning_rate": 1.469891064754177e-05,
      "loss": 0.9419,
      "step": 9266
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2930819988250732,
      "learning_rate": 1.469827173114398e-05,
      "loss": 0.8204,
      "step": 9267
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6937968730926514,
      "learning_rate": 1.4697632814746191e-05,
      "loss": 0.8488,
      "step": 9268
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.6999545097351074,
      "learning_rate": 1.4696993898348401e-05,
      "loss": 0.704,
      "step": 9269
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0256922245025635,
      "learning_rate": 1.4696354981950613e-05,
      "loss": 1.0215,
      "step": 9270
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9535256624221802,
      "learning_rate": 1.4695716065552823e-05,
      "loss": 0.8935,
      "step": 9271
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.032672166824341,
      "learning_rate": 1.4695077149155034e-05,
      "loss": 0.9485,
      "step": 9272
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5696290731430054,
      "learning_rate": 1.4694438232757244e-05,
      "loss": 0.8589,
      "step": 9273
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9537221193313599,
      "learning_rate": 1.4693799316359456e-05,
      "loss": 0.8606,
      "step": 9274
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.471310138702393,
      "learning_rate": 1.4693160399961666e-05,
      "loss": 1.2052,
      "step": 9275
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.533107042312622,
      "learning_rate": 1.4692521483563877e-05,
      "loss": 0.858,
      "step": 9276
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0010826587677,
      "learning_rate": 1.4691882567166087e-05,
      "loss": 1.0472,
      "step": 9277
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5002505779266357,
      "learning_rate": 1.4691243650768298e-05,
      "loss": 1.0211,
      "step": 9278
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5671006441116333,
      "learning_rate": 1.4690604734370508e-05,
      "loss": 0.9998,
      "step": 9279
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3958537578582764,
      "learning_rate": 1.468996581797272e-05,
      "loss": 1.005,
      "step": 9280
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8311854600906372,
      "learning_rate": 1.468932690157493e-05,
      "loss": 0.9222,
      "step": 9281
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7004021406173706,
      "learning_rate": 1.4688687985177141e-05,
      "loss": 0.8697,
      "step": 9282
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7943183183670044,
      "learning_rate": 1.4688049068779351e-05,
      "loss": 0.9489,
      "step": 9283
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.7949023246765137,
      "learning_rate": 1.4687410152381563e-05,
      "loss": 0.7388,
      "step": 9284
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1465630531311035,
      "learning_rate": 1.4686771235983773e-05,
      "loss": 0.9164,
      "step": 9285
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9609947204589844,
      "learning_rate": 1.4686132319585984e-05,
      "loss": 0.8932,
      "step": 9286
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.288604497909546,
      "learning_rate": 1.4685493403188194e-05,
      "loss": 0.9395,
      "step": 9287
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7161821126937866,
      "learning_rate": 1.4684854486790405e-05,
      "loss": 0.9142,
      "step": 9288
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.928987979888916,
      "learning_rate": 1.4684215570392615e-05,
      "loss": 1.3698,
      "step": 9289
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.306593418121338,
      "learning_rate": 1.4683576653994827e-05,
      "loss": 1.0665,
      "step": 9290
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.8474955558776855,
      "learning_rate": 1.4682937737597037e-05,
      "loss": 1.3659,
      "step": 9291
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9063735008239746,
      "learning_rate": 1.4682298821199248e-05,
      "loss": 1.0587,
      "step": 9292
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.192201852798462,
      "learning_rate": 1.4681659904801458e-05,
      "loss": 0.9155,
      "step": 9293
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4952603578567505,
      "learning_rate": 1.468102098840367e-05,
      "loss": 0.4595,
      "step": 9294
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2075698375701904,
      "learning_rate": 1.468038207200588e-05,
      "loss": 0.6418,
      "step": 9295
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8003754615783691,
      "learning_rate": 1.4679743155608091e-05,
      "loss": 0.9608,
      "step": 9296
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.6286094188690186,
      "learning_rate": 1.46791042392103e-05,
      "loss": 0.9089,
      "step": 9297
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7268389463424683,
      "learning_rate": 1.4678465322812512e-05,
      "loss": 0.8791,
      "step": 9298
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8284553289413452,
      "learning_rate": 1.467782640641472e-05,
      "loss": 1.0219,
      "step": 9299
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3822455406188965,
      "learning_rate": 1.4677187490016934e-05,
      "loss": 1.214,
      "step": 9300
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7041871547698975,
      "learning_rate": 1.4676548573619142e-05,
      "loss": 0.3494,
      "step": 9301
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7027111053466797,
      "learning_rate": 1.4675909657221354e-05,
      "loss": 0.5944,
      "step": 9302
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5178883075714111,
      "learning_rate": 1.4675270740823563e-05,
      "loss": 0.4018,
      "step": 9303
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.529539108276367,
      "learning_rate": 1.4674631824425775e-05,
      "loss": 0.7563,
      "step": 9304
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7392048835754395,
      "learning_rate": 1.4673992908027985e-05,
      "loss": 1.0462,
      "step": 9305
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.466613292694092,
      "learning_rate": 1.4673353991630196e-05,
      "loss": 1.1468,
      "step": 9306
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6522964239120483,
      "learning_rate": 1.4672715075232406e-05,
      "loss": 0.9428,
      "step": 9307
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1320347785949707,
      "learning_rate": 1.4672076158834618e-05,
      "loss": 1.0233,
      "step": 9308
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.865311861038208,
      "learning_rate": 1.4671437242436828e-05,
      "loss": 0.8731,
      "step": 9309
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1355175971984863,
      "learning_rate": 1.467079832603904e-05,
      "loss": 0.878,
      "step": 9310
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.669761061668396,
      "learning_rate": 1.4670159409641249e-05,
      "loss": 0.8906,
      "step": 9311
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2922236919403076,
      "learning_rate": 1.466952049324346e-05,
      "loss": 1.0759,
      "step": 9312
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.11773419380188,
      "learning_rate": 1.466888157684567e-05,
      "loss": 0.8778,
      "step": 9313
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9228342771530151,
      "learning_rate": 1.4668242660447882e-05,
      "loss": 0.4357,
      "step": 9314
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8278310298919678,
      "learning_rate": 1.4667603744050092e-05,
      "loss": 0.82,
      "step": 9315
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.48475980758667,
      "learning_rate": 1.4666964827652303e-05,
      "loss": 1.5919,
      "step": 9316
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0429553985595703,
      "learning_rate": 1.4666325911254513e-05,
      "loss": 0.9151,
      "step": 9317
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8715837001800537,
      "learning_rate": 1.4665686994856725e-05,
      "loss": 1.0662,
      "step": 9318
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.756358623504639,
      "learning_rate": 1.4665048078458935e-05,
      "loss": 1.5751,
      "step": 9319
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.1898193359375,
      "learning_rate": 1.4664409162061146e-05,
      "loss": 1.1984,
      "step": 9320
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1986145973205566,
      "learning_rate": 1.4663770245663356e-05,
      "loss": 0.922,
      "step": 9321
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5475322008132935,
      "learning_rate": 1.4663131329265568e-05,
      "loss": 0.9011,
      "step": 9322
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.561713695526123,
      "learning_rate": 1.4662492412867777e-05,
      "loss": 1.0576,
      "step": 9323
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8194886445999146,
      "learning_rate": 1.4661853496469989e-05,
      "loss": 0.881,
      "step": 9324
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9735411405563354,
      "learning_rate": 1.4661214580072199e-05,
      "loss": 0.8626,
      "step": 9325
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0564768314361572,
      "learning_rate": 1.466057566367441e-05,
      "loss": 1.0716,
      "step": 9326
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8048325777053833,
      "learning_rate": 1.465993674727662e-05,
      "loss": 0.9628,
      "step": 9327
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8028018474578857,
      "learning_rate": 1.4659297830878832e-05,
      "loss": 0.8484,
      "step": 9328
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6912161111831665,
      "learning_rate": 1.4658658914481042e-05,
      "loss": 0.9027,
      "step": 9329
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.901737928390503,
      "learning_rate": 1.4658019998083253e-05,
      "loss": 1.2123,
      "step": 9330
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5193790197372437,
      "learning_rate": 1.4657381081685463e-05,
      "loss": 0.8524,
      "step": 9331
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.026797294616699,
      "learning_rate": 1.4656742165287675e-05,
      "loss": 0.821,
      "step": 9332
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.128828525543213,
      "learning_rate": 1.4656103248889883e-05,
      "loss": 1.0088,
      "step": 9333
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2315866947174072,
      "learning_rate": 1.4655464332492096e-05,
      "loss": 0.8484,
      "step": 9334
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.688298225402832,
      "learning_rate": 1.4654825416094304e-05,
      "loss": 0.729,
      "step": 9335
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3413021564483643,
      "learning_rate": 1.4654186499696517e-05,
      "loss": 1.1395,
      "step": 9336
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5445284843444824,
      "learning_rate": 1.4653547583298726e-05,
      "loss": 0.8906,
      "step": 9337
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9045376777648926,
      "learning_rate": 1.4652908666900939e-05,
      "loss": 0.9363,
      "step": 9338
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6349718570709229,
      "learning_rate": 1.4652269750503147e-05,
      "loss": 0.9247,
      "step": 9339
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9129588603973389,
      "learning_rate": 1.4651630834105358e-05,
      "loss": 0.9685,
      "step": 9340
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.729701280593872,
      "learning_rate": 1.4650991917707568e-05,
      "loss": 0.9233,
      "step": 9341
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5767072439193726,
      "learning_rate": 1.465035300130978e-05,
      "loss": 0.916,
      "step": 9342
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2815732955932617,
      "learning_rate": 1.464971408491199e-05,
      "loss": 1.0536,
      "step": 9343
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7273367643356323,
      "learning_rate": 1.4649075168514201e-05,
      "loss": 0.8598,
      "step": 9344
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9517358541488647,
      "learning_rate": 1.4648436252116411e-05,
      "loss": 1.0474,
      "step": 9345
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.485073447227478,
      "learning_rate": 1.4647797335718623e-05,
      "loss": 0.8622,
      "step": 9346
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8376646041870117,
      "learning_rate": 1.4647158419320833e-05,
      "loss": 0.9671,
      "step": 9347
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6365535259246826,
      "learning_rate": 1.4646519502923044e-05,
      "loss": 0.9243,
      "step": 9348
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4269680976867676,
      "learning_rate": 1.4645880586525254e-05,
      "loss": 0.841,
      "step": 9349
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.777911424636841,
      "learning_rate": 1.4645241670127465e-05,
      "loss": 1.0976,
      "step": 9350
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.3654823303222656,
      "learning_rate": 1.4644602753729675e-05,
      "loss": 1.0052,
      "step": 9351
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.342906951904297,
      "learning_rate": 1.4643963837331887e-05,
      "loss": 0.8706,
      "step": 9352
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.720331072807312,
      "learning_rate": 1.4643324920934097e-05,
      "loss": 0.8939,
      "step": 9353
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8439282178878784,
      "learning_rate": 1.4642686004536308e-05,
      "loss": 0.858,
      "step": 9354
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7674230337142944,
      "learning_rate": 1.4642047088138518e-05,
      "loss": 1.236,
      "step": 9355
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.495054244995117,
      "learning_rate": 1.464140817174073e-05,
      "loss": 1.0961,
      "step": 9356
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.0687153339385986,
      "learning_rate": 1.464076925534294e-05,
      "loss": 0.923,
      "step": 9357
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.952184796333313,
      "learning_rate": 1.4640130338945151e-05,
      "loss": 0.8703,
      "step": 9358
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4237186908721924,
      "learning_rate": 1.4639491422547361e-05,
      "loss": 1.0232,
      "step": 9359
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.111898422241211,
      "learning_rate": 1.4638852506149572e-05,
      "loss": 0.9197,
      "step": 9360
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7029049396514893,
      "learning_rate": 1.4638213589751782e-05,
      "loss": 0.7587,
      "step": 9361
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.9802117347717285,
      "learning_rate": 1.4637574673353994e-05,
      "loss": 0.7983,
      "step": 9362
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.905104637145996,
      "learning_rate": 1.4636935756956204e-05,
      "loss": 1.077,
      "step": 9363
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.105781316757202,
      "learning_rate": 1.4636296840558415e-05,
      "loss": 0.9877,
      "step": 9364
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.3372979164123535,
      "learning_rate": 1.4635657924160625e-05,
      "loss": 1.0465,
      "step": 9365
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.387030601501465,
      "learning_rate": 1.4635019007762837e-05,
      "loss": 0.936,
      "step": 9366
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.589940309524536,
      "learning_rate": 1.4634380091365047e-05,
      "loss": 0.9402,
      "step": 9367
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4834251403808594,
      "learning_rate": 1.4633741174967258e-05,
      "loss": 0.883,
      "step": 9368
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8130824565887451,
      "learning_rate": 1.4633102258569466e-05,
      "loss": 0.8468,
      "step": 9369
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.322371244430542,
      "learning_rate": 1.463246334217168e-05,
      "loss": 1.2905,
      "step": 9370
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6738587617874146,
      "learning_rate": 1.4631824425773888e-05,
      "loss": 0.9289,
      "step": 9371
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2284626960754395,
      "learning_rate": 1.4631185509376101e-05,
      "loss": 0.9837,
      "step": 9372
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8594313859939575,
      "learning_rate": 1.4630546592978309e-05,
      "loss": 0.9763,
      "step": 9373
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9372849464416504,
      "learning_rate": 1.4629907676580522e-05,
      "loss": 0.9507,
      "step": 9374
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2111661434173584,
      "learning_rate": 1.462926876018273e-05,
      "loss": 1.1058,
      "step": 9375
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1244306564331055,
      "learning_rate": 1.4628629843784942e-05,
      "loss": 1.0948,
      "step": 9376
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.6543192863464355,
      "learning_rate": 1.4627990927387152e-05,
      "loss": 1.6184,
      "step": 9377
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.775334119796753,
      "learning_rate": 1.4627352010989363e-05,
      "loss": 0.8427,
      "step": 9378
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.192701816558838,
      "learning_rate": 1.4626713094591573e-05,
      "loss": 1.0218,
      "step": 9379
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2303740978240967,
      "learning_rate": 1.4626074178193785e-05,
      "loss": 1.261,
      "step": 9380
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9834054708480835,
      "learning_rate": 1.4625435261795995e-05,
      "loss": 0.8573,
      "step": 9381
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.331815719604492,
      "learning_rate": 1.4624796345398206e-05,
      "loss": 1.0558,
      "step": 9382
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9415638446807861,
      "learning_rate": 1.4624157429000416e-05,
      "loss": 0.8229,
      "step": 9383
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.737497091293335,
      "learning_rate": 1.4623518512602628e-05,
      "loss": 0.9389,
      "step": 9384
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.258723020553589,
      "learning_rate": 1.4622879596204837e-05,
      "loss": 0.9281,
      "step": 9385
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9088647365570068,
      "learning_rate": 1.4622240679807049e-05,
      "loss": 1.0295,
      "step": 9386
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5186015367507935,
      "learning_rate": 1.4621601763409259e-05,
      "loss": 0.9111,
      "step": 9387
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.518421173095703,
      "learning_rate": 1.462096284701147e-05,
      "loss": 1.0205,
      "step": 9388
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.513169527053833,
      "learning_rate": 1.462032393061368e-05,
      "loss": 0.8306,
      "step": 9389
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6462825536727905,
      "learning_rate": 1.4619685014215892e-05,
      "loss": 0.9278,
      "step": 9390
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5021426677703857,
      "learning_rate": 1.4619046097818102e-05,
      "loss": 1.0164,
      "step": 9391
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4990365505218506,
      "learning_rate": 1.4618407181420313e-05,
      "loss": 0.6201,
      "step": 9392
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6432523727416992,
      "learning_rate": 1.4617768265022523e-05,
      "loss": 0.9,
      "step": 9393
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.7420830726623535,
      "learning_rate": 1.4617129348624735e-05,
      "loss": 1.1251,
      "step": 9394
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6913989782333374,
      "learning_rate": 1.4616490432226944e-05,
      "loss": 0.972,
      "step": 9395
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.687713861465454,
      "learning_rate": 1.4615851515829156e-05,
      "loss": 0.945,
      "step": 9396
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7950128316879272,
      "learning_rate": 1.4615212599431366e-05,
      "loss": 0.8999,
      "step": 9397
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.245388984680176,
      "learning_rate": 1.4614573683033577e-05,
      "loss": 1.0812,
      "step": 9398
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6793628931045532,
      "learning_rate": 1.4613934766635787e-05,
      "loss": 0.8805,
      "step": 9399
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.629556655883789,
      "learning_rate": 1.4613295850237999e-05,
      "loss": 0.479,
      "step": 9400
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7809481620788574,
      "learning_rate": 1.4612656933840209e-05,
      "loss": 0.819,
      "step": 9401
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8032550811767578,
      "learning_rate": 1.461201801744242e-05,
      "loss": 0.5193,
      "step": 9402
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.556110143661499,
      "learning_rate": 1.461137910104463e-05,
      "loss": 0.8781,
      "step": 9403
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.812697410583496,
      "learning_rate": 1.4610740184646842e-05,
      "loss": 0.8215,
      "step": 9404
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.335423707962036,
      "learning_rate": 1.4610101268249051e-05,
      "loss": 1.154,
      "step": 9405
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.134973526000977,
      "learning_rate": 1.4609462351851263e-05,
      "loss": 1.5037,
      "step": 9406
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6740516424179077,
      "learning_rate": 1.4608823435453471e-05,
      "loss": 0.8963,
      "step": 9407
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2374165058135986,
      "learning_rate": 1.4608184519055684e-05,
      "loss": 1.0494,
      "step": 9408
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1321096420288086,
      "learning_rate": 1.4607545602657893e-05,
      "loss": 0.9099,
      "step": 9409
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9058141708374023,
      "learning_rate": 1.4606906686260106e-05,
      "loss": 0.8627,
      "step": 9410
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9700140953063965,
      "learning_rate": 1.4606267769862314e-05,
      "loss": 0.9087,
      "step": 9411
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.186570405960083,
      "learning_rate": 1.4605628853464525e-05,
      "loss": 0.6009,
      "step": 9412
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1405439376831055,
      "learning_rate": 1.4604989937066735e-05,
      "loss": 1.0764,
      "step": 9413
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.28561544418335,
      "learning_rate": 1.4604351020668947e-05,
      "loss": 1.2069,
      "step": 9414
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2972373962402344,
      "learning_rate": 1.4603712104271157e-05,
      "loss": 0.7795,
      "step": 9415
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8057838678359985,
      "learning_rate": 1.4603073187873368e-05,
      "loss": 0.9217,
      "step": 9416
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5598946809768677,
      "learning_rate": 1.4602434271475578e-05,
      "loss": 0.9223,
      "step": 9417
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.563494324684143,
      "learning_rate": 1.460179535507779e-05,
      "loss": 0.8741,
      "step": 9418
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4861541986465454,
      "learning_rate": 1.460115643868e-05,
      "loss": 0.8316,
      "step": 9419
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6728029251098633,
      "learning_rate": 1.4600517522282211e-05,
      "loss": 0.886,
      "step": 9420
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9270974397659302,
      "learning_rate": 1.4599878605884421e-05,
      "loss": 0.8849,
      "step": 9421
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9640309810638428,
      "learning_rate": 1.4599239689486632e-05,
      "loss": 0.942,
      "step": 9422
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6616498231887817,
      "learning_rate": 1.4598600773088842e-05,
      "loss": 0.9368,
      "step": 9423
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6996127367019653,
      "learning_rate": 1.4597961856691054e-05,
      "loss": 0.9228,
      "step": 9424
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8459399938583374,
      "learning_rate": 1.4597322940293264e-05,
      "loss": 1.0195,
      "step": 9425
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.9756662845611572,
      "learning_rate": 1.4596684023895475e-05,
      "loss": 1.1355,
      "step": 9426
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.7420804500579834,
      "learning_rate": 1.4596045107497685e-05,
      "loss": 1.0602,
      "step": 9427
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7669228315353394,
      "learning_rate": 1.4595406191099897e-05,
      "loss": 0.9715,
      "step": 9428
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.813452124595642,
      "learning_rate": 1.4594767274702107e-05,
      "loss": 0.8581,
      "step": 9429
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8862966299057007,
      "learning_rate": 1.4594128358304318e-05,
      "loss": 1.0064,
      "step": 9430
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9370274543762207,
      "learning_rate": 1.4593489441906528e-05,
      "loss": 1.0009,
      "step": 9431
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6971265077590942,
      "learning_rate": 1.459285052550874e-05,
      "loss": 0.8682,
      "step": 9432
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.193763494491577,
      "learning_rate": 1.459221160911095e-05,
      "loss": 1.0445,
      "step": 9433
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.503582239151001,
      "learning_rate": 1.4591572692713161e-05,
      "loss": 0.866,
      "step": 9434
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.1318860054016113,
      "learning_rate": 1.459093377631537e-05,
      "loss": 0.9095,
      "step": 9435
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6374977827072144,
      "learning_rate": 1.4590294859917582e-05,
      "loss": 0.8383,
      "step": 9436
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.081817626953125,
      "learning_rate": 1.4589655943519792e-05,
      "loss": 0.8373,
      "step": 9437
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.593733549118042,
      "learning_rate": 1.4589017027122004e-05,
      "loss": 1.0793,
      "step": 9438
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6392178535461426,
      "learning_rate": 1.4588378110724214e-05,
      "loss": 0.8761,
      "step": 9439
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9624098539352417,
      "learning_rate": 1.4587739194326425e-05,
      "loss": 0.881,
      "step": 9440
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.50385856628418,
      "learning_rate": 1.4587100277928635e-05,
      "loss": 1.3605,
      "step": 9441
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6984721422195435,
      "learning_rate": 1.4586461361530846e-05,
      "loss": 0.84,
      "step": 9442
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9135141372680664,
      "learning_rate": 1.4585822445133055e-05,
      "loss": 0.8778,
      "step": 9443
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3679202795028687,
      "learning_rate": 1.4585183528735268e-05,
      "loss": 0.8841,
      "step": 9444
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.19877552986145,
      "learning_rate": 1.4584544612337476e-05,
      "loss": 1.0606,
      "step": 9445
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7776817083358765,
      "learning_rate": 1.458390569593969e-05,
      "loss": 0.88,
      "step": 9446
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7743579149246216,
      "learning_rate": 1.4583266779541897e-05,
      "loss": 0.8779,
      "step": 9447
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.69244122505188,
      "learning_rate": 1.4582627863144109e-05,
      "loss": 0.9161,
      "step": 9448
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.3466999530792236,
      "learning_rate": 1.4581988946746319e-05,
      "loss": 1.034,
      "step": 9449
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7363219261169434,
      "learning_rate": 1.458135003034853e-05,
      "loss": 1.1327,
      "step": 9450
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.035343647003174,
      "learning_rate": 1.458071111395074e-05,
      "loss": 0.9936,
      "step": 9451
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0825343132019043,
      "learning_rate": 1.4580072197552952e-05,
      "loss": 1.1216,
      "step": 9452
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6977574825286865,
      "learning_rate": 1.4579433281155162e-05,
      "loss": 0.7222,
      "step": 9453
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6859676837921143,
      "learning_rate": 1.4578794364757373e-05,
      "loss": 0.9024,
      "step": 9454
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0501582622528076,
      "learning_rate": 1.4578155448359583e-05,
      "loss": 0.9752,
      "step": 9455
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9255603551864624,
      "learning_rate": 1.4577516531961795e-05,
      "loss": 0.8132,
      "step": 9456
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9124794006347656,
      "learning_rate": 1.4576877615564004e-05,
      "loss": 1.0488,
      "step": 9457
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.590882420539856,
      "learning_rate": 1.4576238699166216e-05,
      "loss": 0.7772,
      "step": 9458
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5294491052627563,
      "learning_rate": 1.4575599782768426e-05,
      "loss": 0.886,
      "step": 9459
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.914541482925415,
      "learning_rate": 1.4574960866370637e-05,
      "loss": 0.9578,
      "step": 9460
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2225124835968018,
      "learning_rate": 1.4574321949972847e-05,
      "loss": 0.9101,
      "step": 9461
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9529078006744385,
      "learning_rate": 1.4573683033575059e-05,
      "loss": 0.9729,
      "step": 9462
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.7523348331451416,
      "learning_rate": 1.4573044117177269e-05,
      "loss": 1.5306,
      "step": 9463
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.039616107940674,
      "learning_rate": 1.457240520077948e-05,
      "loss": 0.8589,
      "step": 9464
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.0510077476501465,
      "learning_rate": 1.457176628438169e-05,
      "loss": 1.5609,
      "step": 9465
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.012642860412598,
      "learning_rate": 1.4571127367983902e-05,
      "loss": 1.3067,
      "step": 9466
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8985798358917236,
      "learning_rate": 1.4570488451586111e-05,
      "loss": 0.9401,
      "step": 9467
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0459911823272705,
      "learning_rate": 1.4569849535188323e-05,
      "loss": 1.0102,
      "step": 9468
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6788475513458252,
      "learning_rate": 1.4569210618790533e-05,
      "loss": 0.4405,
      "step": 9469
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.6425235271453857,
      "learning_rate": 1.4568571702392744e-05,
      "loss": 0.8222,
      "step": 9470
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1219606399536133,
      "learning_rate": 1.4567932785994954e-05,
      "loss": 0.9833,
      "step": 9471
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4492650032043457,
      "learning_rate": 1.4567293869597164e-05,
      "loss": 1.1754,
      "step": 9472
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6713721752166748,
      "learning_rate": 1.4566654953199376e-05,
      "loss": 0.8911,
      "step": 9473
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.238527536392212,
      "learning_rate": 1.4566016036801584e-05,
      "loss": 1.0098,
      "step": 9474
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1794273853302,
      "learning_rate": 1.4565377120403797e-05,
      "loss": 1.316,
      "step": 9475
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5855071544647217,
      "learning_rate": 1.4564738204006005e-05,
      "loss": 0.8112,
      "step": 9476
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7356828451156616,
      "learning_rate": 1.4564099287608218e-05,
      "loss": 0.9952,
      "step": 9477
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8808801174163818,
      "learning_rate": 1.4563460371210427e-05,
      "loss": 0.9536,
      "step": 9478
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6067229509353638,
      "learning_rate": 1.4562821454812638e-05,
      "loss": 0.8472,
      "step": 9479
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1610405445098877,
      "learning_rate": 1.4562182538414848e-05,
      "loss": 1.0891,
      "step": 9480
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.419307231903076,
      "learning_rate": 1.456154362201706e-05,
      "loss": 1.1223,
      "step": 9481
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8362724781036377,
      "learning_rate": 1.456090470561927e-05,
      "loss": 1.0357,
      "step": 9482
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6466269493103027,
      "learning_rate": 1.4560265789221481e-05,
      "loss": 0.8923,
      "step": 9483
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7845884561538696,
      "learning_rate": 1.455962687282369e-05,
      "loss": 0.9059,
      "step": 9484
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8374385833740234,
      "learning_rate": 1.4558987956425902e-05,
      "loss": 0.7542,
      "step": 9485
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5889204740524292,
      "learning_rate": 1.4558349040028112e-05,
      "loss": 0.4416,
      "step": 9486
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6054316759109497,
      "learning_rate": 1.4557710123630324e-05,
      "loss": 0.4718,
      "step": 9487
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.655678391456604,
      "learning_rate": 1.4557071207232534e-05,
      "loss": 0.9641,
      "step": 9488
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8288204669952393,
      "learning_rate": 1.4556432290834745e-05,
      "loss": 0.9191,
      "step": 9489
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.421121835708618,
      "learning_rate": 1.4555793374436955e-05,
      "loss": 1.4315,
      "step": 9490
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.102820873260498,
      "learning_rate": 1.4555154458039167e-05,
      "loss": 1.4224,
      "step": 9491
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9651318788528442,
      "learning_rate": 1.4554515541641376e-05,
      "loss": 0.8021,
      "step": 9492
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.868262767791748,
      "learning_rate": 1.4553876625243588e-05,
      "loss": 1.0303,
      "step": 9493
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9873865842819214,
      "learning_rate": 1.4553237708845798e-05,
      "loss": 0.9777,
      "step": 9494
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.5464601516723633,
      "learning_rate": 1.455259879244801e-05,
      "loss": 0.9482,
      "step": 9495
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8667893409729004,
      "learning_rate": 1.455195987605022e-05,
      "loss": 0.9677,
      "step": 9496
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.018280029296875,
      "learning_rate": 1.455132095965243e-05,
      "loss": 0.8626,
      "step": 9497
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9321726560592651,
      "learning_rate": 1.455068204325464e-05,
      "loss": 0.8562,
      "step": 9498
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9348684549331665,
      "learning_rate": 1.4550043126856852e-05,
      "loss": 0.9128,
      "step": 9499
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.368545413017273,
      "learning_rate": 1.4549404210459062e-05,
      "loss": 0.3696,
      "step": 9500
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.702269554138184,
      "learning_rate": 1.4548765294061274e-05,
      "loss": 1.2161,
      "step": 9501
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.202462911605835,
      "learning_rate": 1.4548126377663483e-05,
      "loss": 1.0213,
      "step": 9502
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9005851745605469,
      "learning_rate": 1.4547487461265695e-05,
      "loss": 1.0051,
      "step": 9503
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.0986275672912598,
      "learning_rate": 1.4546848544867905e-05,
      "loss": 1.0278,
      "step": 9504
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.1881585121154785,
      "learning_rate": 1.4546209628470116e-05,
      "loss": 1.1843,
      "step": 9505
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7595758438110352,
      "learning_rate": 1.4545570712072326e-05,
      "loss": 0.9399,
      "step": 9506
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.6144440174102783,
      "learning_rate": 1.4544931795674538e-05,
      "loss": 1.2671,
      "step": 9507
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7274419069290161,
      "learning_rate": 1.4544292879276748e-05,
      "loss": 0.8589,
      "step": 9508
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.975775241851807,
      "learning_rate": 1.454365396287896e-05,
      "loss": 1.3586,
      "step": 9509
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.287332057952881,
      "learning_rate": 1.4543015046481167e-05,
      "loss": 1.3312,
      "step": 9510
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7176003456115723,
      "learning_rate": 1.454237613008338e-05,
      "loss": 0.5619,
      "step": 9511
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.389231204986572,
      "learning_rate": 1.4541737213685589e-05,
      "loss": 1.9422,
      "step": 9512
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0408763885498047,
      "learning_rate": 1.4541098297287802e-05,
      "loss": 0.8471,
      "step": 9513
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8285292387008667,
      "learning_rate": 1.454045938089001e-05,
      "loss": 1.0092,
      "step": 9514
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8467817306518555,
      "learning_rate": 1.4539820464492222e-05,
      "loss": 0.9321,
      "step": 9515
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6227561235427856,
      "learning_rate": 1.4539181548094432e-05,
      "loss": 0.8548,
      "step": 9516
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.257370471954346,
      "learning_rate": 1.4538542631696643e-05,
      "loss": 1.0066,
      "step": 9517
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8161585330963135,
      "learning_rate": 1.4537903715298853e-05,
      "loss": 0.8752,
      "step": 9518
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.130526542663574,
      "learning_rate": 1.4537264798901064e-05,
      "loss": 0.9037,
      "step": 9519
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.327512741088867,
      "learning_rate": 1.4536625882503274e-05,
      "loss": 0.9346,
      "step": 9520
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5893406867980957,
      "learning_rate": 1.4535986966105486e-05,
      "loss": 0.4666,
      "step": 9521
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9181679487228394,
      "learning_rate": 1.4535348049707696e-05,
      "loss": 0.8368,
      "step": 9522
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2741174697875977,
      "learning_rate": 1.4534709133309907e-05,
      "loss": 0.8821,
      "step": 9523
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2206175327301025,
      "learning_rate": 1.4534070216912117e-05,
      "loss": 0.979,
      "step": 9524
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8650949001312256,
      "learning_rate": 1.4533431300514329e-05,
      "loss": 1.1899,
      "step": 9525
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.47202730178833,
      "learning_rate": 1.4532792384116539e-05,
      "loss": 0.7006,
      "step": 9526
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.655836582183838,
      "learning_rate": 1.453215346771875e-05,
      "loss": 0.9823,
      "step": 9527
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1531834602355957,
      "learning_rate": 1.453151455132096e-05,
      "loss": 0.9909,
      "step": 9528
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.086596727371216,
      "learning_rate": 1.4530875634923171e-05,
      "loss": 0.9137,
      "step": 9529
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2049107551574707,
      "learning_rate": 1.4530236718525381e-05,
      "loss": 0.885,
      "step": 9530
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8703421354293823,
      "learning_rate": 1.4529597802127593e-05,
      "loss": 0.97,
      "step": 9531
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.591782331466675,
      "learning_rate": 1.4528958885729803e-05,
      "loss": 1.0733,
      "step": 9532
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0803098678588867,
      "learning_rate": 1.4528319969332014e-05,
      "loss": 1.0503,
      "step": 9533
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0854525566101074,
      "learning_rate": 1.4527681052934224e-05,
      "loss": 0.8698,
      "step": 9534
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.954873561859131,
      "learning_rate": 1.4527042136536436e-05,
      "loss": 1.5502,
      "step": 9535
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3626811504364014,
      "learning_rate": 1.4526403220138646e-05,
      "loss": 1.0247,
      "step": 9536
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.5547847747802734,
      "learning_rate": 1.4525764303740857e-05,
      "loss": 0.8346,
      "step": 9537
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.9182214736938477,
      "learning_rate": 1.4525125387343067e-05,
      "loss": 1.2428,
      "step": 9538
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.771596908569336,
      "learning_rate": 1.4524486470945278e-05,
      "loss": 0.8363,
      "step": 9539
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7727620601654053,
      "learning_rate": 1.4523847554547488e-05,
      "loss": 0.9999,
      "step": 9540
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.255616664886475,
      "learning_rate": 1.45232086381497e-05,
      "loss": 1.0694,
      "step": 9541
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.069644451141357,
      "learning_rate": 1.452256972175191e-05,
      "loss": 1.0976,
      "step": 9542
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.021918535232544,
      "learning_rate": 1.4521930805354121e-05,
      "loss": 0.8311,
      "step": 9543
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7735936641693115,
      "learning_rate": 1.4521291888956331e-05,
      "loss": 0.9023,
      "step": 9544
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.742630958557129,
      "learning_rate": 1.4520652972558543e-05,
      "loss": 0.4508,
      "step": 9545
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1571028232574463,
      "learning_rate": 1.452001405616075e-05,
      "loss": 0.9591,
      "step": 9546
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8621271848678589,
      "learning_rate": 1.4519375139762964e-05,
      "loss": 0.8971,
      "step": 9547
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.799210548400879,
      "learning_rate": 1.4518736223365172e-05,
      "loss": 1.4288,
      "step": 9548
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.052419662475586,
      "learning_rate": 1.4518097306967385e-05,
      "loss": 0.9108,
      "step": 9549
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.757323145866394,
      "learning_rate": 1.4517458390569594e-05,
      "loss": 1.0376,
      "step": 9550
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.021780490875244,
      "learning_rate": 1.4516819474171807e-05,
      "loss": 0.9744,
      "step": 9551
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9198129177093506,
      "learning_rate": 1.4516180557774015e-05,
      "loss": 0.9881,
      "step": 9552
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7658488750457764,
      "learning_rate": 1.4515541641376227e-05,
      "loss": 0.8822,
      "step": 9553
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.530412197113037,
      "learning_rate": 1.4514902724978436e-05,
      "loss": 0.8488,
      "step": 9554
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5398802757263184,
      "learning_rate": 1.4514263808580648e-05,
      "loss": 0.8322,
      "step": 9555
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6554805040359497,
      "learning_rate": 1.4513624892182858e-05,
      "loss": 0.9295,
      "step": 9556
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.824211597442627,
      "learning_rate": 1.451298597578507e-05,
      "loss": 0.9771,
      "step": 9557
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.883328437805176,
      "learning_rate": 1.451234705938728e-05,
      "loss": 0.917,
      "step": 9558
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.6982433795928955,
      "learning_rate": 1.451170814298949e-05,
      "loss": 0.9282,
      "step": 9559
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1195640563964844,
      "learning_rate": 1.45110692265917e-05,
      "loss": 0.9936,
      "step": 9560
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6915274858474731,
      "learning_rate": 1.4510430310193912e-05,
      "loss": 0.8403,
      "step": 9561
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7994319200515747,
      "learning_rate": 1.4509791393796122e-05,
      "loss": 0.9117,
      "step": 9562
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9482568502426147,
      "learning_rate": 1.4509152477398334e-05,
      "loss": 1.1259,
      "step": 9563
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.714587688446045,
      "learning_rate": 1.4508513561000543e-05,
      "loss": 0.8914,
      "step": 9564
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.162871837615967,
      "learning_rate": 1.4507874644602755e-05,
      "loss": 0.9893,
      "step": 9565
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.4026875495910645,
      "learning_rate": 1.4507235728204965e-05,
      "loss": 0.919,
      "step": 9566
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.386727809906006,
      "learning_rate": 1.4506596811807176e-05,
      "loss": 1.7682,
      "step": 9567
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5288194417953491,
      "learning_rate": 1.4505957895409386e-05,
      "loss": 0.4021,
      "step": 9568
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.416987657546997,
      "learning_rate": 1.4505318979011598e-05,
      "loss": 0.8654,
      "step": 9569
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.6951663494110107,
      "learning_rate": 1.4504680062613808e-05,
      "loss": 0.7048,
      "step": 9570
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4492212533950806,
      "learning_rate": 1.450404114621602e-05,
      "loss": 0.4024,
      "step": 9571
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.620422124862671,
      "learning_rate": 1.4503402229818229e-05,
      "loss": 0.9221,
      "step": 9572
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6938279867172241,
      "learning_rate": 1.450276331342044e-05,
      "loss": 0.8072,
      "step": 9573
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8426508903503418,
      "learning_rate": 1.450212439702265e-05,
      "loss": 1.0751,
      "step": 9574
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.098332643508911,
      "learning_rate": 1.4501485480624862e-05,
      "loss": 0.9219,
      "step": 9575
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.239133358001709,
      "learning_rate": 1.4500846564227072e-05,
      "loss": 0.8461,
      "step": 9576
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.802100419998169,
      "learning_rate": 1.4500207647829283e-05,
      "loss": 0.9355,
      "step": 9577
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.494513750076294,
      "learning_rate": 1.4499568731431493e-05,
      "loss": 0.9193,
      "step": 9578
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.892965793609619,
      "learning_rate": 1.4498929815033705e-05,
      "loss": 1.3388,
      "step": 9579
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8717076778411865,
      "learning_rate": 1.4498290898635915e-05,
      "loss": 0.9039,
      "step": 9580
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1641147136688232,
      "learning_rate": 1.4497651982238126e-05,
      "loss": 0.9957,
      "step": 9581
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8107279539108276,
      "learning_rate": 1.4497013065840336e-05,
      "loss": 0.9951,
      "step": 9582
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0631871223449707,
      "learning_rate": 1.4496374149442548e-05,
      "loss": 1.3201,
      "step": 9583
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7832813262939453,
      "learning_rate": 1.4495735233044756e-05,
      "loss": 1.0397,
      "step": 9584
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.137249708175659,
      "learning_rate": 1.4495096316646969e-05,
      "loss": 1.3114,
      "step": 9585
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6519989967346191,
      "learning_rate": 1.4494457400249177e-05,
      "loss": 0.8216,
      "step": 9586
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9984772205352783,
      "learning_rate": 1.449381848385139e-05,
      "loss": 0.9018,
      "step": 9587
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3042588233947754,
      "learning_rate": 1.4493179567453599e-05,
      "loss": 0.9563,
      "step": 9588
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7138452529907227,
      "learning_rate": 1.449254065105581e-05,
      "loss": 0.8714,
      "step": 9589
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7308673858642578,
      "learning_rate": 1.449190173465802e-05,
      "loss": 0.4542,
      "step": 9590
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9118760824203491,
      "learning_rate": 1.4491262818260231e-05,
      "loss": 0.9671,
      "step": 9591
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1202311515808105,
      "learning_rate": 1.4490623901862441e-05,
      "loss": 0.9042,
      "step": 9592
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4582948684692383,
      "learning_rate": 1.4489984985464653e-05,
      "loss": 0.9444,
      "step": 9593
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.204136610031128,
      "learning_rate": 1.4489346069066863e-05,
      "loss": 0.7947,
      "step": 9594
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8455793857574463,
      "learning_rate": 1.4488707152669074e-05,
      "loss": 1.0737,
      "step": 9595
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6502524614334106,
      "learning_rate": 1.4488068236271284e-05,
      "loss": 0.8851,
      "step": 9596
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.849619388580322,
      "learning_rate": 1.4487429319873496e-05,
      "loss": 1.7196,
      "step": 9597
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8178499937057495,
      "learning_rate": 1.4486790403475706e-05,
      "loss": 1.134,
      "step": 9598
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0815396308898926,
      "learning_rate": 1.4486151487077917e-05,
      "loss": 0.9609,
      "step": 9599
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.878502130508423,
      "learning_rate": 1.4485512570680127e-05,
      "loss": 0.7668,
      "step": 9600
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2036336660385132,
      "learning_rate": 1.4484873654282338e-05,
      "loss": 0.4751,
      "step": 9601
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2770400047302246,
      "learning_rate": 1.4484234737884548e-05,
      "loss": 1.2992,
      "step": 9602
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.185612440109253,
      "learning_rate": 1.448359582148676e-05,
      "loss": 0.9079,
      "step": 9603
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8358675241470337,
      "learning_rate": 1.448295690508897e-05,
      "loss": 0.9498,
      "step": 9604
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0979113578796387,
      "learning_rate": 1.4482317988691181e-05,
      "loss": 0.824,
      "step": 9605
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9460577964782715,
      "learning_rate": 1.4481679072293391e-05,
      "loss": 0.8462,
      "step": 9606
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8566678762435913,
      "learning_rate": 1.4481040155895603e-05,
      "loss": 0.6734,
      "step": 9607
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.032158374786377,
      "learning_rate": 1.4480401239497813e-05,
      "loss": 1.4443,
      "step": 9608
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3094217777252197,
      "learning_rate": 1.4479762323100024e-05,
      "loss": 0.9882,
      "step": 9609
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.564684510231018,
      "learning_rate": 1.4479123406702234e-05,
      "loss": 1.0435,
      "step": 9610
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7440643310546875,
      "learning_rate": 1.4478484490304445e-05,
      "loss": 0.8405,
      "step": 9611
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.939921259880066,
      "learning_rate": 1.4477845573906655e-05,
      "loss": 0.9507,
      "step": 9612
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.143404006958008,
      "learning_rate": 1.4477206657508867e-05,
      "loss": 0.9805,
      "step": 9613
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.7716362476348877,
      "learning_rate": 1.4476567741111077e-05,
      "loss": 0.9953,
      "step": 9614
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8056820631027222,
      "learning_rate": 1.4475928824713288e-05,
      "loss": 0.8693,
      "step": 9615
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.356461763381958,
      "learning_rate": 1.4475289908315498e-05,
      "loss": 0.9057,
      "step": 9616
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.436816453933716,
      "learning_rate": 1.447465099191771e-05,
      "loss": 0.9829,
      "step": 9617
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7487910985946655,
      "learning_rate": 1.447401207551992e-05,
      "loss": 0.9287,
      "step": 9618
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3025338649749756,
      "learning_rate": 1.4473373159122131e-05,
      "loss": 1.0492,
      "step": 9619
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3435347080230713,
      "learning_rate": 1.447273424272434e-05,
      "loss": 0.9289,
      "step": 9620
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6932133436203003,
      "learning_rate": 1.4472095326326552e-05,
      "loss": 1.0002,
      "step": 9621
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.084261655807495,
      "learning_rate": 1.447145640992876e-05,
      "loss": 0.9231,
      "step": 9622
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7953150272369385,
      "learning_rate": 1.4470817493530974e-05,
      "loss": 0.9227,
      "step": 9623
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8167216777801514,
      "learning_rate": 1.4470178577133182e-05,
      "loss": 0.8651,
      "step": 9624
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.373026371002197,
      "learning_rate": 1.4469539660735394e-05,
      "loss": 1.4967,
      "step": 9625
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9524978399276733,
      "learning_rate": 1.4468900744337603e-05,
      "loss": 1.0662,
      "step": 9626
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8419429063796997,
      "learning_rate": 1.4468261827939815e-05,
      "loss": 0.9004,
      "step": 9627
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7777018547058105,
      "learning_rate": 1.4467622911542025e-05,
      "loss": 0.9015,
      "step": 9628
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.076643705368042,
      "learning_rate": 1.4466983995144236e-05,
      "loss": 1.0136,
      "step": 9629
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.747370958328247,
      "learning_rate": 1.4466345078746446e-05,
      "loss": 1.254,
      "step": 9630
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.578009009361267,
      "learning_rate": 1.4465706162348658e-05,
      "loss": 0.824,
      "step": 9631
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.021799087524414,
      "learning_rate": 1.4465067245950868e-05,
      "loss": 1.3008,
      "step": 9632
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.856085300445557,
      "learning_rate": 1.446442832955308e-05,
      "loss": 0.7958,
      "step": 9633
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0912938117980957,
      "learning_rate": 1.4463789413155289e-05,
      "loss": 0.6988,
      "step": 9634
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0077717304229736,
      "learning_rate": 1.44631504967575e-05,
      "loss": 0.9427,
      "step": 9635
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.065075159072876,
      "learning_rate": 1.446251158035971e-05,
      "loss": 0.8962,
      "step": 9636
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5988154411315918,
      "learning_rate": 1.4461872663961922e-05,
      "loss": 0.8281,
      "step": 9637
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.125579357147217,
      "learning_rate": 1.4461233747564132e-05,
      "loss": 1.0991,
      "step": 9638
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2566025257110596,
      "learning_rate": 1.4460594831166343e-05,
      "loss": 1.0199,
      "step": 9639
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.816128134727478,
      "learning_rate": 1.4459955914768553e-05,
      "loss": 0.888,
      "step": 9640
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8625112771987915,
      "learning_rate": 1.4459316998370765e-05,
      "loss": 1.2105,
      "step": 9641
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.990647554397583,
      "learning_rate": 1.4458678081972975e-05,
      "loss": 0.9639,
      "step": 9642
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7605355978012085,
      "learning_rate": 1.4458039165575186e-05,
      "loss": 0.9988,
      "step": 9643
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.613037586212158,
      "learning_rate": 1.4457400249177396e-05,
      "loss": 1.2762,
      "step": 9644
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.219513416290283,
      "learning_rate": 1.4456761332779608e-05,
      "loss": 1.1128,
      "step": 9645
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0211286544799805,
      "learning_rate": 1.4456122416381817e-05,
      "loss": 0.7293,
      "step": 9646
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.80349862575531,
      "learning_rate": 1.4455483499984029e-05,
      "loss": 0.9444,
      "step": 9647
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7553480863571167,
      "learning_rate": 1.4454844583586239e-05,
      "loss": 0.9053,
      "step": 9648
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.150404930114746,
      "learning_rate": 1.445420566718845e-05,
      "loss": 1.0533,
      "step": 9649
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8758792877197266,
      "learning_rate": 1.445356675079066e-05,
      "loss": 1.119,
      "step": 9650
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0273120403289795,
      "learning_rate": 1.4452927834392872e-05,
      "loss": 0.922,
      "step": 9651
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2860379219055176,
      "learning_rate": 1.4452288917995082e-05,
      "loss": 0.7498,
      "step": 9652
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0393807888031006,
      "learning_rate": 1.4451650001597293e-05,
      "loss": 1.0166,
      "step": 9653
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.926802396774292,
      "learning_rate": 1.4451011085199503e-05,
      "loss": 0.9774,
      "step": 9654
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.598953127861023,
      "learning_rate": 1.4450372168801715e-05,
      "loss": 0.5515,
      "step": 9655
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.7217869758605957,
      "learning_rate": 1.4449733252403923e-05,
      "loss": 1.1357,
      "step": 9656
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9904992580413818,
      "learning_rate": 1.4449094336006136e-05,
      "loss": 1.2119,
      "step": 9657
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5068857669830322,
      "learning_rate": 1.4448455419608344e-05,
      "loss": 0.8472,
      "step": 9658
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.0761919021606445,
      "learning_rate": 1.4447816503210557e-05,
      "loss": 0.959,
      "step": 9659
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6214526891708374,
      "learning_rate": 1.4447177586812766e-05,
      "loss": 0.8391,
      "step": 9660
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.545346260070801,
      "learning_rate": 1.4446538670414979e-05,
      "loss": 1.7625,
      "step": 9661
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.324949026107788,
      "learning_rate": 1.4445899754017187e-05,
      "loss": 0.966,
      "step": 9662
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.366603374481201,
      "learning_rate": 1.4445260837619398e-05,
      "loss": 1.6603,
      "step": 9663
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.651906490325928,
      "learning_rate": 1.4444621921221608e-05,
      "loss": 1.2863,
      "step": 9664
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.832864284515381,
      "learning_rate": 1.444398300482382e-05,
      "loss": 1.4287,
      "step": 9665
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1282668113708496,
      "learning_rate": 1.444334408842603e-05,
      "loss": 0.8047,
      "step": 9666
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.13478946685791,
      "learning_rate": 1.4442705172028241e-05,
      "loss": 1.1514,
      "step": 9667
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9708715677261353,
      "learning_rate": 1.4442066255630451e-05,
      "loss": 0.91,
      "step": 9668
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.033115863800049,
      "learning_rate": 1.4441427339232663e-05,
      "loss": 0.9399,
      "step": 9669
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6599822044372559,
      "learning_rate": 1.4440788422834873e-05,
      "loss": 0.9341,
      "step": 9670
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0057387351989746,
      "learning_rate": 1.4440149506437084e-05,
      "loss": 1.032,
      "step": 9671
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.515660047531128,
      "learning_rate": 1.4439510590039294e-05,
      "loss": 0.8679,
      "step": 9672
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.604992151260376,
      "learning_rate": 1.4438871673641505e-05,
      "loss": 1.0282,
      "step": 9673
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.858046531677246,
      "learning_rate": 1.4438232757243715e-05,
      "loss": 0.9135,
      "step": 9674
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.123302698135376,
      "learning_rate": 1.4437593840845927e-05,
      "loss": 0.8958,
      "step": 9675
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8609014749526978,
      "learning_rate": 1.4436954924448137e-05,
      "loss": 1.0064,
      "step": 9676
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.193113327026367,
      "learning_rate": 1.4436316008050348e-05,
      "loss": 0.479,
      "step": 9677
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8313143253326416,
      "learning_rate": 1.4435677091652558e-05,
      "loss": 0.9585,
      "step": 9678
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6756659746170044,
      "learning_rate": 1.443503817525477e-05,
      "loss": 0.8701,
      "step": 9679
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.270080089569092,
      "learning_rate": 1.443439925885698e-05,
      "loss": 1.5405,
      "step": 9680
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5561336278915405,
      "learning_rate": 1.4433760342459191e-05,
      "loss": 0.3994,
      "step": 9681
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8279962539672852,
      "learning_rate": 1.4433121426061401e-05,
      "loss": 0.8661,
      "step": 9682
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.914071083068848,
      "learning_rate": 1.4432482509663612e-05,
      "loss": 1.568,
      "step": 9683
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8350250720977783,
      "learning_rate": 1.4431843593265822e-05,
      "loss": 0.8865,
      "step": 9684
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.9785985946655273,
      "learning_rate": 1.4431204676868034e-05,
      "loss": 0.773,
      "step": 9685
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4301955699920654,
      "learning_rate": 1.4430565760470244e-05,
      "loss": 0.798,
      "step": 9686
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4649860858917236,
      "learning_rate": 1.4429926844072455e-05,
      "loss": 0.8302,
      "step": 9687
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6435611248016357,
      "learning_rate": 1.4429287927674665e-05,
      "loss": 0.9529,
      "step": 9688
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2739884853363037,
      "learning_rate": 1.4428649011276877e-05,
      "loss": 0.9162,
      "step": 9689
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.038662910461426,
      "learning_rate": 1.4428010094879087e-05,
      "loss": 0.9688,
      "step": 9690
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.857488989830017,
      "learning_rate": 1.4427371178481298e-05,
      "loss": 0.9685,
      "step": 9691
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.113128185272217,
      "learning_rate": 1.4426732262083506e-05,
      "loss": 1.0569,
      "step": 9692
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9590673446655273,
      "learning_rate": 1.442609334568572e-05,
      "loss": 1.0456,
      "step": 9693
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9571287631988525,
      "learning_rate": 1.4425454429287928e-05,
      "loss": 0.9282,
      "step": 9694
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8297345638275146,
      "learning_rate": 1.4424815512890141e-05,
      "loss": 1.1865,
      "step": 9695
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6407370567321777,
      "learning_rate": 1.4424176596492349e-05,
      "loss": 0.8899,
      "step": 9696
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5134801864624023,
      "learning_rate": 1.4423537680094562e-05,
      "loss": 0.8874,
      "step": 9697
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3269810676574707,
      "learning_rate": 1.442289876369677e-05,
      "loss": 1.0073,
      "step": 9698
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.236811399459839,
      "learning_rate": 1.4422259847298982e-05,
      "loss": 0.9424,
      "step": 9699
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.839684247970581,
      "learning_rate": 1.4421620930901192e-05,
      "loss": 0.922,
      "step": 9700
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.124957323074341,
      "learning_rate": 1.4420982014503403e-05,
      "loss": 1.2509,
      "step": 9701
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8228566646575928,
      "learning_rate": 1.4420343098105613e-05,
      "loss": 1.2099,
      "step": 9702
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.277688980102539,
      "learning_rate": 1.4419704181707825e-05,
      "loss": 1.059,
      "step": 9703
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8613706827163696,
      "learning_rate": 1.4419065265310035e-05,
      "loss": 0.9381,
      "step": 9704
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.896307110786438,
      "learning_rate": 1.4418426348912246e-05,
      "loss": 1.2842,
      "step": 9705
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4928176403045654,
      "learning_rate": 1.4417787432514456e-05,
      "loss": 0.9113,
      "step": 9706
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9300663471221924,
      "learning_rate": 1.4417148516116668e-05,
      "loss": 0.9171,
      "step": 9707
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5494117736816406,
      "learning_rate": 1.4416509599718877e-05,
      "loss": 0.8343,
      "step": 9708
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.943514585494995,
      "learning_rate": 1.4415870683321089e-05,
      "loss": 0.9935,
      "step": 9709
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.050438404083252,
      "learning_rate": 1.4415231766923299e-05,
      "loss": 1.1256,
      "step": 9710
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.523150086402893,
      "learning_rate": 1.441459285052551e-05,
      "loss": 0.8602,
      "step": 9711
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.051981449127197,
      "learning_rate": 1.441395393412772e-05,
      "loss": 1.2376,
      "step": 9712
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7057909965515137,
      "learning_rate": 1.4413315017729932e-05,
      "loss": 0.8923,
      "step": 9713
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8310762643814087,
      "learning_rate": 1.4412676101332142e-05,
      "loss": 0.9362,
      "step": 9714
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.220013380050659,
      "learning_rate": 1.4412037184934353e-05,
      "loss": 0.8604,
      "step": 9715
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.264862537384033,
      "learning_rate": 1.4411398268536563e-05,
      "loss": 1.0916,
      "step": 9716
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.12809419631958,
      "learning_rate": 1.4410759352138775e-05,
      "loss": 0.8981,
      "step": 9717
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.297511100769043,
      "learning_rate": 1.4410120435740984e-05,
      "loss": 1.5011,
      "step": 9718
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.001694440841675,
      "learning_rate": 1.4409481519343196e-05,
      "loss": 0.9517,
      "step": 9719
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8936423063278198,
      "learning_rate": 1.4408842602945406e-05,
      "loss": 0.936,
      "step": 9720
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.915666103363037,
      "learning_rate": 1.4408203686547617e-05,
      "loss": 0.8599,
      "step": 9721
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.4423108100891113,
      "learning_rate": 1.4407564770149827e-05,
      "loss": 1.0312,
      "step": 9722
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6315948963165283,
      "learning_rate": 1.4406925853752039e-05,
      "loss": 0.9363,
      "step": 9723
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.1954867839813232,
      "learning_rate": 1.4406286937354249e-05,
      "loss": 1.1209,
      "step": 9724
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8705394268035889,
      "learning_rate": 1.440564802095646e-05,
      "loss": 1.0079,
      "step": 9725
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.642890214920044,
      "learning_rate": 1.440500910455867e-05,
      "loss": 0.9526,
      "step": 9726
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.253544807434082,
      "learning_rate": 1.4404370188160882e-05,
      "loss": 0.9738,
      "step": 9727
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.886569023132324,
      "learning_rate": 1.4403731271763091e-05,
      "loss": 1.0396,
      "step": 9728
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9201116561889648,
      "learning_rate": 1.4403092355365303e-05,
      "loss": 0.9956,
      "step": 9729
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7310794591903687,
      "learning_rate": 1.4402453438967511e-05,
      "loss": 0.9461,
      "step": 9730
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.568547248840332,
      "learning_rate": 1.4401814522569724e-05,
      "loss": 1.0781,
      "step": 9731
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.723317265510559,
      "learning_rate": 1.4401175606171933e-05,
      "loss": 0.9327,
      "step": 9732
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.7365498542785645,
      "learning_rate": 1.4400536689774146e-05,
      "loss": 1.1979,
      "step": 9733
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.104799747467041,
      "learning_rate": 1.4399897773376354e-05,
      "loss": 0.9563,
      "step": 9734
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9201271533966064,
      "learning_rate": 1.4399258856978565e-05,
      "loss": 0.9962,
      "step": 9735
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7429099082946777,
      "learning_rate": 1.4398619940580775e-05,
      "loss": 0.8371,
      "step": 9736
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9448527097702026,
      "learning_rate": 1.4397981024182987e-05,
      "loss": 1.1024,
      "step": 9737
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1622822284698486,
      "learning_rate": 1.4397342107785197e-05,
      "loss": 0.9098,
      "step": 9738
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7678446769714355,
      "learning_rate": 1.4396703191387408e-05,
      "loss": 0.9055,
      "step": 9739
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.945915699005127,
      "learning_rate": 1.4396064274989618e-05,
      "loss": 0.9533,
      "step": 9740
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8401652574539185,
      "learning_rate": 1.439542535859183e-05,
      "loss": 0.973,
      "step": 9741
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1437859535217285,
      "learning_rate": 1.439478644219404e-05,
      "loss": 0.9131,
      "step": 9742
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.559536933898926,
      "learning_rate": 1.4394147525796251e-05,
      "loss": 0.9433,
      "step": 9743
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5115503072738647,
      "learning_rate": 1.4393508609398461e-05,
      "loss": 0.8969,
      "step": 9744
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.689370632171631,
      "learning_rate": 1.4392869693000672e-05,
      "loss": 1.3536,
      "step": 9745
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.813694953918457,
      "learning_rate": 1.4392230776602882e-05,
      "loss": 1.0028,
      "step": 9746
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5886218547821045,
      "learning_rate": 1.4391591860205094e-05,
      "loss": 0.7485,
      "step": 9747
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.55670428276062,
      "learning_rate": 1.4390952943807304e-05,
      "loss": 0.9786,
      "step": 9748
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.053568363189697,
      "learning_rate": 1.4390314027409515e-05,
      "loss": 1.1593,
      "step": 9749
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7706540822982788,
      "learning_rate": 1.4389675111011725e-05,
      "loss": 0.8526,
      "step": 9750
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.040027379989624,
      "learning_rate": 1.4389036194613937e-05,
      "loss": 1.0645,
      "step": 9751
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7212474346160889,
      "learning_rate": 1.4388397278216147e-05,
      "loss": 0.9118,
      "step": 9752
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0018298625946045,
      "learning_rate": 1.4387758361818358e-05,
      "loss": 0.9814,
      "step": 9753
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8206486701965332,
      "learning_rate": 1.4387119445420568e-05,
      "loss": 0.8615,
      "step": 9754
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4365506172180176,
      "learning_rate": 1.438648052902278e-05,
      "loss": 0.8962,
      "step": 9755
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.865349531173706,
      "learning_rate": 1.438584161262499e-05,
      "loss": 0.973,
      "step": 9756
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.614250659942627,
      "learning_rate": 1.4385202696227201e-05,
      "loss": 1.0825,
      "step": 9757
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6640089750289917,
      "learning_rate": 1.438456377982941e-05,
      "loss": 0.9191,
      "step": 9758
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9927494525909424,
      "learning_rate": 1.4383924863431622e-05,
      "loss": 0.8339,
      "step": 9759
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.2383713722229,
      "learning_rate": 1.4383285947033832e-05,
      "loss": 1.1906,
      "step": 9760
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.843113899230957,
      "learning_rate": 1.4382647030636044e-05,
      "loss": 0.8788,
      "step": 9761
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.013263702392578,
      "learning_rate": 1.4382008114238254e-05,
      "loss": 0.9714,
      "step": 9762
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6921836137771606,
      "learning_rate": 1.4381369197840465e-05,
      "loss": 0.927,
      "step": 9763
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.971214532852173,
      "learning_rate": 1.4380730281442675e-05,
      "loss": 1.0886,
      "step": 9764
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.009404420852661,
      "learning_rate": 1.4380091365044886e-05,
      "loss": 0.8179,
      "step": 9765
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.122480630874634,
      "learning_rate": 1.4379452448647095e-05,
      "loss": 1.0137,
      "step": 9766
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4822635650634766,
      "learning_rate": 1.4378813532249308e-05,
      "loss": 0.6554,
      "step": 9767
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.045894145965576,
      "learning_rate": 1.4378174615851516e-05,
      "loss": 0.8539,
      "step": 9768
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.046196937561035,
      "learning_rate": 1.437753569945373e-05,
      "loss": 1.3451,
      "step": 9769
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8296523094177246,
      "learning_rate": 1.4376896783055937e-05,
      "loss": 0.4106,
      "step": 9770
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.634582996368408,
      "learning_rate": 1.4376257866658149e-05,
      "loss": 1.0904,
      "step": 9771
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7770273685455322,
      "learning_rate": 1.4375618950260359e-05,
      "loss": 0.8882,
      "step": 9772
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.593549370765686,
      "learning_rate": 1.437498003386257e-05,
      "loss": 0.9678,
      "step": 9773
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6636550426483154,
      "learning_rate": 1.437434111746478e-05,
      "loss": 1.1479,
      "step": 9774
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2484846115112305,
      "learning_rate": 1.4373702201066992e-05,
      "loss": 1.2248,
      "step": 9775
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.108334541320801,
      "learning_rate": 1.4373063284669202e-05,
      "loss": 0.9334,
      "step": 9776
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.223235845565796,
      "learning_rate": 1.4372424368271413e-05,
      "loss": 1.1063,
      "step": 9777
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5086833238601685,
      "learning_rate": 1.4371785451873623e-05,
      "loss": 0.9146,
      "step": 9778
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5298656225204468,
      "learning_rate": 1.4371146535475835e-05,
      "loss": 0.4733,
      "step": 9779
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6574987173080444,
      "learning_rate": 1.4370507619078044e-05,
      "loss": 0.9145,
      "step": 9780
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6072368621826172,
      "learning_rate": 1.4369868702680256e-05,
      "loss": 0.9728,
      "step": 9781
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3252017498016357,
      "learning_rate": 1.4369229786282466e-05,
      "loss": 1.0593,
      "step": 9782
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3692632913589478,
      "learning_rate": 1.4368590869884677e-05,
      "loss": 0.3965,
      "step": 9783
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.913387656211853,
      "learning_rate": 1.4367951953486887e-05,
      "loss": 0.9729,
      "step": 9784
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.6972568035125732,
      "learning_rate": 1.4367313037089099e-05,
      "loss": 0.7427,
      "step": 9785
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7391525506973267,
      "learning_rate": 1.4366674120691309e-05,
      "loss": 0.9866,
      "step": 9786
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.52028226852417,
      "learning_rate": 1.436603520429352e-05,
      "loss": 0.4603,
      "step": 9787
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.049278497695923,
      "learning_rate": 1.436539628789573e-05,
      "loss": 0.9377,
      "step": 9788
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.045724391937256,
      "learning_rate": 1.4364757371497942e-05,
      "loss": 1.353,
      "step": 9789
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8245515823364258,
      "learning_rate": 1.4364118455100151e-05,
      "loss": 0.839,
      "step": 9790
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9566231966018677,
      "learning_rate": 1.4363479538702363e-05,
      "loss": 0.8968,
      "step": 9791
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5813605785369873,
      "learning_rate": 1.4362840622304573e-05,
      "loss": 0.8623,
      "step": 9792
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3838406801223755,
      "learning_rate": 1.4362201705906784e-05,
      "loss": 0.8232,
      "step": 9793
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.866628885269165,
      "learning_rate": 1.4361562789508994e-05,
      "loss": 0.7796,
      "step": 9794
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.558326482772827,
      "learning_rate": 1.4360923873111206e-05,
      "loss": 0.7323,
      "step": 9795
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.094616174697876,
      "learning_rate": 1.4360284956713416e-05,
      "loss": 1.1856,
      "step": 9796
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.334962844848633,
      "learning_rate": 1.4359646040315627e-05,
      "loss": 1.0984,
      "step": 9797
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.2981483936309814,
      "learning_rate": 1.4359007123917837e-05,
      "loss": 1.0155,
      "step": 9798
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.1124775409698486,
      "learning_rate": 1.4358368207520049e-05,
      "loss": 0.8438,
      "step": 9799
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.44323468208313,
      "learning_rate": 1.4357729291122258e-05,
      "loss": 0.8735,
      "step": 9800
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0672574043273926,
      "learning_rate": 1.435709037472447e-05,
      "loss": 0.7797,
      "step": 9801
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.6668717861175537,
      "learning_rate": 1.4356451458326678e-05,
      "loss": 1.047,
      "step": 9802
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7102813720703125,
      "learning_rate": 1.4355812541928891e-05,
      "loss": 0.9353,
      "step": 9803
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9094074964523315,
      "learning_rate": 1.43551736255311e-05,
      "loss": 0.9975,
      "step": 9804
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7344486713409424,
      "learning_rate": 1.4354534709133313e-05,
      "loss": 0.8779,
      "step": 9805
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.878889560699463,
      "learning_rate": 1.4353895792735521e-05,
      "loss": 1.1853,
      "step": 9806
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5429754257202148,
      "learning_rate": 1.4353256876337734e-05,
      "loss": 0.8369,
      "step": 9807
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.131084680557251,
      "learning_rate": 1.4352617959939942e-05,
      "loss": 0.9423,
      "step": 9808
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.433657646179199,
      "learning_rate": 1.4351979043542154e-05,
      "loss": 1.2822,
      "step": 9809
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.6846628189086914,
      "learning_rate": 1.4351340127144364e-05,
      "loss": 1.1658,
      "step": 9810
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7700035572052002,
      "learning_rate": 1.4350701210746575e-05,
      "loss": 0.9899,
      "step": 9811
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9086151123046875,
      "learning_rate": 1.4350062294348785e-05,
      "loss": 0.9815,
      "step": 9812
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9556396007537842,
      "learning_rate": 1.4349423377950997e-05,
      "loss": 1.0742,
      "step": 9813
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.688543438911438,
      "learning_rate": 1.4348784461553207e-05,
      "loss": 0.8176,
      "step": 9814
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4717313051223755,
      "learning_rate": 1.4348145545155418e-05,
      "loss": 0.9769,
      "step": 9815
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5752395391464233,
      "learning_rate": 1.4347506628757628e-05,
      "loss": 0.3617,
      "step": 9816
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.175854206085205,
      "learning_rate": 1.434686771235984e-05,
      "loss": 1.3919,
      "step": 9817
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7744605541229248,
      "learning_rate": 1.434622879596205e-05,
      "loss": 0.8442,
      "step": 9818
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9475631713867188,
      "learning_rate": 1.4345589879564261e-05,
      "loss": 1.2509,
      "step": 9819
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.774813175201416,
      "learning_rate": 1.434495096316647e-05,
      "loss": 0.9291,
      "step": 9820
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7863394021987915,
      "learning_rate": 1.4344312046768682e-05,
      "loss": 1.0065,
      "step": 9821
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4712082147598267,
      "learning_rate": 1.4343673130370892e-05,
      "loss": 0.4842,
      "step": 9822
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9276330471038818,
      "learning_rate": 1.4343034213973104e-05,
      "loss": 0.8811,
      "step": 9823
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.929393291473389,
      "learning_rate": 1.4342395297575314e-05,
      "loss": 1.603,
      "step": 9824
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.841848373413086,
      "learning_rate": 1.4341756381177525e-05,
      "loss": 0.9691,
      "step": 9825
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8920764923095703,
      "learning_rate": 1.4341117464779735e-05,
      "loss": 1.0085,
      "step": 9826
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5594383478164673,
      "learning_rate": 1.4340478548381946e-05,
      "loss": 0.9283,
      "step": 9827
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.594815492630005,
      "learning_rate": 1.4339839631984156e-05,
      "loss": 1.1204,
      "step": 9828
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.752673625946045,
      "learning_rate": 1.4339200715586368e-05,
      "loss": 0.8382,
      "step": 9829
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.168417453765869,
      "learning_rate": 1.4338561799188578e-05,
      "loss": 0.9247,
      "step": 9830
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8727667331695557,
      "learning_rate": 1.433792288279079e-05,
      "loss": 0.8683,
      "step": 9831
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0474281311035156,
      "learning_rate": 1.4337283966392999e-05,
      "loss": 0.8867,
      "step": 9832
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.7235546112060547,
      "learning_rate": 1.433664504999521e-05,
      "loss": 0.9335,
      "step": 9833
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0393242835998535,
      "learning_rate": 1.433600613359742e-05,
      "loss": 1.0778,
      "step": 9834
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4806989431381226,
      "learning_rate": 1.4335367217199632e-05,
      "loss": 0.9149,
      "step": 9835
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.595775842666626,
      "learning_rate": 1.4334728300801842e-05,
      "loss": 0.5815,
      "step": 9836
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8153339624404907,
      "learning_rate": 1.4334089384404053e-05,
      "loss": 1.1586,
      "step": 9837
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.2205538749694824,
      "learning_rate": 1.4333450468006262e-05,
      "loss": 1.1567,
      "step": 9838
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0772297382354736,
      "learning_rate": 1.4332811551608475e-05,
      "loss": 0.914,
      "step": 9839
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.602560520172119,
      "learning_rate": 1.4332172635210683e-05,
      "loss": 0.9544,
      "step": 9840
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7604700326919556,
      "learning_rate": 1.4331533718812896e-05,
      "loss": 0.9023,
      "step": 9841
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1240482330322266,
      "learning_rate": 1.4330894802415104e-05,
      "loss": 0.9846,
      "step": 9842
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0741424560546875,
      "learning_rate": 1.4330255886017318e-05,
      "loss": 0.9327,
      "step": 9843
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1284964084625244,
      "learning_rate": 1.4329616969619526e-05,
      "loss": 0.954,
      "step": 9844
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5936031341552734,
      "learning_rate": 1.4328978053221737e-05,
      "loss": 0.4157,
      "step": 9845
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.262885570526123,
      "learning_rate": 1.4328339136823947e-05,
      "loss": 0.9211,
      "step": 9846
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.292668342590332,
      "learning_rate": 1.4327700220426159e-05,
      "loss": 1.0211,
      "step": 9847
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8187661170959473,
      "learning_rate": 1.4327061304028369e-05,
      "loss": 0.9548,
      "step": 9848
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.8672635555267334,
      "learning_rate": 1.432642238763058e-05,
      "loss": 1.2453,
      "step": 9849
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9210968017578125,
      "learning_rate": 1.432578347123279e-05,
      "loss": 0.8219,
      "step": 9850
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9746311902999878,
      "learning_rate": 1.4325144554835002e-05,
      "loss": 0.9113,
      "step": 9851
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.7567520141601562,
      "learning_rate": 1.4324505638437211e-05,
      "loss": 1.1685,
      "step": 9852
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5957647562026978,
      "learning_rate": 1.4323866722039423e-05,
      "loss": 0.838,
      "step": 9853
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6122076511383057,
      "learning_rate": 1.4323227805641633e-05,
      "loss": 0.8941,
      "step": 9854
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.826300024986267,
      "learning_rate": 1.4322588889243844e-05,
      "loss": 0.8417,
      "step": 9855
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0926461219787598,
      "learning_rate": 1.4321949972846054e-05,
      "loss": 0.977,
      "step": 9856
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0484566688537598,
      "learning_rate": 1.4321311056448266e-05,
      "loss": 1.0042,
      "step": 9857
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.058987855911255,
      "learning_rate": 1.4320672140050476e-05,
      "loss": 1.3508,
      "step": 9858
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1328237056732178,
      "learning_rate": 1.4320033223652686e-05,
      "loss": 0.9828,
      "step": 9859
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.9742774963378906,
      "learning_rate": 1.4319394307254897e-05,
      "loss": 1.0106,
      "step": 9860
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.414710521697998,
      "learning_rate": 1.4318755390857107e-05,
      "loss": 1.1893,
      "step": 9861
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.745996356010437,
      "learning_rate": 1.4318116474459318e-05,
      "loss": 0.955,
      "step": 9862
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7882230281829834,
      "learning_rate": 1.4317477558061528e-05,
      "loss": 0.8344,
      "step": 9863
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.477076768875122,
      "learning_rate": 1.431683864166374e-05,
      "loss": 0.8805,
      "step": 9864
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.602447748184204,
      "learning_rate": 1.431619972526595e-05,
      "loss": 0.9669,
      "step": 9865
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0496034622192383,
      "learning_rate": 1.4315560808868161e-05,
      "loss": 0.9848,
      "step": 9866
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.397923707962036,
      "learning_rate": 1.4314921892470371e-05,
      "loss": 0.3959,
      "step": 9867
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.570204734802246,
      "learning_rate": 1.4314282976072583e-05,
      "loss": 1.0509,
      "step": 9868
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3902525901794434,
      "learning_rate": 1.431364405967479e-05,
      "loss": 0.8856,
      "step": 9869
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.218621253967285,
      "learning_rate": 1.4313005143277004e-05,
      "loss": 1.0898,
      "step": 9870
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5169825553894043,
      "learning_rate": 1.4312366226879212e-05,
      "loss": 0.7883,
      "step": 9871
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.423792839050293,
      "learning_rate": 1.4311727310481425e-05,
      "loss": 1.2635,
      "step": 9872
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.63627290725708,
      "learning_rate": 1.4311088394083634e-05,
      "loss": 0.8049,
      "step": 9873
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8873698711395264,
      "learning_rate": 1.4310449477685847e-05,
      "loss": 0.978,
      "step": 9874
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9597200155258179,
      "learning_rate": 1.4309810561288055e-05,
      "loss": 1.01,
      "step": 9875
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1178553104400635,
      "learning_rate": 1.4309171644890267e-05,
      "loss": 1.286,
      "step": 9876
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.85663104057312,
      "learning_rate": 1.4308532728492476e-05,
      "loss": 0.9411,
      "step": 9877
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8592432737350464,
      "learning_rate": 1.4307893812094688e-05,
      "loss": 1.2116,
      "step": 9878
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.873733401298523,
      "learning_rate": 1.4307254895696898e-05,
      "loss": 0.9342,
      "step": 9879
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8304977416992188,
      "learning_rate": 1.430661597929911e-05,
      "loss": 0.8132,
      "step": 9880
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.733483076095581,
      "learning_rate": 1.430597706290132e-05,
      "loss": 0.8149,
      "step": 9881
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.6283345222473145,
      "learning_rate": 1.430533814650353e-05,
      "loss": 1.0145,
      "step": 9882
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8065598011016846,
      "learning_rate": 1.430469923010574e-05,
      "loss": 0.8799,
      "step": 9883
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.237661361694336,
      "learning_rate": 1.4304060313707952e-05,
      "loss": 0.9381,
      "step": 9884
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.468691825866699,
      "learning_rate": 1.4303421397310162e-05,
      "loss": 1.0011,
      "step": 9885
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1379733085632324,
      "learning_rate": 1.4302782480912374e-05,
      "loss": 0.8344,
      "step": 9886
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5906528234481812,
      "learning_rate": 1.4302143564514583e-05,
      "loss": 0.8723,
      "step": 9887
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0119237899780273,
      "learning_rate": 1.4301504648116795e-05,
      "loss": 1.0131,
      "step": 9888
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7948087453842163,
      "learning_rate": 1.4300865731719005e-05,
      "loss": 0.8852,
      "step": 9889
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.800915479660034,
      "learning_rate": 1.4300226815321216e-05,
      "loss": 0.7874,
      "step": 9890
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0591437816619873,
      "learning_rate": 1.4299587898923426e-05,
      "loss": 1.0883,
      "step": 9891
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.043508529663086,
      "learning_rate": 1.4298948982525638e-05,
      "loss": 1.2172,
      "step": 9892
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7560930252075195,
      "learning_rate": 1.4298310066127848e-05,
      "loss": 0.9429,
      "step": 9893
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6576409339904785,
      "learning_rate": 1.429767114973006e-05,
      "loss": 0.8473,
      "step": 9894
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0748231410980225,
      "learning_rate": 1.4297032233332269e-05,
      "loss": 1.0365,
      "step": 9895
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5750703811645508,
      "learning_rate": 1.429639331693448e-05,
      "loss": 0.8548,
      "step": 9896
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4617958068847656,
      "learning_rate": 1.429575440053669e-05,
      "loss": 0.4824,
      "step": 9897
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6654735803604126,
      "learning_rate": 1.4295115484138902e-05,
      "loss": 0.9035,
      "step": 9898
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.707832098007202,
      "learning_rate": 1.4294476567741112e-05,
      "loss": 0.9678,
      "step": 9899
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9062161445617676,
      "learning_rate": 1.4293837651343323e-05,
      "loss": 0.8958,
      "step": 9900
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7682843208312988,
      "learning_rate": 1.4293198734945533e-05,
      "loss": 0.903,
      "step": 9901
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9327090978622437,
      "learning_rate": 1.4292559818547745e-05,
      "loss": 0.9694,
      "step": 9902
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4548776149749756,
      "learning_rate": 1.4291920902149955e-05,
      "loss": 1.3968,
      "step": 9903
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5128881931304932,
      "learning_rate": 1.4291281985752166e-05,
      "loss": 0.9071,
      "step": 9904
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9623914957046509,
      "learning_rate": 1.4290643069354376e-05,
      "loss": 1.098,
      "step": 9905
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.063822031021118,
      "learning_rate": 1.4290004152956588e-05,
      "loss": 0.9629,
      "step": 9906
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.078508138656616,
      "learning_rate": 1.4289365236558796e-05,
      "loss": 0.9222,
      "step": 9907
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.8401453495025635,
      "learning_rate": 1.4288726320161009e-05,
      "loss": 0.6449,
      "step": 9908
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.563340663909912,
      "learning_rate": 1.4288087403763217e-05,
      "loss": 0.873,
      "step": 9909
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7055394649505615,
      "learning_rate": 1.428744848736543e-05,
      "loss": 1.0126,
      "step": 9910
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7628575563430786,
      "learning_rate": 1.4286809570967639e-05,
      "loss": 0.9026,
      "step": 9911
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0841381549835205,
      "learning_rate": 1.428617065456985e-05,
      "loss": 1.1495,
      "step": 9912
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5233047008514404,
      "learning_rate": 1.428553173817206e-05,
      "loss": 0.3855,
      "step": 9913
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.422579050064087,
      "learning_rate": 1.4284892821774271e-05,
      "loss": 0.7657,
      "step": 9914
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.437311053276062,
      "learning_rate": 1.4284253905376481e-05,
      "loss": 0.8167,
      "step": 9915
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.79635488986969,
      "learning_rate": 1.4283614988978693e-05,
      "loss": 1.1424,
      "step": 9916
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.409451961517334,
      "learning_rate": 1.4282976072580903e-05,
      "loss": 0.8036,
      "step": 9917
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7274441719055176,
      "learning_rate": 1.4282337156183114e-05,
      "loss": 0.9209,
      "step": 9918
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.909790515899658,
      "learning_rate": 1.4281698239785324e-05,
      "loss": 1.1699,
      "step": 9919
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7125324010849,
      "learning_rate": 1.4281059323387536e-05,
      "loss": 0.9171,
      "step": 9920
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.781644344329834,
      "learning_rate": 1.4280420406989746e-05,
      "loss": 0.8572,
      "step": 9921
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9786152839660645,
      "learning_rate": 1.4279781490591957e-05,
      "loss": 0.9342,
      "step": 9922
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1112027168273926,
      "learning_rate": 1.4279142574194167e-05,
      "loss": 0.9795,
      "step": 9923
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.371670722961426,
      "learning_rate": 1.4278503657796378e-05,
      "loss": 0.7876,
      "step": 9924
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7048170566558838,
      "learning_rate": 1.4277864741398588e-05,
      "loss": 0.8612,
      "step": 9925
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8563636541366577,
      "learning_rate": 1.42772258250008e-05,
      "loss": 0.8458,
      "step": 9926
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5749948024749756,
      "learning_rate": 1.427658690860301e-05,
      "loss": 0.8403,
      "step": 9927
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.016794204711914,
      "learning_rate": 1.4275947992205221e-05,
      "loss": 1.1678,
      "step": 9928
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7712814807891846,
      "learning_rate": 1.4275309075807431e-05,
      "loss": 0.9073,
      "step": 9929
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.5570523738861084,
      "learning_rate": 1.4274670159409643e-05,
      "loss": 0.9698,
      "step": 9930
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.199373722076416,
      "learning_rate": 1.4274031243011853e-05,
      "loss": 1.0743,
      "step": 9931
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1007144451141357,
      "learning_rate": 1.4273392326614064e-05,
      "loss": 0.8557,
      "step": 9932
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6303045749664307,
      "learning_rate": 1.4272753410216274e-05,
      "loss": 0.8794,
      "step": 9933
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.4337031841278076,
      "learning_rate": 1.4272114493818485e-05,
      "loss": 0.9659,
      "step": 9934
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5678372383117676,
      "learning_rate": 1.4271475577420695e-05,
      "loss": 0.8479,
      "step": 9935
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0166029930114746,
      "learning_rate": 1.4270836661022907e-05,
      "loss": 1.2304,
      "step": 9936
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5767220258712769,
      "learning_rate": 1.4270197744625117e-05,
      "loss": 0.8755,
      "step": 9937
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0005381107330322,
      "learning_rate": 1.4269558828227328e-05,
      "loss": 0.5773,
      "step": 9938
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7595609426498413,
      "learning_rate": 1.4268919911829538e-05,
      "loss": 0.8475,
      "step": 9939
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7980812788009644,
      "learning_rate": 1.426828099543175e-05,
      "loss": 0.8001,
      "step": 9940
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.762535810470581,
      "learning_rate": 1.426764207903396e-05,
      "loss": 0.9413,
      "step": 9941
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.702725648880005,
      "learning_rate": 1.4267003162636171e-05,
      "loss": 0.9911,
      "step": 9942
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8779476881027222,
      "learning_rate": 1.426636424623838e-05,
      "loss": 0.9898,
      "step": 9943
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8328697681427002,
      "learning_rate": 1.4265725329840592e-05,
      "loss": 0.8451,
      "step": 9944
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.247606039047241,
      "learning_rate": 1.42650864134428e-05,
      "loss": 1.0473,
      "step": 9945
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7676984071731567,
      "learning_rate": 1.4264447497045014e-05,
      "loss": 0.9146,
      "step": 9946
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.252305746078491,
      "learning_rate": 1.4263808580647222e-05,
      "loss": 1.052,
      "step": 9947
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5021766424179077,
      "learning_rate": 1.4263169664249434e-05,
      "loss": 0.959,
      "step": 9948
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.972788691520691,
      "learning_rate": 1.4262530747851643e-05,
      "loss": 0.896,
      "step": 9949
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.337329864501953,
      "learning_rate": 1.4261891831453855e-05,
      "loss": 1.0462,
      "step": 9950
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8064674139022827,
      "learning_rate": 1.4261252915056065e-05,
      "loss": 0.9131,
      "step": 9951
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0575644969940186,
      "learning_rate": 1.4260613998658276e-05,
      "loss": 0.9669,
      "step": 9952
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2704391479492188,
      "learning_rate": 1.4259975082260486e-05,
      "loss": 1.0494,
      "step": 9953
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0957303047180176,
      "learning_rate": 1.4259336165862698e-05,
      "loss": 1.1221,
      "step": 9954
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.703325271606445,
      "learning_rate": 1.4258697249464908e-05,
      "loss": 1.3472,
      "step": 9955
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.713658332824707,
      "learning_rate": 1.425805833306712e-05,
      "loss": 0.9728,
      "step": 9956
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7224303483963013,
      "learning_rate": 1.4257419416669329e-05,
      "loss": 0.8324,
      "step": 9957
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.856931209564209,
      "learning_rate": 1.425678050027154e-05,
      "loss": 0.9457,
      "step": 9958
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1079888343811035,
      "learning_rate": 1.425614158387375e-05,
      "loss": 1.008,
      "step": 9959
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0502548217773438,
      "learning_rate": 1.4255502667475962e-05,
      "loss": 0.7768,
      "step": 9960
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.874629259109497,
      "learning_rate": 1.4254863751078172e-05,
      "loss": 0.9775,
      "step": 9961
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5649504661560059,
      "learning_rate": 1.4254224834680383e-05,
      "loss": 0.9031,
      "step": 9962
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.378082752227783,
      "learning_rate": 1.4253585918282593e-05,
      "loss": 1.1248,
      "step": 9963
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.022350549697876,
      "learning_rate": 1.4252947001884805e-05,
      "loss": 1.0104,
      "step": 9964
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1494836807250977,
      "learning_rate": 1.4252308085487015e-05,
      "loss": 1.0029,
      "step": 9965
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5771198272705078,
      "learning_rate": 1.4251669169089226e-05,
      "loss": 0.8108,
      "step": 9966
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3591232299804688,
      "learning_rate": 1.4251030252691436e-05,
      "loss": 0.5414,
      "step": 9967
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1702802181243896,
      "learning_rate": 1.4250391336293648e-05,
      "loss": 0.9928,
      "step": 9968
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7800517082214355,
      "learning_rate": 1.4249752419895857e-05,
      "loss": 0.9449,
      "step": 9969
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.4281153678894043,
      "learning_rate": 1.4249113503498069e-05,
      "loss": 0.9154,
      "step": 9970
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0324935913085938,
      "learning_rate": 1.4248474587100279e-05,
      "loss": 1.1486,
      "step": 9971
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9353127479553223,
      "learning_rate": 1.424783567070249e-05,
      "loss": 0.8431,
      "step": 9972
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0100324153900146,
      "learning_rate": 1.42471967543047e-05,
      "loss": 1.0127,
      "step": 9973
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.658812999725342,
      "learning_rate": 1.4246557837906912e-05,
      "loss": 0.9285,
      "step": 9974
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.644672155380249,
      "learning_rate": 1.4245918921509122e-05,
      "loss": 1.0619,
      "step": 9975
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.120002269744873,
      "learning_rate": 1.4245280005111333e-05,
      "loss": 1.4088,
      "step": 9976
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.202677011489868,
      "learning_rate": 1.4244641088713543e-05,
      "loss": 0.9079,
      "step": 9977
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0564730167388916,
      "learning_rate": 1.4244002172315755e-05,
      "loss": 1.0019,
      "step": 9978
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.072999954223633,
      "learning_rate": 1.4243363255917963e-05,
      "loss": 0.9313,
      "step": 9979
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7421338558197021,
      "learning_rate": 1.4242724339520176e-05,
      "loss": 0.4591,
      "step": 9980
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.123324155807495,
      "learning_rate": 1.4242085423122384e-05,
      "loss": 0.8944,
      "step": 9981
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5799543857574463,
      "learning_rate": 1.4241446506724597e-05,
      "loss": 1.0505,
      "step": 9982
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6407674551010132,
      "learning_rate": 1.4240807590326806e-05,
      "loss": 0.9216,
      "step": 9983
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.970503330230713,
      "learning_rate": 1.4240168673929017e-05,
      "loss": 1.2616,
      "step": 9984
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.017305612564087,
      "learning_rate": 1.4239529757531227e-05,
      "loss": 0.8935,
      "step": 9985
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3451104164123535,
      "learning_rate": 1.4238890841133438e-05,
      "loss": 1.267,
      "step": 9986
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5676316022872925,
      "learning_rate": 1.4238251924735648e-05,
      "loss": 0.9562,
      "step": 9987
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.204711675643921,
      "learning_rate": 1.423761300833786e-05,
      "loss": 0.8476,
      "step": 9988
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9703541994094849,
      "learning_rate": 1.423697409194007e-05,
      "loss": 1.224,
      "step": 9989
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7068620920181274,
      "learning_rate": 1.4236335175542281e-05,
      "loss": 0.9214,
      "step": 9990
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0162692070007324,
      "learning_rate": 1.4235696259144491e-05,
      "loss": 0.9912,
      "step": 9991
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7501486539840698,
      "learning_rate": 1.4235057342746703e-05,
      "loss": 0.9225,
      "step": 9992
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.980907917022705,
      "learning_rate": 1.4234418426348913e-05,
      "loss": 1.2015,
      "step": 9993
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.069812059402466,
      "learning_rate": 1.4233779509951124e-05,
      "loss": 0.9173,
      "step": 9994
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8976565599441528,
      "learning_rate": 1.4233140593553334e-05,
      "loss": 0.9385,
      "step": 9995
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8363796472549438,
      "learning_rate": 1.4232501677155545e-05,
      "loss": 0.9359,
      "step": 9996
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8878796100616455,
      "learning_rate": 1.4231862760757755e-05,
      "loss": 0.8953,
      "step": 9997
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7501814365386963,
      "learning_rate": 1.4231223844359967e-05,
      "loss": 1.0124,
      "step": 9998
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9023489952087402,
      "learning_rate": 1.4230584927962177e-05,
      "loss": 0.9586,
      "step": 9999
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.228534460067749,
      "learning_rate": 1.4229946011564388e-05,
      "loss": 0.8294,
      "step": 10000
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4689722061157227,
      "learning_rate": 1.4229307095166598e-05,
      "loss": 0.997,
      "step": 10001
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9458386898040771,
      "learning_rate": 1.422866817876881e-05,
      "loss": 0.8232,
      "step": 10002
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4006829261779785,
      "learning_rate": 1.422802926237102e-05,
      "loss": 0.9351,
      "step": 10003
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7315698862075806,
      "learning_rate": 1.4227390345973231e-05,
      "loss": 0.8215,
      "step": 10004
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.489503860473633,
      "learning_rate": 1.4226751429575441e-05,
      "loss": 0.9181,
      "step": 10005
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9042972326278687,
      "learning_rate": 1.4226112513177652e-05,
      "loss": 0.9142,
      "step": 10006
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.111321449279785,
      "learning_rate": 1.4225473596779862e-05,
      "loss": 0.9281,
      "step": 10007
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1515421867370605,
      "learning_rate": 1.4224834680382074e-05,
      "loss": 0.9887,
      "step": 10008
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9152321815490723,
      "learning_rate": 1.4224195763984284e-05,
      "loss": 0.9666,
      "step": 10009
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.2452573776245117,
      "learning_rate": 1.4223556847586495e-05,
      "loss": 1.1204,
      "step": 10010
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.208008050918579,
      "learning_rate": 1.4222917931188705e-05,
      "loss": 0.9508,
      "step": 10011
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8646337985992432,
      "learning_rate": 1.4222279014790917e-05,
      "loss": 0.856,
      "step": 10012
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.16745662689209,
      "learning_rate": 1.4221640098393127e-05,
      "loss": 0.6683,
      "step": 10013
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.018596649169922,
      "learning_rate": 1.4221001181995338e-05,
      "loss": 0.9788,
      "step": 10014
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3716768026351929,
      "learning_rate": 1.4220362265597546e-05,
      "loss": 0.7651,
      "step": 10015
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.558870792388916,
      "learning_rate": 1.421972334919976e-05,
      "loss": 1.4519,
      "step": 10016
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8261646032333374,
      "learning_rate": 1.4219084432801968e-05,
      "loss": 0.9499,
      "step": 10017
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.694777250289917,
      "learning_rate": 1.4218445516404181e-05,
      "loss": 0.9338,
      "step": 10018
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8896681070327759,
      "learning_rate": 1.4217806600006389e-05,
      "loss": 0.9704,
      "step": 10019
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.7060558795928955,
      "learning_rate": 1.4217167683608602e-05,
      "loss": 1.2122,
      "step": 10020
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8729053735733032,
      "learning_rate": 1.421652876721081e-05,
      "loss": 0.8048,
      "step": 10021
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.847902774810791,
      "learning_rate": 1.4215889850813022e-05,
      "loss": 1.1952,
      "step": 10022
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1547257900238037,
      "learning_rate": 1.4215250934415232e-05,
      "loss": 0.9775,
      "step": 10023
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5776616334915161,
      "learning_rate": 1.4214612018017443e-05,
      "loss": 0.94,
      "step": 10024
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.893059730529785,
      "learning_rate": 1.4213973101619653e-05,
      "loss": 1.0902,
      "step": 10025
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1287057399749756,
      "learning_rate": 1.4213334185221865e-05,
      "loss": 1.3131,
      "step": 10026
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.104688882827759,
      "learning_rate": 1.4212695268824075e-05,
      "loss": 0.8158,
      "step": 10027
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.148284912109375,
      "learning_rate": 1.4212056352426286e-05,
      "loss": 1.548,
      "step": 10028
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.4530832767486572,
      "learning_rate": 1.4211417436028496e-05,
      "loss": 0.5879,
      "step": 10029
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.099708557128906,
      "learning_rate": 1.4210778519630708e-05,
      "loss": 1.4341,
      "step": 10030
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0610949993133545,
      "learning_rate": 1.4210139603232917e-05,
      "loss": 0.9028,
      "step": 10031
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7522506713867188,
      "learning_rate": 1.4209500686835129e-05,
      "loss": 0.9067,
      "step": 10032
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.208037376403809,
      "learning_rate": 1.4208861770437339e-05,
      "loss": 0.7895,
      "step": 10033
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6517622470855713,
      "learning_rate": 1.420822285403955e-05,
      "loss": 0.9135,
      "step": 10034
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3693578243255615,
      "learning_rate": 1.420758393764176e-05,
      "loss": 0.7782,
      "step": 10035
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9675936698913574,
      "learning_rate": 1.4206945021243972e-05,
      "loss": 0.9479,
      "step": 10036
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4382950067520142,
      "learning_rate": 1.4206306104846182e-05,
      "loss": 0.7588,
      "step": 10037
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8424581289291382,
      "learning_rate": 1.4205667188448393e-05,
      "loss": 0.8465,
      "step": 10038
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4117693901062012,
      "learning_rate": 1.4205028272050603e-05,
      "loss": 0.7609,
      "step": 10039
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7588248252868652,
      "learning_rate": 1.4204389355652815e-05,
      "loss": 0.8158,
      "step": 10040
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3965566158294678,
      "learning_rate": 1.4203750439255024e-05,
      "loss": 0.9177,
      "step": 10041
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3560205698013306,
      "learning_rate": 1.4203111522857236e-05,
      "loss": 0.765,
      "step": 10042
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2123868465423584,
      "learning_rate": 1.4202472606459446e-05,
      "loss": 1.0583,
      "step": 10043
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.740175724029541,
      "learning_rate": 1.4201833690061657e-05,
      "loss": 0.8111,
      "step": 10044
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6161904335021973,
      "learning_rate": 1.4201194773663867e-05,
      "loss": 0.8694,
      "step": 10045
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2379393577575684,
      "learning_rate": 1.4200555857266079e-05,
      "loss": 0.974,
      "step": 10046
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.256174087524414,
      "learning_rate": 1.4199916940868289e-05,
      "loss": 0.9994,
      "step": 10047
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6577848196029663,
      "learning_rate": 1.41992780244705e-05,
      "loss": 0.4576,
      "step": 10048
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6223808526992798,
      "learning_rate": 1.419863910807271e-05,
      "loss": 0.9467,
      "step": 10049
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5852142572402954,
      "learning_rate": 1.4198000191674922e-05,
      "loss": 0.9584,
      "step": 10050
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.247064590454102,
      "learning_rate": 1.4197361275277131e-05,
      "loss": 1.9041,
      "step": 10051
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.024294376373291,
      "learning_rate": 1.4196722358879343e-05,
      "loss": 1.2683,
      "step": 10052
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7625292539596558,
      "learning_rate": 1.4196083442481551e-05,
      "loss": 0.9181,
      "step": 10053
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0953962802886963,
      "learning_rate": 1.4195444526083764e-05,
      "loss": 1.0043,
      "step": 10054
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.700663447380066,
      "learning_rate": 1.4194805609685973e-05,
      "loss": 0.9294,
      "step": 10055
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7911397218704224,
      "learning_rate": 1.4194166693288186e-05,
      "loss": 0.8685,
      "step": 10056
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.635168194770813,
      "learning_rate": 1.4193527776890394e-05,
      "loss": 0.9973,
      "step": 10057
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.6279735565185547,
      "learning_rate": 1.4192888860492605e-05,
      "loss": 0.8464,
      "step": 10058
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.971537470817566,
      "learning_rate": 1.4192249944094815e-05,
      "loss": 1.0674,
      "step": 10059
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.562016725540161,
      "learning_rate": 1.4191611027697027e-05,
      "loss": 0.8957,
      "step": 10060
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.324830055236816,
      "learning_rate": 1.4190972111299237e-05,
      "loss": 1.3485,
      "step": 10061
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9640659093856812,
      "learning_rate": 1.4190333194901448e-05,
      "loss": 1.029,
      "step": 10062
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.98763370513916,
      "learning_rate": 1.4189694278503658e-05,
      "loss": 1.8505,
      "step": 10063
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6647461652755737,
      "learning_rate": 1.418905536210587e-05,
      "loss": 0.8404,
      "step": 10064
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7922557592391968,
      "learning_rate": 1.418841644570808e-05,
      "loss": 0.8267,
      "step": 10065
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6835631132125854,
      "learning_rate": 1.4187777529310291e-05,
      "loss": 0.8705,
      "step": 10066
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4930219650268555,
      "learning_rate": 1.4187138612912501e-05,
      "loss": 0.8547,
      "step": 10067
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.790405035018921,
      "learning_rate": 1.4186499696514712e-05,
      "loss": 0.9467,
      "step": 10068
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.579658031463623,
      "learning_rate": 1.4185860780116922e-05,
      "loss": 0.9111,
      "step": 10069
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3563790321350098,
      "learning_rate": 1.4185221863719134e-05,
      "loss": 0.8847,
      "step": 10070
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6989915370941162,
      "learning_rate": 1.4184582947321344e-05,
      "loss": 0.8705,
      "step": 10071
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9535738229751587,
      "learning_rate": 1.4183944030923555e-05,
      "loss": 1.3035,
      "step": 10072
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.541932463645935,
      "learning_rate": 1.4183305114525765e-05,
      "loss": 0.7982,
      "step": 10073
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.4239659309387207,
      "learning_rate": 1.4182666198127977e-05,
      "loss": 1.1809,
      "step": 10074
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5114338397979736,
      "learning_rate": 1.4182027281730187e-05,
      "loss": 1.2368,
      "step": 10075
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2676539421081543,
      "learning_rate": 1.4181388365332398e-05,
      "loss": 0.8907,
      "step": 10076
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5798308849334717,
      "learning_rate": 1.4180749448934608e-05,
      "loss": 0.916,
      "step": 10077
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1050586700439453,
      "learning_rate": 1.418011053253682e-05,
      "loss": 0.9939,
      "step": 10078
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.320805549621582,
      "learning_rate": 1.417947161613903e-05,
      "loss": 1.0394,
      "step": 10079
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.174942970275879,
      "learning_rate": 1.4178832699741241e-05,
      "loss": 1.368,
      "step": 10080
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6559182405471802,
      "learning_rate": 1.417819378334345e-05,
      "loss": 0.9086,
      "step": 10081
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.030703544616699,
      "learning_rate": 1.4177554866945662e-05,
      "loss": 1.2279,
      "step": 10082
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.016117572784424,
      "learning_rate": 1.4176915950547872e-05,
      "loss": 0.9486,
      "step": 10083
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3737411499023438,
      "learning_rate": 1.4176277034150084e-05,
      "loss": 0.9874,
      "step": 10084
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6885157823562622,
      "learning_rate": 1.4175638117752294e-05,
      "loss": 0.8958,
      "step": 10085
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0457875728607178,
      "learning_rate": 1.4174999201354505e-05,
      "loss": 0.9359,
      "step": 10086
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9399594068527222,
      "learning_rate": 1.4174360284956715e-05,
      "loss": 1.1271,
      "step": 10087
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3182156085968018,
      "learning_rate": 1.4173721368558926e-05,
      "loss": 1.0294,
      "step": 10088
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8066173791885376,
      "learning_rate": 1.4173082452161135e-05,
      "loss": 0.865,
      "step": 10089
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2496206760406494,
      "learning_rate": 1.4172443535763348e-05,
      "loss": 0.9654,
      "step": 10090
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.116414785385132,
      "learning_rate": 1.4171804619365556e-05,
      "loss": 0.9872,
      "step": 10091
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.534878730773926,
      "learning_rate": 1.417116570296777e-05,
      "loss": 1.3572,
      "step": 10092
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.841798186302185,
      "learning_rate": 1.4170526786569977e-05,
      "loss": 0.9455,
      "step": 10093
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2653002738952637,
      "learning_rate": 1.4169887870172189e-05,
      "loss": 1.0127,
      "step": 10094
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7141611576080322,
      "learning_rate": 1.4169248953774399e-05,
      "loss": 0.9324,
      "step": 10095
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7169439792633057,
      "learning_rate": 1.416861003737661e-05,
      "loss": 0.9851,
      "step": 10096
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.164926528930664,
      "learning_rate": 1.416797112097882e-05,
      "loss": 0.8321,
      "step": 10097
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5721579790115356,
      "learning_rate": 1.4167332204581032e-05,
      "loss": 1.0138,
      "step": 10098
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.230963706970215,
      "learning_rate": 1.4166693288183242e-05,
      "loss": 1.1174,
      "step": 10099
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4413048028945923,
      "learning_rate": 1.4166054371785453e-05,
      "loss": 0.9158,
      "step": 10100
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9816111326217651,
      "learning_rate": 1.4165415455387663e-05,
      "loss": 1.0349,
      "step": 10101
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.762782096862793,
      "learning_rate": 1.4164776538989875e-05,
      "loss": 1.3201,
      "step": 10102
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0831191539764404,
      "learning_rate": 1.4164137622592084e-05,
      "loss": 0.9566,
      "step": 10103
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9479128122329712,
      "learning_rate": 1.4163498706194296e-05,
      "loss": 0.8467,
      "step": 10104
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2832696437835693,
      "learning_rate": 1.4162859789796506e-05,
      "loss": 1.3127,
      "step": 10105
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8717069625854492,
      "learning_rate": 1.4162220873398717e-05,
      "loss": 0.931,
      "step": 10106
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9858664274215698,
      "learning_rate": 1.4161581957000927e-05,
      "loss": 0.9304,
      "step": 10107
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.5986204147338867,
      "learning_rate": 1.4160943040603139e-05,
      "loss": 0.9815,
      "step": 10108
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.566482663154602,
      "learning_rate": 1.4160304124205349e-05,
      "loss": 0.8404,
      "step": 10109
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.5094523429870605,
      "learning_rate": 1.415966520780756e-05,
      "loss": 1.1597,
      "step": 10110
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4888408184051514,
      "learning_rate": 1.415902629140977e-05,
      "loss": 1.1193,
      "step": 10111
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.147840738296509,
      "learning_rate": 1.4158387375011982e-05,
      "loss": 1.0214,
      "step": 10112
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5192601680755615,
      "learning_rate": 1.4157748458614191e-05,
      "loss": 0.4499,
      "step": 10113
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8971364498138428,
      "learning_rate": 1.4157109542216403e-05,
      "loss": 0.8708,
      "step": 10114
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6245039701461792,
      "learning_rate": 1.4156470625818613e-05,
      "loss": 1.0249,
      "step": 10115
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5610146522521973,
      "learning_rate": 1.4155831709420824e-05,
      "loss": 0.882,
      "step": 10116
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.718643069267273,
      "learning_rate": 1.4155192793023034e-05,
      "loss": 1.0232,
      "step": 10117
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5435104370117188,
      "learning_rate": 1.4154553876625246e-05,
      "loss": 0.8276,
      "step": 10118
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.558139443397522,
      "learning_rate": 1.4153914960227456e-05,
      "loss": 0.396,
      "step": 10119
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6863248348236084,
      "learning_rate": 1.4153276043829667e-05,
      "loss": 0.9326,
      "step": 10120
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.80594801902771,
      "learning_rate": 1.4152637127431877e-05,
      "loss": 0.9004,
      "step": 10121
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.179840326309204,
      "learning_rate": 1.4151998211034089e-05,
      "loss": 0.9741,
      "step": 10122
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3126533031463623,
      "learning_rate": 1.4151359294636298e-05,
      "loss": 1.0656,
      "step": 10123
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5932117700576782,
      "learning_rate": 1.415072037823851e-05,
      "loss": 0.8071,
      "step": 10124
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3550729751586914,
      "learning_rate": 1.4150081461840718e-05,
      "loss": 0.7567,
      "step": 10125
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.019106388092041,
      "learning_rate": 1.4149442545442931e-05,
      "loss": 1.0506,
      "step": 10126
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.664316177368164,
      "learning_rate": 1.414880362904514e-05,
      "loss": 1.0603,
      "step": 10127
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3840641975402832,
      "learning_rate": 1.4148164712647353e-05,
      "loss": 0.8235,
      "step": 10128
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.353097915649414,
      "learning_rate": 1.4147525796249561e-05,
      "loss": 0.9551,
      "step": 10129
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7706676721572876,
      "learning_rate": 1.4146886879851772e-05,
      "loss": 0.918,
      "step": 10130
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0828633308410645,
      "learning_rate": 1.4146247963453982e-05,
      "loss": 1.0319,
      "step": 10131
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.822155475616455,
      "learning_rate": 1.4145609047056194e-05,
      "loss": 1.1917,
      "step": 10132
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.318507671356201,
      "learning_rate": 1.4144970130658404e-05,
      "loss": 0.8235,
      "step": 10133
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1957433223724365,
      "learning_rate": 1.4144331214260615e-05,
      "loss": 1.0741,
      "step": 10134
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.6767125129699707,
      "learning_rate": 1.4143692297862825e-05,
      "loss": 0.5555,
      "step": 10135
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5289695262908936,
      "learning_rate": 1.4143053381465037e-05,
      "loss": 0.8471,
      "step": 10136
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7776219844818115,
      "learning_rate": 1.4142414465067247e-05,
      "loss": 0.9951,
      "step": 10137
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.062706470489502,
      "learning_rate": 1.4141775548669458e-05,
      "loss": 0.9716,
      "step": 10138
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.4211442470550537,
      "learning_rate": 1.4141136632271668e-05,
      "loss": 0.9801,
      "step": 10139
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.442729473114014,
      "learning_rate": 1.414049771587388e-05,
      "loss": 1.0699,
      "step": 10140
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.107957124710083,
      "learning_rate": 1.413985879947609e-05,
      "loss": 0.6599,
      "step": 10141
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9005647897720337,
      "learning_rate": 1.4139219883078301e-05,
      "loss": 0.8921,
      "step": 10142
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.047894239425659,
      "learning_rate": 1.413858096668051e-05,
      "loss": 0.8102,
      "step": 10143
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.122030019760132,
      "learning_rate": 1.4137942050282722e-05,
      "loss": 0.9949,
      "step": 10144
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0131916999816895,
      "learning_rate": 1.4137303133884932e-05,
      "loss": 0.9173,
      "step": 10145
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.029236316680908,
      "learning_rate": 1.4136664217487144e-05,
      "loss": 1.4103,
      "step": 10146
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5579438209533691,
      "learning_rate": 1.4136025301089354e-05,
      "loss": 0.9074,
      "step": 10147
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.945551872253418,
      "learning_rate": 1.4135386384691565e-05,
      "loss": 0.8437,
      "step": 10148
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.283383846282959,
      "learning_rate": 1.4134747468293775e-05,
      "loss": 0.8531,
      "step": 10149
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.888282060623169,
      "learning_rate": 1.4134108551895986e-05,
      "loss": 1.0515,
      "step": 10150
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9002487659454346,
      "learning_rate": 1.4133469635498196e-05,
      "loss": 0.8442,
      "step": 10151
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7021269798278809,
      "learning_rate": 1.4132830719100408e-05,
      "loss": 0.8524,
      "step": 10152
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7793827056884766,
      "learning_rate": 1.4132191802702618e-05,
      "loss": 0.3864,
      "step": 10153
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9416821002960205,
      "learning_rate": 1.413155288630483e-05,
      "loss": 0.9566,
      "step": 10154
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.687382698059082,
      "learning_rate": 1.4130913969907039e-05,
      "loss": 1.1577,
      "step": 10155
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7466729879379272,
      "learning_rate": 1.413027505350925e-05,
      "loss": 0.8989,
      "step": 10156
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.0094122886657715,
      "learning_rate": 1.412963613711146e-05,
      "loss": 0.794,
      "step": 10157
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7397615909576416,
      "learning_rate": 1.4128997220713672e-05,
      "loss": 0.8568,
      "step": 10158
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7210534811019897,
      "learning_rate": 1.4128358304315882e-05,
      "loss": 0.8067,
      "step": 10159
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.852125406265259,
      "learning_rate": 1.4127719387918093e-05,
      "loss": 1.033,
      "step": 10160
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.218418598175049,
      "learning_rate": 1.4127080471520302e-05,
      "loss": 0.9869,
      "step": 10161
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.153287649154663,
      "learning_rate": 1.4126441555122515e-05,
      "loss": 1.0242,
      "step": 10162
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3972928524017334,
      "learning_rate": 1.4125802638724723e-05,
      "loss": 1.0942,
      "step": 10163
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7713412046432495,
      "learning_rate": 1.4125163722326936e-05,
      "loss": 0.9207,
      "step": 10164
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.126150608062744,
      "learning_rate": 1.4124524805929144e-05,
      "loss": 1.0141,
      "step": 10165
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4928706884384155,
      "learning_rate": 1.4123885889531358e-05,
      "loss": 0.4079,
      "step": 10166
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.250233173370361,
      "learning_rate": 1.4123246973133566e-05,
      "loss": 0.9873,
      "step": 10167
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7196083068847656,
      "learning_rate": 1.4122608056735777e-05,
      "loss": 0.9654,
      "step": 10168
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4189209938049316,
      "learning_rate": 1.4121969140337987e-05,
      "loss": 0.4582,
      "step": 10169
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7969528436660767,
      "learning_rate": 1.4121330223940199e-05,
      "loss": 0.9985,
      "step": 10170
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8855851888656616,
      "learning_rate": 1.4120691307542409e-05,
      "loss": 0.9162,
      "step": 10171
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6714489459991455,
      "learning_rate": 1.412005239114462e-05,
      "loss": 0.9504,
      "step": 10172
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.9003543853759766,
      "learning_rate": 1.411941347474683e-05,
      "loss": 1.0989,
      "step": 10173
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7911229133605957,
      "learning_rate": 1.4118774558349042e-05,
      "loss": 0.8678,
      "step": 10174
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.194007158279419,
      "learning_rate": 1.4118135641951251e-05,
      "loss": 0.9488,
      "step": 10175
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.907588005065918,
      "learning_rate": 1.4117496725553463e-05,
      "loss": 1.2826,
      "step": 10176
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4907859563827515,
      "learning_rate": 1.4116857809155673e-05,
      "loss": 0.4511,
      "step": 10177
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.8811511993408203,
      "learning_rate": 1.4116218892757884e-05,
      "loss": 1.3865,
      "step": 10178
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0347986221313477,
      "learning_rate": 1.4115579976360094e-05,
      "loss": 0.9098,
      "step": 10179
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6639363765716553,
      "learning_rate": 1.4114941059962306e-05,
      "loss": 0.7676,
      "step": 10180
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7277170419692993,
      "learning_rate": 1.4114302143564516e-05,
      "loss": 0.8595,
      "step": 10181
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8859221935272217,
      "learning_rate": 1.4113663227166727e-05,
      "loss": 0.9981,
      "step": 10182
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6592707633972168,
      "learning_rate": 1.4113024310768937e-05,
      "loss": 0.9586,
      "step": 10183
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7985886335372925,
      "learning_rate": 1.4112385394371149e-05,
      "loss": 0.9094,
      "step": 10184
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8835371732711792,
      "learning_rate": 1.4111746477973358e-05,
      "loss": 1.0243,
      "step": 10185
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6495625972747803,
      "learning_rate": 1.411110756157557e-05,
      "loss": 0.9571,
      "step": 10186
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.772687554359436,
      "learning_rate": 1.411046864517778e-05,
      "loss": 0.8881,
      "step": 10187
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.807100772857666,
      "learning_rate": 1.4109829728779991e-05,
      "loss": 1.0111,
      "step": 10188
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1547250747680664,
      "learning_rate": 1.4109190812382201e-05,
      "loss": 0.8448,
      "step": 10189
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4604415893554688,
      "learning_rate": 1.4108551895984413e-05,
      "loss": 0.8259,
      "step": 10190
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.282113790512085,
      "learning_rate": 1.4107912979586623e-05,
      "loss": 0.8099,
      "step": 10191
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.834080696105957,
      "learning_rate": 1.4107274063188834e-05,
      "loss": 1.3298,
      "step": 10192
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7747998237609863,
      "learning_rate": 1.4106635146791044e-05,
      "loss": 0.9417,
      "step": 10193
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0069243907928467,
      "learning_rate": 1.4105996230393256e-05,
      "loss": 1.1722,
      "step": 10194
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5000654458999634,
      "learning_rate": 1.4105357313995465e-05,
      "loss": 0.9026,
      "step": 10195
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2500805854797363,
      "learning_rate": 1.4104718397597677e-05,
      "loss": 0.9066,
      "step": 10196
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8749043941497803,
      "learning_rate": 1.4104079481199887e-05,
      "loss": 0.9874,
      "step": 10197
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0075795650482178,
      "learning_rate": 1.4103440564802098e-05,
      "loss": 0.929,
      "step": 10198
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6077749729156494,
      "learning_rate": 1.4102801648404307e-05,
      "loss": 0.9219,
      "step": 10199
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9510046243667603,
      "learning_rate": 1.410216273200652e-05,
      "loss": 0.8771,
      "step": 10200
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.586672782897949,
      "learning_rate": 1.4101523815608728e-05,
      "loss": 0.7529,
      "step": 10201
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5854530334472656,
      "learning_rate": 1.4100884899210941e-05,
      "loss": 0.8473,
      "step": 10202
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6768840551376343,
      "learning_rate": 1.410024598281315e-05,
      "loss": 0.905,
      "step": 10203
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4634253978729248,
      "learning_rate": 1.4099607066415361e-05,
      "loss": 0.8859,
      "step": 10204
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.272820472717285,
      "learning_rate": 1.409896815001757e-05,
      "loss": 1.081,
      "step": 10205
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.7938642501831055,
      "learning_rate": 1.4098329233619782e-05,
      "loss": 1.2402,
      "step": 10206
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.318352699279785,
      "learning_rate": 1.4097690317221992e-05,
      "loss": 1.4547,
      "step": 10207
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.132844924926758,
      "learning_rate": 1.4097051400824204e-05,
      "loss": 1.2929,
      "step": 10208
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7666884660720825,
      "learning_rate": 1.4096412484426414e-05,
      "loss": 0.9399,
      "step": 10209
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6036639213562012,
      "learning_rate": 1.4095773568028625e-05,
      "loss": 0.9321,
      "step": 10210
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.773476243019104,
      "learning_rate": 1.4095134651630835e-05,
      "loss": 1.0151,
      "step": 10211
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.561821937561035,
      "learning_rate": 1.4094495735233046e-05,
      "loss": 1.0502,
      "step": 10212
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.636492371559143,
      "learning_rate": 1.4093856818835256e-05,
      "loss": 0.512,
      "step": 10213
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1349010467529297,
      "learning_rate": 1.4093217902437468e-05,
      "loss": 1.1149,
      "step": 10214
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6945232152938843,
      "learning_rate": 1.4092578986039678e-05,
      "loss": 0.9292,
      "step": 10215
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5750184059143066,
      "learning_rate": 1.409194006964189e-05,
      "loss": 0.8597,
      "step": 10216
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7830135822296143,
      "learning_rate": 1.4091301153244099e-05,
      "loss": 0.8272,
      "step": 10217
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5280027389526367,
      "learning_rate": 1.409066223684631e-05,
      "loss": 0.9797,
      "step": 10218
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4561980962753296,
      "learning_rate": 1.409002332044852e-05,
      "loss": 0.8496,
      "step": 10219
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.952813982963562,
      "learning_rate": 1.4089384404050732e-05,
      "loss": 0.8962,
      "step": 10220
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9327822923660278,
      "learning_rate": 1.4088745487652942e-05,
      "loss": 0.8599,
      "step": 10221
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7830235958099365,
      "learning_rate": 1.4088106571255153e-05,
      "loss": 0.8926,
      "step": 10222
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6320971250534058,
      "learning_rate": 1.4087467654857363e-05,
      "loss": 0.3743,
      "step": 10223
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9177228212356567,
      "learning_rate": 1.4086828738459575e-05,
      "loss": 0.8832,
      "step": 10224
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.221663475036621,
      "learning_rate": 1.4086189822061785e-05,
      "loss": 0.9395,
      "step": 10225
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.1285336017608643,
      "learning_rate": 1.4085550905663996e-05,
      "loss": 0.9752,
      "step": 10226
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6044354438781738,
      "learning_rate": 1.4084911989266206e-05,
      "loss": 0.8622,
      "step": 10227
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0686707496643066,
      "learning_rate": 1.4084273072868418e-05,
      "loss": 0.8942,
      "step": 10228
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.079598903656006,
      "learning_rate": 1.4083634156470628e-05,
      "loss": 1.1531,
      "step": 10229
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.00360369682312,
      "learning_rate": 1.4082995240072839e-05,
      "loss": 1.2413,
      "step": 10230
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0213794708251953,
      "learning_rate": 1.4082356323675049e-05,
      "loss": 0.8049,
      "step": 10231
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.001426935195923,
      "learning_rate": 1.408171740727726e-05,
      "loss": 0.8163,
      "step": 10232
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5096495151519775,
      "learning_rate": 1.408107849087947e-05,
      "loss": 0.9439,
      "step": 10233
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4846501350402832,
      "learning_rate": 1.4080439574481682e-05,
      "loss": 0.836,
      "step": 10234
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.706347942352295,
      "learning_rate": 1.407980065808389e-05,
      "loss": 0.9453,
      "step": 10235
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7114464044570923,
      "learning_rate": 1.4079161741686103e-05,
      "loss": 0.9513,
      "step": 10236
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.415055751800537,
      "learning_rate": 1.4078522825288311e-05,
      "loss": 1.0141,
      "step": 10237
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1624867916107178,
      "learning_rate": 1.4077883908890525e-05,
      "loss": 0.8606,
      "step": 10238
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1958091259002686,
      "learning_rate": 1.4077244992492733e-05,
      "loss": 0.8968,
      "step": 10239
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.325911045074463,
      "learning_rate": 1.4076606076094944e-05,
      "loss": 1.0688,
      "step": 10240
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.288008689880371,
      "learning_rate": 1.4075967159697154e-05,
      "loss": 0.9305,
      "step": 10241
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1853487491607666,
      "learning_rate": 1.4075328243299366e-05,
      "loss": 1.2552,
      "step": 10242
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.7459957599639893,
      "learning_rate": 1.4074689326901576e-05,
      "loss": 0.838,
      "step": 10243
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7342500686645508,
      "learning_rate": 1.4074050410503786e-05,
      "loss": 0.9312,
      "step": 10244
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.908945322036743,
      "learning_rate": 1.4073411494105997e-05,
      "loss": 0.959,
      "step": 10245
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.905942440032959,
      "learning_rate": 1.4072772577708207e-05,
      "loss": 0.8027,
      "step": 10246
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1995925903320312,
      "learning_rate": 1.4072133661310418e-05,
      "loss": 0.9439,
      "step": 10247
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1618127822875977,
      "learning_rate": 1.4071494744912628e-05,
      "loss": 0.6414,
      "step": 10248
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3729488849639893,
      "learning_rate": 1.407085582851484e-05,
      "loss": 1.0241,
      "step": 10249
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4724094867706299,
      "learning_rate": 1.407021691211705e-05,
      "loss": 0.5089,
      "step": 10250
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0134427547454834,
      "learning_rate": 1.4069577995719261e-05,
      "loss": 0.9916,
      "step": 10251
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.265026092529297,
      "learning_rate": 1.4068939079321471e-05,
      "loss": 1.0194,
      "step": 10252
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9258484840393066,
      "learning_rate": 1.4068300162923683e-05,
      "loss": 0.962,
      "step": 10253
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7641127109527588,
      "learning_rate": 1.4067661246525893e-05,
      "loss": 0.9983,
      "step": 10254
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6783629655838013,
      "learning_rate": 1.4067022330128104e-05,
      "loss": 0.8775,
      "step": 10255
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.790543556213379,
      "learning_rate": 1.4066383413730314e-05,
      "loss": 0.9222,
      "step": 10256
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.902023434638977,
      "learning_rate": 1.4065744497332525e-05,
      "loss": 0.8178,
      "step": 10257
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5614421367645264,
      "learning_rate": 1.4065105580934735e-05,
      "loss": 1.1484,
      "step": 10258
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6224697828292847,
      "learning_rate": 1.4064466664536947e-05,
      "loss": 0.7933,
      "step": 10259
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9427776336669922,
      "learning_rate": 1.4063827748139157e-05,
      "loss": 0.8737,
      "step": 10260
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0011067390441895,
      "learning_rate": 1.4063188831741368e-05,
      "loss": 1.044,
      "step": 10261
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2213587760925293,
      "learning_rate": 1.4062549915343578e-05,
      "loss": 0.9692,
      "step": 10262
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6188675165176392,
      "learning_rate": 1.406191099894579e-05,
      "loss": 0.8645,
      "step": 10263
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9636285305023193,
      "learning_rate": 1.4061272082548e-05,
      "loss": 0.8021,
      "step": 10264
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6644784212112427,
      "learning_rate": 1.4060633166150211e-05,
      "loss": 0.8921,
      "step": 10265
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7150869369506836,
      "learning_rate": 1.405999424975242e-05,
      "loss": 0.8777,
      "step": 10266
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.552696943283081,
      "learning_rate": 1.4059355333354632e-05,
      "loss": 0.9022,
      "step": 10267
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2340009212493896,
      "learning_rate": 1.405871641695684e-05,
      "loss": 0.9051,
      "step": 10268
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.934996485710144,
      "learning_rate": 1.4058077500559054e-05,
      "loss": 0.9631,
      "step": 10269
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7900938987731934,
      "learning_rate": 1.4057438584161262e-05,
      "loss": 0.9,
      "step": 10270
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9885280132293701,
      "learning_rate": 1.4056799667763474e-05,
      "loss": 0.9296,
      "step": 10271
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.018226623535156,
      "learning_rate": 1.4056160751365683e-05,
      "loss": 1.2119,
      "step": 10272
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.256495475769043,
      "learning_rate": 1.4055521834967895e-05,
      "loss": 1.1694,
      "step": 10273
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8958414793014526,
      "learning_rate": 1.4054882918570105e-05,
      "loss": 0.9249,
      "step": 10274
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7233550548553467,
      "learning_rate": 1.4054244002172316e-05,
      "loss": 0.8767,
      "step": 10275
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0399117469787598,
      "learning_rate": 1.4053605085774526e-05,
      "loss": 0.964,
      "step": 10276
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0573763847351074,
      "learning_rate": 1.4052966169376738e-05,
      "loss": 1.0574,
      "step": 10277
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.9064571857452393,
      "learning_rate": 1.4052327252978948e-05,
      "loss": 0.6072,
      "step": 10278
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.162097692489624,
      "learning_rate": 1.405168833658116e-05,
      "loss": 1.2754,
      "step": 10279
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0420632362365723,
      "learning_rate": 1.4051049420183369e-05,
      "loss": 0.9589,
      "step": 10280
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4754633903503418,
      "learning_rate": 1.405041050378558e-05,
      "loss": 0.5133,
      "step": 10281
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.221977710723877,
      "learning_rate": 1.404977158738779e-05,
      "loss": 1.3805,
      "step": 10282
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.760772466659546,
      "learning_rate": 1.4049132670990002e-05,
      "loss": 0.9327,
      "step": 10283
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2141265869140625,
      "learning_rate": 1.4048493754592212e-05,
      "loss": 0.8893,
      "step": 10284
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.990430474281311,
      "learning_rate": 1.4047854838194423e-05,
      "loss": 0.7998,
      "step": 10285
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.134488821029663,
      "learning_rate": 1.4047215921796633e-05,
      "loss": 1.1856,
      "step": 10286
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6167277097702026,
      "learning_rate": 1.4046577005398845e-05,
      "loss": 0.8904,
      "step": 10287
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.451516628265381,
      "learning_rate": 1.4045938089001055e-05,
      "loss": 1.1502,
      "step": 10288
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.22220516204834,
      "learning_rate": 1.4045299172603266e-05,
      "loss": 1.0419,
      "step": 10289
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9091894626617432,
      "learning_rate": 1.4044660256205476e-05,
      "loss": 0.9549,
      "step": 10290
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7270829677581787,
      "learning_rate": 1.4044021339807688e-05,
      "loss": 0.996,
      "step": 10291
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.321183681488037,
      "learning_rate": 1.4043382423409897e-05,
      "loss": 1.4107,
      "step": 10292
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5616778135299683,
      "learning_rate": 1.4042743507012109e-05,
      "loss": 0.8586,
      "step": 10293
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5731104612350464,
      "learning_rate": 1.4042104590614319e-05,
      "loss": 0.9966,
      "step": 10294
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7132470607757568,
      "learning_rate": 1.404146567421653e-05,
      "loss": 0.3822,
      "step": 10295
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.626523494720459,
      "learning_rate": 1.404082675781874e-05,
      "loss": 1.3904,
      "step": 10296
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1528172492980957,
      "learning_rate": 1.4040187841420952e-05,
      "loss": 1.1044,
      "step": 10297
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8183375597000122,
      "learning_rate": 1.4039548925023162e-05,
      "loss": 0.9378,
      "step": 10298
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6886639595031738,
      "learning_rate": 1.4038910008625373e-05,
      "loss": 1.2115,
      "step": 10299
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.455270290374756,
      "learning_rate": 1.4038271092227583e-05,
      "loss": 0.9742,
      "step": 10300
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8767826557159424,
      "learning_rate": 1.4037632175829795e-05,
      "loss": 0.8798,
      "step": 10301
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.25146770477295,
      "learning_rate": 1.4036993259432003e-05,
      "loss": 1.4814,
      "step": 10302
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.783031702041626,
      "learning_rate": 1.4036354343034216e-05,
      "loss": 0.8315,
      "step": 10303
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3543471097946167,
      "learning_rate": 1.4035715426636424e-05,
      "loss": 0.854,
      "step": 10304
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.519171714782715,
      "learning_rate": 1.4035076510238637e-05,
      "loss": 0.652,
      "step": 10305
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.419135332107544,
      "learning_rate": 1.4034437593840846e-05,
      "loss": 1.0406,
      "step": 10306
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9913872480392456,
      "learning_rate": 1.4033798677443057e-05,
      "loss": 0.976,
      "step": 10307
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5996015071868896,
      "learning_rate": 1.4033159761045267e-05,
      "loss": 0.4,
      "step": 10308
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2037277221679688,
      "learning_rate": 1.4032520844647478e-05,
      "loss": 0.8843,
      "step": 10309
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.997036099433899,
      "learning_rate": 1.4031881928249688e-05,
      "loss": 0.8933,
      "step": 10310
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.870177984237671,
      "learning_rate": 1.40312430118519e-05,
      "loss": 0.9124,
      "step": 10311
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.7691917419433594,
      "learning_rate": 1.403060409545411e-05,
      "loss": 1.181,
      "step": 10312
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.47918176651001,
      "learning_rate": 1.4029965179056321e-05,
      "loss": 1.4516,
      "step": 10313
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5829567909240723,
      "learning_rate": 1.4029326262658531e-05,
      "loss": 0.8289,
      "step": 10314
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.14494514465332,
      "learning_rate": 1.4028687346260743e-05,
      "loss": 0.8934,
      "step": 10315
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.4044837951660156,
      "learning_rate": 1.4028048429862953e-05,
      "loss": 1.0887,
      "step": 10316
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1971943378448486,
      "learning_rate": 1.4027409513465164e-05,
      "loss": 1.0826,
      "step": 10317
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3058359622955322,
      "learning_rate": 1.4026770597067374e-05,
      "loss": 1.0154,
      "step": 10318
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0575551986694336,
      "learning_rate": 1.4026131680669585e-05,
      "loss": 0.9204,
      "step": 10319
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5314266681671143,
      "learning_rate": 1.4025492764271795e-05,
      "loss": 0.6127,
      "step": 10320
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5220720767974854,
      "learning_rate": 1.4024853847874007e-05,
      "loss": 0.9164,
      "step": 10321
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.863572120666504,
      "learning_rate": 1.4024214931476217e-05,
      "loss": 0.7168,
      "step": 10322
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1479852199554443,
      "learning_rate": 1.4023576015078428e-05,
      "loss": 0.938,
      "step": 10323
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4117976427078247,
      "learning_rate": 1.4022937098680638e-05,
      "loss": 0.428,
      "step": 10324
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7881959676742554,
      "learning_rate": 1.402229818228285e-05,
      "loss": 0.9683,
      "step": 10325
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9480880498886108,
      "learning_rate": 1.402165926588506e-05,
      "loss": 1.0563,
      "step": 10326
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8244549036026,
      "learning_rate": 1.4021020349487271e-05,
      "loss": 0.9652,
      "step": 10327
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.273041009902954,
      "learning_rate": 1.4020381433089481e-05,
      "loss": 0.8686,
      "step": 10328
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9118566513061523,
      "learning_rate": 1.4019742516691692e-05,
      "loss": 0.9776,
      "step": 10329
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0016939640045166,
      "learning_rate": 1.4019103600293902e-05,
      "loss": 0.9519,
      "step": 10330
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6405950784683228,
      "learning_rate": 1.4018464683896114e-05,
      "loss": 0.8069,
      "step": 10331
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9411495923995972,
      "learning_rate": 1.4017825767498324e-05,
      "loss": 0.9452,
      "step": 10332
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.127468585968018,
      "learning_rate": 1.4017186851100535e-05,
      "loss": 1.3094,
      "step": 10333
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8516216278076172,
      "learning_rate": 1.4016547934702745e-05,
      "loss": 1.2465,
      "step": 10334
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3082284927368164,
      "learning_rate": 1.4015909018304957e-05,
      "loss": 0.9744,
      "step": 10335
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6842395067214966,
      "learning_rate": 1.4015270101907167e-05,
      "loss": 0.3389,
      "step": 10336
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.885157823562622,
      "learning_rate": 1.4014631185509378e-05,
      "loss": 1.0222,
      "step": 10337
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8186452388763428,
      "learning_rate": 1.4013992269111586e-05,
      "loss": 0.8169,
      "step": 10338
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9233534336090088,
      "learning_rate": 1.40133533527138e-05,
      "loss": 0.8945,
      "step": 10339
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7099896669387817,
      "learning_rate": 1.4012714436316008e-05,
      "loss": 0.4628,
      "step": 10340
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.060879230499268,
      "learning_rate": 1.4012075519918221e-05,
      "loss": 1.2506,
      "step": 10341
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.855584979057312,
      "learning_rate": 1.4011436603520429e-05,
      "loss": 0.8521,
      "step": 10342
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.459314823150635,
      "learning_rate": 1.4010797687122642e-05,
      "loss": 0.95,
      "step": 10343
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6895434856414795,
      "learning_rate": 1.401015877072485e-05,
      "loss": 0.8875,
      "step": 10344
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8437741994857788,
      "learning_rate": 1.4009519854327062e-05,
      "loss": 0.9497,
      "step": 10345
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.761582851409912,
      "learning_rate": 1.4008880937929272e-05,
      "loss": 0.7486,
      "step": 10346
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.512362480163574,
      "learning_rate": 1.4008242021531483e-05,
      "loss": 1.3614,
      "step": 10347
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5812325477600098,
      "learning_rate": 1.4007603105133693e-05,
      "loss": 0.8486,
      "step": 10348
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0075089931488037,
      "learning_rate": 1.4006964188735905e-05,
      "loss": 0.9448,
      "step": 10349
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7376575469970703,
      "learning_rate": 1.4006325272338115e-05,
      "loss": 0.8842,
      "step": 10350
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.348649263381958,
      "learning_rate": 1.4005686355940326e-05,
      "loss": 1.0171,
      "step": 10351
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.4512529373168945,
      "learning_rate": 1.4005047439542536e-05,
      "loss": 0.6661,
      "step": 10352
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8258417844772339,
      "learning_rate": 1.4004408523144748e-05,
      "loss": 0.8586,
      "step": 10353
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.615466594696045,
      "learning_rate": 1.4003769606746957e-05,
      "loss": 0.8239,
      "step": 10354
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3690228462219238,
      "learning_rate": 1.4003130690349169e-05,
      "loss": 0.4616,
      "step": 10355
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6693968772888184,
      "learning_rate": 1.4002491773951379e-05,
      "loss": 0.8426,
      "step": 10356
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.995505928993225,
      "learning_rate": 1.400185285755359e-05,
      "loss": 1.0598,
      "step": 10357
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.030379295349121,
      "learning_rate": 1.40012139411558e-05,
      "loss": 0.9372,
      "step": 10358
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0255937576293945,
      "learning_rate": 1.4000575024758012e-05,
      "loss": 1.023,
      "step": 10359
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1523032188415527,
      "learning_rate": 1.3999936108360222e-05,
      "loss": 0.808,
      "step": 10360
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5469040870666504,
      "learning_rate": 1.3999297191962433e-05,
      "loss": 0.9273,
      "step": 10361
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1376430988311768,
      "learning_rate": 1.3998658275564643e-05,
      "loss": 0.9223,
      "step": 10362
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7726569175720215,
      "learning_rate": 1.3998019359166855e-05,
      "loss": 1.0135,
      "step": 10363
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.549740314483643,
      "learning_rate": 1.3997380442769064e-05,
      "loss": 1.5671,
      "step": 10364
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.883939504623413,
      "learning_rate": 1.3996741526371276e-05,
      "loss": 0.8753,
      "step": 10365
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.559539556503296,
      "learning_rate": 1.3996102609973486e-05,
      "loss": 0.9411,
      "step": 10366
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.98416006565094,
      "learning_rate": 1.3995463693575697e-05,
      "loss": 0.9418,
      "step": 10367
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.714121699333191,
      "learning_rate": 1.3994824777177907e-05,
      "loss": 0.8688,
      "step": 10368
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.631244421005249,
      "learning_rate": 1.3994185860780119e-05,
      "loss": 0.9184,
      "step": 10369
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6919872760772705,
      "learning_rate": 1.3993546944382329e-05,
      "loss": 0.948,
      "step": 10370
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.00346040725708,
      "learning_rate": 1.399290802798454e-05,
      "loss": 0.805,
      "step": 10371
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.446659564971924,
      "learning_rate": 1.399226911158675e-05,
      "loss": 0.8863,
      "step": 10372
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5631794929504395,
      "learning_rate": 1.3991630195188962e-05,
      "loss": 0.879,
      "step": 10373
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.834294319152832,
      "learning_rate": 1.3990991278791171e-05,
      "loss": 1.4673,
      "step": 10374
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.507455587387085,
      "learning_rate": 1.3990352362393383e-05,
      "loss": 0.9241,
      "step": 10375
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.427205801010132,
      "learning_rate": 1.3989713445995591e-05,
      "loss": 1.025,
      "step": 10376
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.4638609886169434,
      "learning_rate": 1.3989074529597804e-05,
      "loss": 0.6785,
      "step": 10377
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8100146055221558,
      "learning_rate": 1.3988435613200013e-05,
      "loss": 0.9072,
      "step": 10378
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0678815841674805,
      "learning_rate": 1.3987796696802226e-05,
      "loss": 0.8563,
      "step": 10379
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5974831581115723,
      "learning_rate": 1.3987157780404434e-05,
      "loss": 0.9892,
      "step": 10380
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.9717321395874023,
      "learning_rate": 1.3986518864006645e-05,
      "loss": 0.9793,
      "step": 10381
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3834211826324463,
      "learning_rate": 1.3985879947608855e-05,
      "loss": 0.8912,
      "step": 10382
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.705732822418213,
      "learning_rate": 1.3985241031211067e-05,
      "loss": 1.7091,
      "step": 10383
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.729668140411377,
      "learning_rate": 1.3984602114813277e-05,
      "loss": 1.057,
      "step": 10384
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1980888843536377,
      "learning_rate": 1.3983963198415488e-05,
      "loss": 1.3039,
      "step": 10385
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8899967670440674,
      "learning_rate": 1.3983324282017698e-05,
      "loss": 1.1932,
      "step": 10386
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.244204044342041,
      "learning_rate": 1.398268536561991e-05,
      "loss": 1.0236,
      "step": 10387
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5075041055679321,
      "learning_rate": 1.398204644922212e-05,
      "loss": 0.8251,
      "step": 10388
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.170286178588867,
      "learning_rate": 1.3981407532824331e-05,
      "loss": 0.9196,
      "step": 10389
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6001025438308716,
      "learning_rate": 1.3980768616426541e-05,
      "loss": 0.9634,
      "step": 10390
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.6593017578125,
      "learning_rate": 1.3980129700028752e-05,
      "loss": 1.5834,
      "step": 10391
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.164809465408325,
      "learning_rate": 1.3979490783630962e-05,
      "loss": 0.9773,
      "step": 10392
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.419527530670166,
      "learning_rate": 1.3978851867233174e-05,
      "loss": 1.0789,
      "step": 10393
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8276666402816772,
      "learning_rate": 1.3978212950835384e-05,
      "loss": 1.0243,
      "step": 10394
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9562480449676514,
      "learning_rate": 1.3977574034437595e-05,
      "loss": 0.8706,
      "step": 10395
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.231614351272583,
      "learning_rate": 1.3976935118039805e-05,
      "loss": 0.6755,
      "step": 10396
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.998734474182129,
      "learning_rate": 1.3976296201642017e-05,
      "loss": 0.8094,
      "step": 10397
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.598151445388794,
      "learning_rate": 1.3975657285244227e-05,
      "loss": 0.9054,
      "step": 10398
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8800355195999146,
      "learning_rate": 1.3975018368846438e-05,
      "loss": 0.8411,
      "step": 10399
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.796345829963684,
      "learning_rate": 1.3974379452448648e-05,
      "loss": 0.8716,
      "step": 10400
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7402245998382568,
      "learning_rate": 1.397374053605086e-05,
      "loss": 0.8788,
      "step": 10401
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.591465473175049,
      "learning_rate": 1.397310161965307e-05,
      "loss": 1.128,
      "step": 10402
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9547656774520874,
      "learning_rate": 1.3972462703255281e-05,
      "loss": 0.9545,
      "step": 10403
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8868656158447266,
      "learning_rate": 1.397182378685749e-05,
      "loss": 1.2535,
      "step": 10404
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.745911598205566,
      "learning_rate": 1.3971184870459702e-05,
      "loss": 1.4971,
      "step": 10405
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7607004642486572,
      "learning_rate": 1.3970545954061912e-05,
      "loss": 0.9497,
      "step": 10406
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1078243255615234,
      "learning_rate": 1.3969907037664124e-05,
      "loss": 0.9188,
      "step": 10407
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0540051460266113,
      "learning_rate": 1.3969268121266334e-05,
      "loss": 0.8788,
      "step": 10408
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0462658405303955,
      "learning_rate": 1.3968629204868545e-05,
      "loss": 0.8932,
      "step": 10409
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.158177137374878,
      "learning_rate": 1.3967990288470755e-05,
      "loss": 1.0916,
      "step": 10410
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6023168563842773,
      "learning_rate": 1.3967351372072966e-05,
      "loss": 0.9467,
      "step": 10411
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.355407953262329,
      "learning_rate": 1.3966712455675175e-05,
      "loss": 1.0924,
      "step": 10412
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9577441215515137,
      "learning_rate": 1.3966073539277388e-05,
      "loss": 0.9191,
      "step": 10413
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.883836030960083,
      "learning_rate": 1.3965434622879596e-05,
      "loss": 0.8223,
      "step": 10414
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6074249744415283,
      "learning_rate": 1.396479570648181e-05,
      "loss": 0.9019,
      "step": 10415
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0444085597991943,
      "learning_rate": 1.3964156790084017e-05,
      "loss": 1.0147,
      "step": 10416
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8246880769729614,
      "learning_rate": 1.3963517873686229e-05,
      "loss": 0.8636,
      "step": 10417
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.31794548034668,
      "learning_rate": 1.3962878957288439e-05,
      "loss": 1.4964,
      "step": 10418
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8695261478424072,
      "learning_rate": 1.396224004089065e-05,
      "loss": 0.9366,
      "step": 10419
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0900540351867676,
      "learning_rate": 1.396160112449286e-05,
      "loss": 0.8915,
      "step": 10420
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9106967449188232,
      "learning_rate": 1.3960962208095072e-05,
      "loss": 0.8562,
      "step": 10421
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8019589185714722,
      "learning_rate": 1.3960323291697282e-05,
      "loss": 0.8791,
      "step": 10422
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0580544471740723,
      "learning_rate": 1.3959684375299493e-05,
      "loss": 0.9258,
      "step": 10423
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.001488208770752,
      "learning_rate": 1.3959045458901703e-05,
      "loss": 1.0215,
      "step": 10424
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7937079668045044,
      "learning_rate": 1.3958406542503915e-05,
      "loss": 0.9911,
      "step": 10425
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0751612186431885,
      "learning_rate": 1.3957767626106124e-05,
      "loss": 0.8605,
      "step": 10426
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.258004188537598,
      "learning_rate": 1.3957128709708336e-05,
      "loss": 0.6332,
      "step": 10427
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.583839774131775,
      "learning_rate": 1.3956489793310546e-05,
      "loss": 0.8922,
      "step": 10428
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5858463048934937,
      "learning_rate": 1.3955850876912757e-05,
      "loss": 0.8375,
      "step": 10429
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0024476051330566,
      "learning_rate": 1.3955211960514967e-05,
      "loss": 0.8588,
      "step": 10430
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2254247665405273,
      "learning_rate": 1.3954573044117179e-05,
      "loss": 0.9413,
      "step": 10431
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.87099289894104,
      "learning_rate": 1.3953934127719389e-05,
      "loss": 0.9257,
      "step": 10432
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3661048412323,
      "learning_rate": 1.39532952113216e-05,
      "loss": 0.7437,
      "step": 10433
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0063188076019287,
      "learning_rate": 1.395265629492381e-05,
      "loss": 0.8763,
      "step": 10434
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.473616361618042,
      "learning_rate": 1.3952017378526022e-05,
      "loss": 0.8815,
      "step": 10435
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.702680826187134,
      "learning_rate": 1.3951378462128231e-05,
      "loss": 1.083,
      "step": 10436
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3902482986450195,
      "learning_rate": 1.3950739545730443e-05,
      "loss": 1.3697,
      "step": 10437
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6392465829849243,
      "learning_rate": 1.3950100629332653e-05,
      "loss": 0.8359,
      "step": 10438
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.4368064403533936,
      "learning_rate": 1.3949461712934864e-05,
      "loss": 1.0894,
      "step": 10439
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.5434064865112305,
      "learning_rate": 1.3948822796537074e-05,
      "loss": 1.5344,
      "step": 10440
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8824962377548218,
      "learning_rate": 1.3948183880139286e-05,
      "loss": 0.9564,
      "step": 10441
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.081454277038574,
      "learning_rate": 1.3947544963741496e-05,
      "loss": 1.1559,
      "step": 10442
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5194640159606934,
      "learning_rate": 1.3946906047343707e-05,
      "loss": 1.1942,
      "step": 10443
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7052202224731445,
      "learning_rate": 1.3946267130945917e-05,
      "loss": 0.7949,
      "step": 10444
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6516494750976562,
      "learning_rate": 1.3945628214548129e-05,
      "loss": 0.9091,
      "step": 10445
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.926450490951538,
      "learning_rate": 1.3944989298150338e-05,
      "loss": 0.8411,
      "step": 10446
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.151388168334961,
      "learning_rate": 1.394435038175255e-05,
      "loss": 1.0037,
      "step": 10447
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0508460998535156,
      "learning_rate": 1.3943711465354758e-05,
      "loss": 0.9133,
      "step": 10448
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.35426139831543,
      "learning_rate": 1.3943072548956971e-05,
      "loss": 1.1238,
      "step": 10449
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.356001138687134,
      "learning_rate": 1.394243363255918e-05,
      "loss": 0.6567,
      "step": 10450
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1812198162078857,
      "learning_rate": 1.3941794716161393e-05,
      "loss": 0.904,
      "step": 10451
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3969225883483887,
      "learning_rate": 1.3941155799763601e-05,
      "loss": 0.9002,
      "step": 10452
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.7176244258880615,
      "learning_rate": 1.3940516883365812e-05,
      "loss": 0.9886,
      "step": 10453
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1288697719573975,
      "learning_rate": 1.3939877966968022e-05,
      "loss": 0.96,
      "step": 10454
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9105005264282227,
      "learning_rate": 1.3939239050570234e-05,
      "loss": 0.9359,
      "step": 10455
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.944543480873108,
      "learning_rate": 1.3938600134172444e-05,
      "loss": 0.9355,
      "step": 10456
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8261134624481201,
      "learning_rate": 1.3937961217774655e-05,
      "loss": 0.8734,
      "step": 10457
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.025209426879883,
      "learning_rate": 1.3937322301376865e-05,
      "loss": 0.9248,
      "step": 10458
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9421238899230957,
      "learning_rate": 1.3936683384979077e-05,
      "loss": 0.8866,
      "step": 10459
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9655812978744507,
      "learning_rate": 1.3936044468581287e-05,
      "loss": 0.9408,
      "step": 10460
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0219855308532715,
      "learning_rate": 1.3935405552183498e-05,
      "loss": 1.0078,
      "step": 10461
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9608209133148193,
      "learning_rate": 1.3934766635785708e-05,
      "loss": 0.865,
      "step": 10462
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.7655742168426514,
      "learning_rate": 1.393412771938792e-05,
      "loss": 0.8342,
      "step": 10463
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.589951753616333,
      "learning_rate": 1.393348880299013e-05,
      "loss": 0.4705,
      "step": 10464
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.186577081680298,
      "learning_rate": 1.3932849886592341e-05,
      "loss": 0.8796,
      "step": 10465
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8612176179885864,
      "learning_rate": 1.393221097019455e-05,
      "loss": 0.7733,
      "step": 10466
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.267591714859009,
      "learning_rate": 1.3931572053796762e-05,
      "loss": 0.9718,
      "step": 10467
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9127047061920166,
      "learning_rate": 1.3930933137398972e-05,
      "loss": 0.827,
      "step": 10468
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9448895454406738,
      "learning_rate": 1.3930294221001184e-05,
      "loss": 0.9185,
      "step": 10469
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7845723628997803,
      "learning_rate": 1.3929655304603394e-05,
      "loss": 0.93,
      "step": 10470
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8185224533081055,
      "learning_rate": 1.3929016388205605e-05,
      "loss": 0.9489,
      "step": 10471
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.7047276496887207,
      "learning_rate": 1.3928377471807815e-05,
      "loss": 1.6381,
      "step": 10472
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9695382118225098,
      "learning_rate": 1.3927738555410026e-05,
      "loss": 1.0173,
      "step": 10473
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9933537244796753,
      "learning_rate": 1.3927099639012236e-05,
      "loss": 0.8417,
      "step": 10474
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0986433029174805,
      "learning_rate": 1.3926460722614448e-05,
      "loss": 0.9664,
      "step": 10475
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6690526008605957,
      "learning_rate": 1.3925821806216658e-05,
      "loss": 0.9207,
      "step": 10476
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.181122303009033,
      "learning_rate": 1.392518288981887e-05,
      "loss": 0.9636,
      "step": 10477
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9433071613311768,
      "learning_rate": 1.3924543973421079e-05,
      "loss": 0.7282,
      "step": 10478
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.981124758720398,
      "learning_rate": 1.392390505702329e-05,
      "loss": 0.9399,
      "step": 10479
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6712497472763062,
      "learning_rate": 1.39232661406255e-05,
      "loss": 0.8424,
      "step": 10480
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8254998922348022,
      "learning_rate": 1.3922627224227712e-05,
      "loss": 0.855,
      "step": 10481
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.166668176651001,
      "learning_rate": 1.3921988307829922e-05,
      "loss": 0.9092,
      "step": 10482
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.106520652770996,
      "learning_rate": 1.3921349391432133e-05,
      "loss": 0.933,
      "step": 10483
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7425950765609741,
      "learning_rate": 1.3920710475034342e-05,
      "loss": 0.9558,
      "step": 10484
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3353817462921143,
      "learning_rate": 1.3920071558636555e-05,
      "loss": 1.1406,
      "step": 10485
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0800838470458984,
      "learning_rate": 1.3919432642238763e-05,
      "loss": 0.7657,
      "step": 10486
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.49273681640625,
      "learning_rate": 1.3918793725840976e-05,
      "loss": 0.6667,
      "step": 10487
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7643256187438965,
      "learning_rate": 1.3918154809443184e-05,
      "loss": 0.9422,
      "step": 10488
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.013639211654663,
      "learning_rate": 1.3917515893045398e-05,
      "loss": 1.0265,
      "step": 10489
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1264913082122803,
      "learning_rate": 1.3916876976647606e-05,
      "loss": 0.9297,
      "step": 10490
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.045405149459839,
      "learning_rate": 1.3916238060249817e-05,
      "loss": 0.915,
      "step": 10491
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.865113615989685,
      "learning_rate": 1.3915599143852027e-05,
      "loss": 0.9265,
      "step": 10492
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9594248533248901,
      "learning_rate": 1.3914960227454239e-05,
      "loss": 0.9204,
      "step": 10493
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8717890977859497,
      "learning_rate": 1.3914321311056449e-05,
      "loss": 0.9299,
      "step": 10494
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.868218421936035,
      "learning_rate": 1.391368239465866e-05,
      "loss": 1.0941,
      "step": 10495
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.389040470123291,
      "learning_rate": 1.391304347826087e-05,
      "loss": 1.1192,
      "step": 10496
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.950269103050232,
      "learning_rate": 1.3912404561863082e-05,
      "loss": 0.8786,
      "step": 10497
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.257483959197998,
      "learning_rate": 1.3911765645465291e-05,
      "loss": 0.8248,
      "step": 10498
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.641887903213501,
      "learning_rate": 1.3911126729067503e-05,
      "loss": 0.8222,
      "step": 10499
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8297934532165527,
      "learning_rate": 1.3910487812669713e-05,
      "loss": 0.8129,
      "step": 10500
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6825652122497559,
      "learning_rate": 1.3909848896271924e-05,
      "loss": 0.8772,
      "step": 10501
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8614535331726074,
      "learning_rate": 1.3909209979874134e-05,
      "loss": 0.9259,
      "step": 10502
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6782927513122559,
      "learning_rate": 1.3908571063476346e-05,
      "loss": 0.929,
      "step": 10503
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.6652984619140625,
      "learning_rate": 1.3907932147078556e-05,
      "loss": 1.1245,
      "step": 10504
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8435137271881104,
      "learning_rate": 1.3907293230680767e-05,
      "loss": 0.9031,
      "step": 10505
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.935708999633789,
      "learning_rate": 1.3906654314282977e-05,
      "loss": 0.9423,
      "step": 10506
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.056467056274414,
      "learning_rate": 1.3906015397885189e-05,
      "loss": 0.801,
      "step": 10507
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2468929290771484,
      "learning_rate": 1.3905376481487398e-05,
      "loss": 1.0325,
      "step": 10508
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6802338361740112,
      "learning_rate": 1.390473756508961e-05,
      "loss": 0.9283,
      "step": 10509
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6035493612289429,
      "learning_rate": 1.390409864869182e-05,
      "loss": 0.893,
      "step": 10510
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1554675102233887,
      "learning_rate": 1.3903459732294031e-05,
      "loss": 0.851,
      "step": 10511
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7625912427902222,
      "learning_rate": 1.3902820815896241e-05,
      "loss": 0.964,
      "step": 10512
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6724931001663208,
      "learning_rate": 1.3902181899498453e-05,
      "loss": 0.8509,
      "step": 10513
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8168078660964966,
      "learning_rate": 1.3901542983100663e-05,
      "loss": 0.8641,
      "step": 10514
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6920534372329712,
      "learning_rate": 1.3900904066702874e-05,
      "loss": 1.2358,
      "step": 10515
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.909356117248535,
      "learning_rate": 1.3900265150305084e-05,
      "loss": 0.6713,
      "step": 10516
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5208946466445923,
      "learning_rate": 1.3899626233907296e-05,
      "loss": 0.8619,
      "step": 10517
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.4560022354125977,
      "learning_rate": 1.3898987317509505e-05,
      "loss": 1.1321,
      "step": 10518
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.498199462890625,
      "learning_rate": 1.3898348401111717e-05,
      "loss": 0.768,
      "step": 10519
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1571385860443115,
      "learning_rate": 1.3897709484713927e-05,
      "loss": 1.2719,
      "step": 10520
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.446632742881775,
      "learning_rate": 1.3897070568316138e-05,
      "loss": 0.8481,
      "step": 10521
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.181100368499756,
      "learning_rate": 1.3896431651918347e-05,
      "loss": 0.989,
      "step": 10522
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7481154203414917,
      "learning_rate": 1.389579273552056e-05,
      "loss": 0.9493,
      "step": 10523
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2010321617126465,
      "learning_rate": 1.3895153819122768e-05,
      "loss": 0.8497,
      "step": 10524
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.508849620819092,
      "learning_rate": 1.3894514902724981e-05,
      "loss": 0.9163,
      "step": 10525
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6288305521011353,
      "learning_rate": 1.389387598632719e-05,
      "loss": 0.9315,
      "step": 10526
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.527772903442383,
      "learning_rate": 1.3893237069929401e-05,
      "loss": 0.9441,
      "step": 10527
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2621712684631348,
      "learning_rate": 1.389259815353161e-05,
      "loss": 1.0198,
      "step": 10528
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8935967683792114,
      "learning_rate": 1.3891959237133822e-05,
      "loss": 0.9239,
      "step": 10529
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.668200969696045,
      "learning_rate": 1.3891320320736032e-05,
      "loss": 0.9433,
      "step": 10530
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.361668348312378,
      "learning_rate": 1.3890681404338244e-05,
      "loss": 0.8938,
      "step": 10531
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.444787979125977,
      "learning_rate": 1.3890042487940454e-05,
      "loss": 1.1753,
      "step": 10532
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.030397415161133,
      "learning_rate": 1.3889403571542665e-05,
      "loss": 0.9904,
      "step": 10533
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8418323993682861,
      "learning_rate": 1.3888764655144875e-05,
      "loss": 0.7667,
      "step": 10534
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.246142625808716,
      "learning_rate": 1.3888125738747086e-05,
      "loss": 0.9984,
      "step": 10535
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.33998703956604,
      "learning_rate": 1.3887486822349296e-05,
      "loss": 1.3748,
      "step": 10536
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0584988594055176,
      "learning_rate": 1.3886847905951508e-05,
      "loss": 0.7851,
      "step": 10537
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.684241771697998,
      "learning_rate": 1.3886208989553718e-05,
      "loss": 0.9536,
      "step": 10538
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.928320288658142,
      "learning_rate": 1.388557007315593e-05,
      "loss": 0.9726,
      "step": 10539
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.467958450317383,
      "learning_rate": 1.3884931156758139e-05,
      "loss": 1.4854,
      "step": 10540
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.51210355758667,
      "learning_rate": 1.388429224036035e-05,
      "loss": 0.9461,
      "step": 10541
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0368266105651855,
      "learning_rate": 1.388365332396256e-05,
      "loss": 1.1533,
      "step": 10542
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.5670316219329834,
      "learning_rate": 1.3883014407564772e-05,
      "loss": 1.5729,
      "step": 10543
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7838702201843262,
      "learning_rate": 1.3882375491166982e-05,
      "loss": 0.8858,
      "step": 10544
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8207144737243652,
      "learning_rate": 1.3881736574769193e-05,
      "loss": 0.9417,
      "step": 10545
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3957157135009766,
      "learning_rate": 1.3881097658371403e-05,
      "loss": 1.1008,
      "step": 10546
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1948320865631104,
      "learning_rate": 1.3880458741973615e-05,
      "loss": 0.9433,
      "step": 10547
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1213786602020264,
      "learning_rate": 1.3879819825575825e-05,
      "loss": 0.9916,
      "step": 10548
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.0148024559020996,
      "learning_rate": 1.3879180909178036e-05,
      "loss": 1.1594,
      "step": 10549
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2954235076904297,
      "learning_rate": 1.3878541992780246e-05,
      "loss": 0.3827,
      "step": 10550
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.458425045013428,
      "learning_rate": 1.3877903076382458e-05,
      "loss": 1.2969,
      "step": 10551
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.4719316959381104,
      "learning_rate": 1.3877264159984668e-05,
      "loss": 0.8341,
      "step": 10552
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.5323421955108643,
      "learning_rate": 1.3876625243586879e-05,
      "loss": 1.0311,
      "step": 10553
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3764913082122803,
      "learning_rate": 1.3875986327189089e-05,
      "loss": 0.8669,
      "step": 10554
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.3424794673919678,
      "learning_rate": 1.38753474107913e-05,
      "loss": 1.2632,
      "step": 10555
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.7869982719421387,
      "learning_rate": 1.387470849439351e-05,
      "loss": 0.9641,
      "step": 10556
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.5836260318756104,
      "learning_rate": 1.3874069577995722e-05,
      "loss": 0.9666,
      "step": 10557
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0467166900634766,
      "learning_rate": 1.387343066159793e-05,
      "loss": 0.9367,
      "step": 10558
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0806918144226074,
      "learning_rate": 1.3872791745200143e-05,
      "loss": 0.9197,
      "step": 10559
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.048278570175171,
      "learning_rate": 1.3872152828802351e-05,
      "loss": 0.9446,
      "step": 10560
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.592880129814148,
      "learning_rate": 1.3871513912404565e-05,
      "loss": 0.9369,
      "step": 10561
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5192384719848633,
      "learning_rate": 1.3870874996006773e-05,
      "loss": 0.8534,
      "step": 10562
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.7850215435028076,
      "learning_rate": 1.3870236079608984e-05,
      "loss": 1.2978,
      "step": 10563
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6256036758422852,
      "learning_rate": 1.3869597163211194e-05,
      "loss": 0.9292,
      "step": 10564
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9565117359161377,
      "learning_rate": 1.3868958246813406e-05,
      "loss": 0.881,
      "step": 10565
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7698776721954346,
      "learning_rate": 1.3868319330415616e-05,
      "loss": 0.8528,
      "step": 10566
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6640126705169678,
      "learning_rate": 1.3867680414017827e-05,
      "loss": 0.9772,
      "step": 10567
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9096910953521729,
      "learning_rate": 1.3867041497620037e-05,
      "loss": 0.9009,
      "step": 10568
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1330745220184326,
      "learning_rate": 1.3866402581222249e-05,
      "loss": 0.9739,
      "step": 10569
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.3292837142944336,
      "learning_rate": 1.3865763664824458e-05,
      "loss": 1.0536,
      "step": 10570
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.450551986694336,
      "learning_rate": 1.386512474842667e-05,
      "loss": 1.1291,
      "step": 10571
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4991092681884766,
      "learning_rate": 1.386448583202888e-05,
      "loss": 0.4691,
      "step": 10572
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0516772270202637,
      "learning_rate": 1.3863846915631091e-05,
      "loss": 0.8097,
      "step": 10573
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.9802541732788086,
      "learning_rate": 1.3863207999233301e-05,
      "loss": 0.8678,
      "step": 10574
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.6563408374786377,
      "learning_rate": 1.3862569082835513e-05,
      "loss": 1.0242,
      "step": 10575
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5914111137390137,
      "learning_rate": 1.3861930166437723e-05,
      "loss": 0.503,
      "step": 10576
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.697019100189209,
      "learning_rate": 1.3861291250039934e-05,
      "loss": 0.8667,
      "step": 10577
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6546657085418701,
      "learning_rate": 1.3860652333642144e-05,
      "loss": 0.3957,
      "step": 10578
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.885248899459839,
      "learning_rate": 1.3860013417244356e-05,
      "loss": 0.9227,
      "step": 10579
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8427461385726929,
      "learning_rate": 1.3859374500846565e-05,
      "loss": 0.9273,
      "step": 10580
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0516960620880127,
      "learning_rate": 1.3858735584448777e-05,
      "loss": 0.916,
      "step": 10581
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0259652137756348,
      "learning_rate": 1.3858096668050987e-05,
      "loss": 0.8784,
      "step": 10582
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.869491696357727,
      "learning_rate": 1.3857457751653198e-05,
      "loss": 0.8911,
      "step": 10583
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.932441234588623,
      "learning_rate": 1.3856818835255408e-05,
      "loss": 1.0909,
      "step": 10584
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1461069583892822,
      "learning_rate": 1.385617991885762e-05,
      "loss": 0.8069,
      "step": 10585
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.4756040573120117,
      "learning_rate": 1.385554100245983e-05,
      "loss": 0.9829,
      "step": 10586
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9799197912216187,
      "learning_rate": 1.3854902086062041e-05,
      "loss": 0.9191,
      "step": 10587
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.915560245513916,
      "learning_rate": 1.3854263169664251e-05,
      "loss": 0.802,
      "step": 10588
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.128239631652832,
      "learning_rate": 1.3853624253266463e-05,
      "loss": 0.8515,
      "step": 10589
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7968637943267822,
      "learning_rate": 1.3852985336868672e-05,
      "loss": 0.8822,
      "step": 10590
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.436419725418091,
      "learning_rate": 1.3852346420470884e-05,
      "loss": 0.9654,
      "step": 10591
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9989571571350098,
      "learning_rate": 1.3851707504073094e-05,
      "loss": 1.2417,
      "step": 10592
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.248114824295044,
      "learning_rate": 1.3851068587675305e-05,
      "loss": 1.0281,
      "step": 10593
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0424232482910156,
      "learning_rate": 1.3850429671277514e-05,
      "loss": 0.9736,
      "step": 10594
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1798529624938965,
      "learning_rate": 1.3849790754879727e-05,
      "loss": 1.1056,
      "step": 10595
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.365467071533203,
      "learning_rate": 1.3849151838481935e-05,
      "loss": 0.5896,
      "step": 10596
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.968087911605835,
      "learning_rate": 1.3848512922084148e-05,
      "loss": 0.5912,
      "step": 10597
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8457739353179932,
      "learning_rate": 1.3847874005686356e-05,
      "loss": 0.894,
      "step": 10598
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.42712688446045,
      "learning_rate": 1.3847235089288568e-05,
      "loss": 1.2758,
      "step": 10599
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.5942728519439697,
      "learning_rate": 1.3846596172890778e-05,
      "loss": 0.7204,
      "step": 10600
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9956657886505127,
      "learning_rate": 1.384595725649299e-05,
      "loss": 1.0521,
      "step": 10601
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.232494831085205,
      "learning_rate": 1.38453183400952e-05,
      "loss": 0.4511,
      "step": 10602
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.154442310333252,
      "learning_rate": 1.384467942369741e-05,
      "loss": 1.5049,
      "step": 10603
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6286776065826416,
      "learning_rate": 1.384404050729962e-05,
      "loss": 0.8584,
      "step": 10604
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4723927974700928,
      "learning_rate": 1.3843401590901832e-05,
      "loss": 0.817,
      "step": 10605
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.079639196395874,
      "learning_rate": 1.3842762674504042e-05,
      "loss": 0.9296,
      "step": 10606
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0421061515808105,
      "learning_rate": 1.3842123758106254e-05,
      "loss": 0.7281,
      "step": 10607
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6669944524765015,
      "learning_rate": 1.3841484841708463e-05,
      "loss": 0.9115,
      "step": 10608
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4261960983276367,
      "learning_rate": 1.3840845925310675e-05,
      "loss": 0.8736,
      "step": 10609
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.593008279800415,
      "learning_rate": 1.3840207008912885e-05,
      "loss": 0.6037,
      "step": 10610
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9196468591690063,
      "learning_rate": 1.3839568092515096e-05,
      "loss": 0.9359,
      "step": 10611
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6401278972625732,
      "learning_rate": 1.3838929176117306e-05,
      "loss": 0.9142,
      "step": 10612
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0170950889587402,
      "learning_rate": 1.3838290259719518e-05,
      "loss": 1.2688,
      "step": 10613
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.116971731185913,
      "learning_rate": 1.3837651343321728e-05,
      "loss": 0.931,
      "step": 10614
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.6403331756591797,
      "learning_rate": 1.3837012426923939e-05,
      "loss": 0.8208,
      "step": 10615
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.256291389465332,
      "learning_rate": 1.3836373510526149e-05,
      "loss": 0.9016,
      "step": 10616
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3224186897277832,
      "learning_rate": 1.383573459412836e-05,
      "loss": 0.8469,
      "step": 10617
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.544089913368225,
      "learning_rate": 1.383509567773057e-05,
      "loss": 0.9264,
      "step": 10618
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4890568256378174,
      "learning_rate": 1.3834456761332782e-05,
      "loss": 0.8613,
      "step": 10619
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6194523572921753,
      "learning_rate": 1.3833817844934992e-05,
      "loss": 0.8904,
      "step": 10620
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.907861351966858,
      "learning_rate": 1.3833178928537203e-05,
      "loss": 1.0597,
      "step": 10621
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7650961875915527,
      "learning_rate": 1.3832540012139413e-05,
      "loss": 0.9263,
      "step": 10622
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3354904651641846,
      "learning_rate": 1.3831901095741625e-05,
      "loss": 0.8389,
      "step": 10623
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.885067343711853,
      "learning_rate": 1.3831262179343835e-05,
      "loss": 0.9519,
      "step": 10624
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1069540977478027,
      "learning_rate": 1.3830623262946046e-05,
      "loss": 0.9438,
      "step": 10625
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6745712757110596,
      "learning_rate": 1.3829984346548256e-05,
      "loss": 0.8846,
      "step": 10626
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9856771230697632,
      "learning_rate": 1.3829345430150467e-05,
      "loss": 1.2349,
      "step": 10627
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4334027767181396,
      "learning_rate": 1.3828706513752677e-05,
      "loss": 0.8154,
      "step": 10628
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4965425729751587,
      "learning_rate": 1.3828067597354889e-05,
      "loss": 0.8416,
      "step": 10629
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7321590185165405,
      "learning_rate": 1.3827428680957097e-05,
      "loss": 1.047,
      "step": 10630
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6196905374526978,
      "learning_rate": 1.3826789764559307e-05,
      "loss": 0.8491,
      "step": 10631
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.1817920207977295,
      "learning_rate": 1.3826150848161518e-05,
      "loss": 0.9032,
      "step": 10632
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.631819486618042,
      "learning_rate": 1.3825511931763728e-05,
      "loss": 0.9401,
      "step": 10633
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0044782161712646,
      "learning_rate": 1.382487301536594e-05,
      "loss": 1.0471,
      "step": 10634
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9754377603530884,
      "learning_rate": 1.382423409896815e-05,
      "loss": 0.9621,
      "step": 10635
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.645801305770874,
      "learning_rate": 1.3823595182570361e-05,
      "loss": 0.9544,
      "step": 10636
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.9330625534057617,
      "learning_rate": 1.3822956266172571e-05,
      "loss": 1.1903,
      "step": 10637
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9314743280410767,
      "learning_rate": 1.3822317349774783e-05,
      "loss": 0.8235,
      "step": 10638
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.5157647132873535,
      "learning_rate": 1.3821678433376993e-05,
      "loss": 0.8526,
      "step": 10639
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.286003589630127,
      "learning_rate": 1.3821039516979204e-05,
      "loss": 0.8588,
      "step": 10640
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.629235863685608,
      "learning_rate": 1.3820400600581414e-05,
      "loss": 0.9463,
      "step": 10641
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3088600635528564,
      "learning_rate": 1.3819761684183625e-05,
      "loss": 0.8486,
      "step": 10642
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.076056718826294,
      "learning_rate": 1.3819122767785835e-05,
      "loss": 1.21,
      "step": 10643
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7040627002716064,
      "learning_rate": 1.3818483851388047e-05,
      "loss": 0.8067,
      "step": 10644
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0386104583740234,
      "learning_rate": 1.3817844934990257e-05,
      "loss": 0.9981,
      "step": 10645
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.846911072731018,
      "learning_rate": 1.3817206018592468e-05,
      "loss": 0.9448,
      "step": 10646
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.82279634475708,
      "learning_rate": 1.3816567102194678e-05,
      "loss": 1.3895,
      "step": 10647
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.922927141189575,
      "learning_rate": 1.381592818579689e-05,
      "loss": 1.3107,
      "step": 10648
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7167763710021973,
      "learning_rate": 1.38152892693991e-05,
      "loss": 0.872,
      "step": 10649
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.429297685623169,
      "learning_rate": 1.3814650353001311e-05,
      "loss": 1.0517,
      "step": 10650
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.799133539199829,
      "learning_rate": 1.3814011436603521e-05,
      "loss": 0.8729,
      "step": 10651
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.693031311035156,
      "learning_rate": 1.3813372520205732e-05,
      "loss": 1.589,
      "step": 10652
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.328549861907959,
      "learning_rate": 1.3812733603807942e-05,
      "loss": 1.3995,
      "step": 10653
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9310370683670044,
      "learning_rate": 1.3812094687410154e-05,
      "loss": 1.009,
      "step": 10654
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7496812343597412,
      "learning_rate": 1.3811455771012364e-05,
      "loss": 0.9829,
      "step": 10655
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.375274896621704,
      "learning_rate": 1.3810816854614575e-05,
      "loss": 1.4445,
      "step": 10656
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.02925968170166,
      "learning_rate": 1.3810177938216785e-05,
      "loss": 1.0117,
      "step": 10657
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.602466344833374,
      "learning_rate": 1.3809539021818997e-05,
      "loss": 0.8925,
      "step": 10658
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8951505422592163,
      "learning_rate": 1.3808900105421207e-05,
      "loss": 0.8949,
      "step": 10659
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.2377424240112305,
      "learning_rate": 1.3808261189023418e-05,
      "loss": 0.8964,
      "step": 10660
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.3885223865509033,
      "learning_rate": 1.3807622272625626e-05,
      "loss": 1.2381,
      "step": 10661
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4480260610580444,
      "learning_rate": 1.380698335622784e-05,
      "loss": 0.8247,
      "step": 10662
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6591463088989258,
      "learning_rate": 1.3806344439830048e-05,
      "loss": 0.8512,
      "step": 10663
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7992744445800781,
      "learning_rate": 1.3805705523432261e-05,
      "loss": 0.9716,
      "step": 10664
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.009094715118408,
      "learning_rate": 1.3805066607034469e-05,
      "loss": 1.1072,
      "step": 10665
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9748283624649048,
      "learning_rate": 1.3804427690636682e-05,
      "loss": 0.9922,
      "step": 10666
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.322206974029541,
      "learning_rate": 1.380378877423889e-05,
      "loss": 1.1134,
      "step": 10667
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.242358446121216,
      "learning_rate": 1.3803149857841102e-05,
      "loss": 1.0153,
      "step": 10668
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.083864212036133,
      "learning_rate": 1.3802510941443312e-05,
      "loss": 0.9604,
      "step": 10669
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7213112115859985,
      "learning_rate": 1.3801872025045523e-05,
      "loss": 0.9138,
      "step": 10670
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.121568441390991,
      "learning_rate": 1.3801233108647733e-05,
      "loss": 1.0766,
      "step": 10671
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8883657455444336,
      "learning_rate": 1.3800594192249945e-05,
      "loss": 0.8599,
      "step": 10672
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7576173543930054,
      "learning_rate": 1.3799955275852155e-05,
      "loss": 0.869,
      "step": 10673
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8517743349075317,
      "learning_rate": 1.3799316359454366e-05,
      "loss": 0.9597,
      "step": 10674
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7552740573883057,
      "learning_rate": 1.3798677443056576e-05,
      "loss": 0.9091,
      "step": 10675
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5617305040359497,
      "learning_rate": 1.3798038526658788e-05,
      "loss": 0.867,
      "step": 10676
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5108155012130737,
      "learning_rate": 1.3797399610260997e-05,
      "loss": 0.9689,
      "step": 10677
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6166576147079468,
      "learning_rate": 1.3796760693863209e-05,
      "loss": 0.9117,
      "step": 10678
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9805612564086914,
      "learning_rate": 1.3796121777465419e-05,
      "loss": 1.292,
      "step": 10679
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8558473587036133,
      "learning_rate": 1.379548286106763e-05,
      "loss": 0.9278,
      "step": 10680
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0943057537078857,
      "learning_rate": 1.379484394466984e-05,
      "loss": 1.053,
      "step": 10681
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7278186082839966,
      "learning_rate": 1.3794205028272052e-05,
      "loss": 0.4724,
      "step": 10682
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8777493238449097,
      "learning_rate": 1.3793566111874262e-05,
      "loss": 0.7917,
      "step": 10683
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.917299747467041,
      "learning_rate": 1.3792927195476473e-05,
      "loss": 0.9275,
      "step": 10684
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.028529405593872,
      "learning_rate": 1.3792288279078683e-05,
      "loss": 1.3441,
      "step": 10685
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1285271644592285,
      "learning_rate": 1.3791649362680895e-05,
      "loss": 0.9653,
      "step": 10686
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.0085434913635254,
      "learning_rate": 1.3791010446283104e-05,
      "loss": 1.2635,
      "step": 10687
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7895623445510864,
      "learning_rate": 1.3790371529885316e-05,
      "loss": 0.8256,
      "step": 10688
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.327178478240967,
      "learning_rate": 1.3789732613487526e-05,
      "loss": 0.8874,
      "step": 10689
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6762042045593262,
      "learning_rate": 1.3789093697089737e-05,
      "loss": 0.8574,
      "step": 10690
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5694682598114014,
      "learning_rate": 1.3788454780691947e-05,
      "loss": 0.4001,
      "step": 10691
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9345089197158813,
      "learning_rate": 1.3787815864294159e-05,
      "loss": 0.9353,
      "step": 10692
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.495151400566101,
      "learning_rate": 1.3787176947896369e-05,
      "loss": 0.9551,
      "step": 10693
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.480955123901367,
      "learning_rate": 1.378653803149858e-05,
      "loss": 1.6009,
      "step": 10694
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.269465446472168,
      "learning_rate": 1.378589911510079e-05,
      "loss": 0.9904,
      "step": 10695
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6142762899398804,
      "learning_rate": 1.3785260198703002e-05,
      "loss": 0.9517,
      "step": 10696
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.475356101989746,
      "learning_rate": 1.3784621282305211e-05,
      "loss": 0.8216,
      "step": 10697
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.5358026027679443,
      "learning_rate": 1.3783982365907423e-05,
      "loss": 1.1606,
      "step": 10698
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.889719009399414,
      "learning_rate": 1.3783343449509631e-05,
      "loss": 1.0378,
      "step": 10699
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7504874467849731,
      "learning_rate": 1.3782704533111844e-05,
      "loss": 0.9517,
      "step": 10700
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3911629915237427,
      "learning_rate": 1.3782065616714053e-05,
      "loss": 0.4415,
      "step": 10701
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6396448612213135,
      "learning_rate": 1.3781426700316266e-05,
      "loss": 0.9347,
      "step": 10702
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.170466661453247,
      "learning_rate": 1.3780787783918474e-05,
      "loss": 1.0013,
      "step": 10703
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8380577564239502,
      "learning_rate": 1.3780148867520685e-05,
      "loss": 0.8515,
      "step": 10704
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.20889949798584,
      "learning_rate": 1.3779509951122895e-05,
      "loss": 0.885,
      "step": 10705
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.003274440765381,
      "learning_rate": 1.3778871034725107e-05,
      "loss": 1.2407,
      "step": 10706
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5628077983856201,
      "learning_rate": 1.3778232118327317e-05,
      "loss": 0.8746,
      "step": 10707
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.1165807247161865,
      "learning_rate": 1.3777593201929528e-05,
      "loss": 0.9072,
      "step": 10708
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7161452770233154,
      "learning_rate": 1.3776954285531738e-05,
      "loss": 0.8189,
      "step": 10709
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.609095335006714,
      "learning_rate": 1.377631536913395e-05,
      "loss": 1.1426,
      "step": 10710
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7153620719909668,
      "learning_rate": 1.377567645273616e-05,
      "loss": 0.4239,
      "step": 10711
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3886871337890625,
      "learning_rate": 1.3775037536338371e-05,
      "loss": 1.1032,
      "step": 10712
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1769731044769287,
      "learning_rate": 1.3774398619940581e-05,
      "loss": 0.9382,
      "step": 10713
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9790163040161133,
      "learning_rate": 1.3773759703542792e-05,
      "loss": 0.9941,
      "step": 10714
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8686954975128174,
      "learning_rate": 1.3773120787145002e-05,
      "loss": 0.9199,
      "step": 10715
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.917646884918213,
      "learning_rate": 1.3772481870747214e-05,
      "loss": 0.8564,
      "step": 10716
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2358665466308594,
      "learning_rate": 1.3771842954349424e-05,
      "loss": 0.9006,
      "step": 10717
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.239455699920654,
      "learning_rate": 1.3771204037951635e-05,
      "loss": 1.1676,
      "step": 10718
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8006882667541504,
      "learning_rate": 1.3770565121553845e-05,
      "loss": 0.9287,
      "step": 10719
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4293898344039917,
      "learning_rate": 1.3769926205156057e-05,
      "loss": 0.4193,
      "step": 10720
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2403361797332764,
      "learning_rate": 1.3769287288758267e-05,
      "loss": 0.5724,
      "step": 10721
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2873308658599854,
      "learning_rate": 1.3768648372360478e-05,
      "loss": 1.109,
      "step": 10722
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9766238927841187,
      "learning_rate": 1.3768009455962688e-05,
      "loss": 0.8314,
      "step": 10723
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.101895332336426,
      "learning_rate": 1.37673705395649e-05,
      "loss": 0.91,
      "step": 10724
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8976185321807861,
      "learning_rate": 1.376673162316711e-05,
      "loss": 0.9234,
      "step": 10725
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.813243865966797,
      "learning_rate": 1.3766092706769321e-05,
      "loss": 1.6532,
      "step": 10726
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.86338472366333,
      "learning_rate": 1.376545379037153e-05,
      "loss": 1.0945,
      "step": 10727
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7545806169509888,
      "learning_rate": 1.3764814873973742e-05,
      "loss": 0.8657,
      "step": 10728
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4552558660507202,
      "learning_rate": 1.3764175957575952e-05,
      "loss": 0.9215,
      "step": 10729
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.4785969257354736,
      "learning_rate": 1.3763537041178164e-05,
      "loss": 0.8662,
      "step": 10730
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5415525436401367,
      "learning_rate": 1.3762898124780374e-05,
      "loss": 0.9339,
      "step": 10731
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.968228340148926,
      "learning_rate": 1.3762259208382585e-05,
      "loss": 0.8507,
      "step": 10732
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.451997756958008,
      "learning_rate": 1.3761620291984795e-05,
      "loss": 0.9007,
      "step": 10733
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.904338002204895,
      "learning_rate": 1.3760981375587006e-05,
      "loss": 0.9537,
      "step": 10734
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0401241779327393,
      "learning_rate": 1.3760342459189215e-05,
      "loss": 0.9802,
      "step": 10735
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.71296226978302,
      "learning_rate": 1.3759703542791428e-05,
      "loss": 0.9971,
      "step": 10736
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.880068778991699,
      "learning_rate": 1.3759064626393636e-05,
      "loss": 1.1845,
      "step": 10737
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.9714419841766357,
      "learning_rate": 1.375842570999585e-05,
      "loss": 0.5665,
      "step": 10738
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.85161554813385,
      "learning_rate": 1.3757786793598057e-05,
      "loss": 0.9806,
      "step": 10739
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4973667860031128,
      "learning_rate": 1.3757147877200269e-05,
      "loss": 0.4776,
      "step": 10740
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5830605030059814,
      "learning_rate": 1.3756508960802479e-05,
      "loss": 0.9255,
      "step": 10741
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6133548021316528,
      "learning_rate": 1.375587004440469e-05,
      "loss": 0.3902,
      "step": 10742
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.024075508117676,
      "learning_rate": 1.37552311280069e-05,
      "loss": 0.9164,
      "step": 10743
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8030613660812378,
      "learning_rate": 1.3754592211609112e-05,
      "loss": 0.9219,
      "step": 10744
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.649229049682617,
      "learning_rate": 1.3753953295211322e-05,
      "loss": 0.904,
      "step": 10745
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9560664892196655,
      "learning_rate": 1.3753314378813533e-05,
      "loss": 0.8058,
      "step": 10746
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9537216424942017,
      "learning_rate": 1.3752675462415743e-05,
      "loss": 1.0495,
      "step": 10747
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.704776406288147,
      "learning_rate": 1.3752036546017955e-05,
      "loss": 0.9073,
      "step": 10748
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2048418521881104,
      "learning_rate": 1.3751397629620164e-05,
      "loss": 0.9529,
      "step": 10749
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6764147281646729,
      "learning_rate": 1.3750758713222376e-05,
      "loss": 0.8445,
      "step": 10750
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.653237223625183,
      "learning_rate": 1.3750119796824586e-05,
      "loss": 0.9465,
      "step": 10751
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.787482261657715,
      "learning_rate": 1.3749480880426797e-05,
      "loss": 0.6028,
      "step": 10752
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0150883197784424,
      "learning_rate": 1.3748841964029007e-05,
      "loss": 1.0094,
      "step": 10753
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.65274178981781,
      "learning_rate": 1.3748203047631219e-05,
      "loss": 1.0164,
      "step": 10754
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9837573766708374,
      "learning_rate": 1.3747564131233429e-05,
      "loss": 0.91,
      "step": 10755
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1106457710266113,
      "learning_rate": 1.374692521483564e-05,
      "loss": 0.9426,
      "step": 10756
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3039493560791016,
      "learning_rate": 1.374628629843785e-05,
      "loss": 0.8764,
      "step": 10757
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.382230758666992,
      "learning_rate": 1.3745647382040062e-05,
      "loss": 1.4981,
      "step": 10758
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.64634370803833,
      "learning_rate": 1.3745008465642271e-05,
      "loss": 0.4567,
      "step": 10759
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4279277324676514,
      "learning_rate": 1.3744369549244483e-05,
      "loss": 0.8519,
      "step": 10760
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0539064407348633,
      "learning_rate": 1.3743730632846693e-05,
      "loss": 0.938,
      "step": 10761
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7750074863433838,
      "learning_rate": 1.3743091716448904e-05,
      "loss": 0.9303,
      "step": 10762
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.4638302326202393,
      "learning_rate": 1.3742452800051114e-05,
      "loss": 0.8884,
      "step": 10763
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.143068790435791,
      "learning_rate": 1.3741813883653326e-05,
      "loss": 1.0266,
      "step": 10764
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.469881534576416,
      "learning_rate": 1.3741174967255536e-05,
      "loss": 1.1957,
      "step": 10765
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8980135917663574,
      "learning_rate": 1.3740536050857747e-05,
      "loss": 1.0088,
      "step": 10766
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7651344537734985,
      "learning_rate": 1.3739897134459957e-05,
      "loss": 0.9208,
      "step": 10767
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.777252435684204,
      "learning_rate": 1.3739258218062169e-05,
      "loss": 0.9585,
      "step": 10768
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.3349151611328125,
      "learning_rate": 1.3738619301664378e-05,
      "loss": 1.1989,
      "step": 10769
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.512265920639038,
      "learning_rate": 1.373798038526659e-05,
      "loss": 0.9706,
      "step": 10770
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.0519590377807617,
      "learning_rate": 1.3737341468868798e-05,
      "loss": 0.7182,
      "step": 10771
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.765694499015808,
      "learning_rate": 1.3736702552471011e-05,
      "loss": 0.8712,
      "step": 10772
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.447987079620361,
      "learning_rate": 1.373606363607322e-05,
      "loss": 1.8446,
      "step": 10773
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.476735234260559,
      "learning_rate": 1.3735424719675433e-05,
      "loss": 0.7571,
      "step": 10774
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9188580513000488,
      "learning_rate": 1.3734785803277641e-05,
      "loss": 1.2133,
      "step": 10775
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6997230052947998,
      "learning_rate": 1.3734146886879852e-05,
      "loss": 0.9071,
      "step": 10776
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.102354526519775,
      "learning_rate": 1.3733507970482062e-05,
      "loss": 1.3708,
      "step": 10777
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8097347021102905,
      "learning_rate": 1.3732869054084274e-05,
      "loss": 0.9477,
      "step": 10778
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.3505406379699707,
      "learning_rate": 1.3732230137686484e-05,
      "loss": 0.6347,
      "step": 10779
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.204465866088867,
      "learning_rate": 1.3731591221288695e-05,
      "loss": 1.2353,
      "step": 10780
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5558273792266846,
      "learning_rate": 1.3730952304890905e-05,
      "loss": 0.8293,
      "step": 10781
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.490307092666626,
      "learning_rate": 1.3730313388493117e-05,
      "loss": 0.9463,
      "step": 10782
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3851624727249146,
      "learning_rate": 1.3729674472095327e-05,
      "loss": 0.4552,
      "step": 10783
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.00748348236084,
      "learning_rate": 1.3729035555697538e-05,
      "loss": 1.0308,
      "step": 10784
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.224998474121094,
      "learning_rate": 1.3728396639299748e-05,
      "loss": 1.5161,
      "step": 10785
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9925310611724854,
      "learning_rate": 1.372775772290196e-05,
      "loss": 0.9577,
      "step": 10786
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4989912509918213,
      "learning_rate": 1.372711880650417e-05,
      "loss": 0.4053,
      "step": 10787
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.560319423675537,
      "learning_rate": 1.3726479890106381e-05,
      "loss": 0.8315,
      "step": 10788
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4765841960906982,
      "learning_rate": 1.372584097370859e-05,
      "loss": 0.8356,
      "step": 10789
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6754101514816284,
      "learning_rate": 1.3725202057310802e-05,
      "loss": 0.8062,
      "step": 10790
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1004183292388916,
      "learning_rate": 1.3724563140913012e-05,
      "loss": 1.1178,
      "step": 10791
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4399285316467285,
      "learning_rate": 1.3723924224515224e-05,
      "loss": 0.9138,
      "step": 10792
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7979131937026978,
      "learning_rate": 1.3723285308117434e-05,
      "loss": 0.8682,
      "step": 10793
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.859839677810669,
      "learning_rate": 1.3722646391719645e-05,
      "loss": 0.9473,
      "step": 10794
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6772857904434204,
      "learning_rate": 1.3722007475321855e-05,
      "loss": 0.4555,
      "step": 10795
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8766276836395264,
      "learning_rate": 1.3721368558924066e-05,
      "loss": 0.9255,
      "step": 10796
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7917664051055908,
      "learning_rate": 1.3720729642526276e-05,
      "loss": 1.0229,
      "step": 10797
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.078624725341797,
      "learning_rate": 1.3720090726128488e-05,
      "loss": 0.9988,
      "step": 10798
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0801339149475098,
      "learning_rate": 1.3719451809730698e-05,
      "loss": 0.9275,
      "step": 10799
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.407174825668335,
      "learning_rate": 1.371881289333291e-05,
      "loss": 1.029,
      "step": 10800
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9425169229507446,
      "learning_rate": 1.3718173976935119e-05,
      "loss": 0.9265,
      "step": 10801
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.7819700241088867,
      "learning_rate": 1.371753506053733e-05,
      "loss": 1.3216,
      "step": 10802
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.471883296966553,
      "learning_rate": 1.371689614413954e-05,
      "loss": 1.5521,
      "step": 10803
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8561962842941284,
      "learning_rate": 1.3716257227741752e-05,
      "loss": 0.8862,
      "step": 10804
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9260526895523071,
      "learning_rate": 1.3715618311343962e-05,
      "loss": 0.8498,
      "step": 10805
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.584874153137207,
      "learning_rate": 1.3714979394946173e-05,
      "loss": 1.0285,
      "step": 10806
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7124345302581787,
      "learning_rate": 1.3714340478548382e-05,
      "loss": 0.9113,
      "step": 10807
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8496034145355225,
      "learning_rate": 1.3713701562150595e-05,
      "loss": 0.9269,
      "step": 10808
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7603572607040405,
      "learning_rate": 1.3713062645752803e-05,
      "loss": 1.0272,
      "step": 10809
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.752392292022705,
      "learning_rate": 1.3712423729355016e-05,
      "loss": 0.9977,
      "step": 10810
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.9726760387420654,
      "learning_rate": 1.3711784812957224e-05,
      "loss": 1.0048,
      "step": 10811
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.603327751159668,
      "learning_rate": 1.3711145896559438e-05,
      "loss": 0.9826,
      "step": 10812
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0250234603881836,
      "learning_rate": 1.3710506980161646e-05,
      "loss": 0.9311,
      "step": 10813
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5039753913879395,
      "learning_rate": 1.3709868063763857e-05,
      "loss": 0.7642,
      "step": 10814
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0320897102355957,
      "learning_rate": 1.3709229147366067e-05,
      "loss": 0.934,
      "step": 10815
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.102466106414795,
      "learning_rate": 1.3708590230968279e-05,
      "loss": 1.1265,
      "step": 10816
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.237274408340454,
      "learning_rate": 1.3707951314570489e-05,
      "loss": 0.8354,
      "step": 10817
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.818094253540039,
      "learning_rate": 1.37073123981727e-05,
      "loss": 0.9311,
      "step": 10818
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.237556219100952,
      "learning_rate": 1.370667348177491e-05,
      "loss": 0.6067,
      "step": 10819
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9124921560287476,
      "learning_rate": 1.3706034565377122e-05,
      "loss": 1.0082,
      "step": 10820
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.645174264907837,
      "learning_rate": 1.3705395648979331e-05,
      "loss": 0.7415,
      "step": 10821
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8184627294540405,
      "learning_rate": 1.3704756732581543e-05,
      "loss": 0.8585,
      "step": 10822
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6941157579421997,
      "learning_rate": 1.3704117816183753e-05,
      "loss": 0.8436,
      "step": 10823
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.329411268234253,
      "learning_rate": 1.3703478899785964e-05,
      "loss": 1.2057,
      "step": 10824
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1412436962127686,
      "learning_rate": 1.3702839983388174e-05,
      "loss": 1.1527,
      "step": 10825
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5321311950683594,
      "learning_rate": 1.3702201066990386e-05,
      "loss": 0.843,
      "step": 10826
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8433423042297363,
      "learning_rate": 1.3701562150592596e-05,
      "loss": 1.0027,
      "step": 10827
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7044880390167236,
      "learning_rate": 1.3700923234194807e-05,
      "loss": 0.8921,
      "step": 10828
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.521491527557373,
      "learning_rate": 1.3700284317797017e-05,
      "loss": 1.5925,
      "step": 10829
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.056086540222168,
      "learning_rate": 1.3699645401399229e-05,
      "loss": 1.1295,
      "step": 10830
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.204307794570923,
      "learning_rate": 1.3699006485001438e-05,
      "loss": 1.0783,
      "step": 10831
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4209578037261963,
      "learning_rate": 1.369836756860365e-05,
      "loss": 0.8369,
      "step": 10832
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.3257546424865723,
      "learning_rate": 1.369772865220586e-05,
      "loss": 0.7541,
      "step": 10833
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2041015625,
      "learning_rate": 1.3697089735808071e-05,
      "loss": 0.8916,
      "step": 10834
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9963990449905396,
      "learning_rate": 1.3696450819410281e-05,
      "loss": 0.9173,
      "step": 10835
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9983574151992798,
      "learning_rate": 1.3695811903012493e-05,
      "loss": 1.046,
      "step": 10836
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.657139778137207,
      "learning_rate": 1.3695172986614703e-05,
      "loss": 0.9311,
      "step": 10837
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9426592588424683,
      "learning_rate": 1.3694534070216914e-05,
      "loss": 0.9372,
      "step": 10838
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.858265995979309,
      "learning_rate": 1.3693895153819124e-05,
      "loss": 0.9637,
      "step": 10839
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8026609420776367,
      "learning_rate": 1.3693256237421336e-05,
      "loss": 0.8606,
      "step": 10840
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2614779472351074,
      "learning_rate": 1.3692617321023545e-05,
      "loss": 0.8196,
      "step": 10841
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.21415114402771,
      "learning_rate": 1.3691978404625757e-05,
      "loss": 0.876,
      "step": 10842
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7780094146728516,
      "learning_rate": 1.3691339488227967e-05,
      "loss": 0.4155,
      "step": 10843
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.4066286087036133,
      "learning_rate": 1.3690700571830178e-05,
      "loss": 0.6819,
      "step": 10844
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.4614336490631104,
      "learning_rate": 1.3690061655432387e-05,
      "loss": 1.0246,
      "step": 10845
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8056895732879639,
      "learning_rate": 1.36894227390346e-05,
      "loss": 0.918,
      "step": 10846
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6910439729690552,
      "learning_rate": 1.3688783822636808e-05,
      "loss": 0.8626,
      "step": 10847
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5371851921081543,
      "learning_rate": 1.3688144906239021e-05,
      "loss": 1.0226,
      "step": 10848
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.197813034057617,
      "learning_rate": 1.368750598984123e-05,
      "loss": 1.0668,
      "step": 10849
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.618655800819397,
      "learning_rate": 1.3686867073443441e-05,
      "loss": 0.8453,
      "step": 10850
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.8779778480529785,
      "learning_rate": 1.368622815704565e-05,
      "loss": 1.0093,
      "step": 10851
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.839864730834961,
      "learning_rate": 1.3685589240647862e-05,
      "loss": 1.3365,
      "step": 10852
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9611070156097412,
      "learning_rate": 1.3684950324250072e-05,
      "loss": 0.9978,
      "step": 10853
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8161977529525757,
      "learning_rate": 1.3684311407852284e-05,
      "loss": 0.9415,
      "step": 10854
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.850550651550293,
      "learning_rate": 1.3683672491454494e-05,
      "loss": 1.0159,
      "step": 10855
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.749214768409729,
      "learning_rate": 1.3683033575056705e-05,
      "loss": 0.7912,
      "step": 10856
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6948562860488892,
      "learning_rate": 1.3682394658658915e-05,
      "loss": 0.9626,
      "step": 10857
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5783120393753052,
      "learning_rate": 1.3681755742261126e-05,
      "loss": 0.4416,
      "step": 10858
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.739121675491333,
      "learning_rate": 1.3681116825863336e-05,
      "loss": 0.8872,
      "step": 10859
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6365954875946045,
      "learning_rate": 1.3680477909465548e-05,
      "loss": 0.8701,
      "step": 10860
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.648824691772461,
      "learning_rate": 1.3679838993067758e-05,
      "loss": 0.8687,
      "step": 10861
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.632875442504883,
      "learning_rate": 1.367920007666997e-05,
      "loss": 1.3706,
      "step": 10862
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.669013500213623,
      "learning_rate": 1.3678561160272179e-05,
      "loss": 0.944,
      "step": 10863
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9327518939971924,
      "learning_rate": 1.367792224387439e-05,
      "loss": 0.9225,
      "step": 10864
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9665154218673706,
      "learning_rate": 1.36772833274766e-05,
      "loss": 0.9675,
      "step": 10865
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7663570642471313,
      "learning_rate": 1.3676644411078812e-05,
      "loss": 1.0275,
      "step": 10866
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8963112831115723,
      "learning_rate": 1.3676005494681022e-05,
      "loss": 0.9499,
      "step": 10867
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8562413454055786,
      "learning_rate": 1.3675366578283233e-05,
      "loss": 0.959,
      "step": 10868
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.0593178272247314,
      "learning_rate": 1.3674727661885443e-05,
      "loss": 1.2779,
      "step": 10869
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.811968207359314,
      "learning_rate": 1.3674088745487655e-05,
      "loss": 0.8253,
      "step": 10870
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.742317795753479,
      "learning_rate": 1.3673449829089865e-05,
      "loss": 0.8512,
      "step": 10871
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.499697208404541,
      "learning_rate": 1.3672810912692076e-05,
      "loss": 0.8408,
      "step": 10872
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.775005340576172,
      "learning_rate": 1.3672171996294286e-05,
      "loss": 1.6805,
      "step": 10873
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.728551387786865,
      "learning_rate": 1.3671533079896498e-05,
      "loss": 1.2491,
      "step": 10874
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8737841844558716,
      "learning_rate": 1.3670894163498708e-05,
      "loss": 1.1917,
      "step": 10875
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.637755036354065,
      "learning_rate": 1.3670255247100919e-05,
      "loss": 1.1309,
      "step": 10876
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8774566650390625,
      "learning_rate": 1.3669616330703129e-05,
      "loss": 0.9363,
      "step": 10877
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0812792778015137,
      "learning_rate": 1.366897741430534e-05,
      "loss": 0.9004,
      "step": 10878
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.376432418823242,
      "learning_rate": 1.366833849790755e-05,
      "loss": 0.9678,
      "step": 10879
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.742591142654419,
      "learning_rate": 1.3667699581509762e-05,
      "loss": 0.9119,
      "step": 10880
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1214170455932617,
      "learning_rate": 1.366706066511197e-05,
      "loss": 0.7972,
      "step": 10881
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1198737621307373,
      "learning_rate": 1.3666421748714183e-05,
      "loss": 0.9492,
      "step": 10882
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.8582732677459717,
      "learning_rate": 1.3665782832316391e-05,
      "loss": 1.0169,
      "step": 10883
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.090111255645752,
      "learning_rate": 1.3665143915918605e-05,
      "loss": 1.172,
      "step": 10884
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.342571973800659,
      "learning_rate": 1.3664504999520813e-05,
      "loss": 0.951,
      "step": 10885
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7737438678741455,
      "learning_rate": 1.3663866083123024e-05,
      "loss": 0.9977,
      "step": 10886
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8322449922561646,
      "learning_rate": 1.3663227166725234e-05,
      "loss": 0.9944,
      "step": 10887
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.575533151626587,
      "learning_rate": 1.3662588250327446e-05,
      "loss": 0.9227,
      "step": 10888
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.393839120864868,
      "learning_rate": 1.3661949333929656e-05,
      "loss": 0.7587,
      "step": 10889
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.758030891418457,
      "learning_rate": 1.3661310417531867e-05,
      "loss": 0.9656,
      "step": 10890
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7449723482131958,
      "learning_rate": 1.3660671501134077e-05,
      "loss": 0.9224,
      "step": 10891
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9699302911758423,
      "learning_rate": 1.3660032584736289e-05,
      "loss": 0.8634,
      "step": 10892
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.779856562614441,
      "learning_rate": 1.3659393668338498e-05,
      "loss": 0.9679,
      "step": 10893
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3665649890899658,
      "learning_rate": 1.365875475194071e-05,
      "loss": 0.4426,
      "step": 10894
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1973581314086914,
      "learning_rate": 1.365811583554292e-05,
      "loss": 1.1126,
      "step": 10895
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.237630844116211,
      "learning_rate": 1.3657476919145131e-05,
      "loss": 1.1452,
      "step": 10896
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.8727986812591553,
      "learning_rate": 1.3656838002747341e-05,
      "loss": 1.2182,
      "step": 10897
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.324472427368164,
      "learning_rate": 1.3656199086349553e-05,
      "loss": 1.1637,
      "step": 10898
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.385129928588867,
      "learning_rate": 1.3655560169951763e-05,
      "loss": 0.8043,
      "step": 10899
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.006033182144165,
      "learning_rate": 1.3654921253553974e-05,
      "loss": 1.1615,
      "step": 10900
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3147425651550293,
      "learning_rate": 1.3654282337156184e-05,
      "loss": 1.1457,
      "step": 10901
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8309426307678223,
      "learning_rate": 1.3653643420758396e-05,
      "loss": 0.8834,
      "step": 10902
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9724388122558594,
      "learning_rate": 1.3653004504360605e-05,
      "loss": 1.0032,
      "step": 10903
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6738038063049316,
      "learning_rate": 1.3652365587962817e-05,
      "loss": 0.9603,
      "step": 10904
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5909079313278198,
      "learning_rate": 1.3651726671565027e-05,
      "loss": 0.9369,
      "step": 10905
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.864595890045166,
      "learning_rate": 1.3651087755167238e-05,
      "loss": 0.8784,
      "step": 10906
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.542550802230835,
      "learning_rate": 1.3650448838769448e-05,
      "loss": 0.9347,
      "step": 10907
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.585580825805664,
      "learning_rate": 1.364980992237166e-05,
      "loss": 1.1393,
      "step": 10908
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5340758562088013,
      "learning_rate": 1.364917100597387e-05,
      "loss": 0.425,
      "step": 10909
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.6415252685546875,
      "learning_rate": 1.3648532089576081e-05,
      "loss": 1.2703,
      "step": 10910
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9380601644515991,
      "learning_rate": 1.3647893173178291e-05,
      "loss": 1.1928,
      "step": 10911
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5860097408294678,
      "learning_rate": 1.3647254256780503e-05,
      "loss": 0.87,
      "step": 10912
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3629539012908936,
      "learning_rate": 1.3646615340382712e-05,
      "loss": 0.9911,
      "step": 10913
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.420903205871582,
      "learning_rate": 1.3645976423984924e-05,
      "loss": 0.8037,
      "step": 10914
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1133737564086914,
      "learning_rate": 1.3645337507587134e-05,
      "loss": 1.0077,
      "step": 10915
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.7208964824676514,
      "learning_rate": 1.3644698591189345e-05,
      "loss": 1.1463,
      "step": 10916
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.847043991088867,
      "learning_rate": 1.3644059674791554e-05,
      "loss": 1.1115,
      "step": 10917
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.703069686889648,
      "learning_rate": 1.3643420758393767e-05,
      "loss": 1.3016,
      "step": 10918
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8397942781448364,
      "learning_rate": 1.3642781841995975e-05,
      "loss": 0.9436,
      "step": 10919
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9912439584732056,
      "learning_rate": 1.3642142925598188e-05,
      "loss": 0.4607,
      "step": 10920
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.272810220718384,
      "learning_rate": 1.3641504009200396e-05,
      "loss": 1.0063,
      "step": 10921
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8198657035827637,
      "learning_rate": 1.3640865092802608e-05,
      "loss": 0.9554,
      "step": 10922
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6076327562332153,
      "learning_rate": 1.3640226176404818e-05,
      "loss": 0.8794,
      "step": 10923
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6197360754013062,
      "learning_rate": 1.363958726000703e-05,
      "loss": 0.8875,
      "step": 10924
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8489192724227905,
      "learning_rate": 1.3638948343609239e-05,
      "loss": 0.9701,
      "step": 10925
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.5935542583465576,
      "learning_rate": 1.363830942721145e-05,
      "loss": 1.0718,
      "step": 10926
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9121259450912476,
      "learning_rate": 1.363767051081366e-05,
      "loss": 0.9637,
      "step": 10927
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7925026416778564,
      "learning_rate": 1.3637031594415872e-05,
      "loss": 0.8986,
      "step": 10928
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9421491622924805,
      "learning_rate": 1.3636392678018082e-05,
      "loss": 0.9544,
      "step": 10929
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.332832098007202,
      "learning_rate": 1.3635753761620293e-05,
      "loss": 1.0279,
      "step": 10930
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8962626457214355,
      "learning_rate": 1.3635114845222503e-05,
      "loss": 0.8453,
      "step": 10931
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.749058723449707,
      "learning_rate": 1.3634475928824715e-05,
      "loss": 0.8273,
      "step": 10932
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6212860345840454,
      "learning_rate": 1.3633837012426925e-05,
      "loss": 0.8993,
      "step": 10933
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.6101818084716797,
      "learning_rate": 1.3633198096029136e-05,
      "loss": 1.0385,
      "step": 10934
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.498161792755127,
      "learning_rate": 1.3632559179631346e-05,
      "loss": 1.0752,
      "step": 10935
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9958828687667847,
      "learning_rate": 1.3631920263233558e-05,
      "loss": 0.9749,
      "step": 10936
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5934585332870483,
      "learning_rate": 1.3631281346835768e-05,
      "loss": 0.4594,
      "step": 10937
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.675477504730225,
      "learning_rate": 1.3630642430437979e-05,
      "loss": 1.6803,
      "step": 10938
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6782748699188232,
      "learning_rate": 1.3630003514040189e-05,
      "loss": 0.9214,
      "step": 10939
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.074204921722412,
      "learning_rate": 1.36293645976424e-05,
      "loss": 1.2139,
      "step": 10940
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7518784999847412,
      "learning_rate": 1.362872568124461e-05,
      "loss": 0.906,
      "step": 10941
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5751758813858032,
      "learning_rate": 1.3628086764846822e-05,
      "loss": 0.8451,
      "step": 10942
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.064423084259033,
      "learning_rate": 1.3627447848449032e-05,
      "loss": 0.9547,
      "step": 10943
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3796234130859375,
      "learning_rate": 1.3626808932051243e-05,
      "loss": 1.1229,
      "step": 10944
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.635175347328186,
      "learning_rate": 1.3626170015653453e-05,
      "loss": 0.9653,
      "step": 10945
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1257147789001465,
      "learning_rate": 1.3625531099255665e-05,
      "loss": 1.0265,
      "step": 10946
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6217035055160522,
      "learning_rate": 1.3624892182857875e-05,
      "loss": 0.9655,
      "step": 10947
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.705793023109436,
      "learning_rate": 1.3624253266460086e-05,
      "loss": 0.9902,
      "step": 10948
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3492965698242188,
      "learning_rate": 1.3623614350062296e-05,
      "loss": 0.98,
      "step": 10949
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0195906162261963,
      "learning_rate": 1.3622975433664507e-05,
      "loss": 0.7844,
      "step": 10950
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.232149600982666,
      "learning_rate": 1.3622336517266717e-05,
      "loss": 1.1258,
      "step": 10951
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.177140951156616,
      "learning_rate": 1.3621697600868929e-05,
      "loss": 0.8386,
      "step": 10952
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4926563501358032,
      "learning_rate": 1.3621058684471137e-05,
      "loss": 0.9268,
      "step": 10953
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9741756916046143,
      "learning_rate": 1.362041976807335e-05,
      "loss": 0.9972,
      "step": 10954
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.565171718597412,
      "learning_rate": 1.3619780851675558e-05,
      "loss": 0.9578,
      "step": 10955
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7744934558868408,
      "learning_rate": 1.3619141935277772e-05,
      "loss": 0.8414,
      "step": 10956
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.442237377166748,
      "learning_rate": 1.361850301887998e-05,
      "loss": 0.8955,
      "step": 10957
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.660146951675415,
      "learning_rate": 1.3617864102482193e-05,
      "loss": 0.7559,
      "step": 10958
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.4341325759887695,
      "learning_rate": 1.3617225186084401e-05,
      "loss": 1.018,
      "step": 10959
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9509663581848145,
      "learning_rate": 1.3616586269686613e-05,
      "loss": 1.1171,
      "step": 10960
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.20096492767334,
      "learning_rate": 1.3615947353288823e-05,
      "loss": 1.1549,
      "step": 10961
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7123868465423584,
      "learning_rate": 1.3615308436891034e-05,
      "loss": 0.9856,
      "step": 10962
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.658929824829102,
      "learning_rate": 1.3614669520493244e-05,
      "loss": 1.3148,
      "step": 10963
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.945754051208496,
      "learning_rate": 1.3614030604095456e-05,
      "loss": 1.2281,
      "step": 10964
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0496068000793457,
      "learning_rate": 1.3613391687697665e-05,
      "loss": 0.832,
      "step": 10965
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8826780319213867,
      "learning_rate": 1.3612752771299877e-05,
      "loss": 0.9845,
      "step": 10966
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.46474027633667,
      "learning_rate": 1.3612113854902087e-05,
      "loss": 0.9469,
      "step": 10967
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.3318629264831543,
      "learning_rate": 1.3611474938504298e-05,
      "loss": 0.9726,
      "step": 10968
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.5060904026031494,
      "learning_rate": 1.3610836022106508e-05,
      "loss": 1.1125,
      "step": 10969
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.177656888961792,
      "learning_rate": 1.361019710570872e-05,
      "loss": 1.0619,
      "step": 10970
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5140209197998047,
      "learning_rate": 1.360955818931093e-05,
      "loss": 0.8797,
      "step": 10971
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0959105491638184,
      "learning_rate": 1.3608919272913141e-05,
      "loss": 0.9996,
      "step": 10972
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8347355127334595,
      "learning_rate": 1.3608280356515351e-05,
      "loss": 0.8382,
      "step": 10973
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.879348039627075,
      "learning_rate": 1.3607641440117563e-05,
      "loss": 0.9581,
      "step": 10974
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7492371797561646,
      "learning_rate": 1.3607002523719772e-05,
      "loss": 0.8159,
      "step": 10975
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5267343521118164,
      "learning_rate": 1.3606363607321984e-05,
      "loss": 0.6434,
      "step": 10976
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.827383041381836,
      "learning_rate": 1.3605724690924194e-05,
      "loss": 0.977,
      "step": 10977
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9487992525100708,
      "learning_rate": 1.3605085774526405e-05,
      "loss": 0.9464,
      "step": 10978
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7420306205749512,
      "learning_rate": 1.3604446858128615e-05,
      "loss": 0.9288,
      "step": 10979
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.040494918823242,
      "learning_rate": 1.3603807941730827e-05,
      "loss": 0.8073,
      "step": 10980
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8973512649536133,
      "learning_rate": 1.3603169025333037e-05,
      "loss": 0.915,
      "step": 10981
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.636810064315796,
      "learning_rate": 1.3602530108935248e-05,
      "loss": 0.851,
      "step": 10982
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5057971477508545,
      "learning_rate": 1.3601891192537458e-05,
      "loss": 0.932,
      "step": 10983
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.7678744792938232,
      "learning_rate": 1.360125227613967e-05,
      "loss": 1.0699,
      "step": 10984
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9363174438476562,
      "learning_rate": 1.360061335974188e-05,
      "loss": 0.8837,
      "step": 10985
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3916957378387451,
      "learning_rate": 1.3599974443344091e-05,
      "loss": 0.9328,
      "step": 10986
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0378410816192627,
      "learning_rate": 1.3599335526946301e-05,
      "loss": 0.9061,
      "step": 10987
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0404844284057617,
      "learning_rate": 1.3598696610548512e-05,
      "loss": 0.7281,
      "step": 10988
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.299131155014038,
      "learning_rate": 1.3598057694150722e-05,
      "loss": 0.8003,
      "step": 10989
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.67633855342865,
      "learning_rate": 1.3597418777752934e-05,
      "loss": 0.9173,
      "step": 10990
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9898669719696045,
      "learning_rate": 1.3596779861355142e-05,
      "loss": 0.8467,
      "step": 10991
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0758447647094727,
      "learning_rate": 1.3596140944957355e-05,
      "loss": 0.9106,
      "step": 10992
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1173150539398193,
      "learning_rate": 1.3595502028559563e-05,
      "loss": 0.9472,
      "step": 10993
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8555729389190674,
      "learning_rate": 1.3594863112161777e-05,
      "loss": 0.8425,
      "step": 10994
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.881173849105835,
      "learning_rate": 1.3594224195763985e-05,
      "loss": 0.958,
      "step": 10995
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2227067947387695,
      "learning_rate": 1.3593585279366196e-05,
      "loss": 0.9449,
      "step": 10996
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0758626461029053,
      "learning_rate": 1.3592946362968406e-05,
      "loss": 0.7861,
      "step": 10997
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.351461410522461,
      "learning_rate": 1.3592307446570618e-05,
      "loss": 1.0734,
      "step": 10998
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0202064514160156,
      "learning_rate": 1.3591668530172828e-05,
      "loss": 0.8341,
      "step": 10999
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0354623794555664,
      "learning_rate": 1.3591029613775039e-05,
      "loss": 1.0047,
      "step": 11000
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.0598883628845215,
      "learning_rate": 1.3590390697377249e-05,
      "loss": 1.0531,
      "step": 11001
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0271365642547607,
      "learning_rate": 1.358975178097946e-05,
      "loss": 0.9035,
      "step": 11002
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.8698153495788574,
      "learning_rate": 1.358911286458167e-05,
      "loss": 0.9332,
      "step": 11003
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9003984928131104,
      "learning_rate": 1.3588473948183882e-05,
      "loss": 1.1757,
      "step": 11004
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.902292251586914,
      "learning_rate": 1.3587835031786092e-05,
      "loss": 1.3307,
      "step": 11005
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5142581462860107,
      "learning_rate": 1.3587196115388303e-05,
      "loss": 1.1674,
      "step": 11006
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.4546730518341064,
      "learning_rate": 1.3586557198990513e-05,
      "loss": 1.0905,
      "step": 11007
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2942891120910645,
      "learning_rate": 1.3585918282592725e-05,
      "loss": 1.0086,
      "step": 11008
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.7742764949798584,
      "learning_rate": 1.3585279366194935e-05,
      "loss": 0.8945,
      "step": 11009
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.5673744678497314,
      "learning_rate": 1.3584640449797146e-05,
      "loss": 1.4193,
      "step": 11010
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8446779251098633,
      "learning_rate": 1.3584001533399356e-05,
      "loss": 0.7418,
      "step": 11011
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.773305058479309,
      "learning_rate": 1.3583362617001568e-05,
      "loss": 0.871,
      "step": 11012
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.207526206970215,
      "learning_rate": 1.3582723700603777e-05,
      "loss": 0.785,
      "step": 11013
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9787116050720215,
      "learning_rate": 1.3582084784205989e-05,
      "loss": 0.9757,
      "step": 11014
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.740778684616089,
      "learning_rate": 1.3581445867808199e-05,
      "loss": 0.7797,
      "step": 11015
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.611247181892395,
      "learning_rate": 1.358080695141041e-05,
      "loss": 0.9349,
      "step": 11016
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.579484462738037,
      "learning_rate": 1.358016803501262e-05,
      "loss": 0.7748,
      "step": 11017
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0323216915130615,
      "learning_rate": 1.357952911861483e-05,
      "loss": 0.919,
      "step": 11018
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.936365842819214,
      "learning_rate": 1.3578890202217042e-05,
      "loss": 1.2058,
      "step": 11019
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.72908091545105,
      "learning_rate": 1.357825128581925e-05,
      "loss": 1.0078,
      "step": 11020
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7220216989517212,
      "learning_rate": 1.3577612369421463e-05,
      "loss": 0.9708,
      "step": 11021
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.851152777671814,
      "learning_rate": 1.3576973453023671e-05,
      "loss": 1.1638,
      "step": 11022
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.046102523803711,
      "learning_rate": 1.3576334536625884e-05,
      "loss": 0.9563,
      "step": 11023
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.208440780639648,
      "learning_rate": 1.3575695620228093e-05,
      "loss": 1.0548,
      "step": 11024
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3657231330871582,
      "learning_rate": 1.3575056703830306e-05,
      "loss": 0.5543,
      "step": 11025
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6613181829452515,
      "learning_rate": 1.3574417787432514e-05,
      "loss": 0.9412,
      "step": 11026
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5063536167144775,
      "learning_rate": 1.3573778871034725e-05,
      "loss": 1.0058,
      "step": 11027
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.537776231765747,
      "learning_rate": 1.3573139954636935e-05,
      "loss": 0.8913,
      "step": 11028
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.40036678314209,
      "learning_rate": 1.3572501038239147e-05,
      "loss": 0.6275,
      "step": 11029
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.954033374786377,
      "learning_rate": 1.3571862121841357e-05,
      "loss": 0.8031,
      "step": 11030
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7326879501342773,
      "learning_rate": 1.3571223205443568e-05,
      "loss": 1.0097,
      "step": 11031
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.706757068634033,
      "learning_rate": 1.3570584289045778e-05,
      "loss": 1.4037,
      "step": 11032
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7115702629089355,
      "learning_rate": 1.356994537264799e-05,
      "loss": 0.8603,
      "step": 11033
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.638897180557251,
      "learning_rate": 1.35693064562502e-05,
      "loss": 0.9189,
      "step": 11034
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.583350658416748,
      "learning_rate": 1.3568667539852411e-05,
      "loss": 0.934,
      "step": 11035
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8372677564620972,
      "learning_rate": 1.3568028623454621e-05,
      "loss": 0.8169,
      "step": 11036
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6052806377410889,
      "learning_rate": 1.3567389707056832e-05,
      "loss": 0.9339,
      "step": 11037
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.650564432144165,
      "learning_rate": 1.3566750790659042e-05,
      "loss": 0.9047,
      "step": 11038
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5288492441177368,
      "learning_rate": 1.3566111874261254e-05,
      "loss": 0.8438,
      "step": 11039
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7872440814971924,
      "learning_rate": 1.3565472957863464e-05,
      "loss": 1.0255,
      "step": 11040
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8427964448928833,
      "learning_rate": 1.3564834041465675e-05,
      "loss": 0.9077,
      "step": 11041
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.1202850341796875,
      "learning_rate": 1.3564195125067885e-05,
      "loss": 1.0849,
      "step": 11042
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.4800591468811035,
      "learning_rate": 1.3563556208670097e-05,
      "loss": 0.8416,
      "step": 11043
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.130643844604492,
      "learning_rate": 1.3562917292272307e-05,
      "loss": 1.095,
      "step": 11044
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.797218918800354,
      "learning_rate": 1.3562278375874518e-05,
      "loss": 0.7722,
      "step": 11045
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.94999098777771,
      "learning_rate": 1.3561639459476728e-05,
      "loss": 0.8027,
      "step": 11046
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6858341693878174,
      "learning_rate": 1.356100054307894e-05,
      "loss": 0.8604,
      "step": 11047
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6821298599243164,
      "learning_rate": 1.356036162668115e-05,
      "loss": 0.9564,
      "step": 11048
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8179255723953247,
      "learning_rate": 1.3559722710283361e-05,
      "loss": 0.6567,
      "step": 11049
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9020888805389404,
      "learning_rate": 1.355908379388557e-05,
      "loss": 0.9791,
      "step": 11050
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5563561916351318,
      "learning_rate": 1.3558444877487782e-05,
      "loss": 0.8971,
      "step": 11051
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.134857177734375,
      "learning_rate": 1.3557805961089992e-05,
      "loss": 1.1657,
      "step": 11052
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.532442808151245,
      "learning_rate": 1.3557167044692204e-05,
      "loss": 1.1055,
      "step": 11053
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.17499041557312,
      "learning_rate": 1.3556528128294414e-05,
      "loss": 1.0648,
      "step": 11054
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.050557613372803,
      "learning_rate": 1.3555889211896625e-05,
      "loss": 1.3615,
      "step": 11055
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.0219125747680664,
      "learning_rate": 1.3555250295498835e-05,
      "loss": 1.347,
      "step": 11056
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.527743816375732,
      "learning_rate": 1.3554611379101046e-05,
      "loss": 1.2129,
      "step": 11057
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7971924543380737,
      "learning_rate": 1.3553972462703255e-05,
      "loss": 0.8146,
      "step": 11058
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8661085367202759,
      "learning_rate": 1.3553333546305468e-05,
      "loss": 0.8947,
      "step": 11059
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.401904344558716,
      "learning_rate": 1.3552694629907676e-05,
      "loss": 1.0898,
      "step": 11060
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7620865106582642,
      "learning_rate": 1.355205571350989e-05,
      "loss": 0.8807,
      "step": 11061
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.683542251586914,
      "learning_rate": 1.3551416797112097e-05,
      "loss": 1.1785,
      "step": 11062
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.29140043258667,
      "learning_rate": 1.3550777880714309e-05,
      "loss": 1.0292,
      "step": 11063
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6975923776626587,
      "learning_rate": 1.3550138964316519e-05,
      "loss": 0.8962,
      "step": 11064
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.3324055671691895,
      "learning_rate": 1.354950004791873e-05,
      "loss": 0.8655,
      "step": 11065
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.642617106437683,
      "learning_rate": 1.354886113152094e-05,
      "loss": 0.9127,
      "step": 11066
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4470239877700806,
      "learning_rate": 1.3548222215123152e-05,
      "loss": 0.3882,
      "step": 11067
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.7259154319763184,
      "learning_rate": 1.3547583298725362e-05,
      "loss": 1.1199,
      "step": 11068
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.805960774421692,
      "learning_rate": 1.3546944382327573e-05,
      "loss": 0.9302,
      "step": 11069
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4495930671691895,
      "learning_rate": 1.3546305465929783e-05,
      "loss": 0.4294,
      "step": 11070
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.4007434844970703,
      "learning_rate": 1.3545666549531995e-05,
      "loss": 1.3005,
      "step": 11071
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8771754503250122,
      "learning_rate": 1.3545027633134204e-05,
      "loss": 0.9322,
      "step": 11072
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.163179874420166,
      "learning_rate": 1.3544388716736416e-05,
      "loss": 0.9628,
      "step": 11073
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9853335618972778,
      "learning_rate": 1.3543749800338626e-05,
      "loss": 1.0307,
      "step": 11074
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.2835328578948975,
      "learning_rate": 1.3543110883940837e-05,
      "loss": 1.062,
      "step": 11075
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.051903247833252,
      "learning_rate": 1.3542471967543047e-05,
      "loss": 1.2271,
      "step": 11076
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8446153402328491,
      "learning_rate": 1.3541833051145259e-05,
      "loss": 0.8946,
      "step": 11077
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.125112056732178,
      "learning_rate": 1.3541194134747469e-05,
      "loss": 1.2895,
      "step": 11078
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.250373363494873,
      "learning_rate": 1.354055521834968e-05,
      "loss": 1.2373,
      "step": 11079
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.132481813430786,
      "learning_rate": 1.353991630195189e-05,
      "loss": 0.7384,
      "step": 11080
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1177146434783936,
      "learning_rate": 1.3539277385554102e-05,
      "loss": 0.8544,
      "step": 11081
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7015862464904785,
      "learning_rate": 1.3538638469156311e-05,
      "loss": 0.8556,
      "step": 11082
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4179463386535645,
      "learning_rate": 1.3537999552758523e-05,
      "loss": 0.8745,
      "step": 11083
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.880638837814331,
      "learning_rate": 1.3537360636360733e-05,
      "loss": 1.5572,
      "step": 11084
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4588704109191895,
      "learning_rate": 1.3536721719962944e-05,
      "loss": 0.4727,
      "step": 11085
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9753551483154297,
      "learning_rate": 1.3536082803565154e-05,
      "loss": 0.9052,
      "step": 11086
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7714015245437622,
      "learning_rate": 1.3535443887167366e-05,
      "loss": 0.8804,
      "step": 11087
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5747050046920776,
      "learning_rate": 1.3534804970769576e-05,
      "loss": 0.8601,
      "step": 11088
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8404525518417358,
      "learning_rate": 1.3534166054371787e-05,
      "loss": 0.7715,
      "step": 11089
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.0933709144592285,
      "learning_rate": 1.3533527137973997e-05,
      "loss": 1.1406,
      "step": 11090
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.923982858657837,
      "learning_rate": 1.3532888221576209e-05,
      "loss": 1.0152,
      "step": 11091
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7307549715042114,
      "learning_rate": 1.3532249305178418e-05,
      "loss": 0.8267,
      "step": 11092
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.434888243675232,
      "learning_rate": 1.353161038878063e-05,
      "loss": 0.9194,
      "step": 11093
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1120829582214355,
      "learning_rate": 1.3530971472382838e-05,
      "loss": 1.005,
      "step": 11094
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6405770778656006,
      "learning_rate": 1.3530332555985051e-05,
      "loss": 1.3434,
      "step": 11095
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7575727701187134,
      "learning_rate": 1.352969363958726e-05,
      "loss": 1.0362,
      "step": 11096
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.985528588294983,
      "learning_rate": 1.3529054723189473e-05,
      "loss": 0.9825,
      "step": 11097
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7715439796447754,
      "learning_rate": 1.3528415806791681e-05,
      "loss": 0.8922,
      "step": 11098
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5806341171264648,
      "learning_rate": 1.3527776890393892e-05,
      "loss": 0.8659,
      "step": 11099
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9687118530273438,
      "learning_rate": 1.3527137973996102e-05,
      "loss": 0.8328,
      "step": 11100
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5987106561660767,
      "learning_rate": 1.3526499057598314e-05,
      "loss": 0.9948,
      "step": 11101
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6361830234527588,
      "learning_rate": 1.3525860141200524e-05,
      "loss": 0.8962,
      "step": 11102
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5839555263519287,
      "learning_rate": 1.3525221224802735e-05,
      "loss": 0.9484,
      "step": 11103
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.7932627201080322,
      "learning_rate": 1.3524582308404945e-05,
      "loss": 1.2361,
      "step": 11104
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.039382219314575,
      "learning_rate": 1.3523943392007157e-05,
      "loss": 0.9841,
      "step": 11105
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6792128086090088,
      "learning_rate": 1.3523304475609367e-05,
      "loss": 0.858,
      "step": 11106
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.937814712524414,
      "learning_rate": 1.3522665559211578e-05,
      "loss": 0.9573,
      "step": 11107
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6719995737075806,
      "learning_rate": 1.3522026642813788e-05,
      "loss": 0.9972,
      "step": 11108
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4941325187683105,
      "learning_rate": 1.3521387726416e-05,
      "loss": 0.7724,
      "step": 11109
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0461671352386475,
      "learning_rate": 1.352074881001821e-05,
      "loss": 1.0611,
      "step": 11110
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2359819412231445,
      "learning_rate": 1.3520109893620421e-05,
      "loss": 1.138,
      "step": 11111
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.340365886688232,
      "learning_rate": 1.351947097722263e-05,
      "loss": 1.5899,
      "step": 11112
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.17802095413208,
      "learning_rate": 1.3518832060824842e-05,
      "loss": 0.8825,
      "step": 11113
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.692568063735962,
      "learning_rate": 1.3518193144427052e-05,
      "loss": 0.826,
      "step": 11114
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7878375053405762,
      "learning_rate": 1.3517554228029264e-05,
      "loss": 0.9122,
      "step": 11115
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6083004474639893,
      "learning_rate": 1.3516915311631474e-05,
      "loss": 0.9866,
      "step": 11116
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5941823720932007,
      "learning_rate": 1.3516276395233685e-05,
      "loss": 0.8667,
      "step": 11117
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.184575319290161,
      "learning_rate": 1.3515637478835895e-05,
      "loss": 1.1761,
      "step": 11118
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9253802299499512,
      "learning_rate": 1.3514998562438106e-05,
      "loss": 0.808,
      "step": 11119
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9969878196716309,
      "learning_rate": 1.3514359646040316e-05,
      "loss": 0.8741,
      "step": 11120
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6005662679672241,
      "learning_rate": 1.3513720729642528e-05,
      "loss": 0.4006,
      "step": 11121
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1435601711273193,
      "learning_rate": 1.3513081813244738e-05,
      "loss": 0.911,
      "step": 11122
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7631793022155762,
      "learning_rate": 1.351244289684695e-05,
      "loss": 0.9101,
      "step": 11123
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9444210529327393,
      "learning_rate": 1.3511803980449159e-05,
      "loss": 1.2569,
      "step": 11124
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.608010172843933,
      "learning_rate": 1.351116506405137e-05,
      "loss": 0.8497,
      "step": 11125
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.767702341079712,
      "learning_rate": 1.351052614765358e-05,
      "loss": 0.8847,
      "step": 11126
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1547248363494873,
      "learning_rate": 1.3509887231255792e-05,
      "loss": 1.0026,
      "step": 11127
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.674978256225586,
      "learning_rate": 1.3509248314858002e-05,
      "loss": 0.9908,
      "step": 11128
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5004842281341553,
      "learning_rate": 1.3508609398460213e-05,
      "loss": 0.7909,
      "step": 11129
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6366140842437744,
      "learning_rate": 1.3507970482062422e-05,
      "loss": 0.7648,
      "step": 11130
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.015240430831909,
      "learning_rate": 1.3507331565664635e-05,
      "loss": 1.0255,
      "step": 11131
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.428575038909912,
      "learning_rate": 1.3506692649266843e-05,
      "loss": 1.2636,
      "step": 11132
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3029770851135254,
      "learning_rate": 1.3506053732869056e-05,
      "loss": 0.9743,
      "step": 11133
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9235496520996094,
      "learning_rate": 1.3505414816471264e-05,
      "loss": 1.2425,
      "step": 11134
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.019860029220581,
      "learning_rate": 1.3504775900073478e-05,
      "loss": 0.8974,
      "step": 11135
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.465226173400879,
      "learning_rate": 1.3504136983675686e-05,
      "loss": 0.9887,
      "step": 11136
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.479466438293457,
      "learning_rate": 1.3503498067277897e-05,
      "loss": 1.471,
      "step": 11137
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.279913425445557,
      "learning_rate": 1.3502859150880107e-05,
      "loss": 0.9122,
      "step": 11138
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5871808528900146,
      "learning_rate": 1.3502220234482319e-05,
      "loss": 0.8202,
      "step": 11139
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.081198215484619,
      "learning_rate": 1.3501581318084529e-05,
      "loss": 0.9941,
      "step": 11140
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.585336923599243,
      "learning_rate": 1.350094240168674e-05,
      "loss": 0.938,
      "step": 11141
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8146311044692993,
      "learning_rate": 1.350030348528895e-05,
      "loss": 0.8369,
      "step": 11142
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.672381043434143,
      "learning_rate": 1.3499664568891162e-05,
      "loss": 0.8703,
      "step": 11143
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9259183406829834,
      "learning_rate": 1.3499025652493371e-05,
      "loss": 1.4428,
      "step": 11144
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9908250570297241,
      "learning_rate": 1.3498386736095583e-05,
      "loss": 0.9838,
      "step": 11145
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9116199016571045,
      "learning_rate": 1.3497747819697793e-05,
      "loss": 1.0673,
      "step": 11146
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1961123943328857,
      "learning_rate": 1.3497108903300004e-05,
      "loss": 0.8426,
      "step": 11147
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.035834789276123,
      "learning_rate": 1.3496469986902214e-05,
      "loss": 0.9512,
      "step": 11148
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.5717368125915527,
      "learning_rate": 1.3495831070504426e-05,
      "loss": 0.9466,
      "step": 11149
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.6011030673980713,
      "learning_rate": 1.3495192154106636e-05,
      "loss": 1.0732,
      "step": 11150
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.5464046001434326,
      "learning_rate": 1.3494553237708847e-05,
      "loss": 1.1936,
      "step": 11151
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0617289543151855,
      "learning_rate": 1.3493914321311057e-05,
      "loss": 1.0116,
      "step": 11152
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4997600317001343,
      "learning_rate": 1.3493275404913269e-05,
      "loss": 0.4466,
      "step": 11153
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8489999771118164,
      "learning_rate": 1.3492636488515478e-05,
      "loss": 0.8783,
      "step": 11154
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.508131504058838,
      "learning_rate": 1.349199757211769e-05,
      "loss": 0.9194,
      "step": 11155
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7991009950637817,
      "learning_rate": 1.34913586557199e-05,
      "loss": 0.9114,
      "step": 11156
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7892543077468872,
      "learning_rate": 1.3490719739322111e-05,
      "loss": 0.9415,
      "step": 11157
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7905361652374268,
      "learning_rate": 1.3490080822924321e-05,
      "loss": 0.9577,
      "step": 11158
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8111752271652222,
      "learning_rate": 1.3489441906526533e-05,
      "loss": 0.9373,
      "step": 11159
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8790526390075684,
      "learning_rate": 1.3488802990128743e-05,
      "loss": 0.9365,
      "step": 11160
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.112645387649536,
      "learning_rate": 1.3488164073730954e-05,
      "loss": 0.8421,
      "step": 11161
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.01616096496582,
      "learning_rate": 1.3487525157333164e-05,
      "loss": 1.4251,
      "step": 11162
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.568885326385498,
      "learning_rate": 1.3486886240935376e-05,
      "loss": 0.8745,
      "step": 11163
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.864838600158691,
      "learning_rate": 1.3486247324537585e-05,
      "loss": 1.4446,
      "step": 11164
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.5378634929656982,
      "learning_rate": 1.3485608408139797e-05,
      "loss": 1.0307,
      "step": 11165
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.345799922943115,
      "learning_rate": 1.3484969491742007e-05,
      "loss": 1.1196,
      "step": 11166
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8472096920013428,
      "learning_rate": 1.3484330575344218e-05,
      "loss": 0.9144,
      "step": 11167
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5640603303909302,
      "learning_rate": 1.3483691658946427e-05,
      "loss": 0.8553,
      "step": 11168
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.65435266494751,
      "learning_rate": 1.348305274254864e-05,
      "loss": 1.1628,
      "step": 11169
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.3667609691619873,
      "learning_rate": 1.3482413826150848e-05,
      "loss": 1.0464,
      "step": 11170
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1274890899658203,
      "learning_rate": 1.3481774909753061e-05,
      "loss": 0.9296,
      "step": 11171
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.619362831115723,
      "learning_rate": 1.348113599335527e-05,
      "loss": 1.1388,
      "step": 11172
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6419095993041992,
      "learning_rate": 1.3480497076957481e-05,
      "loss": 1.0254,
      "step": 11173
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.065265417098999,
      "learning_rate": 1.347985816055969e-05,
      "loss": 0.9432,
      "step": 11174
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6951364278793335,
      "learning_rate": 1.3479219244161902e-05,
      "loss": 0.8471,
      "step": 11175
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7324577569961548,
      "learning_rate": 1.3478580327764112e-05,
      "loss": 0.7847,
      "step": 11176
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1178250312805176,
      "learning_rate": 1.3477941411366324e-05,
      "loss": 1.0293,
      "step": 11177
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3211355209350586,
      "learning_rate": 1.3477302494968534e-05,
      "loss": 1.156,
      "step": 11178
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6179163455963135,
      "learning_rate": 1.3476663578570745e-05,
      "loss": 0.8684,
      "step": 11179
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6315388679504395,
      "learning_rate": 1.3476024662172955e-05,
      "loss": 1.0187,
      "step": 11180
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3373990058898926,
      "learning_rate": 1.3475385745775166e-05,
      "loss": 1.0099,
      "step": 11181
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.3290176391601562,
      "learning_rate": 1.3474746829377376e-05,
      "loss": 0.8533,
      "step": 11182
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.34771466255188,
      "learning_rate": 1.3474107912979588e-05,
      "loss": 0.8721,
      "step": 11183
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6025876998901367,
      "learning_rate": 1.3473468996581798e-05,
      "loss": 0.9263,
      "step": 11184
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1062324047088623,
      "learning_rate": 1.347283008018401e-05,
      "loss": 0.8434,
      "step": 11185
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.10094690322876,
      "learning_rate": 1.3472191163786219e-05,
      "loss": 1.5478,
      "step": 11186
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1342217922210693,
      "learning_rate": 1.347155224738843e-05,
      "loss": 0.9803,
      "step": 11187
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.7695415019989014,
      "learning_rate": 1.347091333099064e-05,
      "loss": 0.7989,
      "step": 11188
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.889753818511963,
      "learning_rate": 1.3470274414592852e-05,
      "loss": 1.5886,
      "step": 11189
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.524519443511963,
      "learning_rate": 1.3469635498195062e-05,
      "loss": 0.993,
      "step": 11190
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.200728893280029,
      "learning_rate": 1.3468996581797273e-05,
      "loss": 1.3919,
      "step": 11191
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7499092817306519,
      "learning_rate": 1.3468357665399483e-05,
      "loss": 0.958,
      "step": 11192
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8243345022201538,
      "learning_rate": 1.3467718749001695e-05,
      "loss": 0.5149,
      "step": 11193
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4224276542663574,
      "learning_rate": 1.3467079832603905e-05,
      "loss": 0.815,
      "step": 11194
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7375320196151733,
      "learning_rate": 1.3466440916206116e-05,
      "loss": 0.945,
      "step": 11195
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.104825496673584,
      "learning_rate": 1.3465801999808326e-05,
      "loss": 0.9965,
      "step": 11196
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.885046124458313,
      "learning_rate": 1.3465163083410538e-05,
      "loss": 0.961,
      "step": 11197
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.02209734916687,
      "learning_rate": 1.3464524167012748e-05,
      "loss": 0.8859,
      "step": 11198
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.358358144760132,
      "learning_rate": 1.3463885250614959e-05,
      "loss": 0.7839,
      "step": 11199
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2682371139526367,
      "learning_rate": 1.3463246334217169e-05,
      "loss": 1.0916,
      "step": 11200
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6456316709518433,
      "learning_rate": 1.346260741781938e-05,
      "loss": 0.8806,
      "step": 11201
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.072462558746338,
      "learning_rate": 1.346196850142159e-05,
      "loss": 1.2944,
      "step": 11202
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.1853137016296387,
      "learning_rate": 1.3461329585023802e-05,
      "loss": 1.1902,
      "step": 11203
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.44142746925354,
      "learning_rate": 1.346069066862601e-05,
      "loss": 0.8845,
      "step": 11204
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6792422533035278,
      "learning_rate": 1.3460051752228223e-05,
      "loss": 0.918,
      "step": 11205
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5151699781417847,
      "learning_rate": 1.3459412835830431e-05,
      "loss": 0.9172,
      "step": 11206
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.5046093463897705,
      "learning_rate": 1.3458773919432645e-05,
      "loss": 0.8314,
      "step": 11207
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3108469247817993,
      "learning_rate": 1.3458135003034853e-05,
      "loss": 0.7934,
      "step": 11208
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8687790632247925,
      "learning_rate": 1.3457496086637064e-05,
      "loss": 0.9187,
      "step": 11209
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.200904846191406,
      "learning_rate": 1.3456857170239274e-05,
      "loss": 1.3643,
      "step": 11210
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2861709594726562,
      "learning_rate": 1.3456218253841486e-05,
      "loss": 0.9497,
      "step": 11211
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7624908685684204,
      "learning_rate": 1.3455579337443696e-05,
      "loss": 0.9147,
      "step": 11212
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1329801082611084,
      "learning_rate": 1.3454940421045907e-05,
      "loss": 1.0108,
      "step": 11213
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1757922172546387,
      "learning_rate": 1.3454301504648117e-05,
      "loss": 1.0,
      "step": 11214
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8007218837738037,
      "learning_rate": 1.3453662588250329e-05,
      "loss": 0.9165,
      "step": 11215
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.578979969024658,
      "learning_rate": 1.3453023671852538e-05,
      "loss": 0.6047,
      "step": 11216
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5439443588256836,
      "learning_rate": 1.345238475545475e-05,
      "loss": 0.8951,
      "step": 11217
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7499881982803345,
      "learning_rate": 1.345174583905696e-05,
      "loss": 0.9535,
      "step": 11218
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6491339206695557,
      "learning_rate": 1.3451106922659171e-05,
      "loss": 0.929,
      "step": 11219
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7521077394485474,
      "learning_rate": 1.3450468006261381e-05,
      "loss": 0.7994,
      "step": 11220
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7080124616622925,
      "learning_rate": 1.3449829089863593e-05,
      "loss": 0.8793,
      "step": 11221
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.061020851135254,
      "learning_rate": 1.3449190173465803e-05,
      "loss": 1.128,
      "step": 11222
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.3233954906463623,
      "learning_rate": 1.3448551257068014e-05,
      "loss": 0.8528,
      "step": 11223
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.670438289642334,
      "learning_rate": 1.3447912340670224e-05,
      "loss": 0.9174,
      "step": 11224
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9856312274932861,
      "learning_rate": 1.3447273424272436e-05,
      "loss": 1.0481,
      "step": 11225
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.01250958442688,
      "learning_rate": 1.3446634507874645e-05,
      "loss": 0.9196,
      "step": 11226
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.1574254035949707,
      "learning_rate": 1.3445995591476857e-05,
      "loss": 0.9063,
      "step": 11227
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.245365619659424,
      "learning_rate": 1.3445356675079067e-05,
      "loss": 1.6219,
      "step": 11228
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3817028999328613,
      "learning_rate": 1.3444717758681278e-05,
      "loss": 0.8803,
      "step": 11229
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5005117654800415,
      "learning_rate": 1.3444078842283488e-05,
      "loss": 0.8278,
      "step": 11230
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9820995330810547,
      "learning_rate": 1.34434399258857e-05,
      "loss": 0.9267,
      "step": 11231
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.128056764602661,
      "learning_rate": 1.344280100948791e-05,
      "loss": 0.8481,
      "step": 11232
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.686913013458252,
      "learning_rate": 1.3442162093090121e-05,
      "loss": 0.9697,
      "step": 11233
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.689924716949463,
      "learning_rate": 1.3441523176692331e-05,
      "loss": 0.9148,
      "step": 11234
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.4146933555603027,
      "learning_rate": 1.3440884260294543e-05,
      "loss": 0.9835,
      "step": 11235
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5132672786712646,
      "learning_rate": 1.3440245343896752e-05,
      "loss": 0.774,
      "step": 11236
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1145591735839844,
      "learning_rate": 1.3439606427498964e-05,
      "loss": 1.061,
      "step": 11237
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9333752393722534,
      "learning_rate": 1.3438967511101174e-05,
      "loss": 0.9882,
      "step": 11238
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6690735816955566,
      "learning_rate": 1.3438328594703385e-05,
      "loss": 0.7872,
      "step": 11239
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.408628225326538,
      "learning_rate": 1.3437689678305594e-05,
      "loss": 0.3678,
      "step": 11240
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.746355414390564,
      "learning_rate": 1.3437050761907807e-05,
      "loss": 0.8984,
      "step": 11241
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7062723636627197,
      "learning_rate": 1.3436411845510015e-05,
      "loss": 0.8344,
      "step": 11242
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.456068992614746,
      "learning_rate": 1.3435772929112228e-05,
      "loss": 1.0918,
      "step": 11243
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.301591396331787,
      "learning_rate": 1.3435134012714436e-05,
      "loss": 1.0932,
      "step": 11244
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2200214862823486,
      "learning_rate": 1.3434495096316648e-05,
      "loss": 0.6465,
      "step": 11245
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8811274766921997,
      "learning_rate": 1.3433856179918858e-05,
      "loss": 0.8788,
      "step": 11246
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9706532955169678,
      "learning_rate": 1.343321726352107e-05,
      "loss": 0.9907,
      "step": 11247
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1214816570281982,
      "learning_rate": 1.3432578347123279e-05,
      "loss": 0.889,
      "step": 11248
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5700072050094604,
      "learning_rate": 1.343193943072549e-05,
      "loss": 0.849,
      "step": 11249
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5375736951828003,
      "learning_rate": 1.34313005143277e-05,
      "loss": 0.396,
      "step": 11250
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.756553292274475,
      "learning_rate": 1.3430661597929912e-05,
      "loss": 0.86,
      "step": 11251
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.007826328277588,
      "learning_rate": 1.3430022681532122e-05,
      "loss": 0.9374,
      "step": 11252
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0954973697662354,
      "learning_rate": 1.3429383765134333e-05,
      "loss": 1.0544,
      "step": 11253
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.755009651184082,
      "learning_rate": 1.3428744848736543e-05,
      "loss": 1.6199,
      "step": 11254
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7954323291778564,
      "learning_rate": 1.3428105932338755e-05,
      "loss": 0.8112,
      "step": 11255
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6346304416656494,
      "learning_rate": 1.3427467015940965e-05,
      "loss": 0.8823,
      "step": 11256
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.6504440307617188,
      "learning_rate": 1.3426828099543176e-05,
      "loss": 0.9156,
      "step": 11257
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.016549587249756,
      "learning_rate": 1.3426189183145386e-05,
      "loss": 0.9944,
      "step": 11258
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7742222547531128,
      "learning_rate": 1.3425550266747598e-05,
      "loss": 1.007,
      "step": 11259
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.051705837249756,
      "learning_rate": 1.3424911350349808e-05,
      "loss": 0.9884,
      "step": 11260
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.021075963973999,
      "learning_rate": 1.3424272433952019e-05,
      "loss": 0.9358,
      "step": 11261
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.102750301361084,
      "learning_rate": 1.3423633517554229e-05,
      "loss": 1.1251,
      "step": 11262
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7339977025985718,
      "learning_rate": 1.342299460115644e-05,
      "loss": 0.4936,
      "step": 11263
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.478290557861328,
      "learning_rate": 1.342235568475865e-05,
      "loss": 0.5652,
      "step": 11264
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.884118676185608,
      "learning_rate": 1.3421716768360862e-05,
      "loss": 0.9634,
      "step": 11265
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7016136646270752,
      "learning_rate": 1.3421077851963072e-05,
      "loss": 0.9217,
      "step": 11266
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3758962154388428,
      "learning_rate": 1.3420438935565283e-05,
      "loss": 0.9339,
      "step": 11267
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.071831226348877,
      "learning_rate": 1.3419800019167493e-05,
      "loss": 0.8746,
      "step": 11268
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.238168716430664,
      "learning_rate": 1.3419161102769705e-05,
      "loss": 0.8399,
      "step": 11269
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.554813861846924,
      "learning_rate": 1.3418522186371915e-05,
      "loss": 0.6374,
      "step": 11270
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.313877820968628,
      "learning_rate": 1.3417883269974126e-05,
      "loss": 0.9214,
      "step": 11271
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8652856349945068,
      "learning_rate": 1.3417244353576336e-05,
      "loss": 0.9062,
      "step": 11272
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.482668161392212,
      "learning_rate": 1.3416605437178547e-05,
      "loss": 0.7445,
      "step": 11273
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4205001592636108,
      "learning_rate": 1.3415966520780757e-05,
      "loss": 0.8668,
      "step": 11274
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.516650199890137,
      "learning_rate": 1.3415327604382969e-05,
      "loss": 1.0383,
      "step": 11275
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3964918851852417,
      "learning_rate": 1.3414688687985177e-05,
      "loss": 0.4648,
      "step": 11276
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.298975706100464,
      "learning_rate": 1.341404977158739e-05,
      "loss": 0.8436,
      "step": 11277
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.105269432067871,
      "learning_rate": 1.3413410855189598e-05,
      "loss": 1.1274,
      "step": 11278
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.749925136566162,
      "learning_rate": 1.3412771938791812e-05,
      "loss": 0.8737,
      "step": 11279
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.610685348510742,
      "learning_rate": 1.341213302239402e-05,
      "loss": 0.7823,
      "step": 11280
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.148442268371582,
      "learning_rate": 1.3411494105996233e-05,
      "loss": 1.0516,
      "step": 11281
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6250642538070679,
      "learning_rate": 1.3410855189598441e-05,
      "loss": 0.8619,
      "step": 11282
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3950350284576416,
      "learning_rate": 1.3410216273200653e-05,
      "loss": 0.4168,
      "step": 11283
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.112255811691284,
      "learning_rate": 1.3409577356802863e-05,
      "loss": 0.902,
      "step": 11284
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9691011905670166,
      "learning_rate": 1.3408938440405074e-05,
      "loss": 0.8422,
      "step": 11285
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5039063692092896,
      "learning_rate": 1.3408299524007284e-05,
      "loss": 0.8708,
      "step": 11286
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9970858097076416,
      "learning_rate": 1.3407660607609496e-05,
      "loss": 0.9126,
      "step": 11287
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.319664001464844,
      "learning_rate": 1.3407021691211705e-05,
      "loss": 1.2476,
      "step": 11288
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.98361873626709,
      "learning_rate": 1.3406382774813917e-05,
      "loss": 1.0308,
      "step": 11289
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.000037670135498,
      "learning_rate": 1.3405743858416127e-05,
      "loss": 1.1348,
      "step": 11290
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4870150089263916,
      "learning_rate": 1.3405104942018338e-05,
      "loss": 0.947,
      "step": 11291
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7509448528289795,
      "learning_rate": 1.3404466025620548e-05,
      "loss": 0.8048,
      "step": 11292
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9172219038009644,
      "learning_rate": 1.340382710922276e-05,
      "loss": 0.8413,
      "step": 11293
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8781871795654297,
      "learning_rate": 1.340318819282497e-05,
      "loss": 0.9221,
      "step": 11294
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7938212156295776,
      "learning_rate": 1.3402549276427181e-05,
      "loss": 0.9024,
      "step": 11295
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9941823482513428,
      "learning_rate": 1.3401910360029391e-05,
      "loss": 0.9372,
      "step": 11296
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.721935749053955,
      "learning_rate": 1.3401271443631603e-05,
      "loss": 0.8575,
      "step": 11297
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9466290473937988,
      "learning_rate": 1.3400632527233812e-05,
      "loss": 1.3092,
      "step": 11298
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9189043045043945,
      "learning_rate": 1.3399993610836024e-05,
      "loss": 1.0152,
      "step": 11299
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7345510721206665,
      "learning_rate": 1.3399354694438234e-05,
      "loss": 1.0078,
      "step": 11300
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6064658164978027,
      "learning_rate": 1.3398715778040445e-05,
      "loss": 0.9207,
      "step": 11301
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8558069467544556,
      "learning_rate": 1.3398076861642655e-05,
      "loss": 0.964,
      "step": 11302
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6966081857681274,
      "learning_rate": 1.3397437945244867e-05,
      "loss": 0.8987,
      "step": 11303
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.491665840148926,
      "learning_rate": 1.3396799028847077e-05,
      "loss": 1.1689,
      "step": 11304
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.074939727783203,
      "learning_rate": 1.3396160112449288e-05,
      "loss": 0.6285,
      "step": 11305
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6960687637329102,
      "learning_rate": 1.3395521196051498e-05,
      "loss": 1.0029,
      "step": 11306
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3586134910583496,
      "learning_rate": 1.339488227965371e-05,
      "loss": 0.6027,
      "step": 11307
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6898127794265747,
      "learning_rate": 1.339424336325592e-05,
      "loss": 0.9393,
      "step": 11308
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.012636661529541,
      "learning_rate": 1.3393604446858131e-05,
      "loss": 1.5134,
      "step": 11309
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5309772491455078,
      "learning_rate": 1.339296553046034e-05,
      "loss": 0.8603,
      "step": 11310
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6697145700454712,
      "learning_rate": 1.3392326614062552e-05,
      "loss": 0.9461,
      "step": 11311
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.400026559829712,
      "learning_rate": 1.3391687697664762e-05,
      "loss": 1.0297,
      "step": 11312
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8229738473892212,
      "learning_rate": 1.3391048781266974e-05,
      "loss": 1.0738,
      "step": 11313
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6625806093215942,
      "learning_rate": 1.3390409864869182e-05,
      "loss": 0.8411,
      "step": 11314
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7521551847457886,
      "learning_rate": 1.3389770948471395e-05,
      "loss": 1.1351,
      "step": 11315
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8408455848693848,
      "learning_rate": 1.3389132032073603e-05,
      "loss": 1.0309,
      "step": 11316
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.537325143814087,
      "learning_rate": 1.3388493115675817e-05,
      "loss": 0.9075,
      "step": 11317
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5362935066223145,
      "learning_rate": 1.3387854199278025e-05,
      "loss": 0.7731,
      "step": 11318
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8197578191757202,
      "learning_rate": 1.3387215282880236e-05,
      "loss": 0.9519,
      "step": 11319
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.065701484680176,
      "learning_rate": 1.3386576366482446e-05,
      "loss": 0.9896,
      "step": 11320
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.798181176185608,
      "learning_rate": 1.3385937450084658e-05,
      "loss": 0.9161,
      "step": 11321
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5975571870803833,
      "learning_rate": 1.3385298533686868e-05,
      "loss": 0.8361,
      "step": 11322
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9910402297973633,
      "learning_rate": 1.3384659617289079e-05,
      "loss": 0.974,
      "step": 11323
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0084967613220215,
      "learning_rate": 1.3384020700891289e-05,
      "loss": 1.08,
      "step": 11324
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.036761522293091,
      "learning_rate": 1.33833817844935e-05,
      "loss": 1.1832,
      "step": 11325
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7900851964950562,
      "learning_rate": 1.338274286809571e-05,
      "loss": 0.9699,
      "step": 11326
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.454929828643799,
      "learning_rate": 1.3382103951697922e-05,
      "loss": 0.8228,
      "step": 11327
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8745921850204468,
      "learning_rate": 1.3381465035300132e-05,
      "loss": 0.9382,
      "step": 11328
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.5381336212158203,
      "learning_rate": 1.3380826118902343e-05,
      "loss": 0.8841,
      "step": 11329
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.5550835132598877,
      "learning_rate": 1.3380187202504553e-05,
      "loss": 1.0226,
      "step": 11330
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9712350368499756,
      "learning_rate": 1.3379548286106765e-05,
      "loss": 0.9825,
      "step": 11331
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5268889665603638,
      "learning_rate": 1.3378909369708975e-05,
      "loss": 0.815,
      "step": 11332
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4200754165649414,
      "learning_rate": 1.3378270453311186e-05,
      "loss": 1.0729,
      "step": 11333
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0670666694641113,
      "learning_rate": 1.3377631536913396e-05,
      "loss": 1.1367,
      "step": 11334
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7138686180114746,
      "learning_rate": 1.3376992620515607e-05,
      "loss": 0.81,
      "step": 11335
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1840291023254395,
      "learning_rate": 1.3376353704117817e-05,
      "loss": 0.9412,
      "step": 11336
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.064922571182251,
      "learning_rate": 1.3375714787720029e-05,
      "loss": 1.0322,
      "step": 11337
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.010850429534912,
      "learning_rate": 1.3375075871322239e-05,
      "loss": 0.8239,
      "step": 11338
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6050684452056885,
      "learning_rate": 1.337443695492445e-05,
      "loss": 0.8047,
      "step": 11339
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5493279695510864,
      "learning_rate": 1.337379803852666e-05,
      "loss": 0.473,
      "step": 11340
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1816320419311523,
      "learning_rate": 1.3373159122128872e-05,
      "loss": 1.1311,
      "step": 11341
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.978214740753174,
      "learning_rate": 1.3372520205731082e-05,
      "loss": 1.6265,
      "step": 11342
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.833376407623291,
      "learning_rate": 1.3371881289333293e-05,
      "loss": 0.8687,
      "step": 11343
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9890224933624268,
      "learning_rate": 1.3371242372935503e-05,
      "loss": 0.9861,
      "step": 11344
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.877423048019409,
      "learning_rate": 1.3370603456537714e-05,
      "loss": 0.9998,
      "step": 11345
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.8764307498931885,
      "learning_rate": 1.3369964540139924e-05,
      "loss": 1.3213,
      "step": 11346
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5566402673721313,
      "learning_rate": 1.3369325623742136e-05,
      "loss": 0.8767,
      "step": 11347
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6704140901565552,
      "learning_rate": 1.3368686707344346e-05,
      "loss": 0.8824,
      "step": 11348
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8245961666107178,
      "learning_rate": 1.3368047790946557e-05,
      "loss": 0.7772,
      "step": 11349
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0446622371673584,
      "learning_rate": 1.3367408874548765e-05,
      "loss": 1.208,
      "step": 11350
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0215225219726562,
      "learning_rate": 1.3366769958150979e-05,
      "loss": 0.9266,
      "step": 11351
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5357732772827148,
      "learning_rate": 1.3366131041753187e-05,
      "loss": 0.9617,
      "step": 11352
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.557849884033203,
      "learning_rate": 1.33654921253554e-05,
      "loss": 1.9192,
      "step": 11353
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.502819538116455,
      "learning_rate": 1.3364853208957608e-05,
      "loss": 0.9333,
      "step": 11354
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0345208644866943,
      "learning_rate": 1.336421429255982e-05,
      "loss": 0.9087,
      "step": 11355
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.923114538192749,
      "learning_rate": 1.336357537616203e-05,
      "loss": 0.8867,
      "step": 11356
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9849251508712769,
      "learning_rate": 1.3362936459764241e-05,
      "loss": 0.8764,
      "step": 11357
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7101317644119263,
      "learning_rate": 1.3362297543366451e-05,
      "loss": 1.0021,
      "step": 11358
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1642942428588867,
      "learning_rate": 1.3361658626968663e-05,
      "loss": 1.0775,
      "step": 11359
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5944169759750366,
      "learning_rate": 1.3361019710570872e-05,
      "loss": 0.9197,
      "step": 11360
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9638749361038208,
      "learning_rate": 1.3360380794173084e-05,
      "loss": 0.9729,
      "step": 11361
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.026175022125244,
      "learning_rate": 1.3359741877775294e-05,
      "loss": 0.9595,
      "step": 11362
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3746672868728638,
      "learning_rate": 1.3359102961377505e-05,
      "loss": 0.8216,
      "step": 11363
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6786781549453735,
      "learning_rate": 1.3358464044979715e-05,
      "loss": 0.4451,
      "step": 11364
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6656826734542847,
      "learning_rate": 1.3357825128581927e-05,
      "loss": 1.2161,
      "step": 11365
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6976737976074219,
      "learning_rate": 1.3357186212184137e-05,
      "loss": 0.4348,
      "step": 11366
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8585816621780396,
      "learning_rate": 1.3356547295786348e-05,
      "loss": 0.9956,
      "step": 11367
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7541853189468384,
      "learning_rate": 1.3355908379388558e-05,
      "loss": 1.2854,
      "step": 11368
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4223101139068604,
      "learning_rate": 1.335526946299077e-05,
      "loss": 1.0106,
      "step": 11369
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1325576305389404,
      "learning_rate": 1.335463054659298e-05,
      "loss": 0.9442,
      "step": 11370
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.000711441040039,
      "learning_rate": 1.3353991630195191e-05,
      "loss": 0.4722,
      "step": 11371
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8077847957611084,
      "learning_rate": 1.3353352713797401e-05,
      "loss": 0.9464,
      "step": 11372
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6708426475524902,
      "learning_rate": 1.3352713797399612e-05,
      "loss": 0.9919,
      "step": 11373
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.4657700061798096,
      "learning_rate": 1.3352074881001822e-05,
      "loss": 1.0899,
      "step": 11374
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.8752148151397705,
      "learning_rate": 1.3351435964604034e-05,
      "loss": 0.9434,
      "step": 11375
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6383838653564453,
      "learning_rate": 1.3350797048206244e-05,
      "loss": 0.9655,
      "step": 11376
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.569096326828003,
      "learning_rate": 1.3350158131808455e-05,
      "loss": 1.0005,
      "step": 11377
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.383822441101074,
      "learning_rate": 1.3349519215410665e-05,
      "loss": 0.9388,
      "step": 11378
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7373775243759155,
      "learning_rate": 1.3348880299012877e-05,
      "loss": 0.7998,
      "step": 11379
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0359575748443604,
      "learning_rate": 1.3348241382615086e-05,
      "loss": 1.0302,
      "step": 11380
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2724063396453857,
      "learning_rate": 1.3347602466217298e-05,
      "loss": 1.1632,
      "step": 11381
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9489384889602661,
      "learning_rate": 1.3346963549819508e-05,
      "loss": 0.9384,
      "step": 11382
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.673720359802246,
      "learning_rate": 1.334632463342172e-05,
      "loss": 0.804,
      "step": 11383
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.538339376449585,
      "learning_rate": 1.334568571702393e-05,
      "loss": 0.7957,
      "step": 11384
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0895369052886963,
      "learning_rate": 1.334504680062614e-05,
      "loss": 1.099,
      "step": 11385
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.5621747970581055,
      "learning_rate": 1.3344407884228349e-05,
      "loss": 1.238,
      "step": 11386
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5716720819473267,
      "learning_rate": 1.3343768967830562e-05,
      "loss": 0.8663,
      "step": 11387
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.2730965614318848,
      "learning_rate": 1.334313005143277e-05,
      "loss": 1.0369,
      "step": 11388
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.466626763343811,
      "learning_rate": 1.3342491135034984e-05,
      "loss": 0.8667,
      "step": 11389
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.6771035194396973,
      "learning_rate": 1.3341852218637192e-05,
      "loss": 1.2177,
      "step": 11390
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.6561272144317627,
      "learning_rate": 1.3341213302239403e-05,
      "loss": 0.8506,
      "step": 11391
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6155990362167358,
      "learning_rate": 1.3340574385841613e-05,
      "loss": 0.9844,
      "step": 11392
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7321885824203491,
      "learning_rate": 1.3339935469443825e-05,
      "loss": 0.8067,
      "step": 11393
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.443472385406494,
      "learning_rate": 1.3339296553046035e-05,
      "loss": 1.3095,
      "step": 11394
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5505858659744263,
      "learning_rate": 1.3338657636648246e-05,
      "loss": 0.898,
      "step": 11395
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5243180990219116,
      "learning_rate": 1.3338018720250456e-05,
      "loss": 0.4379,
      "step": 11396
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9181792736053467,
      "learning_rate": 1.3337379803852668e-05,
      "loss": 0.3932,
      "step": 11397
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0738332271575928,
      "learning_rate": 1.3336740887454877e-05,
      "loss": 0.9064,
      "step": 11398
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9788930416107178,
      "learning_rate": 1.3336101971057089e-05,
      "loss": 1.422,
      "step": 11399
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6705347299575806,
      "learning_rate": 1.3335463054659299e-05,
      "loss": 0.8761,
      "step": 11400
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.9175777435302734,
      "learning_rate": 1.333482413826151e-05,
      "loss": 0.6248,
      "step": 11401
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.959710955619812,
      "learning_rate": 1.333418522186372e-05,
      "loss": 0.9811,
      "step": 11402
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7396469116210938,
      "learning_rate": 1.3333546305465932e-05,
      "loss": 0.781,
      "step": 11403
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.4424941539764404,
      "learning_rate": 1.3332907389068142e-05,
      "loss": 0.8892,
      "step": 11404
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.0601837635040283,
      "learning_rate": 1.3332268472670351e-05,
      "loss": 0.6858,
      "step": 11405
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7223707437515259,
      "learning_rate": 1.3331629556272563e-05,
      "loss": 0.8665,
      "step": 11406
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.890233039855957,
      "learning_rate": 1.3330990639874773e-05,
      "loss": 0.7995,
      "step": 11407
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5509175062179565,
      "learning_rate": 1.3330351723476984e-05,
      "loss": 0.9943,
      "step": 11408
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8779271841049194,
      "learning_rate": 1.3329712807079194e-05,
      "loss": 0.9856,
      "step": 11409
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8802237510681152,
      "learning_rate": 1.3329073890681406e-05,
      "loss": 0.9035,
      "step": 11410
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.97085440158844,
      "learning_rate": 1.3328434974283616e-05,
      "loss": 0.996,
      "step": 11411
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6122180223464966,
      "learning_rate": 1.3327796057885827e-05,
      "loss": 0.8794,
      "step": 11412
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.808504819869995,
      "learning_rate": 1.3327157141488037e-05,
      "loss": 0.9016,
      "step": 11413
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.833300828933716,
      "learning_rate": 1.3326518225090249e-05,
      "loss": 1.2003,
      "step": 11414
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.6303892135620117,
      "learning_rate": 1.3325879308692458e-05,
      "loss": 1.2564,
      "step": 11415
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.18479323387146,
      "learning_rate": 1.332524039229467e-05,
      "loss": 0.9277,
      "step": 11416
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9700316190719604,
      "learning_rate": 1.3324601475896878e-05,
      "loss": 0.8341,
      "step": 11417
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6641408205032349,
      "learning_rate": 1.3323962559499091e-05,
      "loss": 0.8507,
      "step": 11418
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.153881788253784,
      "learning_rate": 1.33233236431013e-05,
      "loss": 0.8946,
      "step": 11419
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7970795631408691,
      "learning_rate": 1.3322684726703513e-05,
      "loss": 0.8622,
      "step": 11420
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.069699764251709,
      "learning_rate": 1.3322045810305721e-05,
      "loss": 1.1804,
      "step": 11421
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7035554647445679,
      "learning_rate": 1.3321406893907932e-05,
      "loss": 0.9346,
      "step": 11422
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9427433013916016,
      "learning_rate": 1.3320767977510142e-05,
      "loss": 0.9649,
      "step": 11423
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.729872226715088,
      "learning_rate": 1.3320129061112354e-05,
      "loss": 0.8855,
      "step": 11424
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.885884404182434,
      "learning_rate": 1.3319490144714564e-05,
      "loss": 0.9113,
      "step": 11425
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.387267589569092,
      "learning_rate": 1.3318851228316775e-05,
      "loss": 1.3595,
      "step": 11426
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7057392597198486,
      "learning_rate": 1.3318212311918985e-05,
      "loss": 0.856,
      "step": 11427
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9456737041473389,
      "learning_rate": 1.3317573395521197e-05,
      "loss": 1.0215,
      "step": 11428
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7894657850265503,
      "learning_rate": 1.3316934479123407e-05,
      "loss": 0.9215,
      "step": 11429
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.5385870933532715,
      "learning_rate": 1.3316295562725618e-05,
      "loss": 1.3111,
      "step": 11430
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7639985084533691,
      "learning_rate": 1.3315656646327828e-05,
      "loss": 0.9183,
      "step": 11431
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9769666194915771,
      "learning_rate": 1.331501772993004e-05,
      "loss": 0.8966,
      "step": 11432
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.2451364994049072,
      "learning_rate": 1.331437881353225e-05,
      "loss": 0.9051,
      "step": 11433
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.01127552986145,
      "learning_rate": 1.3313739897134461e-05,
      "loss": 1.0021,
      "step": 11434
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.84160315990448,
      "learning_rate": 1.331310098073667e-05,
      "loss": 0.8172,
      "step": 11435
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3107423782348633,
      "learning_rate": 1.3312462064338882e-05,
      "loss": 1.0834,
      "step": 11436
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7339485883712769,
      "learning_rate": 1.3311823147941092e-05,
      "loss": 0.9479,
      "step": 11437
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.477254867553711,
      "learning_rate": 1.3311184231543304e-05,
      "loss": 0.9425,
      "step": 11438
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9462199211120605,
      "learning_rate": 1.3310545315145514e-05,
      "loss": 0.9435,
      "step": 11439
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3733713626861572,
      "learning_rate": 1.3309906398747725e-05,
      "loss": 0.8261,
      "step": 11440
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.390141010284424,
      "learning_rate": 1.3309267482349935e-05,
      "loss": 0.7522,
      "step": 11441
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.95487904548645,
      "learning_rate": 1.3308628565952146e-05,
      "loss": 0.9152,
      "step": 11442
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1424167156219482,
      "learning_rate": 1.3307989649554356e-05,
      "loss": 0.9426,
      "step": 11443
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.05777645111084,
      "learning_rate": 1.3307350733156568e-05,
      "loss": 0.9346,
      "step": 11444
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1455774307250977,
      "learning_rate": 1.3306711816758778e-05,
      "loss": 0.8261,
      "step": 11445
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.0770368576049805,
      "learning_rate": 1.330607290036099e-05,
      "loss": 0.8337,
      "step": 11446
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.455925226211548,
      "learning_rate": 1.3305433983963199e-05,
      "loss": 0.7794,
      "step": 11447
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.893430233001709,
      "learning_rate": 1.330479506756541e-05,
      "loss": 0.8869,
      "step": 11448
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9376461505889893,
      "learning_rate": 1.330415615116762e-05,
      "loss": 0.8636,
      "step": 11449
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2870101928710938,
      "learning_rate": 1.3303517234769832e-05,
      "loss": 0.8837,
      "step": 11450
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7895911931991577,
      "learning_rate": 1.3302878318372042e-05,
      "loss": 0.8564,
      "step": 11451
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0143861770629883,
      "learning_rate": 1.3302239401974253e-05,
      "loss": 0.8775,
      "step": 11452
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.331804275512695,
      "learning_rate": 1.3301600485576462e-05,
      "loss": 1.3412,
      "step": 11453
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9063202142715454,
      "learning_rate": 1.3300961569178675e-05,
      "loss": 0.9351,
      "step": 11454
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.462031602859497,
      "learning_rate": 1.3300322652780883e-05,
      "loss": 1.1608,
      "step": 11455
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8707895278930664,
      "learning_rate": 1.3299683736383096e-05,
      "loss": 0.8717,
      "step": 11456
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.028371334075928,
      "learning_rate": 1.3299044819985304e-05,
      "loss": 1.1179,
      "step": 11457
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.083116292953491,
      "learning_rate": 1.3298405903587518e-05,
      "loss": 0.9803,
      "step": 11458
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.891249895095825,
      "learning_rate": 1.3297766987189726e-05,
      "loss": 0.9722,
      "step": 11459
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9547909498214722,
      "learning_rate": 1.3297128070791937e-05,
      "loss": 1.0323,
      "step": 11460
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.488748550415039,
      "learning_rate": 1.3296489154394147e-05,
      "loss": 1.2536,
      "step": 11461
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5345364809036255,
      "learning_rate": 1.3295850237996359e-05,
      "loss": 0.8581,
      "step": 11462
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.987899899482727,
      "learning_rate": 1.3295211321598569e-05,
      "loss": 0.9142,
      "step": 11463
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.19736909866333,
      "learning_rate": 1.329457240520078e-05,
      "loss": 0.8812,
      "step": 11464
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.7144036293029785,
      "learning_rate": 1.329393348880299e-05,
      "loss": 1.3363,
      "step": 11465
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.274648427963257,
      "learning_rate": 1.3293294572405202e-05,
      "loss": 1.0497,
      "step": 11466
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8759357929229736,
      "learning_rate": 1.3292655656007411e-05,
      "loss": 0.948,
      "step": 11467
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0485873222351074,
      "learning_rate": 1.3292016739609623e-05,
      "loss": 0.777,
      "step": 11468
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5681812763214111,
      "learning_rate": 1.3291377823211833e-05,
      "loss": 0.8715,
      "step": 11469
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.5837860107421875,
      "learning_rate": 1.3290738906814044e-05,
      "loss": 1.5168,
      "step": 11470
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1874728202819824,
      "learning_rate": 1.3290099990416254e-05,
      "loss": 0.9829,
      "step": 11471
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6422836780548096,
      "learning_rate": 1.3289461074018466e-05,
      "loss": 0.9527,
      "step": 11472
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7413839101791382,
      "learning_rate": 1.3288822157620676e-05,
      "loss": 0.4617,
      "step": 11473
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2299110889434814,
      "learning_rate": 1.3288183241222887e-05,
      "loss": 0.6473,
      "step": 11474
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1304731369018555,
      "learning_rate": 1.3287544324825097e-05,
      "loss": 1.0507,
      "step": 11475
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.820652961730957,
      "learning_rate": 1.3286905408427309e-05,
      "loss": 1.2204,
      "step": 11476
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.0302515029907227,
      "learning_rate": 1.3286266492029518e-05,
      "loss": 1.0465,
      "step": 11477
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7231793403625488,
      "learning_rate": 1.328562757563173e-05,
      "loss": 0.8765,
      "step": 11478
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.220660924911499,
      "learning_rate": 1.328498865923394e-05,
      "loss": 0.9319,
      "step": 11479
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.492126226425171,
      "learning_rate": 1.3284349742836151e-05,
      "loss": 0.9049,
      "step": 11480
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9340410232543945,
      "learning_rate": 1.3283710826438361e-05,
      "loss": 1.0151,
      "step": 11481
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0938212871551514,
      "learning_rate": 1.3283071910040573e-05,
      "loss": 0.9907,
      "step": 11482
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.832479476928711,
      "learning_rate": 1.3282432993642783e-05,
      "loss": 0.9559,
      "step": 11483
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0257081985473633,
      "learning_rate": 1.3281794077244994e-05,
      "loss": 0.8893,
      "step": 11484
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6575967073440552,
      "learning_rate": 1.3281155160847204e-05,
      "loss": 0.7832,
      "step": 11485
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.71138596534729,
      "learning_rate": 1.3280516244449416e-05,
      "loss": 0.9145,
      "step": 11486
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2503325939178467,
      "learning_rate": 1.3279877328051625e-05,
      "loss": 0.8983,
      "step": 11487
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8207446336746216,
      "learning_rate": 1.3279238411653837e-05,
      "loss": 0.9237,
      "step": 11488
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6673744916915894,
      "learning_rate": 1.3278599495256045e-05,
      "loss": 0.8406,
      "step": 11489
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5649940967559814,
      "learning_rate": 1.3277960578858258e-05,
      "loss": 0.9339,
      "step": 11490
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.4043920040130615,
      "learning_rate": 1.3277321662460467e-05,
      "loss": 0.708,
      "step": 11491
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.184693813323975,
      "learning_rate": 1.327668274606268e-05,
      "loss": 1.4055,
      "step": 11492
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.515289783477783,
      "learning_rate": 1.3276043829664888e-05,
      "loss": 1.0148,
      "step": 11493
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7808072566986084,
      "learning_rate": 1.3275404913267101e-05,
      "loss": 0.8409,
      "step": 11494
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8500468730926514,
      "learning_rate": 1.327476599686931e-05,
      "loss": 1.0004,
      "step": 11495
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8208166360855103,
      "learning_rate": 1.3274127080471521e-05,
      "loss": 0.8929,
      "step": 11496
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8851885795593262,
      "learning_rate": 1.327348816407373e-05,
      "loss": 0.8274,
      "step": 11497
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.15073823928833,
      "learning_rate": 1.3272849247675942e-05,
      "loss": 1.8901,
      "step": 11498
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6411381959915161,
      "learning_rate": 1.3272210331278152e-05,
      "loss": 0.7677,
      "step": 11499
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.466811180114746,
      "learning_rate": 1.3271571414880364e-05,
      "loss": 0.8168,
      "step": 11500
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.5656638145446777,
      "learning_rate": 1.3270932498482574e-05,
      "loss": 1.0952,
      "step": 11501
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7245711088180542,
      "learning_rate": 1.3270293582084785e-05,
      "loss": 0.8004,
      "step": 11502
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.8188247680664062,
      "learning_rate": 1.3269654665686995e-05,
      "loss": 1.5522,
      "step": 11503
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.062342166900635,
      "learning_rate": 1.3269015749289206e-05,
      "loss": 1.089,
      "step": 11504
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.13515567779541,
      "learning_rate": 1.3268376832891416e-05,
      "loss": 0.9053,
      "step": 11505
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.097360134124756,
      "learning_rate": 1.3267737916493628e-05,
      "loss": 1.2026,
      "step": 11506
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0415914058685303,
      "learning_rate": 1.3267099000095838e-05,
      "loss": 0.8172,
      "step": 11507
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.234372854232788,
      "learning_rate": 1.326646008369805e-05,
      "loss": 0.9971,
      "step": 11508
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9214814901351929,
      "learning_rate": 1.3265821167300259e-05,
      "loss": 1.0024,
      "step": 11509
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0508882999420166,
      "learning_rate": 1.326518225090247e-05,
      "loss": 0.9621,
      "step": 11510
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.687867522239685,
      "learning_rate": 1.326454333450468e-05,
      "loss": 0.5369,
      "step": 11511
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8275065422058105,
      "learning_rate": 1.3263904418106892e-05,
      "loss": 0.9346,
      "step": 11512
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9588099718093872,
      "learning_rate": 1.3263265501709102e-05,
      "loss": 0.9225,
      "step": 11513
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.772247552871704,
      "learning_rate": 1.3262626585311313e-05,
      "loss": 0.9663,
      "step": 11514
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.131315231323242,
      "learning_rate": 1.3261987668913523e-05,
      "loss": 0.8656,
      "step": 11515
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.2933857440948486,
      "learning_rate": 1.3261348752515735e-05,
      "loss": 0.9599,
      "step": 11516
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.159865379333496,
      "learning_rate": 1.3260709836117945e-05,
      "loss": 0.9236,
      "step": 11517
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.2585649490356445,
      "learning_rate": 1.3260070919720156e-05,
      "loss": 1.0322,
      "step": 11518
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.892124891281128,
      "learning_rate": 1.3259432003322366e-05,
      "loss": 0.9441,
      "step": 11519
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.5852222442626953,
      "learning_rate": 1.3258793086924578e-05,
      "loss": 0.829,
      "step": 11520
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2756967544555664,
      "learning_rate": 1.3258154170526788e-05,
      "loss": 1.0327,
      "step": 11521
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5351483821868896,
      "learning_rate": 1.3257515254128999e-05,
      "loss": 0.8288,
      "step": 11522
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9532092809677124,
      "learning_rate": 1.3256876337731209e-05,
      "loss": 0.9351,
      "step": 11523
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.464479446411133,
      "learning_rate": 1.325623742133342e-05,
      "loss": 0.96,
      "step": 11524
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8714888095855713,
      "learning_rate": 1.325559850493563e-05,
      "loss": 0.9382,
      "step": 11525
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6386598348617554,
      "learning_rate": 1.3254959588537842e-05,
      "loss": 0.9339,
      "step": 11526
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6464190483093262,
      "learning_rate": 1.325432067214005e-05,
      "loss": 0.9265,
      "step": 11527
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.363786220550537,
      "learning_rate": 1.3253681755742263e-05,
      "loss": 0.9035,
      "step": 11528
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.13423752784729,
      "learning_rate": 1.3253042839344471e-05,
      "loss": 1.2015,
      "step": 11529
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0298097133636475,
      "learning_rate": 1.3252403922946685e-05,
      "loss": 1.2688,
      "step": 11530
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8358283042907715,
      "learning_rate": 1.3251765006548893e-05,
      "loss": 0.9233,
      "step": 11531
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.8124284744262695,
      "learning_rate": 1.3251126090151104e-05,
      "loss": 1.2269,
      "step": 11532
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.731958031654358,
      "learning_rate": 1.3250487173753314e-05,
      "loss": 0.9759,
      "step": 11533
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.338682174682617,
      "learning_rate": 1.3249848257355526e-05,
      "loss": 0.765,
      "step": 11534
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.139892578125,
      "learning_rate": 1.3249209340957736e-05,
      "loss": 1.3164,
      "step": 11535
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1682562828063965,
      "learning_rate": 1.3248570424559947e-05,
      "loss": 0.8954,
      "step": 11536
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.7404701709747314,
      "learning_rate": 1.3247931508162157e-05,
      "loss": 0.957,
      "step": 11537
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.005628824234009,
      "learning_rate": 1.3247292591764369e-05,
      "loss": 0.9668,
      "step": 11538
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.417224884033203,
      "learning_rate": 1.3246653675366578e-05,
      "loss": 1.0032,
      "step": 11539
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7814383506774902,
      "learning_rate": 1.324601475896879e-05,
      "loss": 0.8826,
      "step": 11540
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4882872104644775,
      "learning_rate": 1.3245375842571e-05,
      "loss": 0.8563,
      "step": 11541
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9555236101150513,
      "learning_rate": 1.3244736926173211e-05,
      "loss": 0.9252,
      "step": 11542
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9404985904693604,
      "learning_rate": 1.3244098009775421e-05,
      "loss": 1.4468,
      "step": 11543
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7402235269546509,
      "learning_rate": 1.3243459093377633e-05,
      "loss": 0.8503,
      "step": 11544
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7502104043960571,
      "learning_rate": 1.3242820176979843e-05,
      "loss": 0.9428,
      "step": 11545
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.969341516494751,
      "learning_rate": 1.3242181260582054e-05,
      "loss": 1.0011,
      "step": 11546
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2153525352478027,
      "learning_rate": 1.3241542344184264e-05,
      "loss": 0.9253,
      "step": 11547
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9434808492660522,
      "learning_rate": 1.3240903427786476e-05,
      "loss": 0.9287,
      "step": 11548
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5992155075073242,
      "learning_rate": 1.3240264511388685e-05,
      "loss": 0.8886,
      "step": 11549
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.842470645904541,
      "learning_rate": 1.3239625594990897e-05,
      "loss": 1.4783,
      "step": 11550
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8335596323013306,
      "learning_rate": 1.3238986678593107e-05,
      "loss": 0.7954,
      "step": 11551
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.71759831905365,
      "learning_rate": 1.3238347762195318e-05,
      "loss": 0.9213,
      "step": 11552
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8697142601013184,
      "learning_rate": 1.3237708845797528e-05,
      "loss": 0.8008,
      "step": 11553
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9542033672332764,
      "learning_rate": 1.323706992939974e-05,
      "loss": 0.9767,
      "step": 11554
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9670448303222656,
      "learning_rate": 1.323643101300195e-05,
      "loss": 0.918,
      "step": 11555
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2456884384155273,
      "learning_rate": 1.3235792096604161e-05,
      "loss": 0.9556,
      "step": 11556
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5056785345077515,
      "learning_rate": 1.3235153180206371e-05,
      "loss": 0.9605,
      "step": 11557
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.339588165283203,
      "learning_rate": 1.3234514263808583e-05,
      "loss": 1.1205,
      "step": 11558
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9151360988616943,
      "learning_rate": 1.3233875347410792e-05,
      "loss": 0.8628,
      "step": 11559
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6923800706863403,
      "learning_rate": 1.3233236431013004e-05,
      "loss": 0.8902,
      "step": 11560
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.9153239727020264,
      "learning_rate": 1.3232597514615214e-05,
      "loss": 1.0624,
      "step": 11561
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.527435302734375,
      "learning_rate": 1.3231958598217425e-05,
      "loss": 0.8321,
      "step": 11562
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.121295928955078,
      "learning_rate": 1.3231319681819634e-05,
      "loss": 1.1685,
      "step": 11563
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.288329601287842,
      "learning_rate": 1.3230680765421847e-05,
      "loss": 1.083,
      "step": 11564
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.7680859565734863,
      "learning_rate": 1.3230041849024055e-05,
      "loss": 1.0149,
      "step": 11565
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9199275970458984,
      "learning_rate": 1.3229402932626268e-05,
      "loss": 0.8888,
      "step": 11566
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9757963418960571,
      "learning_rate": 1.3228764016228476e-05,
      "loss": 0.9958,
      "step": 11567
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7861977815628052,
      "learning_rate": 1.3228125099830688e-05,
      "loss": 0.9092,
      "step": 11568
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0536458492279053,
      "learning_rate": 1.3227486183432898e-05,
      "loss": 0.874,
      "step": 11569
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.411895751953125,
      "learning_rate": 1.322684726703511e-05,
      "loss": 1.4116,
      "step": 11570
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9592968225479126,
      "learning_rate": 1.3226208350637319e-05,
      "loss": 1.0691,
      "step": 11571
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7017866373062134,
      "learning_rate": 1.322556943423953e-05,
      "loss": 1.0137,
      "step": 11572
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6359919309616089,
      "learning_rate": 1.322493051784174e-05,
      "loss": 0.8803,
      "step": 11573
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.4192557334899902,
      "learning_rate": 1.3224291601443952e-05,
      "loss": 0.6843,
      "step": 11574
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7574273347854614,
      "learning_rate": 1.3223652685046162e-05,
      "loss": 0.9207,
      "step": 11575
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6151310205459595,
      "learning_rate": 1.3223013768648373e-05,
      "loss": 0.9793,
      "step": 11576
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.08335018157959,
      "learning_rate": 1.3222374852250583e-05,
      "loss": 1.015,
      "step": 11577
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6297780275344849,
      "learning_rate": 1.3221735935852795e-05,
      "loss": 0.94,
      "step": 11578
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.276289939880371,
      "learning_rate": 1.3221097019455005e-05,
      "loss": 1.0952,
      "step": 11579
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7855991125106812,
      "learning_rate": 1.3220458103057216e-05,
      "loss": 0.8741,
      "step": 11580
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.259263277053833,
      "learning_rate": 1.3219819186659426e-05,
      "loss": 1.0923,
      "step": 11581
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.453522205352783,
      "learning_rate": 1.3219180270261638e-05,
      "loss": 1.2425,
      "step": 11582
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4322495460510254,
      "learning_rate": 1.3218541353863848e-05,
      "loss": 0.8465,
      "step": 11583
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3563008308410645,
      "learning_rate": 1.3217902437466059e-05,
      "loss": 0.6086,
      "step": 11584
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7140116691589355,
      "learning_rate": 1.3217263521068269e-05,
      "loss": 0.8526,
      "step": 11585
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.097143650054932,
      "learning_rate": 1.321662460467048e-05,
      "loss": 1.159,
      "step": 11586
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7164387702941895,
      "learning_rate": 1.321598568827269e-05,
      "loss": 0.7883,
      "step": 11587
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.2732481956481934,
      "learning_rate": 1.3215346771874902e-05,
      "loss": 1.0737,
      "step": 11588
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6144870519638062,
      "learning_rate": 1.3214707855477112e-05,
      "loss": 0.9011,
      "step": 11589
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3442459106445312,
      "learning_rate": 1.3214068939079323e-05,
      "loss": 0.8489,
      "step": 11590
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.349961519241333,
      "learning_rate": 1.3213430022681533e-05,
      "loss": 1.1253,
      "step": 11591
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5914947986602783,
      "learning_rate": 1.3212791106283745e-05,
      "loss": 0.9239,
      "step": 11592
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.44354248046875,
      "learning_rate": 1.3212152189885955e-05,
      "loss": 1.1006,
      "step": 11593
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.0556576251983643,
      "learning_rate": 1.3211513273488166e-05,
      "loss": 1.0688,
      "step": 11594
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.619483232498169,
      "learning_rate": 1.3210874357090376e-05,
      "loss": 0.9648,
      "step": 11595
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8440277576446533,
      "learning_rate": 1.3210235440692587e-05,
      "loss": 0.8545,
      "step": 11596
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.985986709594727,
      "learning_rate": 1.3209596524294797e-05,
      "loss": 1.2903,
      "step": 11597
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.952159404754639,
      "learning_rate": 1.3208957607897009e-05,
      "loss": 1.1736,
      "step": 11598
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8419970273971558,
      "learning_rate": 1.3208318691499217e-05,
      "loss": 1.2636,
      "step": 11599
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.997692584991455,
      "learning_rate": 1.320767977510143e-05,
      "loss": 1.6682,
      "step": 11600
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6032212972640991,
      "learning_rate": 1.3207040858703638e-05,
      "loss": 0.876,
      "step": 11601
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.182171106338501,
      "learning_rate": 1.3206401942305852e-05,
      "loss": 0.9398,
      "step": 11602
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.7902119159698486,
      "learning_rate": 1.320576302590806e-05,
      "loss": 0.7437,
      "step": 11603
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7511324882507324,
      "learning_rate": 1.3205124109510273e-05,
      "loss": 0.8558,
      "step": 11604
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.441450834274292,
      "learning_rate": 1.3204485193112481e-05,
      "loss": 0.6249,
      "step": 11605
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6209132671356201,
      "learning_rate": 1.3203846276714693e-05,
      "loss": 0.9129,
      "step": 11606
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5098450183868408,
      "learning_rate": 1.3203207360316903e-05,
      "loss": 0.4055,
      "step": 11607
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.2278857231140137,
      "learning_rate": 1.3202568443919114e-05,
      "loss": 0.7031,
      "step": 11608
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8803375959396362,
      "learning_rate": 1.3201929527521324e-05,
      "loss": 1.2383,
      "step": 11609
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.748399019241333,
      "learning_rate": 1.3201290611123536e-05,
      "loss": 0.6847,
      "step": 11610
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9710922241210938,
      "learning_rate": 1.3200651694725745e-05,
      "loss": 1.0192,
      "step": 11611
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9331990480422974,
      "learning_rate": 1.3200012778327957e-05,
      "loss": 0.9388,
      "step": 11612
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7596443891525269,
      "learning_rate": 1.3199373861930167e-05,
      "loss": 0.8576,
      "step": 11613
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4762120246887207,
      "learning_rate": 1.3198734945532378e-05,
      "loss": 1.0074,
      "step": 11614
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.502776622772217,
      "learning_rate": 1.3198096029134588e-05,
      "loss": 1.0329,
      "step": 11615
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.05591082572937,
      "learning_rate": 1.31974571127368e-05,
      "loss": 1.3166,
      "step": 11616
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.706862449645996,
      "learning_rate": 1.319681819633901e-05,
      "loss": 1.1912,
      "step": 11617
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.9322967529296875,
      "learning_rate": 1.3196179279941221e-05,
      "loss": 2.1025,
      "step": 11618
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6138209104537964,
      "learning_rate": 1.3195540363543431e-05,
      "loss": 0.8125,
      "step": 11619
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.406589984893799,
      "learning_rate": 1.3194901447145643e-05,
      "loss": 0.9867,
      "step": 11620
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6387354135513306,
      "learning_rate": 1.3194262530747852e-05,
      "loss": 0.4938,
      "step": 11621
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.480146050453186,
      "learning_rate": 1.3193623614350064e-05,
      "loss": 0.9496,
      "step": 11622
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.740891456604004,
      "learning_rate": 1.3192984697952274e-05,
      "loss": 1.2254,
      "step": 11623
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.6174042224884033,
      "learning_rate": 1.3192345781554485e-05,
      "loss": 1.0566,
      "step": 11624
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.386199712753296,
      "learning_rate": 1.3191706865156695e-05,
      "loss": 0.8588,
      "step": 11625
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7668043375015259,
      "learning_rate": 1.3191067948758907e-05,
      "loss": 0.9755,
      "step": 11626
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4642452001571655,
      "learning_rate": 1.3190429032361117e-05,
      "loss": 0.8117,
      "step": 11627
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.832568883895874,
      "learning_rate": 1.3189790115963328e-05,
      "loss": 0.8802,
      "step": 11628
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.473855018615723,
      "learning_rate": 1.3189151199565538e-05,
      "loss": 1.3444,
      "step": 11629
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.7024149894714355,
      "learning_rate": 1.318851228316775e-05,
      "loss": 1.3617,
      "step": 11630
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6342027187347412,
      "learning_rate": 1.318787336676996e-05,
      "loss": 0.8695,
      "step": 11631
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0793826580047607,
      "learning_rate": 1.3187234450372171e-05,
      "loss": 0.8066,
      "step": 11632
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.131520986557007,
      "learning_rate": 1.318659553397438e-05,
      "loss": 0.9648,
      "step": 11633
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.90264892578125,
      "learning_rate": 1.3185956617576592e-05,
      "loss": 0.8274,
      "step": 11634
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9055604934692383,
      "learning_rate": 1.31853177011788e-05,
      "loss": 1.0504,
      "step": 11635
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.824027180671692,
      "learning_rate": 1.3184678784781014e-05,
      "loss": 0.9242,
      "step": 11636
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.51969051361084,
      "learning_rate": 1.3184039868383222e-05,
      "loss": 1.8476,
      "step": 11637
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8351563215255737,
      "learning_rate": 1.3183400951985435e-05,
      "loss": 0.9137,
      "step": 11638
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4032365083694458,
      "learning_rate": 1.3182762035587643e-05,
      "loss": 0.9194,
      "step": 11639
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7258810997009277,
      "learning_rate": 1.3182123119189857e-05,
      "loss": 1.1204,
      "step": 11640
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8048348426818848,
      "learning_rate": 1.3181484202792065e-05,
      "loss": 0.9917,
      "step": 11641
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3414196968078613,
      "learning_rate": 1.3180845286394276e-05,
      "loss": 0.7174,
      "step": 11642
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.801238775253296,
      "learning_rate": 1.3180206369996486e-05,
      "loss": 0.8418,
      "step": 11643
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.028923511505127,
      "learning_rate": 1.3179567453598698e-05,
      "loss": 0.9554,
      "step": 11644
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.812513828277588,
      "learning_rate": 1.3178928537200908e-05,
      "loss": 0.8025,
      "step": 11645
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8595035076141357,
      "learning_rate": 1.3178289620803119e-05,
      "loss": 0.8885,
      "step": 11646
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7625722885131836,
      "learning_rate": 1.3177650704405329e-05,
      "loss": 0.8818,
      "step": 11647
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6236552000045776,
      "learning_rate": 1.317701178800754e-05,
      "loss": 0.5059,
      "step": 11648
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.658239364624023,
      "learning_rate": 1.317637287160975e-05,
      "loss": 1.5438,
      "step": 11649
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9467236995697021,
      "learning_rate": 1.3175733955211962e-05,
      "loss": 0.9443,
      "step": 11650
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.066556692123413,
      "learning_rate": 1.3175095038814172e-05,
      "loss": 1.1495,
      "step": 11651
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1308798789978027,
      "learning_rate": 1.3174456122416383e-05,
      "loss": 1.0024,
      "step": 11652
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9160267114639282,
      "learning_rate": 1.3173817206018593e-05,
      "loss": 1.041,
      "step": 11653
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9515537023544312,
      "learning_rate": 1.3173178289620805e-05,
      "loss": 0.8415,
      "step": 11654
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0927507877349854,
      "learning_rate": 1.3172539373223015e-05,
      "loss": 0.9317,
      "step": 11655
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8972713947296143,
      "learning_rate": 1.3171900456825226e-05,
      "loss": 1.1812,
      "step": 11656
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.0724165439605713,
      "learning_rate": 1.3171261540427436e-05,
      "loss": 1.0343,
      "step": 11657
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9885915517807007,
      "learning_rate": 1.3170622624029647e-05,
      "loss": 0.8617,
      "step": 11658
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0135409832000732,
      "learning_rate": 1.3169983707631857e-05,
      "loss": 0.9737,
      "step": 11659
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9243438243865967,
      "learning_rate": 1.3169344791234069e-05,
      "loss": 1.4081,
      "step": 11660
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9174162149429321,
      "learning_rate": 1.3168705874836279e-05,
      "loss": 1.0548,
      "step": 11661
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6195905208587646,
      "learning_rate": 1.316806695843849e-05,
      "loss": 0.9302,
      "step": 11662
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.742984652519226,
      "learning_rate": 1.31674280420407e-05,
      "loss": 0.912,
      "step": 11663
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.999999761581421,
      "learning_rate": 1.3166789125642912e-05,
      "loss": 0.8487,
      "step": 11664
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8386355638504028,
      "learning_rate": 1.3166150209245122e-05,
      "loss": 0.9309,
      "step": 11665
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6384072303771973,
      "learning_rate": 1.3165511292847333e-05,
      "loss": 0.9258,
      "step": 11666
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7559930086135864,
      "learning_rate": 1.3164872376449543e-05,
      "loss": 0.7869,
      "step": 11667
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.208798885345459,
      "learning_rate": 1.3164233460051754e-05,
      "loss": 0.754,
      "step": 11668
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.37187123298645,
      "learning_rate": 1.3163594543653964e-05,
      "loss": 1.2627,
      "step": 11669
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.971518635749817,
      "learning_rate": 1.3162955627256176e-05,
      "loss": 0.7085,
      "step": 11670
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9679993391036987,
      "learning_rate": 1.3162316710858386e-05,
      "loss": 0.9986,
      "step": 11671
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7871803045272827,
      "learning_rate": 1.3161677794460597e-05,
      "loss": 0.926,
      "step": 11672
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9237157106399536,
      "learning_rate": 1.3161038878062805e-05,
      "loss": 0.9979,
      "step": 11673
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6706595420837402,
      "learning_rate": 1.3160399961665019e-05,
      "loss": 0.3884,
      "step": 11674
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9739598035812378,
      "learning_rate": 1.3159761045267227e-05,
      "loss": 0.977,
      "step": 11675
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4263393878936768,
      "learning_rate": 1.315912212886944e-05,
      "loss": 1.0348,
      "step": 11676
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5829521417617798,
      "learning_rate": 1.3158483212471648e-05,
      "loss": 0.9275,
      "step": 11677
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8008450269699097,
      "learning_rate": 1.315784429607386e-05,
      "loss": 0.9069,
      "step": 11678
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.9933958053588867,
      "learning_rate": 1.315720537967607e-05,
      "loss": 0.9577,
      "step": 11679
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5053167343139648,
      "learning_rate": 1.3156566463278281e-05,
      "loss": 0.9127,
      "step": 11680
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6562715768814087,
      "learning_rate": 1.3155927546880491e-05,
      "loss": 0.9334,
      "step": 11681
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.649672508239746,
      "learning_rate": 1.3155288630482703e-05,
      "loss": 1.0895,
      "step": 11682
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.098354339599609,
      "learning_rate": 1.3154649714084912e-05,
      "loss": 1.4079,
      "step": 11683
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1781203746795654,
      "learning_rate": 1.3154010797687124e-05,
      "loss": 1.0337,
      "step": 11684
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5763779878616333,
      "learning_rate": 1.3153371881289334e-05,
      "loss": 0.7416,
      "step": 11685
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7435901165008545,
      "learning_rate": 1.3152732964891545e-05,
      "loss": 0.8268,
      "step": 11686
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9979697465896606,
      "learning_rate": 1.3152094048493755e-05,
      "loss": 1.2327,
      "step": 11687
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6332132816314697,
      "learning_rate": 1.3151455132095967e-05,
      "loss": 0.9397,
      "step": 11688
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.6491425037384033,
      "learning_rate": 1.3150816215698177e-05,
      "loss": 1.0937,
      "step": 11689
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9517276287078857,
      "learning_rate": 1.3150177299300388e-05,
      "loss": 0.9222,
      "step": 11690
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8566722869873047,
      "learning_rate": 1.3149538382902598e-05,
      "loss": 0.8722,
      "step": 11691
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.861008644104004,
      "learning_rate": 1.314889946650481e-05,
      "loss": 0.7389,
      "step": 11692
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6487642526626587,
      "learning_rate": 1.314826055010702e-05,
      "loss": 0.7386,
      "step": 11693
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9085241556167603,
      "learning_rate": 1.3147621633709231e-05,
      "loss": 0.9666,
      "step": 11694
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.943665504455566,
      "learning_rate": 1.314698271731144e-05,
      "loss": 1.7455,
      "step": 11695
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.786914825439453,
      "learning_rate": 1.3146343800913652e-05,
      "loss": 0.8044,
      "step": 11696
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7312867641448975,
      "learning_rate": 1.3145704884515862e-05,
      "loss": 0.9465,
      "step": 11697
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.84209680557251,
      "learning_rate": 1.3145065968118074e-05,
      "loss": 1.9074,
      "step": 11698
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9676055908203125,
      "learning_rate": 1.3144427051720284e-05,
      "loss": 0.719,
      "step": 11699
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.5359394550323486,
      "learning_rate": 1.3143788135322495e-05,
      "loss": 1.0625,
      "step": 11700
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4117226600646973,
      "learning_rate": 1.3143149218924705e-05,
      "loss": 0.84,
      "step": 11701
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8484028577804565,
      "learning_rate": 1.3142510302526917e-05,
      "loss": 0.964,
      "step": 11702
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8280876874923706,
      "learning_rate": 1.3141871386129126e-05,
      "loss": 0.8313,
      "step": 11703
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7660350799560547,
      "learning_rate": 1.3141232469731338e-05,
      "loss": 0.8954,
      "step": 11704
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8035483360290527,
      "learning_rate": 1.3140593553333548e-05,
      "loss": 0.7939,
      "step": 11705
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6510518789291382,
      "learning_rate": 1.313995463693576e-05,
      "loss": 0.9411,
      "step": 11706
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8658899068832397,
      "learning_rate": 1.313931572053797e-05,
      "loss": 0.8669,
      "step": 11707
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4759399890899658,
      "learning_rate": 1.313867680414018e-05,
      "loss": 0.3716,
      "step": 11708
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6990728378295898,
      "learning_rate": 1.3138037887742389e-05,
      "loss": 0.82,
      "step": 11709
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.873074531555176,
      "learning_rate": 1.3137398971344602e-05,
      "loss": 1.2253,
      "step": 11710
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.777682304382324,
      "learning_rate": 1.313676005494681e-05,
      "loss": 0.9658,
      "step": 11711
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7410567998886108,
      "learning_rate": 1.3136121138549024e-05,
      "loss": 0.9142,
      "step": 11712
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8739205598831177,
      "learning_rate": 1.3135482222151232e-05,
      "loss": 0.9657,
      "step": 11713
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.775665521621704,
      "learning_rate": 1.3134843305753443e-05,
      "loss": 1.0731,
      "step": 11714
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.904299020767212,
      "learning_rate": 1.3134204389355653e-05,
      "loss": 1.0201,
      "step": 11715
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.49307918548584,
      "learning_rate": 1.3133565472957865e-05,
      "loss": 1.1028,
      "step": 11716
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.8858754634857178,
      "learning_rate": 1.3132926556560075e-05,
      "loss": 0.7407,
      "step": 11717
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.051678419113159,
      "learning_rate": 1.3132287640162286e-05,
      "loss": 0.8946,
      "step": 11718
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.472311496734619,
      "learning_rate": 1.3131648723764496e-05,
      "loss": 1.1133,
      "step": 11719
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6796772480010986,
      "learning_rate": 1.3131009807366707e-05,
      "loss": 0.9581,
      "step": 11720
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0565526485443115,
      "learning_rate": 1.3130370890968917e-05,
      "loss": 0.9003,
      "step": 11721
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5453481674194336,
      "learning_rate": 1.3129731974571129e-05,
      "loss": 0.9033,
      "step": 11722
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.612988829612732,
      "learning_rate": 1.3129093058173339e-05,
      "loss": 0.9265,
      "step": 11723
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.3420515060424805,
      "learning_rate": 1.312845414177555e-05,
      "loss": 1.6875,
      "step": 11724
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.806586742401123,
      "learning_rate": 1.312781522537776e-05,
      "loss": 0.9826,
      "step": 11725
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.770424842834473,
      "learning_rate": 1.3127176308979972e-05,
      "loss": 0.5391,
      "step": 11726
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5141749382019043,
      "learning_rate": 1.3126537392582182e-05,
      "loss": 0.8707,
      "step": 11727
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5908746719360352,
      "learning_rate": 1.3125898476184393e-05,
      "loss": 1.0047,
      "step": 11728
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5392014980316162,
      "learning_rate": 1.3125259559786603e-05,
      "loss": 0.8951,
      "step": 11729
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.7872817516326904,
      "learning_rate": 1.3124620643388814e-05,
      "loss": 1.2972,
      "step": 11730
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.010842800140381,
      "learning_rate": 1.3123981726991024e-05,
      "loss": 1.352,
      "step": 11731
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8125346899032593,
      "learning_rate": 1.3123342810593236e-05,
      "loss": 0.954,
      "step": 11732
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0884199142456055,
      "learning_rate": 1.3122703894195446e-05,
      "loss": 1.2875,
      "step": 11733
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.842063307762146,
      "learning_rate": 1.3122064977797657e-05,
      "loss": 0.6998,
      "step": 11734
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8317292928695679,
      "learning_rate": 1.3121426061399867e-05,
      "loss": 0.8459,
      "step": 11735
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6021260023117065,
      "learning_rate": 1.3120787145002079e-05,
      "loss": 0.8377,
      "step": 11736
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.76022469997406,
      "learning_rate": 1.3120148228604289e-05,
      "loss": 0.8872,
      "step": 11737
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.231987476348877,
      "learning_rate": 1.31195093122065e-05,
      "loss": 1.564,
      "step": 11738
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7625758647918701,
      "learning_rate": 1.311887039580871e-05,
      "loss": 0.9267,
      "step": 11739
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.667756080627441,
      "learning_rate": 1.3118231479410921e-05,
      "loss": 1.4736,
      "step": 11740
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.220001220703125,
      "learning_rate": 1.3117592563013131e-05,
      "loss": 1.4537,
      "step": 11741
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8824759721755981,
      "learning_rate": 1.3116953646615343e-05,
      "loss": 1.0218,
      "step": 11742
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.870725154876709,
      "learning_rate": 1.3116314730217553e-05,
      "loss": 0.4384,
      "step": 11743
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9087209701538086,
      "learning_rate": 1.3115675813819764e-05,
      "loss": 0.9139,
      "step": 11744
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1105175018310547,
      "learning_rate": 1.3115036897421972e-05,
      "loss": 0.9498,
      "step": 11745
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8723870515823364,
      "learning_rate": 1.3114397981024186e-05,
      "loss": 0.9979,
      "step": 11746
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.615027904510498,
      "learning_rate": 1.3113759064626394e-05,
      "loss": 0.8819,
      "step": 11747
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.087948322296143,
      "learning_rate": 1.3113120148228607e-05,
      "loss": 1.1198,
      "step": 11748
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3641107082366943,
      "learning_rate": 1.3112481231830815e-05,
      "loss": 0.8409,
      "step": 11749
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.756911516189575,
      "learning_rate": 1.3111842315433028e-05,
      "loss": 0.8956,
      "step": 11750
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8237740993499756,
      "learning_rate": 1.3111203399035237e-05,
      "loss": 0.7611,
      "step": 11751
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7199100255966187,
      "learning_rate": 1.3110564482637448e-05,
      "loss": 0.9889,
      "step": 11752
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7413376569747925,
      "learning_rate": 1.3109925566239658e-05,
      "loss": 0.9854,
      "step": 11753
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7963714599609375,
      "learning_rate": 1.310928664984187e-05,
      "loss": 0.9384,
      "step": 11754
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4527488946914673,
      "learning_rate": 1.310864773344408e-05,
      "loss": 0.8889,
      "step": 11755
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5851049423217773,
      "learning_rate": 1.3108008817046291e-05,
      "loss": 0.8419,
      "step": 11756
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5445270538330078,
      "learning_rate": 1.3107369900648501e-05,
      "loss": 0.8933,
      "step": 11757
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.230769157409668,
      "learning_rate": 1.3106730984250712e-05,
      "loss": 1.2508,
      "step": 11758
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7823501825332642,
      "learning_rate": 1.3106092067852922e-05,
      "loss": 0.8786,
      "step": 11759
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.9526255130767822,
      "learning_rate": 1.3105453151455134e-05,
      "loss": 1.0364,
      "step": 11760
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6586015224456787,
      "learning_rate": 1.3104814235057344e-05,
      "loss": 0.4394,
      "step": 11761
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7577613592147827,
      "learning_rate": 1.3104175318659555e-05,
      "loss": 0.9705,
      "step": 11762
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.793851137161255,
      "learning_rate": 1.3103536402261765e-05,
      "loss": 0.7046,
      "step": 11763
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.082944869995117,
      "learning_rate": 1.3102897485863977e-05,
      "loss": 0.8585,
      "step": 11764
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5422446727752686,
      "learning_rate": 1.3102258569466186e-05,
      "loss": 0.9556,
      "step": 11765
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.137214183807373,
      "learning_rate": 1.3101619653068398e-05,
      "loss": 1.1353,
      "step": 11766
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.785733938217163,
      "learning_rate": 1.3100980736670608e-05,
      "loss": 0.5796,
      "step": 11767
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.896355390548706,
      "learning_rate": 1.310034182027282e-05,
      "loss": 0.9621,
      "step": 11768
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.364790678024292,
      "learning_rate": 1.309970290387503e-05,
      "loss": 1.064,
      "step": 11769
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7236335277557373,
      "learning_rate": 1.309906398747724e-05,
      "loss": 0.8215,
      "step": 11770
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.384242296218872,
      "learning_rate": 1.309842507107945e-05,
      "loss": 0.769,
      "step": 11771
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2144837379455566,
      "learning_rate": 1.3097786154681662e-05,
      "loss": 0.889,
      "step": 11772
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5654256343841553,
      "learning_rate": 1.3097147238283872e-05,
      "loss": 0.8111,
      "step": 11773
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9494727849960327,
      "learning_rate": 1.3096508321886084e-05,
      "loss": 0.9275,
      "step": 11774
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5435677766799927,
      "learning_rate": 1.3095869405488293e-05,
      "loss": 0.9144,
      "step": 11775
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6142492294311523,
      "learning_rate": 1.3095230489090505e-05,
      "loss": 0.7731,
      "step": 11776
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.5111491680145264,
      "learning_rate": 1.3094591572692715e-05,
      "loss": 0.8539,
      "step": 11777
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.813402771949768,
      "learning_rate": 1.3093952656294926e-05,
      "loss": 0.9206,
      "step": 11778
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.107124090194702,
      "learning_rate": 1.3093313739897136e-05,
      "loss": 0.8597,
      "step": 11779
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7028443813323975,
      "learning_rate": 1.3092674823499348e-05,
      "loss": 0.4453,
      "step": 11780
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.91331148147583,
      "learning_rate": 1.3092035907101558e-05,
      "loss": 1.0294,
      "step": 11781
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6859134435653687,
      "learning_rate": 1.309139699070377e-05,
      "loss": 0.9707,
      "step": 11782
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7472423315048218,
      "learning_rate": 1.3090758074305977e-05,
      "loss": 0.885,
      "step": 11783
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1844217777252197,
      "learning_rate": 1.309011915790819e-05,
      "loss": 0.9148,
      "step": 11784
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2372772693634033,
      "learning_rate": 1.3089480241510399e-05,
      "loss": 0.9353,
      "step": 11785
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4586989879608154,
      "learning_rate": 1.3088841325112612e-05,
      "loss": 0.7805,
      "step": 11786
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.000888347625732,
      "learning_rate": 1.308820240871482e-05,
      "loss": 1.5047,
      "step": 11787
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7782877683639526,
      "learning_rate": 1.3087563492317032e-05,
      "loss": 0.921,
      "step": 11788
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.771545171737671,
      "learning_rate": 1.3086924575919242e-05,
      "loss": 0.9259,
      "step": 11789
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.27877140045166,
      "learning_rate": 1.3086285659521451e-05,
      "loss": 0.9352,
      "step": 11790
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7245203256607056,
      "learning_rate": 1.3085646743123663e-05,
      "loss": 0.8311,
      "step": 11791
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7158267498016357,
      "learning_rate": 1.3085007826725873e-05,
      "loss": 0.9382,
      "step": 11792
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.201122522354126,
      "learning_rate": 1.3084368910328084e-05,
      "loss": 0.9809,
      "step": 11793
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.192127227783203,
      "learning_rate": 1.3083729993930294e-05,
      "loss": 1.0242,
      "step": 11794
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7420272827148438,
      "learning_rate": 1.3083091077532506e-05,
      "loss": 0.8863,
      "step": 11795
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8139081001281738,
      "learning_rate": 1.3082452161134716e-05,
      "loss": 0.7304,
      "step": 11796
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.433877944946289,
      "learning_rate": 1.3081813244736927e-05,
      "loss": 0.834,
      "step": 11797
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.267179489135742,
      "learning_rate": 1.3081174328339137e-05,
      "loss": 1.2069,
      "step": 11798
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0326099395751953,
      "learning_rate": 1.3080535411941349e-05,
      "loss": 1.0675,
      "step": 11799
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.6192922592163086,
      "learning_rate": 1.3079896495543558e-05,
      "loss": 1.1537,
      "step": 11800
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9973129034042358,
      "learning_rate": 1.307925757914577e-05,
      "loss": 1.0013,
      "step": 11801
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.842564582824707,
      "learning_rate": 1.307861866274798e-05,
      "loss": 0.8649,
      "step": 11802
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5929267406463623,
      "learning_rate": 1.3077979746350191e-05,
      "loss": 0.8329,
      "step": 11803
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9325242042541504,
      "learning_rate": 1.3077340829952401e-05,
      "loss": 0.8325,
      "step": 11804
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1508171558380127,
      "learning_rate": 1.3076701913554613e-05,
      "loss": 1.033,
      "step": 11805
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.467550277709961,
      "learning_rate": 1.3076062997156823e-05,
      "loss": 0.4252,
      "step": 11806
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2466087341308594,
      "learning_rate": 1.3075424080759034e-05,
      "loss": 0.7062,
      "step": 11807
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.568370819091797,
      "learning_rate": 1.3074785164361244e-05,
      "loss": 0.9327,
      "step": 11808
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6494908332824707,
      "learning_rate": 1.3074146247963456e-05,
      "loss": 0.8327,
      "step": 11809
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.67742919921875,
      "learning_rate": 1.3073507331565665e-05,
      "loss": 0.9426,
      "step": 11810
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7630655765533447,
      "learning_rate": 1.3072868415167877e-05,
      "loss": 0.8645,
      "step": 11811
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6553804874420166,
      "learning_rate": 1.3072229498770085e-05,
      "loss": 0.7839,
      "step": 11812
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0369720458984375,
      "learning_rate": 1.3071590582372298e-05,
      "loss": 0.7925,
      "step": 11813
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8566734790802002,
      "learning_rate": 1.3070951665974507e-05,
      "loss": 0.7994,
      "step": 11814
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.164170265197754,
      "learning_rate": 1.307031274957672e-05,
      "loss": 0.916,
      "step": 11815
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.796424388885498,
      "learning_rate": 1.3069673833178928e-05,
      "loss": 0.8563,
      "step": 11816
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.873624801635742,
      "learning_rate": 1.3069034916781141e-05,
      "loss": 1.3045,
      "step": 11817
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.7574429512023926,
      "learning_rate": 1.306839600038335e-05,
      "loss": 0.8925,
      "step": 11818
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2418298721313477,
      "learning_rate": 1.3067757083985561e-05,
      "loss": 1.219,
      "step": 11819
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8811933994293213,
      "learning_rate": 1.306711816758777e-05,
      "loss": 1.1271,
      "step": 11820
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6560195684432983,
      "learning_rate": 1.3066479251189982e-05,
      "loss": 0.9043,
      "step": 11821
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0543694496154785,
      "learning_rate": 1.3065840334792192e-05,
      "loss": 0.9816,
      "step": 11822
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6170285940170288,
      "learning_rate": 1.3065201418394404e-05,
      "loss": 0.8376,
      "step": 11823
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2772586345672607,
      "learning_rate": 1.3064562501996614e-05,
      "loss": 0.984,
      "step": 11824
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7751749753952026,
      "learning_rate": 1.3063923585598825e-05,
      "loss": 0.9456,
      "step": 11825
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.477843165397644,
      "learning_rate": 1.3063284669201035e-05,
      "loss": 0.4244,
      "step": 11826
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.006448984146118,
      "learning_rate": 1.3062645752803246e-05,
      "loss": 0.8902,
      "step": 11827
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9252243041992188,
      "learning_rate": 1.3062006836405456e-05,
      "loss": 1.2545,
      "step": 11828
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.02289080619812,
      "learning_rate": 1.3061367920007668e-05,
      "loss": 1.096,
      "step": 11829
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8292081356048584,
      "learning_rate": 1.3060729003609878e-05,
      "loss": 1.0013,
      "step": 11830
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8674904108047485,
      "learning_rate": 1.306009008721209e-05,
      "loss": 1.1186,
      "step": 11831
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.355349063873291,
      "learning_rate": 1.3059451170814299e-05,
      "loss": 0.7721,
      "step": 11832
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1709144115448,
      "learning_rate": 1.305881225441651e-05,
      "loss": 0.9202,
      "step": 11833
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.8930082321167,
      "learning_rate": 1.305817333801872e-05,
      "loss": 1.2933,
      "step": 11834
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7682850360870361,
      "learning_rate": 1.3057534421620932e-05,
      "loss": 0.7622,
      "step": 11835
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7966904640197754,
      "learning_rate": 1.3056895505223142e-05,
      "loss": 0.8646,
      "step": 11836
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.135413408279419,
      "learning_rate": 1.3056256588825353e-05,
      "loss": 0.9714,
      "step": 11837
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6793168783187866,
      "learning_rate": 1.3055617672427563e-05,
      "loss": 0.8619,
      "step": 11838
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0874338150024414,
      "learning_rate": 1.3054978756029775e-05,
      "loss": 1.1055,
      "step": 11839
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6189703941345215,
      "learning_rate": 1.3054339839631985e-05,
      "loss": 0.9411,
      "step": 11840
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.125667572021484,
      "learning_rate": 1.3053700923234196e-05,
      "loss": 1.1399,
      "step": 11841
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.322723388671875,
      "learning_rate": 1.3053062006836406e-05,
      "loss": 0.8511,
      "step": 11842
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7052363157272339,
      "learning_rate": 1.3052423090438618e-05,
      "loss": 0.9772,
      "step": 11843
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.940385341644287,
      "learning_rate": 1.3051784174040828e-05,
      "loss": 0.9109,
      "step": 11844
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0092835426330566,
      "learning_rate": 1.3051145257643039e-05,
      "loss": 0.9001,
      "step": 11845
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.68351411819458,
      "learning_rate": 1.3050506341245249e-05,
      "loss": 0.933,
      "step": 11846
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6028387546539307,
      "learning_rate": 1.304986742484746e-05,
      "loss": 0.4616,
      "step": 11847
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.4404051303863525,
      "learning_rate": 1.304922850844967e-05,
      "loss": 0.5056,
      "step": 11848
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.863973617553711,
      "learning_rate": 1.3048589592051882e-05,
      "loss": 1.0425,
      "step": 11849
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0894742012023926,
      "learning_rate": 1.304795067565409e-05,
      "loss": 0.9799,
      "step": 11850
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.3308043479919434,
      "learning_rate": 1.3047311759256303e-05,
      "loss": 0.7547,
      "step": 11851
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2908644676208496,
      "learning_rate": 1.3046672842858511e-05,
      "loss": 1.005,
      "step": 11852
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.063439846038818,
      "learning_rate": 1.3046033926460725e-05,
      "loss": 1.4746,
      "step": 11853
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2869324684143066,
      "learning_rate": 1.3045395010062933e-05,
      "loss": 0.9013,
      "step": 11854
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.184493541717529,
      "learning_rate": 1.3044756093665144e-05,
      "loss": 1.0928,
      "step": 11855
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6594706773757935,
      "learning_rate": 1.3044117177267354e-05,
      "loss": 0.9093,
      "step": 11856
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8078101873397827,
      "learning_rate": 1.3043478260869566e-05,
      "loss": 0.9643,
      "step": 11857
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3955501317977905,
      "learning_rate": 1.3042839344471776e-05,
      "loss": 0.4875,
      "step": 11858
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8276612758636475,
      "learning_rate": 1.3042200428073987e-05,
      "loss": 1.1631,
      "step": 11859
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.829421281814575,
      "learning_rate": 1.3041561511676197e-05,
      "loss": 1.3357,
      "step": 11860
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9217756986618042,
      "learning_rate": 1.3040922595278409e-05,
      "loss": 0.9781,
      "step": 11861
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7029567956924438,
      "learning_rate": 1.3040283678880618e-05,
      "loss": 0.8133,
      "step": 11862
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.654362678527832,
      "learning_rate": 1.303964476248283e-05,
      "loss": 0.8933,
      "step": 11863
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.7565059661865234,
      "learning_rate": 1.303900584608504e-05,
      "loss": 1.0305,
      "step": 11864
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0400712490081787,
      "learning_rate": 1.3038366929687251e-05,
      "loss": 0.9202,
      "step": 11865
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.724531412124634,
      "learning_rate": 1.3037728013289461e-05,
      "loss": 0.7943,
      "step": 11866
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.248870849609375,
      "learning_rate": 1.3037089096891673e-05,
      "loss": 0.8276,
      "step": 11867
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.518836498260498,
      "learning_rate": 1.3036450180493883e-05,
      "loss": 0.9483,
      "step": 11868
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9005670547485352,
      "learning_rate": 1.3035811264096094e-05,
      "loss": 1.0091,
      "step": 11869
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8075059652328491,
      "learning_rate": 1.3035172347698304e-05,
      "loss": 0.8796,
      "step": 11870
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0736653804779053,
      "learning_rate": 1.3034533431300516e-05,
      "loss": 0.8167,
      "step": 11871
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7069900035858154,
      "learning_rate": 1.3033894514902725e-05,
      "loss": 0.954,
      "step": 11872
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6290478706359863,
      "learning_rate": 1.3033255598504937e-05,
      "loss": 0.4037,
      "step": 11873
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1600873470306396,
      "learning_rate": 1.3032616682107147e-05,
      "loss": 1.1991,
      "step": 11874
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9310351610183716,
      "learning_rate": 1.3031977765709358e-05,
      "loss": 0.8652,
      "step": 11875
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.4941906929016113,
      "learning_rate": 1.3031338849311568e-05,
      "loss": 0.7844,
      "step": 11876
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9882245063781738,
      "learning_rate": 1.303069993291378e-05,
      "loss": 0.9548,
      "step": 11877
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7306749820709229,
      "learning_rate": 1.303006101651599e-05,
      "loss": 0.8816,
      "step": 11878
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.498416423797607,
      "learning_rate": 1.3029422100118201e-05,
      "loss": 1.9882,
      "step": 11879
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.155329704284668,
      "learning_rate": 1.3028783183720411e-05,
      "loss": 1.1908,
      "step": 11880
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.2793939113616943,
      "learning_rate": 1.3028144267322623e-05,
      "loss": 1.2724,
      "step": 11881
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.61622953414917,
      "learning_rate": 1.3027505350924832e-05,
      "loss": 0.9978,
      "step": 11882
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.074183702468872,
      "learning_rate": 1.3026866434527044e-05,
      "loss": 0.8911,
      "step": 11883
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8790913820266724,
      "learning_rate": 1.3026227518129254e-05,
      "loss": 0.8782,
      "step": 11884
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5507405996322632,
      "learning_rate": 1.3025588601731465e-05,
      "loss": 0.8218,
      "step": 11885
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5966376066207886,
      "learning_rate": 1.3024949685333674e-05,
      "loss": 0.8411,
      "step": 11886
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8931456804275513,
      "learning_rate": 1.3024310768935887e-05,
      "loss": 0.9225,
      "step": 11887
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.7231392860412598,
      "learning_rate": 1.3023671852538095e-05,
      "loss": 1.4668,
      "step": 11888
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9445019960403442,
      "learning_rate": 1.3023032936140308e-05,
      "loss": 1.0073,
      "step": 11889
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.22160267829895,
      "learning_rate": 1.3022394019742516e-05,
      "loss": 0.9499,
      "step": 11890
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7521204948425293,
      "learning_rate": 1.3021755103344728e-05,
      "loss": 0.928,
      "step": 11891
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6339713335037231,
      "learning_rate": 1.3021116186946938e-05,
      "loss": 0.8969,
      "step": 11892
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.055034875869751,
      "learning_rate": 1.302047727054915e-05,
      "loss": 0.9178,
      "step": 11893
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.6174163818359375,
      "learning_rate": 1.3019838354151359e-05,
      "loss": 1.1201,
      "step": 11894
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.999316692352295,
      "learning_rate": 1.301919943775357e-05,
      "loss": 1.1304,
      "step": 11895
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.89453649520874,
      "learning_rate": 1.301856052135578e-05,
      "loss": 1.6288,
      "step": 11896
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7575263977050781,
      "learning_rate": 1.3017921604957992e-05,
      "loss": 0.9201,
      "step": 11897
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1895978450775146,
      "learning_rate": 1.3017282688560202e-05,
      "loss": 0.9268,
      "step": 11898
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6068321466445923,
      "learning_rate": 1.3016643772162413e-05,
      "loss": 0.8738,
      "step": 11899
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.146052360534668,
      "learning_rate": 1.3016004855764623e-05,
      "loss": 1.0099,
      "step": 11900
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.443310260772705,
      "learning_rate": 1.3015365939366835e-05,
      "loss": 0.9111,
      "step": 11901
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5766689777374268,
      "learning_rate": 1.3014727022969045e-05,
      "loss": 0.9302,
      "step": 11902
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.834784507751465,
      "learning_rate": 1.3014088106571256e-05,
      "loss": 1.1235,
      "step": 11903
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.842116832733154,
      "learning_rate": 1.3013449190173466e-05,
      "loss": 1.2713,
      "step": 11904
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7196290493011475,
      "learning_rate": 1.3012810273775678e-05,
      "loss": 0.9103,
      "step": 11905
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1962502002716064,
      "learning_rate": 1.3012171357377888e-05,
      "loss": 0.7858,
      "step": 11906
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5163805484771729,
      "learning_rate": 1.3011532440980099e-05,
      "loss": 0.7759,
      "step": 11907
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8368587493896484,
      "learning_rate": 1.3010893524582309e-05,
      "loss": 0.942,
      "step": 11908
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5068191289901733,
      "learning_rate": 1.301025460818452e-05,
      "loss": 0.819,
      "step": 11909
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.547909736633301,
      "learning_rate": 1.300961569178673e-05,
      "loss": 1.1634,
      "step": 11910
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.779679775238037,
      "learning_rate": 1.3008976775388942e-05,
      "loss": 1.0089,
      "step": 11911
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.816509485244751,
      "learning_rate": 1.3008337858991152e-05,
      "loss": 0.6487,
      "step": 11912
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3507872819900513,
      "learning_rate": 1.3007698942593363e-05,
      "loss": 0.8134,
      "step": 11913
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8625845909118652,
      "learning_rate": 1.3007060026195573e-05,
      "loss": 0.9339,
      "step": 11914
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6515223979949951,
      "learning_rate": 1.3006421109797785e-05,
      "loss": 0.9471,
      "step": 11915
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9774903059005737,
      "learning_rate": 1.3005782193399995e-05,
      "loss": 0.8963,
      "step": 11916
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.145824670791626,
      "learning_rate": 1.3005143277002206e-05,
      "loss": 0.855,
      "step": 11917
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7113310098648071,
      "learning_rate": 1.3004504360604416e-05,
      "loss": 0.8991,
      "step": 11918
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.593105435371399,
      "learning_rate": 1.3003865444206627e-05,
      "loss": 1.0398,
      "step": 11919
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7385327816009521,
      "learning_rate": 1.3003226527808837e-05,
      "loss": 0.9187,
      "step": 11920
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.837897777557373,
      "learning_rate": 1.3002587611411049e-05,
      "loss": 0.942,
      "step": 11921
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.037360668182373,
      "learning_rate": 1.3001948695013257e-05,
      "loss": 0.9605,
      "step": 11922
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.261868715286255,
      "learning_rate": 1.300130977861547e-05,
      "loss": 0.8701,
      "step": 11923
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8734992742538452,
      "learning_rate": 1.3000670862217678e-05,
      "loss": 1.0991,
      "step": 11924
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.058974742889404,
      "learning_rate": 1.3000031945819892e-05,
      "loss": 1.1787,
      "step": 11925
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.946089506149292,
      "learning_rate": 1.29993930294221e-05,
      "loss": 0.8082,
      "step": 11926
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8564276695251465,
      "learning_rate": 1.2998754113024313e-05,
      "loss": 0.6158,
      "step": 11927
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9523354768753052,
      "learning_rate": 1.2998115196626521e-05,
      "loss": 0.9786,
      "step": 11928
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.100039482116699,
      "learning_rate": 1.2997476280228733e-05,
      "loss": 1.2495,
      "step": 11929
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.036036014556885,
      "learning_rate": 1.2996837363830943e-05,
      "loss": 1.3707,
      "step": 11930
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6966462135314941,
      "learning_rate": 1.2996198447433154e-05,
      "loss": 1.0887,
      "step": 11931
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8306055068969727,
      "learning_rate": 1.2995559531035364e-05,
      "loss": 0.844,
      "step": 11932
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1759114265441895,
      "learning_rate": 1.2994920614637576e-05,
      "loss": 0.8796,
      "step": 11933
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.5440971851348877,
      "learning_rate": 1.2994281698239785e-05,
      "loss": 0.9207,
      "step": 11934
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6434391736984253,
      "learning_rate": 1.2993642781841997e-05,
      "loss": 1.0003,
      "step": 11935
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.5801398754119873,
      "learning_rate": 1.2993003865444207e-05,
      "loss": 0.8934,
      "step": 11936
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0777077674865723,
      "learning_rate": 1.2992364949046418e-05,
      "loss": 0.9121,
      "step": 11937
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.325361251831055,
      "learning_rate": 1.2991726032648628e-05,
      "loss": 0.9954,
      "step": 11938
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8927205801010132,
      "learning_rate": 1.299108711625084e-05,
      "loss": 0.9955,
      "step": 11939
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8171515464782715,
      "learning_rate": 1.299044819985305e-05,
      "loss": 1.1451,
      "step": 11940
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7429463863372803,
      "learning_rate": 1.2989809283455261e-05,
      "loss": 1.2109,
      "step": 11941
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.586277961730957,
      "learning_rate": 1.2989170367057471e-05,
      "loss": 1.3971,
      "step": 11942
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.81758189201355,
      "learning_rate": 1.2988531450659683e-05,
      "loss": 1.0636,
      "step": 11943
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.981562614440918,
      "learning_rate": 1.2987892534261892e-05,
      "loss": 1.3009,
      "step": 11944
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9487494230270386,
      "learning_rate": 1.2987253617864104e-05,
      "loss": 1.2153,
      "step": 11945
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9670621156692505,
      "learning_rate": 1.2986614701466314e-05,
      "loss": 1.1191,
      "step": 11946
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8109325170516968,
      "learning_rate": 1.2985975785068525e-05,
      "loss": 0.8844,
      "step": 11947
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.785812258720398,
      "learning_rate": 1.2985336868670735e-05,
      "loss": 0.4672,
      "step": 11948
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.267160654067993,
      "learning_rate": 1.2984697952272947e-05,
      "loss": 0.88,
      "step": 11949
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.22766375541687,
      "learning_rate": 1.2984059035875157e-05,
      "loss": 0.954,
      "step": 11950
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.067612409591675,
      "learning_rate": 1.2983420119477368e-05,
      "loss": 1.1169,
      "step": 11951
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.077824115753174,
      "learning_rate": 1.2982781203079578e-05,
      "loss": 1.1559,
      "step": 11952
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6716125011444092,
      "learning_rate": 1.298214228668179e-05,
      "loss": 0.9089,
      "step": 11953
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.252061367034912,
      "learning_rate": 1.2981503370284e-05,
      "loss": 1.1557,
      "step": 11954
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8870197534561157,
      "learning_rate": 1.2980864453886211e-05,
      "loss": 0.9544,
      "step": 11955
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.413244247436523,
      "learning_rate": 1.298022553748842e-05,
      "loss": 1.7702,
      "step": 11956
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.686819553375244,
      "learning_rate": 1.2979586621090632e-05,
      "loss": 0.9918,
      "step": 11957
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6980503797531128,
      "learning_rate": 1.297894770469284e-05,
      "loss": 1.0082,
      "step": 11958
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6573717594146729,
      "learning_rate": 1.2978308788295054e-05,
      "loss": 0.7236,
      "step": 11959
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0826587677001953,
      "learning_rate": 1.2977669871897262e-05,
      "loss": 0.9529,
      "step": 11960
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.5312843322753906,
      "learning_rate": 1.2977030955499475e-05,
      "loss": 0.7426,
      "step": 11961
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.235902309417725,
      "learning_rate": 1.2976392039101683e-05,
      "loss": 0.7917,
      "step": 11962
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6159626245498657,
      "learning_rate": 1.2975753122703897e-05,
      "loss": 0.9015,
      "step": 11963
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.745251178741455,
      "learning_rate": 1.2975114206306105e-05,
      "loss": 0.5095,
      "step": 11964
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5758163928985596,
      "learning_rate": 1.2974475289908316e-05,
      "loss": 0.9325,
      "step": 11965
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6901397705078125,
      "learning_rate": 1.2973836373510526e-05,
      "loss": 1.0097,
      "step": 11966
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4613621234893799,
      "learning_rate": 1.2973197457112738e-05,
      "loss": 0.8976,
      "step": 11967
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.011033058166504,
      "learning_rate": 1.2972558540714948e-05,
      "loss": 0.847,
      "step": 11968
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.407106637954712,
      "learning_rate": 1.2971919624317159e-05,
      "loss": 0.8111,
      "step": 11969
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.028012752532959,
      "learning_rate": 1.2971280707919369e-05,
      "loss": 0.9028,
      "step": 11970
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4626879692077637,
      "learning_rate": 1.297064179152158e-05,
      "loss": 0.8958,
      "step": 11971
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8485368490219116,
      "learning_rate": 1.297000287512379e-05,
      "loss": 0.874,
      "step": 11972
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.63603401184082,
      "learning_rate": 1.2969363958726002e-05,
      "loss": 1.5744,
      "step": 11973
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.972250938415527,
      "learning_rate": 1.2968725042328212e-05,
      "loss": 0.8236,
      "step": 11974
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8129116296768188,
      "learning_rate": 1.2968086125930423e-05,
      "loss": 0.8629,
      "step": 11975
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.5260050296783447,
      "learning_rate": 1.2967447209532633e-05,
      "loss": 1.2961,
      "step": 11976
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.918577671051025,
      "learning_rate": 1.2966808293134845e-05,
      "loss": 1.6075,
      "step": 11977
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.887810707092285,
      "learning_rate": 1.2966169376737055e-05,
      "loss": 1.1867,
      "step": 11978
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9626684188842773,
      "learning_rate": 1.2965530460339266e-05,
      "loss": 0.5188,
      "step": 11979
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3059990406036377,
      "learning_rate": 1.2964891543941476e-05,
      "loss": 1.2305,
      "step": 11980
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0065665245056152,
      "learning_rate": 1.2964252627543687e-05,
      "loss": 0.9158,
      "step": 11981
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9105397462844849,
      "learning_rate": 1.2963613711145897e-05,
      "loss": 0.9772,
      "step": 11982
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.088238477706909,
      "learning_rate": 1.2962974794748109e-05,
      "loss": 0.9025,
      "step": 11983
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.846367835998535,
      "learning_rate": 1.2962335878350319e-05,
      "loss": 0.8116,
      "step": 11984
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5199404954910278,
      "learning_rate": 1.296169696195253e-05,
      "loss": 0.418,
      "step": 11985
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.722244143486023,
      "learning_rate": 1.296105804555474e-05,
      "loss": 0.8529,
      "step": 11986
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.559225082397461,
      "learning_rate": 1.2960419129156952e-05,
      "loss": 0.8618,
      "step": 11987
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.931257724761963,
      "learning_rate": 1.2959780212759162e-05,
      "loss": 0.7956,
      "step": 11988
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3858704566955566,
      "learning_rate": 1.2959141296361373e-05,
      "loss": 0.9598,
      "step": 11989
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5266170501708984,
      "learning_rate": 1.2958502379963583e-05,
      "loss": 0.9008,
      "step": 11990
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.188183546066284,
      "learning_rate": 1.2957863463565794e-05,
      "loss": 0.4951,
      "step": 11991
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5124469995498657,
      "learning_rate": 1.2957224547168004e-05,
      "loss": 0.8863,
      "step": 11992
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4213480949401855,
      "learning_rate": 1.2956585630770216e-05,
      "loss": 0.7905,
      "step": 11993
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0029821395874023,
      "learning_rate": 1.2955946714372426e-05,
      "loss": 1.1514,
      "step": 11994
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6507090330123901,
      "learning_rate": 1.2955307797974637e-05,
      "loss": 0.9524,
      "step": 11995
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5490267276763916,
      "learning_rate": 1.2954668881576845e-05,
      "loss": 0.9085,
      "step": 11996
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.585325837135315,
      "learning_rate": 1.2954029965179059e-05,
      "loss": 0.9349,
      "step": 11997
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9004555940628052,
      "learning_rate": 1.2953391048781267e-05,
      "loss": 1.0173,
      "step": 11998
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1532070636749268,
      "learning_rate": 1.295275213238348e-05,
      "loss": 0.8322,
      "step": 11999
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.122051954269409,
      "learning_rate": 1.2952113215985688e-05,
      "loss": 1.0151,
      "step": 12000
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8689634799957275,
      "learning_rate": 1.29514742995879e-05,
      "loss": 0.9685,
      "step": 12001
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5807931423187256,
      "learning_rate": 1.295083538319011e-05,
      "loss": 0.9871,
      "step": 12002
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.843798279762268,
      "learning_rate": 1.2950196466792321e-05,
      "loss": 0.8226,
      "step": 12003
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9298830032348633,
      "learning_rate": 1.2949557550394531e-05,
      "loss": 0.8973,
      "step": 12004
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7183669805526733,
      "learning_rate": 1.2948918633996743e-05,
      "loss": 0.8988,
      "step": 12005
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7255682945251465,
      "learning_rate": 1.2948279717598952e-05,
      "loss": 0.9015,
      "step": 12006
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0383827686309814,
      "learning_rate": 1.2947640801201164e-05,
      "loss": 0.7314,
      "step": 12007
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.608938455581665,
      "learning_rate": 1.2947001884803374e-05,
      "loss": 0.9885,
      "step": 12008
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1732075214385986,
      "learning_rate": 1.2946362968405585e-05,
      "loss": 1.0986,
      "step": 12009
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0530612468719482,
      "learning_rate": 1.2945724052007795e-05,
      "loss": 0.9677,
      "step": 12010
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.027953624725342,
      "learning_rate": 1.2945085135610007e-05,
      "loss": 0.9328,
      "step": 12011
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8135401010513306,
      "learning_rate": 1.2944446219212217e-05,
      "loss": 0.9214,
      "step": 12012
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.080538272857666,
      "learning_rate": 1.2943807302814428e-05,
      "loss": 0.7797,
      "step": 12013
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5565611124038696,
      "learning_rate": 1.2943168386416638e-05,
      "loss": 0.7975,
      "step": 12014
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9911458492279053,
      "learning_rate": 1.294252947001885e-05,
      "loss": 0.5954,
      "step": 12015
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.619872808456421,
      "learning_rate": 1.294189055362106e-05,
      "loss": 0.9068,
      "step": 12016
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3189806938171387,
      "learning_rate": 1.2941251637223271e-05,
      "loss": 0.9599,
      "step": 12017
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4771416187286377,
      "learning_rate": 1.294061272082548e-05,
      "loss": 0.8551,
      "step": 12018
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.945656657218933,
      "learning_rate": 1.2939973804427692e-05,
      "loss": 0.962,
      "step": 12019
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6183284521102905,
      "learning_rate": 1.2939334888029902e-05,
      "loss": 0.9587,
      "step": 12020
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1483631134033203,
      "learning_rate": 1.2938695971632114e-05,
      "loss": 0.9502,
      "step": 12021
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8012046813964844,
      "learning_rate": 1.2938057055234324e-05,
      "loss": 1.1527,
      "step": 12022
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0281002521514893,
      "learning_rate": 1.2937418138836535e-05,
      "loss": 0.8195,
      "step": 12023
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6340221166610718,
      "learning_rate": 1.2936779222438745e-05,
      "loss": 0.9059,
      "step": 12024
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0372443199157715,
      "learning_rate": 1.2936140306040957e-05,
      "loss": 0.9294,
      "step": 12025
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0869736671447754,
      "learning_rate": 1.2935501389643166e-05,
      "loss": 0.9114,
      "step": 12026
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2470250129699707,
      "learning_rate": 1.2934862473245378e-05,
      "loss": 0.9603,
      "step": 12027
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.737924575805664,
      "learning_rate": 1.2934223556847588e-05,
      "loss": 1.0655,
      "step": 12028
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8449105024337769,
      "learning_rate": 1.29335846404498e-05,
      "loss": 1.2665,
      "step": 12029
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0085270404815674,
      "learning_rate": 1.293294572405201e-05,
      "loss": 1.1326,
      "step": 12030
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8910787105560303,
      "learning_rate": 1.293230680765422e-05,
      "loss": 0.9728,
      "step": 12031
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.955151557922363,
      "learning_rate": 1.2931667891256429e-05,
      "loss": 1.3248,
      "step": 12032
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3461480140686035,
      "learning_rate": 1.2931028974858642e-05,
      "loss": 0.7788,
      "step": 12033
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.891310214996338,
      "learning_rate": 1.293039005846085e-05,
      "loss": 1.6766,
      "step": 12034
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.202017307281494,
      "learning_rate": 1.2929751142063064e-05,
      "loss": 0.9853,
      "step": 12035
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.864564299583435,
      "learning_rate": 1.2929112225665272e-05,
      "loss": 0.9189,
      "step": 12036
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7348312139511108,
      "learning_rate": 1.2928473309267483e-05,
      "loss": 0.8924,
      "step": 12037
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8926960229873657,
      "learning_rate": 1.2927834392869693e-05,
      "loss": 0.9146,
      "step": 12038
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.807793140411377,
      "learning_rate": 1.2927195476471905e-05,
      "loss": 1.3238,
      "step": 12039
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.1695668697357178,
      "learning_rate": 1.2926556560074115e-05,
      "loss": 0.5994,
      "step": 12040
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0118253231048584,
      "learning_rate": 1.2925917643676326e-05,
      "loss": 0.9005,
      "step": 12041
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.880375862121582,
      "learning_rate": 1.2925278727278536e-05,
      "loss": 0.9407,
      "step": 12042
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.4470949172973633,
      "learning_rate": 1.2924639810880747e-05,
      "loss": 1.0466,
      "step": 12043
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9685752391815186,
      "learning_rate": 1.2924000894482957e-05,
      "loss": 0.9404,
      "step": 12044
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.846402883529663,
      "learning_rate": 1.2923361978085169e-05,
      "loss": 0.9163,
      "step": 12045
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.353781223297119,
      "learning_rate": 1.2922723061687379e-05,
      "loss": 0.847,
      "step": 12046
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.2490193843841553,
      "learning_rate": 1.292208414528959e-05,
      "loss": 1.1137,
      "step": 12047
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.659864902496338,
      "learning_rate": 1.29214452288918e-05,
      "loss": 1.0518,
      "step": 12048
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8332977294921875,
      "learning_rate": 1.2920806312494012e-05,
      "loss": 1.1202,
      "step": 12049
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8342092037200928,
      "learning_rate": 1.2920167396096222e-05,
      "loss": 0.8476,
      "step": 12050
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.000527858734131,
      "learning_rate": 1.2919528479698433e-05,
      "loss": 0.8579,
      "step": 12051
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6935476064682007,
      "learning_rate": 1.2918889563300643e-05,
      "loss": 1.1126,
      "step": 12052
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.274216890335083,
      "learning_rate": 1.2918250646902854e-05,
      "loss": 0.8621,
      "step": 12053
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.6342415809631348,
      "learning_rate": 1.2917611730505064e-05,
      "loss": 1.1309,
      "step": 12054
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5874230861663818,
      "learning_rate": 1.2916972814107276e-05,
      "loss": 0.7566,
      "step": 12055
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.540462851524353,
      "learning_rate": 1.2916333897709486e-05,
      "loss": 0.4843,
      "step": 12056
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.592586040496826,
      "learning_rate": 1.2915694981311697e-05,
      "loss": 1.2546,
      "step": 12057
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0127294063568115,
      "learning_rate": 1.2915056064913907e-05,
      "loss": 0.9953,
      "step": 12058
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.14174222946167,
      "learning_rate": 1.2914417148516119e-05,
      "loss": 1.0377,
      "step": 12059
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.5007548332214355,
      "learning_rate": 1.2913778232118329e-05,
      "loss": 0.9591,
      "step": 12060
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0292470455169678,
      "learning_rate": 1.291313931572054e-05,
      "loss": 1.0,
      "step": 12061
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3701932430267334,
      "learning_rate": 1.291250039932275e-05,
      "loss": 0.9136,
      "step": 12062
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.663713812828064,
      "learning_rate": 1.2911861482924961e-05,
      "loss": 0.9335,
      "step": 12063
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7412891387939453,
      "learning_rate": 1.2911222566527171e-05,
      "loss": 0.829,
      "step": 12064
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.796931028366089,
      "learning_rate": 1.2910583650129383e-05,
      "loss": 1.2265,
      "step": 12065
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4379544258117676,
      "learning_rate": 1.2909944733731593e-05,
      "loss": 0.7391,
      "step": 12066
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.728380560874939,
      "learning_rate": 1.2909305817333804e-05,
      "loss": 0.8774,
      "step": 12067
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.990215539932251,
      "learning_rate": 1.2908666900936012e-05,
      "loss": 0.895,
      "step": 12068
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.4248762130737305,
      "learning_rate": 1.2908027984538226e-05,
      "loss": 1.25,
      "step": 12069
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7116107940673828,
      "learning_rate": 1.2907389068140434e-05,
      "loss": 0.4756,
      "step": 12070
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9928569793701172,
      "learning_rate": 1.2906750151742647e-05,
      "loss": 0.9025,
      "step": 12071
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.85136079788208,
      "learning_rate": 1.2906111235344855e-05,
      "loss": 0.9387,
      "step": 12072
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.540343999862671,
      "learning_rate": 1.2905472318947068e-05,
      "loss": 0.6202,
      "step": 12073
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9228758811950684,
      "learning_rate": 1.2904833402549277e-05,
      "loss": 0.8668,
      "step": 12074
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9092942476272583,
      "learning_rate": 1.2904194486151488e-05,
      "loss": 0.8859,
      "step": 12075
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1926255226135254,
      "learning_rate": 1.2903555569753698e-05,
      "loss": 0.8475,
      "step": 12076
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9215307235717773,
      "learning_rate": 1.290291665335591e-05,
      "loss": 0.872,
      "step": 12077
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4972318410873413,
      "learning_rate": 1.290227773695812e-05,
      "loss": 0.4174,
      "step": 12078
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6356418132781982,
      "learning_rate": 1.2901638820560331e-05,
      "loss": 0.8528,
      "step": 12079
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6828937530517578,
      "learning_rate": 1.2900999904162541e-05,
      "loss": 0.9599,
      "step": 12080
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.6008732318878174,
      "learning_rate": 1.2900360987764752e-05,
      "loss": 1.1061,
      "step": 12081
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.411982536315918,
      "learning_rate": 1.2899722071366962e-05,
      "loss": 0.7488,
      "step": 12082
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.995482087135315,
      "learning_rate": 1.2899083154969174e-05,
      "loss": 0.8962,
      "step": 12083
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4593979120254517,
      "learning_rate": 1.2898444238571384e-05,
      "loss": 0.7586,
      "step": 12084
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6140451431274414,
      "learning_rate": 1.2897805322173595e-05,
      "loss": 0.4341,
      "step": 12085
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.011963129043579,
      "learning_rate": 1.2897166405775805e-05,
      "loss": 0.9688,
      "step": 12086
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.234196662902832,
      "learning_rate": 1.2896527489378017e-05,
      "loss": 1.0184,
      "step": 12087
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.058353900909424,
      "learning_rate": 1.2895888572980226e-05,
      "loss": 1.0692,
      "step": 12088
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9077701568603516,
      "learning_rate": 1.2895249656582438e-05,
      "loss": 0.9819,
      "step": 12089
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.452291488647461,
      "learning_rate": 1.2894610740184648e-05,
      "loss": 0.8697,
      "step": 12090
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.104031562805176,
      "learning_rate": 1.289397182378686e-05,
      "loss": 1.0035,
      "step": 12091
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.879932165145874,
      "learning_rate": 1.289333290738907e-05,
      "loss": 0.9902,
      "step": 12092
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4324994087219238,
      "learning_rate": 1.289269399099128e-05,
      "loss": 0.7345,
      "step": 12093
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6618374586105347,
      "learning_rate": 1.289205507459349e-05,
      "loss": 0.8882,
      "step": 12094
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.7324135303497314,
      "learning_rate": 1.2891416158195702e-05,
      "loss": 1.027,
      "step": 12095
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.622244358062744,
      "learning_rate": 1.2890777241797912e-05,
      "loss": 1.4385,
      "step": 12096
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.6110188961029053,
      "learning_rate": 1.2890138325400124e-05,
      "loss": 1.3715,
      "step": 12097
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.603397250175476,
      "learning_rate": 1.2889499409002333e-05,
      "loss": 0.9348,
      "step": 12098
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0359506607055664,
      "learning_rate": 1.2888860492604545e-05,
      "loss": 0.8891,
      "step": 12099
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.616774320602417,
      "learning_rate": 1.2888221576206755e-05,
      "loss": 0.8971,
      "step": 12100
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6884629726409912,
      "learning_rate": 1.2887582659808966e-05,
      "loss": 0.8462,
      "step": 12101
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.923370361328125,
      "learning_rate": 1.2886943743411176e-05,
      "loss": 0.9081,
      "step": 12102
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.586700677871704,
      "learning_rate": 1.2886304827013388e-05,
      "loss": 0.9552,
      "step": 12103
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8975744247436523,
      "learning_rate": 1.2885665910615596e-05,
      "loss": 1.3091,
      "step": 12104
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9152921438217163,
      "learning_rate": 1.288502699421781e-05,
      "loss": 1.0066,
      "step": 12105
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.6171867847442627,
      "learning_rate": 1.2884388077820017e-05,
      "loss": 0.8785,
      "step": 12106
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.14458966255188,
      "learning_rate": 1.288374916142223e-05,
      "loss": 0.953,
      "step": 12107
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6266636848449707,
      "learning_rate": 1.2883110245024439e-05,
      "loss": 0.794,
      "step": 12108
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8833792209625244,
      "learning_rate": 1.2882471328626652e-05,
      "loss": 1.145,
      "step": 12109
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3198275566101074,
      "learning_rate": 1.288183241222886e-05,
      "loss": 1.1204,
      "step": 12110
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.447911024093628,
      "learning_rate": 1.2881193495831072e-05,
      "loss": 1.2349,
      "step": 12111
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7788499593734741,
      "learning_rate": 1.2880554579433282e-05,
      "loss": 0.9028,
      "step": 12112
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.0053486824035645,
      "learning_rate": 1.2879915663035493e-05,
      "loss": 0.8297,
      "step": 12113
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.9785969257354736,
      "learning_rate": 1.2879276746637703e-05,
      "loss": 0.9517,
      "step": 12114
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8685939311981201,
      "learning_rate": 1.2878637830239914e-05,
      "loss": 1.053,
      "step": 12115
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.245591402053833,
      "learning_rate": 1.2877998913842124e-05,
      "loss": 0.7944,
      "step": 12116
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1080548763275146,
      "learning_rate": 1.2877359997444336e-05,
      "loss": 0.5826,
      "step": 12117
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9294131994247437,
      "learning_rate": 1.2876721081046546e-05,
      "loss": 0.9932,
      "step": 12118
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6417473554611206,
      "learning_rate": 1.2876082164648757e-05,
      "loss": 0.9201,
      "step": 12119
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.460930347442627,
      "learning_rate": 1.2875443248250967e-05,
      "loss": 0.8581,
      "step": 12120
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1456613540649414,
      "learning_rate": 1.2874804331853179e-05,
      "loss": 0.9557,
      "step": 12121
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.510359287261963,
      "learning_rate": 1.2874165415455389e-05,
      "loss": 1.1035,
      "step": 12122
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.3492181301116943,
      "learning_rate": 1.28735264990576e-05,
      "loss": 1.0084,
      "step": 12123
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.926077127456665,
      "learning_rate": 1.287288758265981e-05,
      "loss": 0.5155,
      "step": 12124
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7234315872192383,
      "learning_rate": 1.2872248666262021e-05,
      "loss": 0.9388,
      "step": 12125
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8198014497756958,
      "learning_rate": 1.2871609749864231e-05,
      "loss": 0.9984,
      "step": 12126
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.51127028465271,
      "learning_rate": 1.2870970833466443e-05,
      "loss": 0.9272,
      "step": 12127
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5538331270217896,
      "learning_rate": 1.2870331917068653e-05,
      "loss": 0.8401,
      "step": 12128
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.3015332221984863,
      "learning_rate": 1.2869693000670864e-05,
      "loss": 1.0799,
      "step": 12129
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7179207801818848,
      "learning_rate": 1.2869054084273074e-05,
      "loss": 0.887,
      "step": 12130
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1528093814849854,
      "learning_rate": 1.2868415167875286e-05,
      "loss": 0.9434,
      "step": 12131
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7772434949874878,
      "learning_rate": 1.2867776251477496e-05,
      "loss": 1.0821,
      "step": 12132
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1215319633483887,
      "learning_rate": 1.2867137335079707e-05,
      "loss": 0.7786,
      "step": 12133
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.252045154571533,
      "learning_rate": 1.2866498418681917e-05,
      "loss": 1.0045,
      "step": 12134
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5261478424072266,
      "learning_rate": 1.2865859502284128e-05,
      "loss": 1.2482,
      "step": 12135
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.439889430999756,
      "learning_rate": 1.2865220585886338e-05,
      "loss": 1.2035,
      "step": 12136
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.313833236694336,
      "learning_rate": 1.286458166948855e-05,
      "loss": 1.5326,
      "step": 12137
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6643353700637817,
      "learning_rate": 1.286394275309076e-05,
      "loss": 0.9632,
      "step": 12138
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7454034090042114,
      "learning_rate": 1.2863303836692971e-05,
      "loss": 0.9791,
      "step": 12139
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9640792608261108,
      "learning_rate": 1.2862664920295181e-05,
      "loss": 0.8464,
      "step": 12140
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1724302768707275,
      "learning_rate": 1.2862026003897393e-05,
      "loss": 0.9963,
      "step": 12141
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.64204740524292,
      "learning_rate": 1.2861387087499601e-05,
      "loss": 0.812,
      "step": 12142
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2168309688568115,
      "learning_rate": 1.2860748171101814e-05,
      "loss": 0.9239,
      "step": 12143
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.010765314102173,
      "learning_rate": 1.2860109254704022e-05,
      "loss": 0.9322,
      "step": 12144
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.2149407863616943,
      "learning_rate": 1.2859470338306235e-05,
      "loss": 0.8366,
      "step": 12145
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.783332109451294,
      "learning_rate": 1.2858831421908444e-05,
      "loss": 0.9696,
      "step": 12146
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.204005241394043,
      "learning_rate": 1.2858192505510655e-05,
      "loss": 0.731,
      "step": 12147
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2413864135742188,
      "learning_rate": 1.2857553589112865e-05,
      "loss": 1.082,
      "step": 12148
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.770148277282715,
      "learning_rate": 1.2856914672715077e-05,
      "loss": 1.0929,
      "step": 12149
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1409759521484375,
      "learning_rate": 1.2856275756317286e-05,
      "loss": 1.094,
      "step": 12150
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.014094114303589,
      "learning_rate": 1.2855636839919498e-05,
      "loss": 0.8825,
      "step": 12151
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1144535541534424,
      "learning_rate": 1.2854997923521708e-05,
      "loss": 0.8507,
      "step": 12152
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.479804515838623,
      "learning_rate": 1.285435900712392e-05,
      "loss": 0.7402,
      "step": 12153
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8037774562835693,
      "learning_rate": 1.285372009072613e-05,
      "loss": 0.9223,
      "step": 12154
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5654113292694092,
      "learning_rate": 1.285308117432834e-05,
      "loss": 0.7308,
      "step": 12155
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5853617191314697,
      "learning_rate": 1.285244225793055e-05,
      "loss": 0.394,
      "step": 12156
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.489802360534668,
      "learning_rate": 1.2851803341532762e-05,
      "loss": 0.7715,
      "step": 12157
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.0415117740631104,
      "learning_rate": 1.2851164425134972e-05,
      "loss": 1.1301,
      "step": 12158
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9031882286071777,
      "learning_rate": 1.2850525508737184e-05,
      "loss": 1.0218,
      "step": 12159
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9446840286254883,
      "learning_rate": 1.2849886592339393e-05,
      "loss": 0.9307,
      "step": 12160
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5860258340835571,
      "learning_rate": 1.2849247675941605e-05,
      "loss": 0.9272,
      "step": 12161
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.485558032989502,
      "learning_rate": 1.2848608759543815e-05,
      "loss": 0.8889,
      "step": 12162
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7294458150863647,
      "learning_rate": 1.2847969843146026e-05,
      "loss": 0.8647,
      "step": 12163
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7615145444869995,
      "learning_rate": 1.2847330926748236e-05,
      "loss": 1.13,
      "step": 12164
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.139611005783081,
      "learning_rate": 1.2846692010350448e-05,
      "loss": 0.9636,
      "step": 12165
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8223087787628174,
      "learning_rate": 1.2846053093952658e-05,
      "loss": 0.926,
      "step": 12166
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.188450336456299,
      "learning_rate": 1.284541417755487e-05,
      "loss": 0.8883,
      "step": 12167
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3586654663085938,
      "learning_rate": 1.2844775261157079e-05,
      "loss": 0.9453,
      "step": 12168
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0982794761657715,
      "learning_rate": 1.284413634475929e-05,
      "loss": 1.0356,
      "step": 12169
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7018606662750244,
      "learning_rate": 1.28434974283615e-05,
      "loss": 1.0272,
      "step": 12170
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9375194311141968,
      "learning_rate": 1.2842858511963712e-05,
      "loss": 0.9028,
      "step": 12171
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7274525165557861,
      "learning_rate": 1.2842219595565922e-05,
      "loss": 1.0342,
      "step": 12172
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.662451982498169,
      "learning_rate": 1.2841580679168133e-05,
      "loss": 0.9068,
      "step": 12173
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0056934356689453,
      "learning_rate": 1.2840941762770343e-05,
      "loss": 1.1504,
      "step": 12174
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6774375438690186,
      "learning_rate": 1.2840302846372555e-05,
      "loss": 0.972,
      "step": 12175
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8379498720169067,
      "learning_rate": 1.2839663929974765e-05,
      "loss": 0.8913,
      "step": 12176
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.848231554031372,
      "learning_rate": 1.2839025013576973e-05,
      "loss": 0.9037,
      "step": 12177
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7921054363250732,
      "learning_rate": 1.2838386097179184e-05,
      "loss": 0.9573,
      "step": 12178
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5159128904342651,
      "learning_rate": 1.2837747180781394e-05,
      "loss": 0.8526,
      "step": 12179
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.020064115524292,
      "learning_rate": 1.2837108264383606e-05,
      "loss": 0.9208,
      "step": 12180
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7612663507461548,
      "learning_rate": 1.2836469347985816e-05,
      "loss": 0.9429,
      "step": 12181
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4225578308105469,
      "learning_rate": 1.2835830431588027e-05,
      "loss": 0.8508,
      "step": 12182
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5637701749801636,
      "learning_rate": 1.2835191515190237e-05,
      "loss": 0.954,
      "step": 12183
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4996867179870605,
      "learning_rate": 1.2834552598792449e-05,
      "loss": 1.0024,
      "step": 12184
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.2763054370880127,
      "learning_rate": 1.2833913682394658e-05,
      "loss": 1.108,
      "step": 12185
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.558328628540039,
      "learning_rate": 1.283327476599687e-05,
      "loss": 0.7538,
      "step": 12186
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.657880187034607,
      "learning_rate": 1.283263584959908e-05,
      "loss": 0.9571,
      "step": 12187
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.792616367340088,
      "learning_rate": 1.2831996933201291e-05,
      "loss": 0.9423,
      "step": 12188
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1120829582214355,
      "learning_rate": 1.2831358016803501e-05,
      "loss": 1.1561,
      "step": 12189
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.8086845874786377,
      "learning_rate": 1.2830719100405713e-05,
      "loss": 1.1149,
      "step": 12190
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8527846336364746,
      "learning_rate": 1.2830080184007923e-05,
      "loss": 1.2615,
      "step": 12191
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.085754156112671,
      "learning_rate": 1.2829441267610134e-05,
      "loss": 1.0567,
      "step": 12192
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8978508710861206,
      "learning_rate": 1.2828802351212344e-05,
      "loss": 0.9396,
      "step": 12193
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8601642847061157,
      "learning_rate": 1.2828163434814556e-05,
      "loss": 0.8984,
      "step": 12194
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9580448865890503,
      "learning_rate": 1.2827524518416765e-05,
      "loss": 0.9766,
      "step": 12195
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0012199878692627,
      "learning_rate": 1.2826885602018977e-05,
      "loss": 0.9287,
      "step": 12196
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8945646286010742,
      "learning_rate": 1.2826246685621187e-05,
      "loss": 0.931,
      "step": 12197
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.381449222564697,
      "learning_rate": 1.2825607769223398e-05,
      "loss": 1.4285,
      "step": 12198
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.2334132194519043,
      "learning_rate": 1.2824968852825608e-05,
      "loss": 0.705,
      "step": 12199
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6046361923217773,
      "learning_rate": 1.282432993642782e-05,
      "loss": 1.0024,
      "step": 12200
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9149712324142456,
      "learning_rate": 1.282369102003003e-05,
      "loss": 1.124,
      "step": 12201
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8027018308639526,
      "learning_rate": 1.2823052103632241e-05,
      "loss": 0.8511,
      "step": 12202
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6573498249053955,
      "learning_rate": 1.2822413187234451e-05,
      "loss": 0.9268,
      "step": 12203
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.685242772102356,
      "learning_rate": 1.2821774270836663e-05,
      "loss": 0.9409,
      "step": 12204
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6081942319869995,
      "learning_rate": 1.2821135354438872e-05,
      "loss": 0.9302,
      "step": 12205
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5078456401824951,
      "learning_rate": 1.2820496438041084e-05,
      "loss": 0.894,
      "step": 12206
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3875300884246826,
      "learning_rate": 1.2819857521643294e-05,
      "loss": 0.8216,
      "step": 12207
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.814551591873169,
      "learning_rate": 1.2819218605245505e-05,
      "loss": 0.6974,
      "step": 12208
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.8071863651275635,
      "learning_rate": 1.2818579688847714e-05,
      "loss": 0.9232,
      "step": 12209
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0384559631347656,
      "learning_rate": 1.2817940772449927e-05,
      "loss": 0.9436,
      "step": 12210
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.613912582397461,
      "learning_rate": 1.2817301856052135e-05,
      "loss": 0.9224,
      "step": 12211
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5880967378616333,
      "learning_rate": 1.2816662939654348e-05,
      "loss": 0.9067,
      "step": 12212
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9569289684295654,
      "learning_rate": 1.2816024023256556e-05,
      "loss": 0.9782,
      "step": 12213
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1107494831085205,
      "learning_rate": 1.2815385106858768e-05,
      "loss": 0.9114,
      "step": 12214
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.7102386951446533,
      "learning_rate": 1.2814746190460978e-05,
      "loss": 0.8453,
      "step": 12215
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.70437753200531,
      "learning_rate": 1.281410727406319e-05,
      "loss": 0.9119,
      "step": 12216
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6446655988693237,
      "learning_rate": 1.2813468357665399e-05,
      "loss": 0.8493,
      "step": 12217
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9626680612564087,
      "learning_rate": 1.281282944126761e-05,
      "loss": 0.9118,
      "step": 12218
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.109064817428589,
      "learning_rate": 1.281219052486982e-05,
      "loss": 1.3065,
      "step": 12219
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8925645351409912,
      "learning_rate": 1.2811551608472032e-05,
      "loss": 1.0158,
      "step": 12220
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0133774280548096,
      "learning_rate": 1.2810912692074242e-05,
      "loss": 0.9188,
      "step": 12221
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4608979225158691,
      "learning_rate": 1.2810273775676453e-05,
      "loss": 0.8455,
      "step": 12222
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5873241424560547,
      "learning_rate": 1.2809634859278663e-05,
      "loss": 0.854,
      "step": 12223
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4376527070999146,
      "learning_rate": 1.2808995942880875e-05,
      "loss": 0.7556,
      "step": 12224
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6403166055679321,
      "learning_rate": 1.2808357026483085e-05,
      "loss": 0.3684,
      "step": 12225
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.223483085632324,
      "learning_rate": 1.2807718110085296e-05,
      "loss": 1.0226,
      "step": 12226
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5060458183288574,
      "learning_rate": 1.2807079193687506e-05,
      "loss": 0.8641,
      "step": 12227
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9857444763183594,
      "learning_rate": 1.2806440277289718e-05,
      "loss": 1.3656,
      "step": 12228
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6399379968643188,
      "learning_rate": 1.2805801360891928e-05,
      "loss": 0.9301,
      "step": 12229
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6352624893188477,
      "learning_rate": 1.2805162444494139e-05,
      "loss": 0.3676,
      "step": 12230
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4879714250564575,
      "learning_rate": 1.2804523528096349e-05,
      "loss": 0.8697,
      "step": 12231
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9186992645263672,
      "learning_rate": 1.280388461169856e-05,
      "loss": 1.0142,
      "step": 12232
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5867615938186646,
      "learning_rate": 1.280324569530077e-05,
      "loss": 0.879,
      "step": 12233
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7288987636566162,
      "learning_rate": 1.2802606778902982e-05,
      "loss": 0.9302,
      "step": 12234
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.479572296142578,
      "learning_rate": 1.2801967862505192e-05,
      "loss": 1.3142,
      "step": 12235
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.607945203781128,
      "learning_rate": 1.2801328946107403e-05,
      "loss": 0.8145,
      "step": 12236
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6284769773483276,
      "learning_rate": 1.2800690029709613e-05,
      "loss": 0.9204,
      "step": 12237
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.629593849182129,
      "learning_rate": 1.2800051113311825e-05,
      "loss": 0.793,
      "step": 12238
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7800909280776978,
      "learning_rate": 1.2799412196914035e-05,
      "loss": 0.8797,
      "step": 12239
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.672151803970337,
      "learning_rate": 1.2798773280516246e-05,
      "loss": 0.8893,
      "step": 12240
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.240436553955078,
      "learning_rate": 1.2798134364118456e-05,
      "loss": 0.9174,
      "step": 12241
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.750404953956604,
      "learning_rate": 1.2797495447720667e-05,
      "loss": 0.9504,
      "step": 12242
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.826821804046631,
      "learning_rate": 1.2796856531322877e-05,
      "loss": 0.942,
      "step": 12243
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.998044013977051,
      "learning_rate": 1.2796217614925089e-05,
      "loss": 0.8284,
      "step": 12244
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.105917453765869,
      "learning_rate": 1.2795578698527297e-05,
      "loss": 0.8985,
      "step": 12245
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0572733879089355,
      "learning_rate": 1.279493978212951e-05,
      "loss": 1.0774,
      "step": 12246
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.647104501724243,
      "learning_rate": 1.2794300865731718e-05,
      "loss": 1.3383,
      "step": 12247
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.6705896854400635,
      "learning_rate": 1.2793661949333932e-05,
      "loss": 1.0943,
      "step": 12248
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8178393840789795,
      "learning_rate": 1.279302303293614e-05,
      "loss": 0.9521,
      "step": 12249
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.744377613067627,
      "learning_rate": 1.2792384116538351e-05,
      "loss": 0.9721,
      "step": 12250
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5941089391708374,
      "learning_rate": 1.2791745200140561e-05,
      "loss": 0.8207,
      "step": 12251
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7649954557418823,
      "learning_rate": 1.2791106283742773e-05,
      "loss": 0.9013,
      "step": 12252
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6215510368347168,
      "learning_rate": 1.2790467367344983e-05,
      "loss": 0.8705,
      "step": 12253
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0551977157592773,
      "learning_rate": 1.2789828450947194e-05,
      "loss": 1.046,
      "step": 12254
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.8521389961242676,
      "learning_rate": 1.2789189534549404e-05,
      "loss": 0.8361,
      "step": 12255
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5126546621322632,
      "learning_rate": 1.2788550618151616e-05,
      "loss": 0.9116,
      "step": 12256
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9286000728607178,
      "learning_rate": 1.2787911701753825e-05,
      "loss": 1.0184,
      "step": 12257
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4129934310913086,
      "learning_rate": 1.2787272785356037e-05,
      "loss": 0.9809,
      "step": 12258
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6642762422561646,
      "learning_rate": 1.2786633868958247e-05,
      "loss": 0.8006,
      "step": 12259
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9939035177230835,
      "learning_rate": 1.2785994952560458e-05,
      "loss": 0.9087,
      "step": 12260
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6617075204849243,
      "learning_rate": 1.2785356036162668e-05,
      "loss": 0.9152,
      "step": 12261
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6497550010681152,
      "learning_rate": 1.278471711976488e-05,
      "loss": 0.9338,
      "step": 12262
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6069220304489136,
      "learning_rate": 1.278407820336709e-05,
      "loss": 0.8752,
      "step": 12263
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6126850843429565,
      "learning_rate": 1.2783439286969301e-05,
      "loss": 0.8834,
      "step": 12264
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.980302333831787,
      "learning_rate": 1.2782800370571511e-05,
      "loss": 0.9437,
      "step": 12265
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6433591842651367,
      "learning_rate": 1.2782161454173723e-05,
      "loss": 0.9085,
      "step": 12266
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.003199577331543,
      "learning_rate": 1.2781522537775932e-05,
      "loss": 1.0915,
      "step": 12267
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.879270315170288,
      "learning_rate": 1.2780883621378144e-05,
      "loss": 0.8986,
      "step": 12268
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0685558319091797,
      "learning_rate": 1.2780244704980354e-05,
      "loss": 1.0218,
      "step": 12269
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0561795234680176,
      "learning_rate": 1.2779605788582565e-05,
      "loss": 0.8768,
      "step": 12270
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.89109206199646,
      "learning_rate": 1.2778966872184775e-05,
      "loss": 0.9813,
      "step": 12271
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7066630125045776,
      "learning_rate": 1.2778327955786987e-05,
      "loss": 0.9861,
      "step": 12272
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.206606149673462,
      "learning_rate": 1.2777689039389197e-05,
      "loss": 1.2558,
      "step": 12273
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.3314337730407715,
      "learning_rate": 1.2777050122991408e-05,
      "loss": 1.3695,
      "step": 12274
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8919041156768799,
      "learning_rate": 1.2776411206593618e-05,
      "loss": 0.9581,
      "step": 12275
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.541405200958252,
      "learning_rate": 1.277577229019583e-05,
      "loss": 0.9672,
      "step": 12276
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5069254636764526,
      "learning_rate": 1.277513337379804e-05,
      "loss": 0.8721,
      "step": 12277
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.88157320022583,
      "learning_rate": 1.2774494457400251e-05,
      "loss": 0.9055,
      "step": 12278
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6347233057022095,
      "learning_rate": 1.277385554100246e-05,
      "loss": 1.0109,
      "step": 12279
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6078320741653442,
      "learning_rate": 1.2773216624604672e-05,
      "loss": 0.9367,
      "step": 12280
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.915609359741211,
      "learning_rate": 1.277257770820688e-05,
      "loss": 1.3087,
      "step": 12281
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.173398017883301,
      "learning_rate": 1.2771938791809094e-05,
      "loss": 1.055,
      "step": 12282
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8513641357421875,
      "learning_rate": 1.2771299875411302e-05,
      "loss": 1.0605,
      "step": 12283
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1675925254821777,
      "learning_rate": 1.2770660959013515e-05,
      "loss": 1.0145,
      "step": 12284
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3784444332122803,
      "learning_rate": 1.2770022042615723e-05,
      "loss": 0.8766,
      "step": 12285
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6053428649902344,
      "learning_rate": 1.2769383126217937e-05,
      "loss": 0.9416,
      "step": 12286
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9123793840408325,
      "learning_rate": 1.2768744209820145e-05,
      "loss": 0.8309,
      "step": 12287
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3070156574249268,
      "learning_rate": 1.2768105293422356e-05,
      "loss": 0.9533,
      "step": 12288
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.6988723278045654,
      "learning_rate": 1.2767466377024566e-05,
      "loss": 1.1213,
      "step": 12289
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.722232699394226,
      "learning_rate": 1.2766827460626778e-05,
      "loss": 0.9778,
      "step": 12290
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.514192819595337,
      "learning_rate": 1.2766188544228988e-05,
      "loss": 0.801,
      "step": 12291
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7782210111618042,
      "learning_rate": 1.2765549627831199e-05,
      "loss": 0.9703,
      "step": 12292
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3459455966949463,
      "learning_rate": 1.2764910711433409e-05,
      "loss": 1.0813,
      "step": 12293
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8560127019882202,
      "learning_rate": 1.276427179503562e-05,
      "loss": 1.1386,
      "step": 12294
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4101885557174683,
      "learning_rate": 1.276363287863783e-05,
      "loss": 0.3414,
      "step": 12295
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1491761207580566,
      "learning_rate": 1.2762993962240042e-05,
      "loss": 0.9332,
      "step": 12296
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4471092224121094,
      "learning_rate": 1.2762355045842252e-05,
      "loss": 0.3922,
      "step": 12297
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.144716739654541,
      "learning_rate": 1.2761716129444463e-05,
      "loss": 1.152,
      "step": 12298
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.03854513168335,
      "learning_rate": 1.2761077213046673e-05,
      "loss": 1.6829,
      "step": 12299
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0884289741516113,
      "learning_rate": 1.2760438296648885e-05,
      "loss": 0.9855,
      "step": 12300
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5932214260101318,
      "learning_rate": 1.2759799380251095e-05,
      "loss": 0.8197,
      "step": 12301
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.297097682952881,
      "learning_rate": 1.2759160463853306e-05,
      "loss": 1.1262,
      "step": 12302
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7526228427886963,
      "learning_rate": 1.2758521547455516e-05,
      "loss": 0.8018,
      "step": 12303
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7469886541366577,
      "learning_rate": 1.2757882631057727e-05,
      "loss": 0.9633,
      "step": 12304
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7966864109039307,
      "learning_rate": 1.2757243714659937e-05,
      "loss": 0.8425,
      "step": 12305
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.788142681121826,
      "learning_rate": 1.2756604798262149e-05,
      "loss": 1.0566,
      "step": 12306
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7290116548538208,
      "learning_rate": 1.2755965881864359e-05,
      "loss": 0.8997,
      "step": 12307
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8190046548843384,
      "learning_rate": 1.275532696546657e-05,
      "loss": 0.8306,
      "step": 12308
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6608973741531372,
      "learning_rate": 1.275468804906878e-05,
      "loss": 0.8811,
      "step": 12309
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7382571697235107,
      "learning_rate": 1.2754049132670992e-05,
      "loss": 0.92,
      "step": 12310
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.892630577087402,
      "learning_rate": 1.2753410216273202e-05,
      "loss": 1.2418,
      "step": 12311
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.361379623413086,
      "learning_rate": 1.2752771299875413e-05,
      "loss": 0.4523,
      "step": 12312
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.15789532661438,
      "learning_rate": 1.2752132383477623e-05,
      "loss": 0.6137,
      "step": 12313
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.857919692993164,
      "learning_rate": 1.2751493467079834e-05,
      "loss": 1.0621,
      "step": 12314
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.692958116531372,
      "learning_rate": 1.2750854550682044e-05,
      "loss": 0.7611,
      "step": 12315
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.80396568775177,
      "learning_rate": 1.2750215634284256e-05,
      "loss": 0.8731,
      "step": 12316
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0401461124420166,
      "learning_rate": 1.2749576717886466e-05,
      "loss": 0.9678,
      "step": 12317
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.896329641342163,
      "learning_rate": 1.2748937801488677e-05,
      "loss": 1.1339,
      "step": 12318
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.492574453353882,
      "learning_rate": 1.2748298885090885e-05,
      "loss": 0.9019,
      "step": 12319
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6512852907180786,
      "learning_rate": 1.2747659968693099e-05,
      "loss": 0.7433,
      "step": 12320
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.219367742538452,
      "learning_rate": 1.2747021052295307e-05,
      "loss": 0.9913,
      "step": 12321
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9012823104858398,
      "learning_rate": 1.274638213589752e-05,
      "loss": 1.0752,
      "step": 12322
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.196340084075928,
      "learning_rate": 1.2745743219499728e-05,
      "loss": 1.373,
      "step": 12323
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9352909326553345,
      "learning_rate": 1.274510430310194e-05,
      "loss": 0.9559,
      "step": 12324
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.79793119430542,
      "learning_rate": 1.274446538670415e-05,
      "loss": 0.8045,
      "step": 12325
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.164776086807251,
      "learning_rate": 1.2743826470306361e-05,
      "loss": 0.8068,
      "step": 12326
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9698179960250854,
      "learning_rate": 1.2743187553908571e-05,
      "loss": 0.4296,
      "step": 12327
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4170145988464355,
      "learning_rate": 1.2742548637510783e-05,
      "loss": 0.8794,
      "step": 12328
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.823772668838501,
      "learning_rate": 1.2741909721112992e-05,
      "loss": 0.9597,
      "step": 12329
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.0788068771362305,
      "learning_rate": 1.2741270804715204e-05,
      "loss": 1.2784,
      "step": 12330
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1055147647857666,
      "learning_rate": 1.2740631888317414e-05,
      "loss": 0.8492,
      "step": 12331
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0757532119750977,
      "learning_rate": 1.2739992971919625e-05,
      "loss": 1.1627,
      "step": 12332
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6404643058776855,
      "learning_rate": 1.2739354055521835e-05,
      "loss": 0.9539,
      "step": 12333
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.834159255027771,
      "learning_rate": 1.2738715139124047e-05,
      "loss": 0.9159,
      "step": 12334
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1227548122406006,
      "learning_rate": 1.2738076222726257e-05,
      "loss": 0.9559,
      "step": 12335
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5660414695739746,
      "learning_rate": 1.2737437306328468e-05,
      "loss": 0.8921,
      "step": 12336
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5283868312835693,
      "learning_rate": 1.2736798389930678e-05,
      "loss": 1.0613,
      "step": 12337
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5388091802597046,
      "learning_rate": 1.273615947353289e-05,
      "loss": 0.939,
      "step": 12338
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.610562801361084,
      "learning_rate": 1.27355205571351e-05,
      "loss": 1.5318,
      "step": 12339
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.005325794219971,
      "learning_rate": 1.2734881640737311e-05,
      "loss": 0.7504,
      "step": 12340
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.8736133575439453,
      "learning_rate": 1.273424272433952e-05,
      "loss": 1.0268,
      "step": 12341
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0820536613464355,
      "learning_rate": 1.2733603807941732e-05,
      "loss": 0.9313,
      "step": 12342
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.804659605026245,
      "learning_rate": 1.2732964891543942e-05,
      "loss": 1.2701,
      "step": 12343
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4240665435791016,
      "learning_rate": 1.2732325975146154e-05,
      "loss": 0.9367,
      "step": 12344
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6049617528915405,
      "learning_rate": 1.2731687058748364e-05,
      "loss": 0.4492,
      "step": 12345
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7025964260101318,
      "learning_rate": 1.2731048142350575e-05,
      "loss": 0.8153,
      "step": 12346
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2783212661743164,
      "learning_rate": 1.2730409225952785e-05,
      "loss": 0.8484,
      "step": 12347
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.571676254272461,
      "learning_rate": 1.2729770309554997e-05,
      "loss": 1.5059,
      "step": 12348
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.314732074737549,
      "learning_rate": 1.2729131393157206e-05,
      "loss": 1.2772,
      "step": 12349
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.313446998596191,
      "learning_rate": 1.2728492476759418e-05,
      "loss": 1.708,
      "step": 12350
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0405962467193604,
      "learning_rate": 1.2727853560361628e-05,
      "loss": 0.8857,
      "step": 12351
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1919078826904297,
      "learning_rate": 1.272721464396384e-05,
      "loss": 0.8373,
      "step": 12352
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7229231595993042,
      "learning_rate": 1.272657572756605e-05,
      "loss": 0.9814,
      "step": 12353
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7213200330734253,
      "learning_rate": 1.272593681116826e-05,
      "loss": 0.8649,
      "step": 12354
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.137463331222534,
      "learning_rate": 1.2725297894770469e-05,
      "loss": 1.0067,
      "step": 12355
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7194185256958008,
      "learning_rate": 1.2724658978372682e-05,
      "loss": 0.8558,
      "step": 12356
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.070915222167969,
      "learning_rate": 1.272402006197489e-05,
      "loss": 1.1737,
      "step": 12357
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8337699174880981,
      "learning_rate": 1.2723381145577104e-05,
      "loss": 0.9984,
      "step": 12358
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.2541797161102295,
      "learning_rate": 1.2722742229179312e-05,
      "loss": 0.8569,
      "step": 12359
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8030551671981812,
      "learning_rate": 1.2722103312781523e-05,
      "loss": 0.9278,
      "step": 12360
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.369694948196411,
      "learning_rate": 1.2721464396383733e-05,
      "loss": 1.0703,
      "step": 12361
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.1924848556518555,
      "learning_rate": 1.2720825479985945e-05,
      "loss": 1.1762,
      "step": 12362
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6846858263015747,
      "learning_rate": 1.2720186563588155e-05,
      "loss": 0.9989,
      "step": 12363
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4938613176345825,
      "learning_rate": 1.2719547647190366e-05,
      "loss": 0.8674,
      "step": 12364
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6789718866348267,
      "learning_rate": 1.2718908730792576e-05,
      "loss": 0.8624,
      "step": 12365
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4803075790405273,
      "learning_rate": 1.2718269814394787e-05,
      "loss": 0.8384,
      "step": 12366
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8550704717636108,
      "learning_rate": 1.2717630897996997e-05,
      "loss": 0.6064,
      "step": 12367
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.9783549308776855,
      "learning_rate": 1.2716991981599209e-05,
      "loss": 1.0171,
      "step": 12368
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7888816595077515,
      "learning_rate": 1.2716353065201419e-05,
      "loss": 0.9031,
      "step": 12369
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6354914903640747,
      "learning_rate": 1.271571414880363e-05,
      "loss": 0.8359,
      "step": 12370
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7246077060699463,
      "learning_rate": 1.271507523240584e-05,
      "loss": 1.2825,
      "step": 12371
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2625436782836914,
      "learning_rate": 1.2714436316008052e-05,
      "loss": 1.0271,
      "step": 12372
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.72385311126709,
      "learning_rate": 1.2713797399610262e-05,
      "loss": 1.6686,
      "step": 12373
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7792059183120728,
      "learning_rate": 1.2713158483212473e-05,
      "loss": 0.8984,
      "step": 12374
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9145907163619995,
      "learning_rate": 1.2712519566814683e-05,
      "loss": 0.8975,
      "step": 12375
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.147592306137085,
      "learning_rate": 1.2711880650416894e-05,
      "loss": 0.9199,
      "step": 12376
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1991164684295654,
      "learning_rate": 1.2711241734019104e-05,
      "loss": 0.8973,
      "step": 12377
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7584190368652344,
      "learning_rate": 1.2710602817621316e-05,
      "loss": 0.5177,
      "step": 12378
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0741961002349854,
      "learning_rate": 1.2709963901223526e-05,
      "loss": 1.021,
      "step": 12379
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.4683570861816406,
      "learning_rate": 1.2709324984825737e-05,
      "loss": 0.7936,
      "step": 12380
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.485546827316284,
      "learning_rate": 1.2708686068427947e-05,
      "loss": 0.8069,
      "step": 12381
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.0647575855255127,
      "learning_rate": 1.2708047152030159e-05,
      "loss": 0.924,
      "step": 12382
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.8075404167175293,
      "learning_rate": 1.2707408235632369e-05,
      "loss": 1.4289,
      "step": 12383
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.7382166385650635,
      "learning_rate": 1.270676931923458e-05,
      "loss": 0.9422,
      "step": 12384
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1101536750793457,
      "learning_rate": 1.270613040283679e-05,
      "loss": 1.23,
      "step": 12385
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3996121883392334,
      "learning_rate": 1.2705491486439001e-05,
      "loss": 0.9824,
      "step": 12386
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.723903179168701,
      "learning_rate": 1.2704852570041211e-05,
      "loss": 0.9743,
      "step": 12387
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4425597190856934,
      "learning_rate": 1.2704213653643423e-05,
      "loss": 0.726,
      "step": 12388
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8291871547698975,
      "learning_rate": 1.2703574737245633e-05,
      "loss": 0.8397,
      "step": 12389
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0882492065429688,
      "learning_rate": 1.2702935820847844e-05,
      "loss": 0.9618,
      "step": 12390
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6096110343933105,
      "learning_rate": 1.2702296904450052e-05,
      "loss": 1.0004,
      "step": 12391
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.798305869102478,
      "learning_rate": 1.2701657988052266e-05,
      "loss": 0.9464,
      "step": 12392
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7434862852096558,
      "learning_rate": 1.2701019071654474e-05,
      "loss": 0.765,
      "step": 12393
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.018092632293701,
      "learning_rate": 1.2700380155256687e-05,
      "loss": 0.9506,
      "step": 12394
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6933467388153076,
      "learning_rate": 1.2699741238858895e-05,
      "loss": 0.9975,
      "step": 12395
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.006582260131836,
      "learning_rate": 1.2699102322461107e-05,
      "loss": 1.0359,
      "step": 12396
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4507958889007568,
      "learning_rate": 1.2698463406063317e-05,
      "loss": 0.8552,
      "step": 12397
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5530812740325928,
      "learning_rate": 1.2697824489665528e-05,
      "loss": 0.9287,
      "step": 12398
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.12003755569458,
      "learning_rate": 1.2697185573267738e-05,
      "loss": 1.1141,
      "step": 12399
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.222811222076416,
      "learning_rate": 1.269654665686995e-05,
      "loss": 1.0021,
      "step": 12400
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.164952039718628,
      "learning_rate": 1.269590774047216e-05,
      "loss": 0.8855,
      "step": 12401
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6418991088867188,
      "learning_rate": 1.2695268824074371e-05,
      "loss": 0.8407,
      "step": 12402
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.741798996925354,
      "learning_rate": 1.269462990767658e-05,
      "loss": 1.0363,
      "step": 12403
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.812913179397583,
      "learning_rate": 1.2693990991278792e-05,
      "loss": 0.925,
      "step": 12404
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.462903022766113,
      "learning_rate": 1.2693352074881002e-05,
      "loss": 1.2607,
      "step": 12405
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0044777393341064,
      "learning_rate": 1.2692713158483214e-05,
      "loss": 1.2656,
      "step": 12406
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1275100708007812,
      "learning_rate": 1.2692074242085424e-05,
      "loss": 0.937,
      "step": 12407
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3931386470794678,
      "learning_rate": 1.2691435325687635e-05,
      "loss": 0.4183,
      "step": 12408
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3669769763946533,
      "learning_rate": 1.2690796409289845e-05,
      "loss": 1.0495,
      "step": 12409
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7624549865722656,
      "learning_rate": 1.2690157492892057e-05,
      "loss": 0.8329,
      "step": 12410
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.94768226146698,
      "learning_rate": 1.2689518576494266e-05,
      "loss": 0.8967,
      "step": 12411
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8076075315475464,
      "learning_rate": 1.2688879660096478e-05,
      "loss": 0.9458,
      "step": 12412
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.192690372467041,
      "learning_rate": 1.2688240743698688e-05,
      "loss": 1.8266,
      "step": 12413
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5859107971191406,
      "learning_rate": 1.26876018273009e-05,
      "loss": 0.9321,
      "step": 12414
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9514005184173584,
      "learning_rate": 1.268696291090311e-05,
      "loss": 0.9619,
      "step": 12415
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5738576650619507,
      "learning_rate": 1.268632399450532e-05,
      "loss": 0.8796,
      "step": 12416
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5901877880096436,
      "learning_rate": 1.268568507810753e-05,
      "loss": 1.0606,
      "step": 12417
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6295446157455444,
      "learning_rate": 1.2685046161709742e-05,
      "loss": 0.9121,
      "step": 12418
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6094355583190918,
      "learning_rate": 1.2684407245311952e-05,
      "loss": 0.9245,
      "step": 12419
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.9698994159698486,
      "learning_rate": 1.2683768328914164e-05,
      "loss": 0.9813,
      "step": 12420
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.673811435699463,
      "learning_rate": 1.2683129412516373e-05,
      "loss": 0.8972,
      "step": 12421
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5079847574234009,
      "learning_rate": 1.2682490496118585e-05,
      "loss": 0.8802,
      "step": 12422
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.971513032913208,
      "learning_rate": 1.2681851579720795e-05,
      "loss": 1.1451,
      "step": 12423
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7993758916854858,
      "learning_rate": 1.2681212663323006e-05,
      "loss": 0.9659,
      "step": 12424
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7900407314300537,
      "learning_rate": 1.2680573746925216e-05,
      "loss": 0.8876,
      "step": 12425
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.967383623123169,
      "learning_rate": 1.2679934830527428e-05,
      "loss": 1.1087,
      "step": 12426
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.381737470626831,
      "learning_rate": 1.2679295914129636e-05,
      "loss": 0.7707,
      "step": 12427
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9722955226898193,
      "learning_rate": 1.267865699773185e-05,
      "loss": 0.849,
      "step": 12428
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0363528728485107,
      "learning_rate": 1.2678018081334057e-05,
      "loss": 1.368,
      "step": 12429
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.887609839439392,
      "learning_rate": 1.267737916493627e-05,
      "loss": 0.8835,
      "step": 12430
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9849544763565063,
      "learning_rate": 1.2676740248538479e-05,
      "loss": 0.9714,
      "step": 12431
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.4316422939300537,
      "learning_rate": 1.2676101332140692e-05,
      "loss": 1.2111,
      "step": 12432
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5862784385681152,
      "learning_rate": 1.26754624157429e-05,
      "loss": 0.866,
      "step": 12433
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5820374488830566,
      "learning_rate": 1.2674823499345112e-05,
      "loss": 0.9142,
      "step": 12434
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9402533769607544,
      "learning_rate": 1.2674184582947322e-05,
      "loss": 0.8981,
      "step": 12435
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.5846500396728516,
      "learning_rate": 1.2673545666549533e-05,
      "loss": 1.203,
      "step": 12436
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.079002380371094,
      "learning_rate": 1.2672906750151743e-05,
      "loss": 1.1836,
      "step": 12437
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.014631509780884,
      "learning_rate": 1.2672267833753954e-05,
      "loss": 0.9408,
      "step": 12438
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.082446336746216,
      "learning_rate": 1.2671628917356164e-05,
      "loss": 0.7413,
      "step": 12439
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.417935848236084,
      "learning_rate": 1.2670990000958376e-05,
      "loss": 0.7807,
      "step": 12440
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.976146697998047,
      "learning_rate": 1.2670351084560586e-05,
      "loss": 1.6217,
      "step": 12441
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.7374086380004883,
      "learning_rate": 1.2669712168162797e-05,
      "loss": 1.065,
      "step": 12442
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6135644912719727,
      "learning_rate": 1.2669073251765007e-05,
      "loss": 0.8237,
      "step": 12443
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5575917959213257,
      "learning_rate": 1.2668434335367219e-05,
      "loss": 0.4649,
      "step": 12444
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8137768507003784,
      "learning_rate": 1.2667795418969429e-05,
      "loss": 0.8993,
      "step": 12445
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.639553189277649,
      "learning_rate": 1.266715650257164e-05,
      "loss": 0.8182,
      "step": 12446
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0621087551116943,
      "learning_rate": 1.266651758617385e-05,
      "loss": 0.886,
      "step": 12447
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9368538856506348,
      "learning_rate": 1.2665878669776061e-05,
      "loss": 0.9909,
      "step": 12448
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.478239893913269,
      "learning_rate": 1.2665239753378271e-05,
      "loss": 0.8814,
      "step": 12449
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.971853256225586,
      "learning_rate": 1.2664600836980483e-05,
      "loss": 1.0501,
      "step": 12450
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7919752597808838,
      "learning_rate": 1.2663961920582693e-05,
      "loss": 0.9836,
      "step": 12451
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6511788368225098,
      "learning_rate": 1.2663323004184904e-05,
      "loss": 0.8104,
      "step": 12452
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6334151029586792,
      "learning_rate": 1.2662684087787114e-05,
      "loss": 0.8348,
      "step": 12453
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.472893476486206,
      "learning_rate": 1.2662045171389326e-05,
      "loss": 1.0612,
      "step": 12454
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.082319974899292,
      "learning_rate": 1.2661406254991536e-05,
      "loss": 0.8278,
      "step": 12455
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7969696521759033,
      "learning_rate": 1.2660767338593747e-05,
      "loss": 0.8334,
      "step": 12456
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8158091306686401,
      "learning_rate": 1.2660128422195957e-05,
      "loss": 1.0496,
      "step": 12457
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7332079410552979,
      "learning_rate": 1.2659489505798168e-05,
      "loss": 0.8534,
      "step": 12458
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0016138553619385,
      "learning_rate": 1.2658850589400378e-05,
      "loss": 0.9278,
      "step": 12459
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.700035810470581,
      "learning_rate": 1.265821167300259e-05,
      "loss": 0.9018,
      "step": 12460
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.749159336090088,
      "learning_rate": 1.26575727566048e-05,
      "loss": 0.8762,
      "step": 12461
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8408582210540771,
      "learning_rate": 1.2656933840207011e-05,
      "loss": 0.9717,
      "step": 12462
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0026233196258545,
      "learning_rate": 1.2656294923809221e-05,
      "loss": 0.9196,
      "step": 12463
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5272685289382935,
      "learning_rate": 1.2655656007411433e-05,
      "loss": 0.8279,
      "step": 12464
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6900880336761475,
      "learning_rate": 1.2655017091013641e-05,
      "loss": 0.9094,
      "step": 12465
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.2004923820495605,
      "learning_rate": 1.2654378174615854e-05,
      "loss": 1.2598,
      "step": 12466
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6095587015151978,
      "learning_rate": 1.2653739258218062e-05,
      "loss": 0.8993,
      "step": 12467
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4450690746307373,
      "learning_rate": 1.2653100341820275e-05,
      "loss": 0.9047,
      "step": 12468
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.932100296020508,
      "learning_rate": 1.2652461425422484e-05,
      "loss": 1.1878,
      "step": 12469
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.833923101425171,
      "learning_rate": 1.2651822509024695e-05,
      "loss": 0.9278,
      "step": 12470
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8115991353988647,
      "learning_rate": 1.2651183592626905e-05,
      "loss": 0.9162,
      "step": 12471
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5481668710708618,
      "learning_rate": 1.2650544676229117e-05,
      "loss": 0.8833,
      "step": 12472
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8600711822509766,
      "learning_rate": 1.2649905759831326e-05,
      "loss": 0.8022,
      "step": 12473
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.341932535171509,
      "learning_rate": 1.2649266843433538e-05,
      "loss": 1.2158,
      "step": 12474
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6415616273880005,
      "learning_rate": 1.2648627927035748e-05,
      "loss": 0.8117,
      "step": 12475
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0962486267089844,
      "learning_rate": 1.264798901063796e-05,
      "loss": 0.8117,
      "step": 12476
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5747675895690918,
      "learning_rate": 1.264735009424017e-05,
      "loss": 0.8905,
      "step": 12477
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.73646080493927,
      "learning_rate": 1.264671117784238e-05,
      "loss": 0.8334,
      "step": 12478
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9386982917785645,
      "learning_rate": 1.264607226144459e-05,
      "loss": 0.8864,
      "step": 12479
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.384253740310669,
      "learning_rate": 1.2645433345046802e-05,
      "loss": 0.6217,
      "step": 12480
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4074574708938599,
      "learning_rate": 1.2644794428649012e-05,
      "loss": 0.8233,
      "step": 12481
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.137904167175293,
      "learning_rate": 1.2644155512251224e-05,
      "loss": 0.7528,
      "step": 12482
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.345731258392334,
      "learning_rate": 1.2643516595853433e-05,
      "loss": 0.92,
      "step": 12483
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8846179246902466,
      "learning_rate": 1.2642877679455645e-05,
      "loss": 0.8998,
      "step": 12484
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7866731882095337,
      "learning_rate": 1.2642238763057855e-05,
      "loss": 0.9152,
      "step": 12485
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6043094396591187,
      "learning_rate": 1.2641599846660066e-05,
      "loss": 0.9585,
      "step": 12486
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0429577827453613,
      "learning_rate": 1.2640960930262276e-05,
      "loss": 0.8011,
      "step": 12487
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0658211708068848,
      "learning_rate": 1.2640322013864488e-05,
      "loss": 1.0974,
      "step": 12488
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.18941593170166,
      "learning_rate": 1.2639683097466698e-05,
      "loss": 1.4061,
      "step": 12489
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.124619960784912,
      "learning_rate": 1.263904418106891e-05,
      "loss": 1.0783,
      "step": 12490
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.8031697273254395,
      "learning_rate": 1.2638405264671119e-05,
      "loss": 1.0365,
      "step": 12491
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9180442094802856,
      "learning_rate": 1.263776634827333e-05,
      "loss": 1.1164,
      "step": 12492
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.61001718044281,
      "learning_rate": 1.263712743187554e-05,
      "loss": 0.848,
      "step": 12493
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.091442584991455,
      "learning_rate": 1.2636488515477752e-05,
      "loss": 0.8988,
      "step": 12494
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0669775009155273,
      "learning_rate": 1.2635849599079962e-05,
      "loss": 0.9344,
      "step": 12495
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7982819080352783,
      "learning_rate": 1.2635210682682173e-05,
      "loss": 0.9768,
      "step": 12496
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.202545404434204,
      "learning_rate": 1.2634571766284383e-05,
      "loss": 1.2596,
      "step": 12497
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5221688747406006,
      "learning_rate": 1.2633932849886595e-05,
      "loss": 0.9032,
      "step": 12498
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3239645957946777,
      "learning_rate": 1.2633293933488805e-05,
      "loss": 0.5694,
      "step": 12499
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.8795759677886963,
      "learning_rate": 1.2632655017091016e-05,
      "loss": 0.9511,
      "step": 12500
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.340665340423584,
      "learning_rate": 1.2632016100693224e-05,
      "loss": 1.8127,
      "step": 12501
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.656393051147461,
      "learning_rate": 1.2631377184295438e-05,
      "loss": 0.9423,
      "step": 12502
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.02667498588562,
      "learning_rate": 1.2630738267897646e-05,
      "loss": 0.9403,
      "step": 12503
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1988296508789062,
      "learning_rate": 1.2630099351499859e-05,
      "loss": 1.2623,
      "step": 12504
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9655237197875977,
      "learning_rate": 1.2629460435102067e-05,
      "loss": 1.0876,
      "step": 12505
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.289867401123047,
      "learning_rate": 1.2628821518704279e-05,
      "loss": 1.0197,
      "step": 12506
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7455625534057617,
      "learning_rate": 1.2628182602306489e-05,
      "loss": 0.9041,
      "step": 12507
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.639108419418335,
      "learning_rate": 1.26275436859087e-05,
      "loss": 0.9038,
      "step": 12508
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2441494464874268,
      "learning_rate": 1.262690476951091e-05,
      "loss": 0.9021,
      "step": 12509
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4235246181488037,
      "learning_rate": 1.2626265853113122e-05,
      "loss": 0.9522,
      "step": 12510
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5591353178024292,
      "learning_rate": 1.2625626936715331e-05,
      "loss": 1.0345,
      "step": 12511
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4115973711013794,
      "learning_rate": 1.2624988020317543e-05,
      "loss": 0.9104,
      "step": 12512
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.0927958488464355,
      "learning_rate": 1.2624349103919753e-05,
      "loss": 1.1188,
      "step": 12513
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.5024075508117676,
      "learning_rate": 1.2623710187521964e-05,
      "loss": 1.0623,
      "step": 12514
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5373371839523315,
      "learning_rate": 1.2623071271124174e-05,
      "loss": 0.8976,
      "step": 12515
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.7683587074279785,
      "learning_rate": 1.2622432354726386e-05,
      "loss": 0.9274,
      "step": 12516
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3765828609466553,
      "learning_rate": 1.2621793438328596e-05,
      "loss": 0.9147,
      "step": 12517
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.6563215255737305,
      "learning_rate": 1.2621154521930807e-05,
      "loss": 1.0452,
      "step": 12518
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.74420428276062,
      "learning_rate": 1.2620515605533017e-05,
      "loss": 0.8084,
      "step": 12519
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3492779731750488,
      "learning_rate": 1.2619876689135229e-05,
      "loss": 0.8207,
      "step": 12520
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.057889223098755,
      "learning_rate": 1.2619237772737438e-05,
      "loss": 1.2604,
      "step": 12521
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0363664627075195,
      "learning_rate": 1.261859885633965e-05,
      "loss": 1.0171,
      "step": 12522
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.481377363204956,
      "learning_rate": 1.261795993994186e-05,
      "loss": 0.7067,
      "step": 12523
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.955172300338745,
      "learning_rate": 1.2617321023544071e-05,
      "loss": 1.2909,
      "step": 12524
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7329710721969604,
      "learning_rate": 1.2616682107146281e-05,
      "loss": 0.8818,
      "step": 12525
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.938712239265442,
      "learning_rate": 1.2616043190748493e-05,
      "loss": 0.9819,
      "step": 12526
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9987250566482544,
      "learning_rate": 1.2615404274350703e-05,
      "loss": 0.9555,
      "step": 12527
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0515875816345215,
      "learning_rate": 1.2614765357952914e-05,
      "loss": 1.0118,
      "step": 12528
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9211711883544922,
      "learning_rate": 1.2614126441555124e-05,
      "loss": 0.9741,
      "step": 12529
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.624450922012329,
      "learning_rate": 1.2613487525157336e-05,
      "loss": 0.8687,
      "step": 12530
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2689907550811768,
      "learning_rate": 1.2612848608759545e-05,
      "loss": 1.0082,
      "step": 12531
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.325786590576172,
      "learning_rate": 1.2612209692361757e-05,
      "loss": 1.17,
      "step": 12532
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5004998445510864,
      "learning_rate": 1.2611570775963967e-05,
      "loss": 0.8092,
      "step": 12533
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.73143470287323,
      "learning_rate": 1.2610931859566178e-05,
      "loss": 0.9844,
      "step": 12534
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3858113288879395,
      "learning_rate": 1.2610292943168388e-05,
      "loss": 0.9035,
      "step": 12535
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7823121547698975,
      "learning_rate": 1.26096540267706e-05,
      "loss": 0.9096,
      "step": 12536
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9251251220703125,
      "learning_rate": 1.2609015110372808e-05,
      "loss": 0.9087,
      "step": 12537
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.3249146938323975,
      "learning_rate": 1.2608376193975021e-05,
      "loss": 0.9482,
      "step": 12538
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5934324264526367,
      "learning_rate": 1.260773727757723e-05,
      "loss": 0.4314,
      "step": 12539
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0275652408599854,
      "learning_rate": 1.2607098361179443e-05,
      "loss": 1.0482,
      "step": 12540
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.581784963607788,
      "learning_rate": 1.260645944478165e-05,
      "loss": 0.9013,
      "step": 12541
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.5497710704803467,
      "learning_rate": 1.2605820528383864e-05,
      "loss": 1.1872,
      "step": 12542
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.991984486579895,
      "learning_rate": 1.2605181611986072e-05,
      "loss": 0.9735,
      "step": 12543
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7771843671798706,
      "learning_rate": 1.2604542695588284e-05,
      "loss": 0.9284,
      "step": 12544
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.538525104522705,
      "learning_rate": 1.2603903779190493e-05,
      "loss": 1.1604,
      "step": 12545
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2278544902801514,
      "learning_rate": 1.2603264862792705e-05,
      "loss": 1.3161,
      "step": 12546
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2330515384674072,
      "learning_rate": 1.2602625946394915e-05,
      "loss": 0.9367,
      "step": 12547
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3927464485168457,
      "learning_rate": 1.2601987029997126e-05,
      "loss": 0.4341,
      "step": 12548
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.962349534034729,
      "learning_rate": 1.2601348113599336e-05,
      "loss": 1.1965,
      "step": 12549
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.452695369720459,
      "learning_rate": 1.2600709197201548e-05,
      "loss": 1.0167,
      "step": 12550
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8737924098968506,
      "learning_rate": 1.2600070280803758e-05,
      "loss": 0.8447,
      "step": 12551
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5067561864852905,
      "learning_rate": 1.259943136440597e-05,
      "loss": 0.8822,
      "step": 12552
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5332776308059692,
      "learning_rate": 1.2598792448008179e-05,
      "loss": 0.4582,
      "step": 12553
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9351997375488281,
      "learning_rate": 1.259815353161039e-05,
      "loss": 0.9584,
      "step": 12554
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9168775081634521,
      "learning_rate": 1.25975146152126e-05,
      "loss": 0.8246,
      "step": 12555
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0191822052001953,
      "learning_rate": 1.2596875698814812e-05,
      "loss": 1.2287,
      "step": 12556
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5586519241333008,
      "learning_rate": 1.2596236782417022e-05,
      "loss": 0.9016,
      "step": 12557
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.5951507091522217,
      "learning_rate": 1.2595597866019233e-05,
      "loss": 0.9273,
      "step": 12558
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5662143230438232,
      "learning_rate": 1.2594958949621443e-05,
      "loss": 0.3964,
      "step": 12559
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2604095935821533,
      "learning_rate": 1.2594320033223655e-05,
      "loss": 0.9453,
      "step": 12560
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.441014766693115,
      "learning_rate": 1.2593681116825865e-05,
      "loss": 0.9957,
      "step": 12561
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.8218626976013184,
      "learning_rate": 1.2593042200428076e-05,
      "loss": 1.1215,
      "step": 12562
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.4671008586883545,
      "learning_rate": 1.2592403284030286e-05,
      "loss": 0.9768,
      "step": 12563
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7827504873275757,
      "learning_rate": 1.2591764367632496e-05,
      "loss": 0.9391,
      "step": 12564
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8747506141662598,
      "learning_rate": 1.2591125451234707e-05,
      "loss": 0.915,
      "step": 12565
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.4182028770446777,
      "learning_rate": 1.2590486534836917e-05,
      "loss": 1.3856,
      "step": 12566
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4026128053665161,
      "learning_rate": 1.2589847618439129e-05,
      "loss": 0.8172,
      "step": 12567
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3328816890716553,
      "learning_rate": 1.2589208702041337e-05,
      "loss": 1.1545,
      "step": 12568
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.387129545211792,
      "learning_rate": 1.258856978564355e-05,
      "loss": 0.9301,
      "step": 12569
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7257767915725708,
      "learning_rate": 1.2587930869245758e-05,
      "loss": 0.9894,
      "step": 12570
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6559982299804688,
      "learning_rate": 1.2587291952847972e-05,
      "loss": 0.4252,
      "step": 12571
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.90320885181427,
      "learning_rate": 1.258665303645018e-05,
      "loss": 1.1454,
      "step": 12572
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1864728927612305,
      "learning_rate": 1.2586014120052391e-05,
      "loss": 1.2576,
      "step": 12573
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.285032033920288,
      "learning_rate": 1.2585375203654601e-05,
      "loss": 0.8975,
      "step": 12574
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.472481608390808,
      "learning_rate": 1.2584736287256813e-05,
      "loss": 0.812,
      "step": 12575
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.960155487060547,
      "learning_rate": 1.2584097370859023e-05,
      "loss": 0.8029,
      "step": 12576
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7946338653564453,
      "learning_rate": 1.2583458454461234e-05,
      "loss": 1.1931,
      "step": 12577
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.120724678039551,
      "learning_rate": 1.2582819538063444e-05,
      "loss": 1.0305,
      "step": 12578
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.301706314086914,
      "learning_rate": 1.2582180621665656e-05,
      "loss": 0.8904,
      "step": 12579
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.887384057044983,
      "learning_rate": 1.2581541705267865e-05,
      "loss": 1.2118,
      "step": 12580
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.582756519317627,
      "learning_rate": 1.2580902788870077e-05,
      "loss": 0.7635,
      "step": 12581
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6743944883346558,
      "learning_rate": 1.2580263872472287e-05,
      "loss": 0.843,
      "step": 12582
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7376388311386108,
      "learning_rate": 1.2579624956074498e-05,
      "loss": 0.725,
      "step": 12583
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.808789849281311,
      "learning_rate": 1.2578986039676708e-05,
      "loss": 0.9471,
      "step": 12584
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.413468360900879,
      "learning_rate": 1.257834712327892e-05,
      "loss": 1.4773,
      "step": 12585
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7239288091659546,
      "learning_rate": 1.257770820688113e-05,
      "loss": 0.7838,
      "step": 12586
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1718766689300537,
      "learning_rate": 1.2577069290483341e-05,
      "loss": 0.9987,
      "step": 12587
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.464411735534668,
      "learning_rate": 1.2576430374085551e-05,
      "loss": 0.8962,
      "step": 12588
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.001994609832764,
      "learning_rate": 1.2575791457687763e-05,
      "loss": 1.0262,
      "step": 12589
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.237583637237549,
      "learning_rate": 1.2575152541289972e-05,
      "loss": 1.0935,
      "step": 12590
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.977739930152893,
      "learning_rate": 1.2574513624892184e-05,
      "loss": 1.0483,
      "step": 12591
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6367379426956177,
      "learning_rate": 1.2573874708494394e-05,
      "loss": 0.9017,
      "step": 12592
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.270744562149048,
      "learning_rate": 1.2573235792096605e-05,
      "loss": 0.9639,
      "step": 12593
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.1315205097198486,
      "learning_rate": 1.2572596875698815e-05,
      "loss": 0.8466,
      "step": 12594
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.190086603164673,
      "learning_rate": 1.2571957959301027e-05,
      "loss": 0.9761,
      "step": 12595
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9648631811141968,
      "learning_rate": 1.2571319042903237e-05,
      "loss": 0.9874,
      "step": 12596
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.364261627197266,
      "learning_rate": 1.2570680126505448e-05,
      "loss": 0.8491,
      "step": 12597
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1425864696502686,
      "learning_rate": 1.2570041210107658e-05,
      "loss": 1.1288,
      "step": 12598
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8932133913040161,
      "learning_rate": 1.256940229370987e-05,
      "loss": 1.172,
      "step": 12599
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.268538475036621,
      "learning_rate": 1.256876337731208e-05,
      "loss": 0.8969,
      "step": 12600
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.034142255783081,
      "learning_rate": 1.2568124460914291e-05,
      "loss": 0.9196,
      "step": 12601
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5303722620010376,
      "learning_rate": 1.25674855445165e-05,
      "loss": 0.3682,
      "step": 12602
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.626251220703125,
      "learning_rate": 1.2566846628118712e-05,
      "loss": 0.84,
      "step": 12603
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.605983018875122,
      "learning_rate": 1.256620771172092e-05,
      "loss": 0.7293,
      "step": 12604
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7356032133102417,
      "learning_rate": 1.2565568795323134e-05,
      "loss": 0.9195,
      "step": 12605
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.3387959003448486,
      "learning_rate": 1.2564929878925342e-05,
      "loss": 1.0468,
      "step": 12606
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6532360315322876,
      "learning_rate": 1.2564290962527555e-05,
      "loss": 0.9446,
      "step": 12607
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.1467368602752686,
      "learning_rate": 1.2563652046129763e-05,
      "loss": 0.8888,
      "step": 12608
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1189842224121094,
      "learning_rate": 1.2563013129731977e-05,
      "loss": 0.7447,
      "step": 12609
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1034579277038574,
      "learning_rate": 1.2562374213334185e-05,
      "loss": 0.9548,
      "step": 12610
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6731494665145874,
      "learning_rate": 1.2561735296936396e-05,
      "loss": 0.4256,
      "step": 12611
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9030253887176514,
      "learning_rate": 1.2561096380538606e-05,
      "loss": 0.8671,
      "step": 12612
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8293898105621338,
      "learning_rate": 1.2560457464140818e-05,
      "loss": 0.9784,
      "step": 12613
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1972873210906982,
      "learning_rate": 1.2559818547743028e-05,
      "loss": 1.0749,
      "step": 12614
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.707090139389038,
      "learning_rate": 1.2559179631345239e-05,
      "loss": 0.9311,
      "step": 12615
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.288861274719238,
      "learning_rate": 1.2558540714947449e-05,
      "loss": 1.0415,
      "step": 12616
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8356126546859741,
      "learning_rate": 1.255790179854966e-05,
      "loss": 0.957,
      "step": 12617
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5734506845474243,
      "learning_rate": 1.255726288215187e-05,
      "loss": 0.3593,
      "step": 12618
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6299149990081787,
      "learning_rate": 1.2556623965754082e-05,
      "loss": 0.8623,
      "step": 12619
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3309211730957031,
      "learning_rate": 1.2555985049356292e-05,
      "loss": 0.4994,
      "step": 12620
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3421146869659424,
      "learning_rate": 1.2555346132958503e-05,
      "loss": 0.9401,
      "step": 12621
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8469916582107544,
      "learning_rate": 1.2554707216560713e-05,
      "loss": 0.8717,
      "step": 12622
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.240648031234741,
      "learning_rate": 1.2554068300162925e-05,
      "loss": 0.9541,
      "step": 12623
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0281436443328857,
      "learning_rate": 1.2553429383765135e-05,
      "loss": 0.8419,
      "step": 12624
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.894509792327881,
      "learning_rate": 1.2552790467367346e-05,
      "loss": 1.0292,
      "step": 12625
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.883959174156189,
      "learning_rate": 1.2552151550969556e-05,
      "loss": 0.9798,
      "step": 12626
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0922086238861084,
      "learning_rate": 1.2551512634571767e-05,
      "loss": 0.9379,
      "step": 12627
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.415177583694458,
      "learning_rate": 1.2550873718173977e-05,
      "loss": 0.8136,
      "step": 12628
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8239699602127075,
      "learning_rate": 1.2550234801776189e-05,
      "loss": 0.9563,
      "step": 12629
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4964098930358887,
      "learning_rate": 1.2549595885378399e-05,
      "loss": 0.9267,
      "step": 12630
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8349124193191528,
      "learning_rate": 1.254895696898061e-05,
      "loss": 0.9635,
      "step": 12631
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6526999473571777,
      "learning_rate": 1.254831805258282e-05,
      "loss": 1.0125,
      "step": 12632
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1188724040985107,
      "learning_rate": 1.2547679136185032e-05,
      "loss": 0.9933,
      "step": 12633
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.26436185836792,
      "learning_rate": 1.2547040219787242e-05,
      "loss": 0.7794,
      "step": 12634
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1141624450683594,
      "learning_rate": 1.2546401303389453e-05,
      "loss": 0.8481,
      "step": 12635
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7960546016693115,
      "learning_rate": 1.2545762386991663e-05,
      "loss": 0.9146,
      "step": 12636
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.783001184463501,
      "learning_rate": 1.2545123470593874e-05,
      "loss": 0.938,
      "step": 12637
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4202702045440674,
      "learning_rate": 1.2544484554196084e-05,
      "loss": 0.4433,
      "step": 12638
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6245205402374268,
      "learning_rate": 1.2543845637798296e-05,
      "loss": 0.846,
      "step": 12639
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8670668601989746,
      "learning_rate": 1.2543206721400506e-05,
      "loss": 1.0227,
      "step": 12640
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9510353803634644,
      "learning_rate": 1.2542567805002717e-05,
      "loss": 0.9414,
      "step": 12641
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.3502309322357178,
      "learning_rate": 1.2541928888604925e-05,
      "loss": 1.1735,
      "step": 12642
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8925248384475708,
      "learning_rate": 1.2541289972207139e-05,
      "loss": 0.992,
      "step": 12643
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.502620816230774,
      "learning_rate": 1.2540651055809347e-05,
      "loss": 0.4927,
      "step": 12644
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7999465465545654,
      "learning_rate": 1.254001213941156e-05,
      "loss": 1.2332,
      "step": 12645
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7541018724441528,
      "learning_rate": 1.2539373223013768e-05,
      "loss": 0.9528,
      "step": 12646
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8683016300201416,
      "learning_rate": 1.253873430661598e-05,
      "loss": 1.1375,
      "step": 12647
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9115495681762695,
      "learning_rate": 1.253809539021819e-05,
      "loss": 0.9508,
      "step": 12648
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5110185146331787,
      "learning_rate": 1.2537456473820401e-05,
      "loss": 0.446,
      "step": 12649
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6531572341918945,
      "learning_rate": 1.2536817557422611e-05,
      "loss": 0.4311,
      "step": 12650
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4826372861862183,
      "learning_rate": 1.2536178641024823e-05,
      "loss": 0.8729,
      "step": 12651
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7370685338974,
      "learning_rate": 1.2535539724627032e-05,
      "loss": 0.9125,
      "step": 12652
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4841432571411133,
      "learning_rate": 1.2534900808229244e-05,
      "loss": 0.5026,
      "step": 12653
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1123268604278564,
      "learning_rate": 1.2534261891831454e-05,
      "loss": 0.9689,
      "step": 12654
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6634092330932617,
      "learning_rate": 1.2533622975433665e-05,
      "loss": 0.8799,
      "step": 12655
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6330361366271973,
      "learning_rate": 1.2532984059035875e-05,
      "loss": 0.9202,
      "step": 12656
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.723940372467041,
      "learning_rate": 1.2532345142638087e-05,
      "loss": 0.8285,
      "step": 12657
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0008251667022705,
      "learning_rate": 1.2531706226240297e-05,
      "loss": 0.9912,
      "step": 12658
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1276848316192627,
      "learning_rate": 1.2531067309842508e-05,
      "loss": 0.9045,
      "step": 12659
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8831981420516968,
      "learning_rate": 1.2530428393444718e-05,
      "loss": 0.9629,
      "step": 12660
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.812437057495117,
      "learning_rate": 1.252978947704693e-05,
      "loss": 1.3644,
      "step": 12661
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1079328060150146,
      "learning_rate": 1.252915056064914e-05,
      "loss": 0.98,
      "step": 12662
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4361965656280518,
      "learning_rate": 1.2528511644251351e-05,
      "loss": 0.4355,
      "step": 12663
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0974786281585693,
      "learning_rate": 1.252787272785356e-05,
      "loss": 0.8744,
      "step": 12664
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.937349796295166,
      "learning_rate": 1.2527233811455772e-05,
      "loss": 0.9774,
      "step": 12665
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.138737201690674,
      "learning_rate": 1.2526594895057982e-05,
      "loss": 1.0217,
      "step": 12666
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6652626991271973,
      "learning_rate": 1.2525955978660194e-05,
      "loss": 0.813,
      "step": 12667
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9935301542282104,
      "learning_rate": 1.2525317062262404e-05,
      "loss": 0.9949,
      "step": 12668
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.825082540512085,
      "learning_rate": 1.2524678145864615e-05,
      "loss": 0.4334,
      "step": 12669
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8466416597366333,
      "learning_rate": 1.2524039229466825e-05,
      "loss": 0.9146,
      "step": 12670
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7902437448501587,
      "learning_rate": 1.2523400313069037e-05,
      "loss": 0.9284,
      "step": 12671
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0136232376098633,
      "learning_rate": 1.2522761396671246e-05,
      "loss": 0.7908,
      "step": 12672
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.318166971206665,
      "learning_rate": 1.2522122480273458e-05,
      "loss": 0.9584,
      "step": 12673
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.446347951889038,
      "learning_rate": 1.2521483563875668e-05,
      "loss": 1.0144,
      "step": 12674
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.5592660903930664,
      "learning_rate": 1.252084464747788e-05,
      "loss": 1.1493,
      "step": 12675
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.482245683670044,
      "learning_rate": 1.252020573108009e-05,
      "loss": 0.7843,
      "step": 12676
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1088619232177734,
      "learning_rate": 1.25195668146823e-05,
      "loss": 0.9588,
      "step": 12677
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8637468814849854,
      "learning_rate": 1.2518927898284509e-05,
      "loss": 0.9391,
      "step": 12678
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.794018268585205,
      "learning_rate": 1.2518288981886722e-05,
      "loss": 0.5031,
      "step": 12679
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9055616855621338,
      "learning_rate": 1.251765006548893e-05,
      "loss": 1.0357,
      "step": 12680
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.166942834854126,
      "learning_rate": 1.2517011149091144e-05,
      "loss": 1.0752,
      "step": 12681
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9275822639465332,
      "learning_rate": 1.2516372232693352e-05,
      "loss": 1.0536,
      "step": 12682
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1136767864227295,
      "learning_rate": 1.2515733316295563e-05,
      "loss": 1.0573,
      "step": 12683
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5585697889328003,
      "learning_rate": 1.2515094399897773e-05,
      "loss": 0.3973,
      "step": 12684
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7703956365585327,
      "learning_rate": 1.2514455483499985e-05,
      "loss": 0.903,
      "step": 12685
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.378159523010254,
      "learning_rate": 1.2513816567102195e-05,
      "loss": 0.769,
      "step": 12686
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.830751895904541,
      "learning_rate": 1.2513177650704406e-05,
      "loss": 0.887,
      "step": 12687
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.214845895767212,
      "learning_rate": 1.2512538734306616e-05,
      "loss": 0.9426,
      "step": 12688
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5045254230499268,
      "learning_rate": 1.2511899817908827e-05,
      "loss": 0.8759,
      "step": 12689
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1296393871307373,
      "learning_rate": 1.2511260901511037e-05,
      "loss": 0.8422,
      "step": 12690
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5870354175567627,
      "learning_rate": 1.2510621985113249e-05,
      "loss": 1.0319,
      "step": 12691
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.33455228805542,
      "learning_rate": 1.2509983068715459e-05,
      "loss": 0.947,
      "step": 12692
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9474912881851196,
      "learning_rate": 1.250934415231767e-05,
      "loss": 1.0362,
      "step": 12693
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3616231679916382,
      "learning_rate": 1.250870523591988e-05,
      "loss": 0.4242,
      "step": 12694
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5789294242858887,
      "learning_rate": 1.2508066319522092e-05,
      "loss": 0.7597,
      "step": 12695
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4905542135238647,
      "learning_rate": 1.2507427403124302e-05,
      "loss": 0.8609,
      "step": 12696
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.5027823448181152,
      "learning_rate": 1.2506788486726513e-05,
      "loss": 1.0367,
      "step": 12697
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7046914100646973,
      "learning_rate": 1.2506149570328723e-05,
      "loss": 0.8744,
      "step": 12698
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5880928039550781,
      "learning_rate": 1.2505510653930934e-05,
      "loss": 0.9241,
      "step": 12699
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.150968074798584,
      "learning_rate": 1.2504871737533144e-05,
      "loss": 0.9642,
      "step": 12700
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.779347896575928,
      "learning_rate": 1.2504232821135356e-05,
      "loss": 1.2987,
      "step": 12701
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9328415393829346,
      "learning_rate": 1.2503593904737566e-05,
      "loss": 0.9648,
      "step": 12702
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4400793313980103,
      "learning_rate": 1.2502954988339777e-05,
      "loss": 0.4183,
      "step": 12703
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.209974527359009,
      "learning_rate": 1.2502316071941987e-05,
      "loss": 0.8342,
      "step": 12704
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.214282274246216,
      "learning_rate": 1.2501677155544199e-05,
      "loss": 0.8529,
      "step": 12705
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7196030616760254,
      "learning_rate": 1.2501038239146409e-05,
      "loss": 0.7864,
      "step": 12706
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0685861110687256,
      "learning_rate": 1.250039932274862e-05,
      "loss": 0.8325,
      "step": 12707
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.21183180809021,
      "learning_rate": 1.249976040635083e-05,
      "loss": 0.9795,
      "step": 12708
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.5330114364624023,
      "learning_rate": 1.2499121489953041e-05,
      "loss": 1.0584,
      "step": 12709
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.69338059425354,
      "learning_rate": 1.2498482573555251e-05,
      "loss": 0.8837,
      "step": 12710
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1976048946380615,
      "learning_rate": 1.2497843657157463e-05,
      "loss": 0.7722,
      "step": 12711
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5168348550796509,
      "learning_rate": 1.2497204740759673e-05,
      "loss": 0.7726,
      "step": 12712
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.929617166519165,
      "learning_rate": 1.2496565824361884e-05,
      "loss": 1.1321,
      "step": 12713
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1473023891448975,
      "learning_rate": 1.2495926907964092e-05,
      "loss": 1.1761,
      "step": 12714
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.324769973754883,
      "learning_rate": 1.2495287991566306e-05,
      "loss": 1.0596,
      "step": 12715
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8301948308944702,
      "learning_rate": 1.2494649075168514e-05,
      "loss": 0.9221,
      "step": 12716
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6518080234527588,
      "learning_rate": 1.2494010158770727e-05,
      "loss": 0.9971,
      "step": 12717
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.245253324508667,
      "learning_rate": 1.2493371242372935e-05,
      "loss": 0.7647,
      "step": 12718
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2077949047088623,
      "learning_rate": 1.2492732325975147e-05,
      "loss": 1.1981,
      "step": 12719
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.956848382949829,
      "learning_rate": 1.2492093409577357e-05,
      "loss": 0.7994,
      "step": 12720
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.326511859893799,
      "learning_rate": 1.2491454493179568e-05,
      "loss": 1.0179,
      "step": 12721
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4773696660995483,
      "learning_rate": 1.2490815576781778e-05,
      "loss": 0.781,
      "step": 12722
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9846768379211426,
      "learning_rate": 1.249017666038399e-05,
      "loss": 0.8319,
      "step": 12723
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4163966178894043,
      "learning_rate": 1.24895377439862e-05,
      "loss": 1.0451,
      "step": 12724
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.370760440826416,
      "learning_rate": 1.2488898827588411e-05,
      "loss": 0.8312,
      "step": 12725
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.586594581604004,
      "learning_rate": 1.248825991119062e-05,
      "loss": 0.9203,
      "step": 12726
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.804632306098938,
      "learning_rate": 1.2487620994792832e-05,
      "loss": 1.0739,
      "step": 12727
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7583884000778198,
      "learning_rate": 1.2486982078395042e-05,
      "loss": 1.156,
      "step": 12728
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7280938625335693,
      "learning_rate": 1.2486343161997254e-05,
      "loss": 1.046,
      "step": 12729
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9636974334716797,
      "learning_rate": 1.2485704245599464e-05,
      "loss": 0.8895,
      "step": 12730
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0802478790283203,
      "learning_rate": 1.2485065329201675e-05,
      "loss": 0.9055,
      "step": 12731
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2021892070770264,
      "learning_rate": 1.2484426412803885e-05,
      "loss": 0.9849,
      "step": 12732
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7430038452148438,
      "learning_rate": 1.2483787496406097e-05,
      "loss": 0.7274,
      "step": 12733
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.130640983581543,
      "learning_rate": 1.2483148580008306e-05,
      "loss": 0.8588,
      "step": 12734
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7934606075286865,
      "learning_rate": 1.2482509663610518e-05,
      "loss": 1.2955,
      "step": 12735
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5984488725662231,
      "learning_rate": 1.2481870747212728e-05,
      "loss": 0.8751,
      "step": 12736
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.011483669281006,
      "learning_rate": 1.248123183081494e-05,
      "loss": 0.922,
      "step": 12737
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.526628851890564,
      "learning_rate": 1.248059291441715e-05,
      "loss": 0.9309,
      "step": 12738
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9092304706573486,
      "learning_rate": 1.247995399801936e-05,
      "loss": 0.934,
      "step": 12739
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.5338101387023926,
      "learning_rate": 1.247931508162157e-05,
      "loss": 1.0453,
      "step": 12740
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4956398010253906,
      "learning_rate": 1.2478676165223782e-05,
      "loss": 0.9485,
      "step": 12741
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.29697585105896,
      "learning_rate": 1.2478037248825992e-05,
      "loss": 1.0713,
      "step": 12742
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7227846384048462,
      "learning_rate": 1.2477398332428204e-05,
      "loss": 0.893,
      "step": 12743
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6408860683441162,
      "learning_rate": 1.2476759416030413e-05,
      "loss": 0.8335,
      "step": 12744
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.654013752937317,
      "learning_rate": 1.2476120499632625e-05,
      "loss": 0.8788,
      "step": 12745
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8799506425857544,
      "learning_rate": 1.2475481583234835e-05,
      "loss": 0.9612,
      "step": 12746
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.5875847339630127,
      "learning_rate": 1.2474842666837046e-05,
      "loss": 0.9674,
      "step": 12747
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.141791582107544,
      "learning_rate": 1.2474203750439256e-05,
      "loss": 0.9958,
      "step": 12748
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3067917823791504,
      "learning_rate": 1.2473564834041468e-05,
      "loss": 0.4079,
      "step": 12749
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8040461540222168,
      "learning_rate": 1.2472925917643676e-05,
      "loss": 0.842,
      "step": 12750
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4179346561431885,
      "learning_rate": 1.247228700124589e-05,
      "loss": 0.8314,
      "step": 12751
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8407864570617676,
      "learning_rate": 1.2471648084848097e-05,
      "loss": 1.0675,
      "step": 12752
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5135232210159302,
      "learning_rate": 1.247100916845031e-05,
      "loss": 0.4082,
      "step": 12753
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0852925777435303,
      "learning_rate": 1.2470370252052519e-05,
      "loss": 0.94,
      "step": 12754
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8048826456069946,
      "learning_rate": 1.2469731335654732e-05,
      "loss": 0.9263,
      "step": 12755
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0914957523345947,
      "learning_rate": 1.246909241925694e-05,
      "loss": 1.3346,
      "step": 12756
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9002635478973389,
      "learning_rate": 1.2468453502859152e-05,
      "loss": 1.008,
      "step": 12757
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.647634983062744,
      "learning_rate": 1.2467814586461362e-05,
      "loss": 1.4022,
      "step": 12758
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7521766424179077,
      "learning_rate": 1.2467175670063573e-05,
      "loss": 1.2563,
      "step": 12759
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.3593502044677734,
      "learning_rate": 1.2466536753665783e-05,
      "loss": 1.0507,
      "step": 12760
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.584661602973938,
      "learning_rate": 1.2465897837267994e-05,
      "loss": 0.752,
      "step": 12761
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.517840027809143,
      "learning_rate": 1.2465258920870204e-05,
      "loss": 0.8104,
      "step": 12762
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.07007098197937,
      "learning_rate": 1.2464620004472416e-05,
      "loss": 0.9106,
      "step": 12763
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6356509923934937,
      "learning_rate": 1.2463981088074626e-05,
      "loss": 0.9512,
      "step": 12764
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.4375593662261963,
      "learning_rate": 1.2463342171676837e-05,
      "loss": 0.8834,
      "step": 12765
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9817278385162354,
      "learning_rate": 1.2462703255279047e-05,
      "loss": 0.7929,
      "step": 12766
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.304265975952148,
      "learning_rate": 1.2462064338881259e-05,
      "loss": 1.194,
      "step": 12767
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9002680778503418,
      "learning_rate": 1.2461425422483469e-05,
      "loss": 0.8841,
      "step": 12768
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.610252380371094,
      "learning_rate": 1.246078650608568e-05,
      "loss": 1.3288,
      "step": 12769
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.56862473487854,
      "learning_rate": 1.246014758968789e-05,
      "loss": 0.6705,
      "step": 12770
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.1294267177581787,
      "learning_rate": 1.2459508673290101e-05,
      "loss": 0.6853,
      "step": 12771
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.063683271408081,
      "learning_rate": 1.2458869756892311e-05,
      "loss": 0.7741,
      "step": 12772
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.668927788734436,
      "learning_rate": 1.2458230840494523e-05,
      "loss": 0.9613,
      "step": 12773
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2128958702087402,
      "learning_rate": 1.2457591924096733e-05,
      "loss": 0.9381,
      "step": 12774
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7140324115753174,
      "learning_rate": 1.2456953007698944e-05,
      "loss": 0.6241,
      "step": 12775
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9435285329818726,
      "learning_rate": 1.2456314091301154e-05,
      "loss": 1.0585,
      "step": 12776
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0652902126312256,
      "learning_rate": 1.2455675174903366e-05,
      "loss": 0.8395,
      "step": 12777
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9905072450637817,
      "learning_rate": 1.2455036258505576e-05,
      "loss": 0.8582,
      "step": 12778
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1821489334106445,
      "learning_rate": 1.2454397342107787e-05,
      "loss": 0.6751,
      "step": 12779
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8184341192245483,
      "learning_rate": 1.2453758425709997e-05,
      "loss": 0.9132,
      "step": 12780
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7376642227172852,
      "learning_rate": 1.2453119509312208e-05,
      "loss": 0.9118,
      "step": 12781
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7563718557357788,
      "learning_rate": 1.2452480592914418e-05,
      "loss": 1.1721,
      "step": 12782
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8817138671875,
      "learning_rate": 1.245184167651663e-05,
      "loss": 1.2907,
      "step": 12783
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6223100423812866,
      "learning_rate": 1.245120276011884e-05,
      "loss": 0.4154,
      "step": 12784
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.53192138671875,
      "learning_rate": 1.2450563843721051e-05,
      "loss": 0.8122,
      "step": 12785
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8124587535858154,
      "learning_rate": 1.2449924927323261e-05,
      "loss": 1.2654,
      "step": 12786
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.875922441482544,
      "learning_rate": 1.2449286010925473e-05,
      "loss": 0.9359,
      "step": 12787
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.951640248298645,
      "learning_rate": 1.244864709452768e-05,
      "loss": 0.9334,
      "step": 12788
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6476998329162598,
      "learning_rate": 1.2448008178129894e-05,
      "loss": 0.9861,
      "step": 12789
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9413836002349854,
      "learning_rate": 1.2447369261732102e-05,
      "loss": 0.525,
      "step": 12790
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5441559553146362,
      "learning_rate": 1.2446730345334315e-05,
      "loss": 0.7923,
      "step": 12791
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.3616127967834473,
      "learning_rate": 1.2446091428936524e-05,
      "loss": 0.91,
      "step": 12792
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5676840543746948,
      "learning_rate": 1.2445452512538735e-05,
      "loss": 0.888,
      "step": 12793
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.6472349166870117,
      "learning_rate": 1.2444813596140945e-05,
      "loss": 1.2025,
      "step": 12794
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.651803970336914,
      "learning_rate": 1.2444174679743157e-05,
      "loss": 0.4593,
      "step": 12795
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9500970840454102,
      "learning_rate": 1.2443535763345366e-05,
      "loss": 0.9573,
      "step": 12796
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.3090460300445557,
      "learning_rate": 1.2442896846947578e-05,
      "loss": 1.0112,
      "step": 12797
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5203001499176025,
      "learning_rate": 1.2442257930549788e-05,
      "loss": 0.7285,
      "step": 12798
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2807509899139404,
      "learning_rate": 1.2441619014152e-05,
      "loss": 0.9818,
      "step": 12799
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6366922855377197,
      "learning_rate": 1.244098009775421e-05,
      "loss": 0.6606,
      "step": 12800
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9524317979812622,
      "learning_rate": 1.244034118135642e-05,
      "loss": 0.9213,
      "step": 12801
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.6826162338256836,
      "learning_rate": 1.243970226495863e-05,
      "loss": 1.3421,
      "step": 12802
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.675188660621643,
      "learning_rate": 1.2439063348560842e-05,
      "loss": 0.9957,
      "step": 12803
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6584879159927368,
      "learning_rate": 1.2438424432163052e-05,
      "loss": 0.9854,
      "step": 12804
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.212299346923828,
      "learning_rate": 1.2437785515765264e-05,
      "loss": 1.0456,
      "step": 12805
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7181733846664429,
      "learning_rate": 1.2437146599367473e-05,
      "loss": 0.9683,
      "step": 12806
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4064157009124756,
      "learning_rate": 1.2436507682969685e-05,
      "loss": 0.9887,
      "step": 12807
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6985210180282593,
      "learning_rate": 1.2435868766571895e-05,
      "loss": 0.972,
      "step": 12808
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6939094066619873,
      "learning_rate": 1.2435229850174106e-05,
      "loss": 0.9718,
      "step": 12809
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8014618158340454,
      "learning_rate": 1.2434590933776316e-05,
      "loss": 0.9567,
      "step": 12810
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5041366815567017,
      "learning_rate": 1.2433952017378528e-05,
      "loss": 0.8731,
      "step": 12811
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6455442905426025,
      "learning_rate": 1.2433313100980738e-05,
      "loss": 0.9104,
      "step": 12812
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6819779872894287,
      "learning_rate": 1.243267418458295e-05,
      "loss": 0.8615,
      "step": 12813
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.35369610786438,
      "learning_rate": 1.2432035268185159e-05,
      "loss": 1.076,
      "step": 12814
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.903929591178894,
      "learning_rate": 1.243139635178737e-05,
      "loss": 0.8153,
      "step": 12815
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7213283777236938,
      "learning_rate": 1.243075743538958e-05,
      "loss": 0.9437,
      "step": 12816
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.850743293762207,
      "learning_rate": 1.2430118518991792e-05,
      "loss": 0.9489,
      "step": 12817
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.014003038406372,
      "learning_rate": 1.2429479602594002e-05,
      "loss": 0.8895,
      "step": 12818
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7317737340927124,
      "learning_rate": 1.2428840686196213e-05,
      "loss": 0.8543,
      "step": 12819
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9247123003005981,
      "learning_rate": 1.2428201769798423e-05,
      "loss": 1.0428,
      "step": 12820
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.87116813659668,
      "learning_rate": 1.2427562853400635e-05,
      "loss": 1.6292,
      "step": 12821
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.426037073135376,
      "learning_rate": 1.2426923937002845e-05,
      "loss": 1.1735,
      "step": 12822
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.988600730895996,
      "learning_rate": 1.2426285020605056e-05,
      "loss": 0.9839,
      "step": 12823
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7214868068695068,
      "learning_rate": 1.2425646104207264e-05,
      "loss": 0.8835,
      "step": 12824
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.4467475414276123,
      "learning_rate": 1.2425007187809478e-05,
      "loss": 1.089,
      "step": 12825
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4772692918777466,
      "learning_rate": 1.2424368271411686e-05,
      "loss": 0.3855,
      "step": 12826
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7858898639678955,
      "learning_rate": 1.2423729355013899e-05,
      "loss": 0.933,
      "step": 12827
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4625580310821533,
      "learning_rate": 1.2423090438616107e-05,
      "loss": 0.8598,
      "step": 12828
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4073222875595093,
      "learning_rate": 1.2422451522218319e-05,
      "loss": 0.4153,
      "step": 12829
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.4634711742401123,
      "learning_rate": 1.2421812605820529e-05,
      "loss": 1.1913,
      "step": 12830
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7158055305480957,
      "learning_rate": 1.242117368942274e-05,
      "loss": 0.8565,
      "step": 12831
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1505746841430664,
      "learning_rate": 1.242053477302495e-05,
      "loss": 1.0779,
      "step": 12832
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6634676456451416,
      "learning_rate": 1.2419895856627161e-05,
      "loss": 0.8668,
      "step": 12833
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.38077449798584,
      "learning_rate": 1.2419256940229371e-05,
      "loss": 1.1812,
      "step": 12834
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5265449285507202,
      "learning_rate": 1.2418618023831583e-05,
      "loss": 0.8322,
      "step": 12835
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.727464437484741,
      "learning_rate": 1.2417979107433793e-05,
      "loss": 0.9654,
      "step": 12836
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.751494288444519,
      "learning_rate": 1.2417340191036004e-05,
      "loss": 0.8267,
      "step": 12837
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.962936282157898,
      "learning_rate": 1.2416701274638214e-05,
      "loss": 1.2793,
      "step": 12838
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.233238697052002,
      "learning_rate": 1.2416062358240426e-05,
      "loss": 0.9076,
      "step": 12839
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9158048629760742,
      "learning_rate": 1.2415423441842636e-05,
      "loss": 0.8686,
      "step": 12840
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0391952991485596,
      "learning_rate": 1.2414784525444847e-05,
      "loss": 0.9965,
      "step": 12841
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0860228538513184,
      "learning_rate": 1.2414145609047057e-05,
      "loss": 0.9666,
      "step": 12842
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5160776376724243,
      "learning_rate": 1.2413506692649268e-05,
      "loss": 0.7926,
      "step": 12843
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6614128351211548,
      "learning_rate": 1.2412867776251478e-05,
      "loss": 1.0327,
      "step": 12844
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.002971649169922,
      "learning_rate": 1.241222885985369e-05,
      "loss": 1.2134,
      "step": 12845
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1483774185180664,
      "learning_rate": 1.24115899434559e-05,
      "loss": 0.9849,
      "step": 12846
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1387932300567627,
      "learning_rate": 1.2410951027058111e-05,
      "loss": 0.841,
      "step": 12847
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.183038234710693,
      "learning_rate": 1.2410312110660321e-05,
      "loss": 1.3218,
      "step": 12848
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.16706919670105,
      "learning_rate": 1.2409673194262533e-05,
      "loss": 0.9166,
      "step": 12849
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6946810483932495,
      "learning_rate": 1.2409034277864743e-05,
      "loss": 0.933,
      "step": 12850
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6577445268630981,
      "learning_rate": 1.2408395361466954e-05,
      "loss": 0.8253,
      "step": 12851
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.965099573135376,
      "learning_rate": 1.2407756445069164e-05,
      "loss": 1.0164,
      "step": 12852
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.5941720008850098,
      "learning_rate": 1.2407117528671375e-05,
      "loss": 0.8625,
      "step": 12853
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.969860076904297,
      "learning_rate": 1.2406478612273585e-05,
      "loss": 1.393,
      "step": 12854
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7003393173217773,
      "learning_rate": 1.2405839695875797e-05,
      "loss": 1.0705,
      "step": 12855
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7003463506698608,
      "learning_rate": 1.2405200779478007e-05,
      "loss": 0.9268,
      "step": 12856
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6770715713500977,
      "learning_rate": 1.2404561863080218e-05,
      "loss": 0.9589,
      "step": 12857
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4137120246887207,
      "learning_rate": 1.2403922946682428e-05,
      "loss": 0.8281,
      "step": 12858
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.62447190284729,
      "learning_rate": 1.240328403028464e-05,
      "loss": 1.0542,
      "step": 12859
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.801358699798584,
      "learning_rate": 1.2402645113886848e-05,
      "loss": 0.9505,
      "step": 12860
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8995009660720825,
      "learning_rate": 1.2402006197489061e-05,
      "loss": 1.151,
      "step": 12861
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.944484233856201,
      "learning_rate": 1.240136728109127e-05,
      "loss": 1.2724,
      "step": 12862
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6502593755722046,
      "learning_rate": 1.2400728364693482e-05,
      "loss": 0.893,
      "step": 12863
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.549618721008301,
      "learning_rate": 1.240008944829569e-05,
      "loss": 1.0333,
      "step": 12864
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4437925815582275,
      "learning_rate": 1.2399450531897902e-05,
      "loss": 0.8416,
      "step": 12865
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.072605609893799,
      "learning_rate": 1.2398811615500112e-05,
      "loss": 0.8485,
      "step": 12866
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5540452003479004,
      "learning_rate": 1.2398172699102324e-05,
      "loss": 0.9167,
      "step": 12867
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8611862659454346,
      "learning_rate": 1.2397533782704533e-05,
      "loss": 0.8355,
      "step": 12868
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0276007652282715,
      "learning_rate": 1.2396894866306745e-05,
      "loss": 0.8483,
      "step": 12869
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5083186626434326,
      "learning_rate": 1.2396255949908955e-05,
      "loss": 0.9344,
      "step": 12870
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6975429058074951,
      "learning_rate": 1.2395617033511166e-05,
      "loss": 0.8374,
      "step": 12871
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6808676719665527,
      "learning_rate": 1.2394978117113376e-05,
      "loss": 0.9516,
      "step": 12872
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6134575605392456,
      "learning_rate": 1.2394339200715588e-05,
      "loss": 0.8909,
      "step": 12873
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.605062007904053,
      "learning_rate": 1.2393700284317798e-05,
      "loss": 1.0571,
      "step": 12874
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.7768659591674805,
      "learning_rate": 1.239306136792001e-05,
      "loss": 1.1838,
      "step": 12875
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.9765372276306152,
      "learning_rate": 1.2392422451522219e-05,
      "loss": 0.8374,
      "step": 12876
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2351486682891846,
      "learning_rate": 1.239178353512443e-05,
      "loss": 0.8668,
      "step": 12877
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.24526047706604,
      "learning_rate": 1.239114461872664e-05,
      "loss": 1.0462,
      "step": 12878
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.379411458969116,
      "learning_rate": 1.2390505702328852e-05,
      "loss": 0.9974,
      "step": 12879
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1802642345428467,
      "learning_rate": 1.2389866785931062e-05,
      "loss": 0.9009,
      "step": 12880
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6357803344726562,
      "learning_rate": 1.2389227869533273e-05,
      "loss": 0.5728,
      "step": 12881
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8332560062408447,
      "learning_rate": 1.2388588953135483e-05,
      "loss": 0.9118,
      "step": 12882
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.580399513244629,
      "learning_rate": 1.2387950036737695e-05,
      "loss": 0.3819,
      "step": 12883
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.462045907974243,
      "learning_rate": 1.2387311120339905e-05,
      "loss": 0.9085,
      "step": 12884
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3844420909881592,
      "learning_rate": 1.2386672203942116e-05,
      "loss": 0.8532,
      "step": 12885
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.067425489425659,
      "learning_rate": 1.2386033287544326e-05,
      "loss": 0.8493,
      "step": 12886
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1515896320343018,
      "learning_rate": 1.2385394371146538e-05,
      "loss": 0.8669,
      "step": 12887
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.413327693939209,
      "learning_rate": 1.2384755454748747e-05,
      "loss": 0.8345,
      "step": 12888
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.92875075340271,
      "learning_rate": 1.2384116538350959e-05,
      "loss": 1.4991,
      "step": 12889
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8101348876953125,
      "learning_rate": 1.2383477621953169e-05,
      "loss": 1.1058,
      "step": 12890
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6882364749908447,
      "learning_rate": 1.238283870555538e-05,
      "loss": 0.8311,
      "step": 12891
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6042592525482178,
      "learning_rate": 1.238219978915759e-05,
      "loss": 1.1826,
      "step": 12892
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7045518159866333,
      "learning_rate": 1.2381560872759802e-05,
      "loss": 0.396,
      "step": 12893
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5282706022262573,
      "learning_rate": 1.2380921956362012e-05,
      "loss": 0.3808,
      "step": 12894
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.332470178604126,
      "learning_rate": 1.2380283039964223e-05,
      "loss": 0.815,
      "step": 12895
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7929073572158813,
      "learning_rate": 1.2379644123566431e-05,
      "loss": 0.821,
      "step": 12896
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.826159954071045,
      "learning_rate": 1.2379005207168645e-05,
      "loss": 0.9611,
      "step": 12897
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6214643716812134,
      "learning_rate": 1.2378366290770853e-05,
      "loss": 0.9722,
      "step": 12898
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.1846532821655273,
      "learning_rate": 1.2377727374373066e-05,
      "loss": 1.0651,
      "step": 12899
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.658127784729004,
      "learning_rate": 1.2377088457975274e-05,
      "loss": 0.8777,
      "step": 12900
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8603668212890625,
      "learning_rate": 1.2376449541577487e-05,
      "loss": 1.1403,
      "step": 12901
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.113517999649048,
      "learning_rate": 1.2375810625179696e-05,
      "loss": 1.071,
      "step": 12902
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8069684505462646,
      "learning_rate": 1.2375171708781907e-05,
      "loss": 0.6197,
      "step": 12903
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.2592692375183105,
      "learning_rate": 1.2374532792384117e-05,
      "loss": 1.4332,
      "step": 12904
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9934812784194946,
      "learning_rate": 1.2373893875986329e-05,
      "loss": 0.945,
      "step": 12905
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.782056450843811,
      "learning_rate": 1.2373254959588538e-05,
      "loss": 0.9454,
      "step": 12906
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.9358575344085693,
      "learning_rate": 1.237261604319075e-05,
      "loss": 0.9001,
      "step": 12907
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.386629343032837,
      "learning_rate": 1.237197712679296e-05,
      "loss": 0.865,
      "step": 12908
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.728744387626648,
      "learning_rate": 1.2371338210395171e-05,
      "loss": 0.9585,
      "step": 12909
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.394631862640381,
      "learning_rate": 1.2370699293997381e-05,
      "loss": 0.4074,
      "step": 12910
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4408897161483765,
      "learning_rate": 1.2370060377599593e-05,
      "loss": 0.8339,
      "step": 12911
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.006666898727417,
      "learning_rate": 1.2369421461201803e-05,
      "loss": 0.9386,
      "step": 12912
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.867082953453064,
      "learning_rate": 1.2368782544804014e-05,
      "loss": 0.8292,
      "step": 12913
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.212883234024048,
      "learning_rate": 1.2368143628406224e-05,
      "loss": 1.093,
      "step": 12914
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5660685300827026,
      "learning_rate": 1.2367504712008436e-05,
      "loss": 0.9396,
      "step": 12915
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.090182065963745,
      "learning_rate": 1.2366865795610645e-05,
      "loss": 1.0609,
      "step": 12916
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.9456300735473633,
      "learning_rate": 1.2366226879212857e-05,
      "loss": 1.1608,
      "step": 12917
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.947461128234863,
      "learning_rate": 1.2365587962815067e-05,
      "loss": 1.375,
      "step": 12918
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3932533264160156,
      "learning_rate": 1.2364949046417278e-05,
      "loss": 0.3925,
      "step": 12919
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5051448345184326,
      "learning_rate": 1.2364310130019488e-05,
      "loss": 0.9308,
      "step": 12920
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.654913067817688,
      "learning_rate": 1.23636712136217e-05,
      "loss": 0.9639,
      "step": 12921
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9346215724945068,
      "learning_rate": 1.236303229722391e-05,
      "loss": 1.0747,
      "step": 12922
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.222416400909424,
      "learning_rate": 1.2362393380826121e-05,
      "loss": 1.8215,
      "step": 12923
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7771345376968384,
      "learning_rate": 1.2361754464428331e-05,
      "loss": 0.9819,
      "step": 12924
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3891180753707886,
      "learning_rate": 1.2361115548030543e-05,
      "loss": 0.8186,
      "step": 12925
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4841980934143066,
      "learning_rate": 1.2360476631632752e-05,
      "loss": 0.9487,
      "step": 12926
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7672122716903687,
      "learning_rate": 1.2359837715234964e-05,
      "loss": 0.9043,
      "step": 12927
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.758098125457764,
      "learning_rate": 1.2359198798837174e-05,
      "loss": 1.2449,
      "step": 12928
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7248642444610596,
      "learning_rate": 1.2358559882439385e-05,
      "loss": 0.8954,
      "step": 12929
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9860731363296509,
      "learning_rate": 1.2357920966041595e-05,
      "loss": 0.9265,
      "step": 12930
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2016875743865967,
      "learning_rate": 1.2357282049643807e-05,
      "loss": 0.9249,
      "step": 12931
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.593507170677185,
      "learning_rate": 1.2356643133246017e-05,
      "loss": 0.8691,
      "step": 12932
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.8177876472473145,
      "learning_rate": 1.2356004216848228e-05,
      "loss": 1.3403,
      "step": 12933
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.1433188915252686,
      "learning_rate": 1.2355365300450436e-05,
      "loss": 1.0864,
      "step": 12934
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7992658615112305,
      "learning_rate": 1.235472638405265e-05,
      "loss": 1.171,
      "step": 12935
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5401875972747803,
      "learning_rate": 1.2354087467654858e-05,
      "loss": 0.9071,
      "step": 12936
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.963144063949585,
      "learning_rate": 1.2353448551257071e-05,
      "loss": 0.8826,
      "step": 12937
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8963743448257446,
      "learning_rate": 1.2352809634859279e-05,
      "loss": 0.9218,
      "step": 12938
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1406362056732178,
      "learning_rate": 1.235217071846149e-05,
      "loss": 1.3111,
      "step": 12939
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5624240636825562,
      "learning_rate": 1.23515318020637e-05,
      "loss": 0.8368,
      "step": 12940
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.068795680999756,
      "learning_rate": 1.2350892885665912e-05,
      "loss": 0.843,
      "step": 12941
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2857775688171387,
      "learning_rate": 1.2350253969268122e-05,
      "loss": 0.9394,
      "step": 12942
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0701451301574707,
      "learning_rate": 1.2349615052870333e-05,
      "loss": 0.7447,
      "step": 12943
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7703838348388672,
      "learning_rate": 1.2348976136472543e-05,
      "loss": 0.9463,
      "step": 12944
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6297552585601807,
      "learning_rate": 1.2348337220074755e-05,
      "loss": 0.9852,
      "step": 12945
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7505146265029907,
      "learning_rate": 1.2347698303676965e-05,
      "loss": 1.0127,
      "step": 12946
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1798417568206787,
      "learning_rate": 1.2347059387279176e-05,
      "loss": 1.0432,
      "step": 12947
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0476882457733154,
      "learning_rate": 1.2346420470881386e-05,
      "loss": 1.0183,
      "step": 12948
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.139754295349121,
      "learning_rate": 1.2345781554483598e-05,
      "loss": 0.9116,
      "step": 12949
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7388814687728882,
      "learning_rate": 1.2345142638085807e-05,
      "loss": 0.9563,
      "step": 12950
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.059478521347046,
      "learning_rate": 1.2344503721688017e-05,
      "loss": 0.9715,
      "step": 12951
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6522741317749023,
      "learning_rate": 1.2343864805290229e-05,
      "loss": 0.8705,
      "step": 12952
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7456905841827393,
      "learning_rate": 1.2343225888892439e-05,
      "loss": 0.9168,
      "step": 12953
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8718327283859253,
      "learning_rate": 1.234258697249465e-05,
      "loss": 0.8685,
      "step": 12954
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.154707670211792,
      "learning_rate": 1.234194805609686e-05,
      "loss": 0.7703,
      "step": 12955
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.511270523071289,
      "learning_rate": 1.2341309139699072e-05,
      "loss": 0.8832,
      "step": 12956
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.24944806098938,
      "learning_rate": 1.2340670223301282e-05,
      "loss": 0.9173,
      "step": 12957
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9793564081192017,
      "learning_rate": 1.2340031306903493e-05,
      "loss": 0.5487,
      "step": 12958
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.9720282554626465,
      "learning_rate": 1.2339392390505703e-05,
      "loss": 1.6303,
      "step": 12959
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.04091739654541,
      "learning_rate": 1.2338753474107914e-05,
      "loss": 1.0237,
      "step": 12960
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0968499183654785,
      "learning_rate": 1.2338114557710124e-05,
      "loss": 0.7211,
      "step": 12961
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.071835517883301,
      "learning_rate": 1.2337475641312336e-05,
      "loss": 0.9258,
      "step": 12962
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5645265579223633,
      "learning_rate": 1.2336836724914546e-05,
      "loss": 0.9064,
      "step": 12963
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9225867986679077,
      "learning_rate": 1.2336197808516757e-05,
      "loss": 0.9796,
      "step": 12964
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.570369005203247,
      "learning_rate": 1.2335558892118965e-05,
      "loss": 0.5685,
      "step": 12965
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.176539182662964,
      "learning_rate": 1.2334919975721179e-05,
      "loss": 0.9652,
      "step": 12966
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.26038122177124,
      "learning_rate": 1.2334281059323387e-05,
      "loss": 1.6069,
      "step": 12967
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8564039468765259,
      "learning_rate": 1.23336421429256e-05,
      "loss": 0.8809,
      "step": 12968
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.152256488800049,
      "learning_rate": 1.2333003226527808e-05,
      "loss": 1.1662,
      "step": 12969
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1761770248413086,
      "learning_rate": 1.233236431013002e-05,
      "loss": 0.815,
      "step": 12970
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8811250925064087,
      "learning_rate": 1.233172539373223e-05,
      "loss": 0.905,
      "step": 12971
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.587638258934021,
      "learning_rate": 1.2331086477334441e-05,
      "loss": 0.8624,
      "step": 12972
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5101661682128906,
      "learning_rate": 1.2330447560936651e-05,
      "loss": 0.7646,
      "step": 12973
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.059460163116455,
      "learning_rate": 1.2329808644538863e-05,
      "loss": 1.1493,
      "step": 12974
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7340667247772217,
      "learning_rate": 1.2329169728141072e-05,
      "loss": 0.9481,
      "step": 12975
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.3593947887420654,
      "learning_rate": 1.2328530811743284e-05,
      "loss": 1.0431,
      "step": 12976
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7876006364822388,
      "learning_rate": 1.2327891895345494e-05,
      "loss": 0.8454,
      "step": 12977
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7249901294708252,
      "learning_rate": 1.2327252978947705e-05,
      "loss": 0.9064,
      "step": 12978
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.674725532531738,
      "learning_rate": 1.2326614062549915e-05,
      "loss": 1.2771,
      "step": 12979
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.84266996383667,
      "learning_rate": 1.2325975146152127e-05,
      "loss": 0.8979,
      "step": 12980
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6619360446929932,
      "learning_rate": 1.2325336229754337e-05,
      "loss": 0.8563,
      "step": 12981
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6831971406936646,
      "learning_rate": 1.2324697313356548e-05,
      "loss": 0.8412,
      "step": 12982
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6489976644515991,
      "learning_rate": 1.2324058396958758e-05,
      "loss": 0.7887,
      "step": 12983
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5891882181167603,
      "learning_rate": 1.232341948056097e-05,
      "loss": 0.8463,
      "step": 12984
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.5066497325897217,
      "learning_rate": 1.232278056416318e-05,
      "loss": 1.2841,
      "step": 12985
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.4060864448547363,
      "learning_rate": 1.2322141647765391e-05,
      "loss": 1.0323,
      "step": 12986
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9615981578826904,
      "learning_rate": 1.23215027313676e-05,
      "loss": 0.8925,
      "step": 12987
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.10486102104187,
      "learning_rate": 1.2320863814969812e-05,
      "loss": 0.9712,
      "step": 12988
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8081485033035278,
      "learning_rate": 1.2320224898572022e-05,
      "loss": 0.7404,
      "step": 12989
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.12995982170105,
      "learning_rate": 1.2319585982174234e-05,
      "loss": 0.9349,
      "step": 12990
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.299459457397461,
      "learning_rate": 1.2318947065776444e-05,
      "loss": 0.8981,
      "step": 12991
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.511040687561035,
      "learning_rate": 1.2318308149378655e-05,
      "loss": 1.1219,
      "step": 12992
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.015839099884033,
      "learning_rate": 1.2317669232980865e-05,
      "loss": 0.9108,
      "step": 12993
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.776403546333313,
      "learning_rate": 1.2317030316583077e-05,
      "loss": 0.8396,
      "step": 12994
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.728687763214111,
      "learning_rate": 1.2316391400185286e-05,
      "loss": 1.3019,
      "step": 12995
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.0749125480651855,
      "learning_rate": 1.2315752483787498e-05,
      "loss": 1.2918,
      "step": 12996
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5339893102645874,
      "learning_rate": 1.2315113567389708e-05,
      "loss": 0.812,
      "step": 12997
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.654159665107727,
      "learning_rate": 1.231447465099192e-05,
      "loss": 0.8362,
      "step": 12998
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7904027700424194,
      "learning_rate": 1.231383573459413e-05,
      "loss": 0.943,
      "step": 12999
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5752339363098145,
      "learning_rate": 1.231319681819634e-05,
      "loss": 0.8225,
      "step": 13000
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9971122741699219,
      "learning_rate": 1.2312557901798549e-05,
      "loss": 0.8074,
      "step": 13001
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.125556468963623,
      "learning_rate": 1.2311918985400762e-05,
      "loss": 0.7471,
      "step": 13002
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5194944143295288,
      "learning_rate": 1.231128006900297e-05,
      "loss": 0.8682,
      "step": 13003
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.062089204788208,
      "learning_rate": 1.2310641152605184e-05,
      "loss": 0.7447,
      "step": 13004
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2798502445220947,
      "learning_rate": 1.2310002236207392e-05,
      "loss": 0.7805,
      "step": 13005
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8575234413146973,
      "learning_rate": 1.2309363319809603e-05,
      "loss": 1.6083,
      "step": 13006
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.729602336883545,
      "learning_rate": 1.2308724403411813e-05,
      "loss": 1.5008,
      "step": 13007
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6762804985046387,
      "learning_rate": 1.2308085487014025e-05,
      "loss": 0.847,
      "step": 13008
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7208071947097778,
      "learning_rate": 1.2307446570616235e-05,
      "loss": 0.8628,
      "step": 13009
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.913527011871338,
      "learning_rate": 1.2306807654218446e-05,
      "loss": 1.2082,
      "step": 13010
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6591848134994507,
      "learning_rate": 1.2306168737820656e-05,
      "loss": 0.9193,
      "step": 13011
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.521482467651367,
      "learning_rate": 1.2305529821422867e-05,
      "loss": 0.8583,
      "step": 13012
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2730205059051514,
      "learning_rate": 1.2304890905025077e-05,
      "loss": 1.1186,
      "step": 13013
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5615057945251465,
      "learning_rate": 1.2304251988627289e-05,
      "loss": 1.0498,
      "step": 13014
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.048600196838379,
      "learning_rate": 1.2303613072229499e-05,
      "loss": 0.8505,
      "step": 13015
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8161035776138306,
      "learning_rate": 1.230297415583171e-05,
      "loss": 0.8175,
      "step": 13016
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6457992792129517,
      "learning_rate": 1.230233523943392e-05,
      "loss": 0.3732,
      "step": 13017
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7296704053878784,
      "learning_rate": 1.2301696323036132e-05,
      "loss": 0.8317,
      "step": 13018
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6765979528427124,
      "learning_rate": 1.2301057406638342e-05,
      "loss": 0.9136,
      "step": 13019
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.371997356414795,
      "learning_rate": 1.2300418490240553e-05,
      "loss": 1.181,
      "step": 13020
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.729280471801758,
      "learning_rate": 1.2299779573842763e-05,
      "loss": 1.0775,
      "step": 13021
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9828628301620483,
      "learning_rate": 1.2299140657444974e-05,
      "loss": 1.0244,
      "step": 13022
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8885293006896973,
      "learning_rate": 1.2298501741047184e-05,
      "loss": 1.2162,
      "step": 13023
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3657633066177368,
      "learning_rate": 1.2297862824649396e-05,
      "loss": 0.8991,
      "step": 13024
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.100614309310913,
      "learning_rate": 1.2297223908251606e-05,
      "loss": 0.8394,
      "step": 13025
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.3531644344329834,
      "learning_rate": 1.2296584991853817e-05,
      "loss": 0.9703,
      "step": 13026
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.2039999961853027,
      "learning_rate": 1.2295946075456027e-05,
      "loss": 0.9435,
      "step": 13027
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.113316059112549,
      "learning_rate": 1.2295307159058239e-05,
      "loss": 1.0297,
      "step": 13028
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.697801947593689,
      "learning_rate": 1.2294668242660449e-05,
      "loss": 0.8356,
      "step": 13029
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6471816301345825,
      "learning_rate": 1.229402932626266e-05,
      "loss": 0.8857,
      "step": 13030
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.347407341003418,
      "learning_rate": 1.229339040986487e-05,
      "loss": 1.6816,
      "step": 13031
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6820354461669922,
      "learning_rate": 1.2292751493467081e-05,
      "loss": 0.9033,
      "step": 13032
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1632213592529297,
      "learning_rate": 1.2292112577069291e-05,
      "loss": 0.8742,
      "step": 13033
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.490932822227478,
      "learning_rate": 1.2291473660671503e-05,
      "loss": 0.8645,
      "step": 13034
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.007333755493164,
      "learning_rate": 1.2290834744273713e-05,
      "loss": 1.2783,
      "step": 13035
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.800122618675232,
      "learning_rate": 1.2290195827875924e-05,
      "loss": 1.1027,
      "step": 13036
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8839672803878784,
      "learning_rate": 1.2289556911478132e-05,
      "loss": 0.855,
      "step": 13037
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.021589279174805,
      "learning_rate": 1.2288917995080346e-05,
      "loss": 1.4488,
      "step": 13038
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5966144800186157,
      "learning_rate": 1.2288279078682554e-05,
      "loss": 0.8913,
      "step": 13039
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9712841510772705,
      "learning_rate": 1.2287640162284767e-05,
      "loss": 0.7679,
      "step": 13040
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.159715175628662,
      "learning_rate": 1.2287001245886975e-05,
      "loss": 1.187,
      "step": 13041
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9562691450119019,
      "learning_rate": 1.2286362329489187e-05,
      "loss": 0.9721,
      "step": 13042
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.059119462966919,
      "learning_rate": 1.2285723413091397e-05,
      "loss": 1.0737,
      "step": 13043
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5581910610198975,
      "learning_rate": 1.2285084496693608e-05,
      "loss": 1.1276,
      "step": 13044
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1130716800689697,
      "learning_rate": 1.2284445580295818e-05,
      "loss": 0.5388,
      "step": 13045
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.3070480823516846,
      "learning_rate": 1.228380666389803e-05,
      "loss": 1.0562,
      "step": 13046
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.436941146850586,
      "learning_rate": 1.228316774750024e-05,
      "loss": 0.7331,
      "step": 13047
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.35026741027832,
      "learning_rate": 1.2282528831102451e-05,
      "loss": 1.6243,
      "step": 13048
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7078760862350464,
      "learning_rate": 1.228188991470466e-05,
      "loss": 0.8685,
      "step": 13049
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.055332899093628,
      "learning_rate": 1.2281250998306872e-05,
      "loss": 0.957,
      "step": 13050
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.943593978881836,
      "learning_rate": 1.2280612081909082e-05,
      "loss": 1.4342,
      "step": 13051
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8948578834533691,
      "learning_rate": 1.2279973165511294e-05,
      "loss": 0.9166,
      "step": 13052
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.9852046966552734,
      "learning_rate": 1.2279334249113504e-05,
      "loss": 0.8521,
      "step": 13053
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.745776891708374,
      "learning_rate": 1.2278695332715715e-05,
      "loss": 0.9549,
      "step": 13054
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6438723802566528,
      "learning_rate": 1.2278056416317925e-05,
      "loss": 0.9335,
      "step": 13055
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9856408834457397,
      "learning_rate": 1.2277417499920137e-05,
      "loss": 0.9359,
      "step": 13056
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.010788917541504,
      "learning_rate": 1.2276778583522346e-05,
      "loss": 1.0074,
      "step": 13057
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2160959243774414,
      "learning_rate": 1.2276139667124558e-05,
      "loss": 0.9104,
      "step": 13058
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.966389536857605,
      "learning_rate": 1.2275500750726768e-05,
      "loss": 0.9887,
      "step": 13059
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5211588144302368,
      "learning_rate": 1.227486183432898e-05,
      "loss": 0.8532,
      "step": 13060
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.468356609344482,
      "learning_rate": 1.227422291793119e-05,
      "loss": 1.2618,
      "step": 13061
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8213539123535156,
      "learning_rate": 1.22735840015334e-05,
      "loss": 0.9297,
      "step": 13062
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0949833393096924,
      "learning_rate": 1.227294508513561e-05,
      "loss": 1.1844,
      "step": 13063
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.183213472366333,
      "learning_rate": 1.2272306168737822e-05,
      "loss": 0.8764,
      "step": 13064
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0395925045013428,
      "learning_rate": 1.2271667252340032e-05,
      "loss": 0.9332,
      "step": 13065
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.3660478591918945,
      "learning_rate": 1.2271028335942244e-05,
      "loss": 0.9107,
      "step": 13066
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.883944034576416,
      "learning_rate": 1.2270389419544453e-05,
      "loss": 0.8244,
      "step": 13067
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.1473517417907715,
      "learning_rate": 1.2269750503146665e-05,
      "loss": 1.3597,
      "step": 13068
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4874212741851807,
      "learning_rate": 1.2269111586748875e-05,
      "loss": 0.4605,
      "step": 13069
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9831985235214233,
      "learning_rate": 1.2268472670351086e-05,
      "loss": 0.8731,
      "step": 13070
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8090169429779053,
      "learning_rate": 1.2267833753953296e-05,
      "loss": 1.2224,
      "step": 13071
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.9989421367645264,
      "learning_rate": 1.2267194837555508e-05,
      "loss": 1.1431,
      "step": 13072
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.211723566055298,
      "learning_rate": 1.2266555921157716e-05,
      "loss": 1.2175,
      "step": 13073
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9389793872833252,
      "learning_rate": 1.226591700475993e-05,
      "loss": 0.9663,
      "step": 13074
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8772850036621094,
      "learning_rate": 1.2265278088362137e-05,
      "loss": 0.9453,
      "step": 13075
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0895514488220215,
      "learning_rate": 1.226463917196435e-05,
      "loss": 1.0326,
      "step": 13076
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1851606369018555,
      "learning_rate": 1.2264000255566559e-05,
      "loss": 1.012,
      "step": 13077
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.383420705795288,
      "learning_rate": 1.2263361339168772e-05,
      "loss": 0.9907,
      "step": 13078
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0797905921936035,
      "learning_rate": 1.226272242277098e-05,
      "loss": 0.9324,
      "step": 13079
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3494699001312256,
      "learning_rate": 1.2262083506373192e-05,
      "loss": 0.8368,
      "step": 13080
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.135662078857422,
      "learning_rate": 1.2261444589975402e-05,
      "loss": 1.4004,
      "step": 13081
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5504683256149292,
      "learning_rate": 1.2260805673577613e-05,
      "loss": 0.9946,
      "step": 13082
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6503474712371826,
      "learning_rate": 1.2260166757179823e-05,
      "loss": 0.7191,
      "step": 13083
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.137383460998535,
      "learning_rate": 1.2259527840782034e-05,
      "loss": 1.1976,
      "step": 13084
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9869791269302368,
      "learning_rate": 1.2258888924384244e-05,
      "loss": 0.9847,
      "step": 13085
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7665457725524902,
      "learning_rate": 1.2258250007986456e-05,
      "loss": 0.4106,
      "step": 13086
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5076358318328857,
      "learning_rate": 1.2257611091588666e-05,
      "loss": 0.968,
      "step": 13087
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8795111179351807,
      "learning_rate": 1.2256972175190877e-05,
      "loss": 0.8695,
      "step": 13088
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.119591236114502,
      "learning_rate": 1.2256333258793087e-05,
      "loss": 1.3396,
      "step": 13089
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8158519268035889,
      "learning_rate": 1.2255694342395299e-05,
      "loss": 0.8271,
      "step": 13090
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6738334894180298,
      "learning_rate": 1.2255055425997509e-05,
      "loss": 0.8627,
      "step": 13091
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.879573941230774,
      "learning_rate": 1.225441650959972e-05,
      "loss": 0.9149,
      "step": 13092
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8635225296020508,
      "learning_rate": 1.225377759320193e-05,
      "loss": 0.8637,
      "step": 13093
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9261343479156494,
      "learning_rate": 1.2253138676804141e-05,
      "loss": 0.9277,
      "step": 13094
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8678785562515259,
      "learning_rate": 1.2252499760406351e-05,
      "loss": 0.7871,
      "step": 13095
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.85406231880188,
      "learning_rate": 1.2251860844008563e-05,
      "loss": 0.7016,
      "step": 13096
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7833341360092163,
      "learning_rate": 1.2251221927610773e-05,
      "loss": 0.9974,
      "step": 13097
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1934244632720947,
      "learning_rate": 1.2250583011212984e-05,
      "loss": 0.8177,
      "step": 13098
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8189952373504639,
      "learning_rate": 1.2249944094815194e-05,
      "loss": 0.8491,
      "step": 13099
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2066309452056885,
      "learning_rate": 1.2249305178417406e-05,
      "loss": 0.9003,
      "step": 13100
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7889450788497925,
      "learning_rate": 1.2248666262019616e-05,
      "loss": 0.842,
      "step": 13101
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.718543529510498,
      "learning_rate": 1.2248027345621827e-05,
      "loss": 0.9015,
      "step": 13102
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.972418785095215,
      "learning_rate": 1.2247388429224037e-05,
      "loss": 1.5581,
      "step": 13103
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8349647521972656,
      "learning_rate": 1.2246749512826248e-05,
      "loss": 1.1421,
      "step": 13104
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.564955949783325,
      "learning_rate": 1.2246110596428458e-05,
      "loss": 0.8062,
      "step": 13105
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8946598768234253,
      "learning_rate": 1.224547168003067e-05,
      "loss": 0.9418,
      "step": 13106
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5384665727615356,
      "learning_rate": 1.224483276363288e-05,
      "loss": 0.8885,
      "step": 13107
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0730628967285156,
      "learning_rate": 1.2244193847235091e-05,
      "loss": 0.864,
      "step": 13108
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.213136672973633,
      "learning_rate": 1.2243554930837301e-05,
      "loss": 1.1157,
      "step": 13109
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8695734739303589,
      "learning_rate": 1.2242916014439513e-05,
      "loss": 0.4682,
      "step": 13110
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.8109848499298096,
      "learning_rate": 1.224227709804172e-05,
      "loss": 1.1942,
      "step": 13111
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3381805419921875,
      "learning_rate": 1.2241638181643934e-05,
      "loss": 0.7472,
      "step": 13112
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2040491104125977,
      "learning_rate": 1.2240999265246142e-05,
      "loss": 0.8296,
      "step": 13113
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6490859985351562,
      "learning_rate": 1.2240360348848355e-05,
      "loss": 0.4741,
      "step": 13114
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7354494333267212,
      "learning_rate": 1.2239721432450564e-05,
      "loss": 0.772,
      "step": 13115
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3481000661849976,
      "learning_rate": 1.2239082516052775e-05,
      "loss": 0.8563,
      "step": 13116
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5392297506332397,
      "learning_rate": 1.2238443599654985e-05,
      "loss": 0.897,
      "step": 13117
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6266374588012695,
      "learning_rate": 1.2237804683257197e-05,
      "loss": 0.8063,
      "step": 13118
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0375993251800537,
      "learning_rate": 1.2237165766859406e-05,
      "loss": 0.7136,
      "step": 13119
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8208085298538208,
      "learning_rate": 1.2236526850461618e-05,
      "loss": 0.891,
      "step": 13120
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8224647045135498,
      "learning_rate": 1.2235887934063828e-05,
      "loss": 0.7826,
      "step": 13121
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.864117980003357,
      "learning_rate": 1.223524901766604e-05,
      "loss": 0.9835,
      "step": 13122
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9673670530319214,
      "learning_rate": 1.223461010126825e-05,
      "loss": 0.9512,
      "step": 13123
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.197037935256958,
      "learning_rate": 1.223397118487046e-05,
      "loss": 0.83,
      "step": 13124
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3875283002853394,
      "learning_rate": 1.223333226847267e-05,
      "loss": 0.4477,
      "step": 13125
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.550888180732727,
      "learning_rate": 1.2232693352074882e-05,
      "loss": 0.9271,
      "step": 13126
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.4034135341644287,
      "learning_rate": 1.2232054435677092e-05,
      "loss": 1.4956,
      "step": 13127
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8899648189544678,
      "learning_rate": 1.2231415519279304e-05,
      "loss": 1.1913,
      "step": 13128
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.743515968322754,
      "learning_rate": 1.2230776602881513e-05,
      "loss": 0.8758,
      "step": 13129
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5071600675582886,
      "learning_rate": 1.2230137686483725e-05,
      "loss": 0.8569,
      "step": 13130
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.984382152557373,
      "learning_rate": 1.2229498770085935e-05,
      "loss": 1.0593,
      "step": 13131
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9095218181610107,
      "learning_rate": 1.2228859853688146e-05,
      "loss": 0.9508,
      "step": 13132
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6439327001571655,
      "learning_rate": 1.2228220937290356e-05,
      "loss": 0.9817,
      "step": 13133
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3081107139587402,
      "learning_rate": 1.2227582020892568e-05,
      "loss": 1.0327,
      "step": 13134
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6412612199783325,
      "learning_rate": 1.2226943104494778e-05,
      "loss": 0.8755,
      "step": 13135
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.191013813018799,
      "learning_rate": 1.222630418809699e-05,
      "loss": 1.0131,
      "step": 13136
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7531235218048096,
      "learning_rate": 1.2225665271699199e-05,
      "loss": 0.8359,
      "step": 13137
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.0244364738464355,
      "learning_rate": 1.222502635530141e-05,
      "loss": 1.1023,
      "step": 13138
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9031533002853394,
      "learning_rate": 1.222438743890362e-05,
      "loss": 0.8652,
      "step": 13139
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.115415334701538,
      "learning_rate": 1.2223748522505832e-05,
      "loss": 1.4965,
      "step": 13140
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8411459922790527,
      "learning_rate": 1.2223109606108042e-05,
      "loss": 1.216,
      "step": 13141
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5394192934036255,
      "learning_rate": 1.2222470689710253e-05,
      "loss": 0.819,
      "step": 13142
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.669119119644165,
      "learning_rate": 1.2221831773312463e-05,
      "loss": 1.0442,
      "step": 13143
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.809686183929443,
      "learning_rate": 1.2221192856914675e-05,
      "loss": 0.9279,
      "step": 13144
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.051444053649902,
      "learning_rate": 1.2220553940516885e-05,
      "loss": 0.9857,
      "step": 13145
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.02677845954895,
      "learning_rate": 1.2219915024119096e-05,
      "loss": 1.0459,
      "step": 13146
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7576631307601929,
      "learning_rate": 1.2219276107721304e-05,
      "loss": 0.9222,
      "step": 13147
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7472100257873535,
      "learning_rate": 1.2218637191323518e-05,
      "loss": 0.9303,
      "step": 13148
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6088923215866089,
      "learning_rate": 1.2217998274925726e-05,
      "loss": 0.4346,
      "step": 13149
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3099758625030518,
      "learning_rate": 1.2217359358527939e-05,
      "loss": 0.8296,
      "step": 13150
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9802626371383667,
      "learning_rate": 1.2216720442130147e-05,
      "loss": 0.9679,
      "step": 13151
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8123884201049805,
      "learning_rate": 1.2216081525732359e-05,
      "loss": 0.7952,
      "step": 13152
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5126389265060425,
      "learning_rate": 1.2215442609334569e-05,
      "loss": 0.3967,
      "step": 13153
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6778606176376343,
      "learning_rate": 1.221480369293678e-05,
      "loss": 0.8587,
      "step": 13154
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.003875732421875,
      "learning_rate": 1.221416477653899e-05,
      "loss": 0.8986,
      "step": 13155
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8566044569015503,
      "learning_rate": 1.2213525860141201e-05,
      "loss": 0.9672,
      "step": 13156
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9612616300582886,
      "learning_rate": 1.2212886943743411e-05,
      "loss": 1.1643,
      "step": 13157
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7201659679412842,
      "learning_rate": 1.2212248027345623e-05,
      "loss": 0.9215,
      "step": 13158
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8144716024398804,
      "learning_rate": 1.2211609110947833e-05,
      "loss": 0.6921,
      "step": 13159
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.547993183135986,
      "learning_rate": 1.2210970194550044e-05,
      "loss": 1.1458,
      "step": 13160
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.7569565773010254,
      "learning_rate": 1.2210331278152254e-05,
      "loss": 0.723,
      "step": 13161
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.385334014892578,
      "learning_rate": 1.2209692361754466e-05,
      "loss": 1.0832,
      "step": 13162
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6225672960281372,
      "learning_rate": 1.2209053445356676e-05,
      "loss": 0.8808,
      "step": 13163
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8731005191802979,
      "learning_rate": 1.2208414528958887e-05,
      "loss": 0.8673,
      "step": 13164
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7248661518096924,
      "learning_rate": 1.2207775612561097e-05,
      "loss": 0.8306,
      "step": 13165
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0767881870269775,
      "learning_rate": 1.2207136696163308e-05,
      "loss": 0.5214,
      "step": 13166
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.6294665336608887,
      "learning_rate": 1.2206497779765518e-05,
      "loss": 1.0417,
      "step": 13167
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.244400501251221,
      "learning_rate": 1.220585886336773e-05,
      "loss": 1.0468,
      "step": 13168
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7244133949279785,
      "learning_rate": 1.220521994696994e-05,
      "loss": 1.0038,
      "step": 13169
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8518118858337402,
      "learning_rate": 1.2204581030572151e-05,
      "loss": 0.3943,
      "step": 13170
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7107726335525513,
      "learning_rate": 1.2203942114174361e-05,
      "loss": 0.9408,
      "step": 13171
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.6700077056884766,
      "learning_rate": 1.2203303197776573e-05,
      "loss": 1.5079,
      "step": 13172
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0922815799713135,
      "learning_rate": 1.2202664281378783e-05,
      "loss": 0.9637,
      "step": 13173
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.852456092834473,
      "learning_rate": 1.2202025364980994e-05,
      "loss": 0.9699,
      "step": 13174
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8283979892730713,
      "learning_rate": 1.2201386448583204e-05,
      "loss": 0.8143,
      "step": 13175
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.677298903465271,
      "learning_rate": 1.2200747532185415e-05,
      "loss": 0.7987,
      "step": 13176
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.738047122955322,
      "learning_rate": 1.2200108615787625e-05,
      "loss": 1.8392,
      "step": 13177
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.783157467842102,
      "learning_rate": 1.2199469699389837e-05,
      "loss": 0.7734,
      "step": 13178
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8504819869995117,
      "learning_rate": 1.2198830782992047e-05,
      "loss": 0.9812,
      "step": 13179
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5078357458114624,
      "learning_rate": 1.2198191866594258e-05,
      "loss": 0.829,
      "step": 13180
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.648658275604248,
      "learning_rate": 1.2197552950196468e-05,
      "loss": 0.9616,
      "step": 13181
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.690677642822266,
      "learning_rate": 1.219691403379868e-05,
      "loss": 1.456,
      "step": 13182
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.0036749839782715,
      "learning_rate": 1.2196275117400888e-05,
      "loss": 1.419,
      "step": 13183
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.633451223373413,
      "learning_rate": 1.2195636201003101e-05,
      "loss": 0.902,
      "step": 13184
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.791731595993042,
      "learning_rate": 1.219499728460531e-05,
      "loss": 1.0,
      "step": 13185
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.105440855026245,
      "learning_rate": 1.2194358368207522e-05,
      "loss": 1.0377,
      "step": 13186
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7059088945388794,
      "learning_rate": 1.219371945180973e-05,
      "loss": 0.9146,
      "step": 13187
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9572724103927612,
      "learning_rate": 1.2193080535411942e-05,
      "loss": 0.9873,
      "step": 13188
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.636572003364563,
      "learning_rate": 1.2192441619014152e-05,
      "loss": 0.9405,
      "step": 13189
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6885571479797363,
      "learning_rate": 1.2191802702616364e-05,
      "loss": 0.8259,
      "step": 13190
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.946063995361328,
      "learning_rate": 1.2191163786218573e-05,
      "loss": 1.0193,
      "step": 13191
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.981276512145996,
      "learning_rate": 1.2190524869820785e-05,
      "loss": 1.1471,
      "step": 13192
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.7986409664154053,
      "learning_rate": 1.2189885953422995e-05,
      "loss": 1.1667,
      "step": 13193
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9139097929000854,
      "learning_rate": 1.2189247037025206e-05,
      "loss": 1.0085,
      "step": 13194
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4534730911254883,
      "learning_rate": 1.2188608120627416e-05,
      "loss": 0.4014,
      "step": 13195
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.22556471824646,
      "learning_rate": 1.2187969204229628e-05,
      "loss": 1.1919,
      "step": 13196
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7514418363571167,
      "learning_rate": 1.2187330287831838e-05,
      "loss": 1.0171,
      "step": 13197
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8397817611694336,
      "learning_rate": 1.218669137143405e-05,
      "loss": 0.9217,
      "step": 13198
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6922504901885986,
      "learning_rate": 1.2186052455036259e-05,
      "loss": 0.8988,
      "step": 13199
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.179844379425049,
      "learning_rate": 1.218541353863847e-05,
      "loss": 1.1988,
      "step": 13200
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9257254600524902,
      "learning_rate": 1.218477462224068e-05,
      "loss": 0.9112,
      "step": 13201
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8176482915878296,
      "learning_rate": 1.2184135705842892e-05,
      "loss": 0.9145,
      "step": 13202
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.075136661529541,
      "learning_rate": 1.2183496789445102e-05,
      "loss": 0.7304,
      "step": 13203
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1804215908050537,
      "learning_rate": 1.2182857873047313e-05,
      "loss": 0.9271,
      "step": 13204
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.1936235427856445,
      "learning_rate": 1.2182218956649523e-05,
      "loss": 1.5829,
      "step": 13205
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7415657043457031,
      "learning_rate": 1.2181580040251735e-05,
      "loss": 0.9665,
      "step": 13206
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.299539089202881,
      "learning_rate": 1.2180941123853945e-05,
      "loss": 0.9658,
      "step": 13207
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8916699886322021,
      "learning_rate": 1.2180302207456156e-05,
      "loss": 1.031,
      "step": 13208
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0106217861175537,
      "learning_rate": 1.2179663291058366e-05,
      "loss": 1.2309,
      "step": 13209
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.149495840072632,
      "learning_rate": 1.2179024374660578e-05,
      "loss": 0.9495,
      "step": 13210
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.961303472518921,
      "learning_rate": 1.2178385458262787e-05,
      "loss": 1.1564,
      "step": 13211
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0283901691436768,
      "learning_rate": 1.2177746541864999e-05,
      "loss": 0.9467,
      "step": 13212
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9954986572265625,
      "learning_rate": 1.2177107625467209e-05,
      "loss": 0.7724,
      "step": 13213
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0827207565307617,
      "learning_rate": 1.217646870906942e-05,
      "loss": 1.101,
      "step": 13214
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.099780559539795,
      "learning_rate": 1.217582979267163e-05,
      "loss": 0.8965,
      "step": 13215
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.783059000968933,
      "learning_rate": 1.2175190876273842e-05,
      "loss": 0.8638,
      "step": 13216
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.4662699699401855,
      "learning_rate": 1.2174551959876052e-05,
      "loss": 0.9904,
      "step": 13217
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7132606506347656,
      "learning_rate": 1.2173913043478263e-05,
      "loss": 0.8132,
      "step": 13218
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9450287818908691,
      "learning_rate": 1.2173274127080471e-05,
      "loss": 1.0761,
      "step": 13219
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0954277515411377,
      "learning_rate": 1.2172635210682685e-05,
      "loss": 0.9757,
      "step": 13220
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.210468292236328,
      "learning_rate": 1.2171996294284893e-05,
      "loss": 0.7888,
      "step": 13221
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7024205923080444,
      "learning_rate": 1.2171357377887106e-05,
      "loss": 0.9017,
      "step": 13222
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.431452751159668,
      "learning_rate": 1.2170718461489314e-05,
      "loss": 0.9173,
      "step": 13223
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.3191585540771484,
      "learning_rate": 1.2170079545091527e-05,
      "loss": 0.7782,
      "step": 13224
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.207603931427002,
      "learning_rate": 1.2169440628693736e-05,
      "loss": 1.1968,
      "step": 13225
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7946977615356445,
      "learning_rate": 1.2168801712295947e-05,
      "loss": 0.8316,
      "step": 13226
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8940666913986206,
      "learning_rate": 1.2168162795898157e-05,
      "loss": 0.7968,
      "step": 13227
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2690160274505615,
      "learning_rate": 1.2167523879500368e-05,
      "loss": 0.943,
      "step": 13228
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.557418704032898,
      "learning_rate": 1.2166884963102578e-05,
      "loss": 0.3471,
      "step": 13229
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5234793424606323,
      "learning_rate": 1.216624604670479e-05,
      "loss": 0.9688,
      "step": 13230
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5387308597564697,
      "learning_rate": 1.2165607130307e-05,
      "loss": 0.8513,
      "step": 13231
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8248502016067505,
      "learning_rate": 1.2164968213909211e-05,
      "loss": 0.8891,
      "step": 13232
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.3180489540100098,
      "learning_rate": 1.2164329297511421e-05,
      "loss": 1.1854,
      "step": 13233
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8900346755981445,
      "learning_rate": 1.2163690381113633e-05,
      "loss": 0.7335,
      "step": 13234
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5906765460968018,
      "learning_rate": 1.2163051464715843e-05,
      "loss": 0.4893,
      "step": 13235
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8799901008605957,
      "learning_rate": 1.2162412548318054e-05,
      "loss": 0.9099,
      "step": 13236
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0154850482940674,
      "learning_rate": 1.2161773631920264e-05,
      "loss": 1.0284,
      "step": 13237
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7129968404769897,
      "learning_rate": 1.2161134715522475e-05,
      "loss": 0.9141,
      "step": 13238
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0242605209350586,
      "learning_rate": 1.2160495799124685e-05,
      "loss": 0.9239,
      "step": 13239
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0129506587982178,
      "learning_rate": 1.2159856882726897e-05,
      "loss": 0.8955,
      "step": 13240
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1071553230285645,
      "learning_rate": 1.2159217966329107e-05,
      "loss": 0.8951,
      "step": 13241
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.451317548751831,
      "learning_rate": 1.2158579049931318e-05,
      "loss": 0.5181,
      "step": 13242
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.179595947265625,
      "learning_rate": 1.2157940133533528e-05,
      "loss": 1.5685,
      "step": 13243
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7269258499145508,
      "learning_rate": 1.215730121713574e-05,
      "loss": 1.0098,
      "step": 13244
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.476412534713745,
      "learning_rate": 1.215666230073795e-05,
      "loss": 0.9824,
      "step": 13245
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0044140815734863,
      "learning_rate": 1.2156023384340161e-05,
      "loss": 0.9476,
      "step": 13246
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6340303421020508,
      "learning_rate": 1.2155384467942371e-05,
      "loss": 1.0313,
      "step": 13247
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8675237894058228,
      "learning_rate": 1.2154745551544582e-05,
      "loss": 0.9869,
      "step": 13248
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.079171657562256,
      "learning_rate": 1.2154106635146792e-05,
      "loss": 0.8516,
      "step": 13249
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.658995509147644,
      "learning_rate": 1.2153467718749004e-05,
      "loss": 0.8904,
      "step": 13250
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2168943881988525,
      "learning_rate": 1.2152828802351214e-05,
      "loss": 1.0785,
      "step": 13251
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9323997497558594,
      "learning_rate": 1.2152189885953425e-05,
      "loss": 1.0528,
      "step": 13252
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4888219833374023,
      "learning_rate": 1.2151550969555635e-05,
      "loss": 1.1224,
      "step": 13253
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.726893424987793,
      "learning_rate": 1.2150912053157847e-05,
      "loss": 0.8876,
      "step": 13254
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8918005228042603,
      "learning_rate": 1.2150273136760057e-05,
      "loss": 0.8823,
      "step": 13255
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.632519483566284,
      "learning_rate": 1.2149634220362268e-05,
      "loss": 0.832,
      "step": 13256
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6181179285049438,
      "learning_rate": 1.2148995303964476e-05,
      "loss": 0.8362,
      "step": 13257
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.992048978805542,
      "learning_rate": 1.214835638756669e-05,
      "loss": 0.9452,
      "step": 13258
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6187318563461304,
      "learning_rate": 1.2147717471168898e-05,
      "loss": 0.8387,
      "step": 13259
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.47888445854187,
      "learning_rate": 1.2147078554771111e-05,
      "loss": 0.6381,
      "step": 13260
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8467679023742676,
      "learning_rate": 1.2146439638373319e-05,
      "loss": 0.9345,
      "step": 13261
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4065396785736084,
      "learning_rate": 1.214580072197553e-05,
      "loss": 0.8296,
      "step": 13262
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.889417052268982,
      "learning_rate": 1.214516180557774e-05,
      "loss": 0.8811,
      "step": 13263
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8295130729675293,
      "learning_rate": 1.2144522889179952e-05,
      "loss": 1.0979,
      "step": 13264
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6649320125579834,
      "learning_rate": 1.2143883972782162e-05,
      "loss": 0.8424,
      "step": 13265
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6198467016220093,
      "learning_rate": 1.2143245056384373e-05,
      "loss": 0.8528,
      "step": 13266
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.757673263549805,
      "learning_rate": 1.2142606139986583e-05,
      "loss": 1.7259,
      "step": 13267
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.9714457988739014,
      "learning_rate": 1.2141967223588795e-05,
      "loss": 1.144,
      "step": 13268
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7452101707458496,
      "learning_rate": 1.2141328307191005e-05,
      "loss": 0.9859,
      "step": 13269
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8345528841018677,
      "learning_rate": 1.2140689390793216e-05,
      "loss": 0.844,
      "step": 13270
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7181214094161987,
      "learning_rate": 1.2140050474395426e-05,
      "loss": 0.871,
      "step": 13271
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5966310501098633,
      "learning_rate": 1.2139411557997638e-05,
      "loss": 0.8872,
      "step": 13272
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.304180860519409,
      "learning_rate": 1.2138772641599847e-05,
      "loss": 1.06,
      "step": 13273
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6663075685501099,
      "learning_rate": 1.2138133725202059e-05,
      "loss": 0.8995,
      "step": 13274
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.7606399059295654,
      "learning_rate": 1.2137494808804269e-05,
      "loss": 0.8353,
      "step": 13275
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9974257946014404,
      "learning_rate": 1.213685589240648e-05,
      "loss": 1.0437,
      "step": 13276
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9930413961410522,
      "learning_rate": 1.213621697600869e-05,
      "loss": 0.9606,
      "step": 13277
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7394838333129883,
      "learning_rate": 1.2135578059610902e-05,
      "loss": 0.847,
      "step": 13278
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.321866989135742,
      "learning_rate": 1.2134939143213112e-05,
      "loss": 1.0504,
      "step": 13279
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.670001268386841,
      "learning_rate": 1.2134300226815323e-05,
      "loss": 1.0061,
      "step": 13280
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0108554363250732,
      "learning_rate": 1.2133661310417533e-05,
      "loss": 0.8967,
      "step": 13281
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5684489011764526,
      "learning_rate": 1.2133022394019745e-05,
      "loss": 0.8649,
      "step": 13282
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.830304503440857,
      "learning_rate": 1.2132383477621954e-05,
      "loss": 0.9317,
      "step": 13283
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8067888021469116,
      "learning_rate": 1.2131744561224166e-05,
      "loss": 0.9719,
      "step": 13284
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.62339186668396,
      "learning_rate": 1.2131105644826376e-05,
      "loss": 0.8294,
      "step": 13285
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7016304731369019,
      "learning_rate": 1.2130466728428587e-05,
      "loss": 0.9046,
      "step": 13286
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5812792778015137,
      "learning_rate": 1.2129827812030797e-05,
      "loss": 0.9892,
      "step": 13287
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.562019348144531,
      "learning_rate": 1.2129188895633009e-05,
      "loss": 1.6026,
      "step": 13288
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6886957883834839,
      "learning_rate": 1.2128549979235219e-05,
      "loss": 0.8855,
      "step": 13289
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7865549325942993,
      "learning_rate": 1.212791106283743e-05,
      "loss": 0.9182,
      "step": 13290
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.016882419586182,
      "learning_rate": 1.212727214643964e-05,
      "loss": 0.9792,
      "step": 13291
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5810415744781494,
      "learning_rate": 1.2126633230041852e-05,
      "loss": 0.8288,
      "step": 13292
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2955400943756104,
      "learning_rate": 1.212599431364406e-05,
      "loss": 0.8874,
      "step": 13293
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.81016206741333,
      "learning_rate": 1.2125355397246273e-05,
      "loss": 0.8168,
      "step": 13294
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1721785068511963,
      "learning_rate": 1.2124716480848481e-05,
      "loss": 0.9329,
      "step": 13295
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.223114967346191,
      "learning_rate": 1.2124077564450694e-05,
      "loss": 1.1094,
      "step": 13296
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3765242099761963,
      "learning_rate": 1.2123438648052903e-05,
      "loss": 0.7678,
      "step": 13297
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.729144334793091,
      "learning_rate": 1.2122799731655114e-05,
      "loss": 1.3145,
      "step": 13298
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9735394716262817,
      "learning_rate": 1.2122160815257324e-05,
      "loss": 0.9948,
      "step": 13299
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6692798137664795,
      "learning_rate": 1.2121521898859536e-05,
      "loss": 0.8402,
      "step": 13300
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.133818626403809,
      "learning_rate": 1.2120882982461745e-05,
      "loss": 1.2007,
      "step": 13301
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5052738189697266,
      "learning_rate": 1.2120244066063957e-05,
      "loss": 0.9347,
      "step": 13302
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3469243049621582,
      "learning_rate": 1.2119605149666167e-05,
      "loss": 0.4896,
      "step": 13303
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5288517475128174,
      "learning_rate": 1.2118966233268378e-05,
      "loss": 0.897,
      "step": 13304
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.046967029571533,
      "learning_rate": 1.2118327316870588e-05,
      "loss": 1.1422,
      "step": 13305
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2373085021972656,
      "learning_rate": 1.21176884004728e-05,
      "loss": 1.0661,
      "step": 13306
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6675013303756714,
      "learning_rate": 1.211704948407501e-05,
      "loss": 0.9554,
      "step": 13307
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7288371324539185,
      "learning_rate": 1.2116410567677221e-05,
      "loss": 0.854,
      "step": 13308
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4682257175445557,
      "learning_rate": 1.2115771651279431e-05,
      "loss": 0.4134,
      "step": 13309
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.825305700302124,
      "learning_rate": 1.2115132734881643e-05,
      "loss": 0.8734,
      "step": 13310
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.645984172821045,
      "learning_rate": 1.2114493818483852e-05,
      "loss": 0.8076,
      "step": 13311
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.098602771759033,
      "learning_rate": 1.2113854902086064e-05,
      "loss": 0.7677,
      "step": 13312
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.691880702972412,
      "learning_rate": 1.2113215985688274e-05,
      "loss": 0.767,
      "step": 13313
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.210737705230713,
      "learning_rate": 1.2112577069290485e-05,
      "loss": 1.3175,
      "step": 13314
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6839293241500854,
      "learning_rate": 1.2111938152892695e-05,
      "loss": 0.9385,
      "step": 13315
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2262256145477295,
      "learning_rate": 1.2111299236494907e-05,
      "loss": 0.5534,
      "step": 13316
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.890878438949585,
      "learning_rate": 1.2110660320097117e-05,
      "loss": 1.1709,
      "step": 13317
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8210043907165527,
      "learning_rate": 1.2110021403699328e-05,
      "loss": 0.9103,
      "step": 13318
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.751368522644043,
      "learning_rate": 1.2109382487301538e-05,
      "loss": 1.1543,
      "step": 13319
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3336679935455322,
      "learning_rate": 1.210874357090375e-05,
      "loss": 1.0001,
      "step": 13320
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6875932216644287,
      "learning_rate": 1.210810465450596e-05,
      "loss": 0.8425,
      "step": 13321
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8357222080230713,
      "learning_rate": 1.2107465738108171e-05,
      "loss": 0.9314,
      "step": 13322
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4364306926727295,
      "learning_rate": 1.210682682171038e-05,
      "loss": 0.8276,
      "step": 13323
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1882829666137695,
      "learning_rate": 1.2106187905312592e-05,
      "loss": 0.8835,
      "step": 13324
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.018002033233643,
      "learning_rate": 1.2105548988914802e-05,
      "loss": 1.3388,
      "step": 13325
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9811153411865234,
      "learning_rate": 1.2104910072517014e-05,
      "loss": 0.9875,
      "step": 13326
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.924100637435913,
      "learning_rate": 1.2104271156119224e-05,
      "loss": 1.0169,
      "step": 13327
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5000792741775513,
      "learning_rate": 1.2103632239721435e-05,
      "loss": 0.8629,
      "step": 13328
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2988317012786865,
      "learning_rate": 1.2102993323323643e-05,
      "loss": 0.9335,
      "step": 13329
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9642189741134644,
      "learning_rate": 1.2102354406925857e-05,
      "loss": 0.8847,
      "step": 13330
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.873995304107666,
      "learning_rate": 1.2101715490528065e-05,
      "loss": 0.6695,
      "step": 13331
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.865316390991211,
      "learning_rate": 1.2101076574130278e-05,
      "loss": 0.9336,
      "step": 13332
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.841606378555298,
      "learning_rate": 1.2100437657732486e-05,
      "loss": 1.2481,
      "step": 13333
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.825446605682373,
      "learning_rate": 1.2099798741334698e-05,
      "loss": 0.9543,
      "step": 13334
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.785764217376709,
      "learning_rate": 1.2099159824936907e-05,
      "loss": 1.1659,
      "step": 13335
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4498687982559204,
      "learning_rate": 1.2098520908539117e-05,
      "loss": 0.8613,
      "step": 13336
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.0412163734436035,
      "learning_rate": 1.2097881992141329e-05,
      "loss": 0.847,
      "step": 13337
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.5566673278808594,
      "learning_rate": 1.2097243075743539e-05,
      "loss": 0.9826,
      "step": 13338
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5555977821350098,
      "learning_rate": 1.209660415934575e-05,
      "loss": 0.4475,
      "step": 13339
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.166346788406372,
      "learning_rate": 1.209596524294796e-05,
      "loss": 1.1092,
      "step": 13340
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8494727611541748,
      "learning_rate": 1.2095326326550172e-05,
      "loss": 0.9532,
      "step": 13341
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.751941680908203,
      "learning_rate": 1.2094687410152382e-05,
      "loss": 0.6068,
      "step": 13342
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.884567379951477,
      "learning_rate": 1.2094048493754593e-05,
      "loss": 0.8629,
      "step": 13343
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6915452480316162,
      "learning_rate": 1.2093409577356803e-05,
      "loss": 0.8213,
      "step": 13344
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.819406509399414,
      "learning_rate": 1.2092770660959014e-05,
      "loss": 1.364,
      "step": 13345
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0894360542297363,
      "learning_rate": 1.2092131744561224e-05,
      "loss": 0.775,
      "step": 13346
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.640460729598999,
      "learning_rate": 1.2091492828163436e-05,
      "loss": 0.8665,
      "step": 13347
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8180227279663086,
      "learning_rate": 1.2090853911765646e-05,
      "loss": 0.9778,
      "step": 13348
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0880730152130127,
      "learning_rate": 1.2090214995367857e-05,
      "loss": 0.9105,
      "step": 13349
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.306706666946411,
      "learning_rate": 1.2089576078970067e-05,
      "loss": 0.7188,
      "step": 13350
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.347391128540039,
      "learning_rate": 1.2088937162572279e-05,
      "loss": 1.0132,
      "step": 13351
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2108607292175293,
      "learning_rate": 1.2088298246174489e-05,
      "loss": 0.8661,
      "step": 13352
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6740648746490479,
      "learning_rate": 1.20876593297767e-05,
      "loss": 0.9089,
      "step": 13353
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9023596048355103,
      "learning_rate": 1.208702041337891e-05,
      "loss": 1.2892,
      "step": 13354
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0721068382263184,
      "learning_rate": 1.2086381496981121e-05,
      "loss": 0.78,
      "step": 13355
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9074488878250122,
      "learning_rate": 1.2085742580583331e-05,
      "loss": 1.2265,
      "step": 13356
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4662023782730103,
      "learning_rate": 1.2085103664185543e-05,
      "loss": 0.8548,
      "step": 13357
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0774102210998535,
      "learning_rate": 1.2084464747787753e-05,
      "loss": 1.0079,
      "step": 13358
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8015434741973877,
      "learning_rate": 1.2083825831389964e-05,
      "loss": 0.7894,
      "step": 13359
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.417388677597046,
      "learning_rate": 1.2083186914992172e-05,
      "loss": 1.192,
      "step": 13360
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.855283498764038,
      "learning_rate": 1.2082547998594386e-05,
      "loss": 0.837,
      "step": 13361
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.5592269897460938,
      "learning_rate": 1.2081909082196594e-05,
      "loss": 1.2395,
      "step": 13362
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6768193244934082,
      "learning_rate": 1.2081270165798807e-05,
      "loss": 0.8457,
      "step": 13363
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9309749603271484,
      "learning_rate": 1.2080631249401015e-05,
      "loss": 0.8524,
      "step": 13364
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.939246654510498,
      "learning_rate": 1.2079992333003227e-05,
      "loss": 0.866,
      "step": 13365
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6179054975509644,
      "learning_rate": 1.2079353416605437e-05,
      "loss": 0.8315,
      "step": 13366
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5706195831298828,
      "learning_rate": 1.2078714500207648e-05,
      "loss": 0.8459,
      "step": 13367
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4130775928497314,
      "learning_rate": 1.2078075583809858e-05,
      "loss": 1.1953,
      "step": 13368
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8365625143051147,
      "learning_rate": 1.207743666741207e-05,
      "loss": 0.8816,
      "step": 13369
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5851335525512695,
      "learning_rate": 1.207679775101428e-05,
      "loss": 0.8721,
      "step": 13370
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1748929023742676,
      "learning_rate": 1.2076158834616491e-05,
      "loss": 0.7505,
      "step": 13371
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.783746361732483,
      "learning_rate": 1.20755199182187e-05,
      "loss": 0.4192,
      "step": 13372
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9932897090911865,
      "learning_rate": 1.2074881001820912e-05,
      "loss": 0.9732,
      "step": 13373
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7972551584243774,
      "learning_rate": 1.2074242085423122e-05,
      "loss": 0.9382,
      "step": 13374
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0725958347320557,
      "learning_rate": 1.2073603169025334e-05,
      "loss": 1.2816,
      "step": 13375
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9546061754226685,
      "learning_rate": 1.2072964252627544e-05,
      "loss": 0.8969,
      "step": 13376
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.891783356666565,
      "learning_rate": 1.2072325336229755e-05,
      "loss": 1.003,
      "step": 13377
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8480629920959473,
      "learning_rate": 1.2071686419831965e-05,
      "loss": 1.2256,
      "step": 13378
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6648240089416504,
      "learning_rate": 1.2071047503434177e-05,
      "loss": 0.8239,
      "step": 13379
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7449294328689575,
      "learning_rate": 1.2070408587036386e-05,
      "loss": 0.8529,
      "step": 13380
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8841875791549683,
      "learning_rate": 1.2069769670638598e-05,
      "loss": 0.7789,
      "step": 13381
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.468878149986267,
      "learning_rate": 1.2069130754240808e-05,
      "loss": 0.434,
      "step": 13382
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.741089344024658,
      "learning_rate": 1.206849183784302e-05,
      "loss": 0.8609,
      "step": 13383
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.783539295196533,
      "learning_rate": 1.206785292144523e-05,
      "loss": 1.0339,
      "step": 13384
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.61716365814209,
      "learning_rate": 1.206721400504744e-05,
      "loss": 1.1772,
      "step": 13385
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2565791606903076,
      "learning_rate": 1.206657508864965e-05,
      "loss": 0.7374,
      "step": 13386
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3175472021102905,
      "learning_rate": 1.2065936172251862e-05,
      "loss": 0.8547,
      "step": 13387
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8214199542999268,
      "learning_rate": 1.2065297255854072e-05,
      "loss": 0.8977,
      "step": 13388
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6282057762145996,
      "learning_rate": 1.2064658339456284e-05,
      "loss": 0.8366,
      "step": 13389
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5563091039657593,
      "learning_rate": 1.2064019423058493e-05,
      "loss": 0.9132,
      "step": 13390
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7099823951721191,
      "learning_rate": 1.2063380506660705e-05,
      "loss": 0.874,
      "step": 13391
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9177948236465454,
      "learning_rate": 1.2062741590262915e-05,
      "loss": 1.1272,
      "step": 13392
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.708631992340088,
      "learning_rate": 1.2062102673865126e-05,
      "loss": 0.979,
      "step": 13393
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.703598141670227,
      "learning_rate": 1.2061463757467336e-05,
      "loss": 0.8325,
      "step": 13394
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.5875964164733887,
      "learning_rate": 1.2060824841069548e-05,
      "loss": 1.1089,
      "step": 13395
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8342647552490234,
      "learning_rate": 1.2060185924671756e-05,
      "loss": 1.2414,
      "step": 13396
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.173753261566162,
      "learning_rate": 1.205954700827397e-05,
      "loss": 1.7684,
      "step": 13397
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.4972660541534424,
      "learning_rate": 1.2058908091876177e-05,
      "loss": 1.0888,
      "step": 13398
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3519320487976074,
      "learning_rate": 1.205826917547839e-05,
      "loss": 0.8843,
      "step": 13399
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6678403615951538,
      "learning_rate": 1.2057630259080599e-05,
      "loss": 0.9505,
      "step": 13400
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5921204090118408,
      "learning_rate": 1.2056991342682812e-05,
      "loss": 0.8293,
      "step": 13401
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.0284225940704346,
      "learning_rate": 1.205635242628502e-05,
      "loss": 0.8104,
      "step": 13402
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.383233070373535,
      "learning_rate": 1.2055713509887232e-05,
      "loss": 1.0826,
      "step": 13403
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6929614543914795,
      "learning_rate": 1.2055074593489442e-05,
      "loss": 0.9398,
      "step": 13404
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4814393520355225,
      "learning_rate": 1.2054435677091653e-05,
      "loss": 0.8279,
      "step": 13405
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2369766235351562,
      "learning_rate": 1.2053796760693863e-05,
      "loss": 0.9316,
      "step": 13406
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.166731595993042,
      "learning_rate": 1.2053157844296074e-05,
      "loss": 0.8701,
      "step": 13407
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.410275459289551,
      "learning_rate": 1.2052518927898284e-05,
      "loss": 1.2859,
      "step": 13408
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.358532190322876,
      "learning_rate": 1.2051880011500496e-05,
      "loss": 0.7084,
      "step": 13409
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.658136010169983,
      "learning_rate": 1.2051241095102706e-05,
      "loss": 0.8677,
      "step": 13410
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.9330790042877197,
      "learning_rate": 1.2050602178704917e-05,
      "loss": 1.388,
      "step": 13411
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3841091394424438,
      "learning_rate": 1.2049963262307127e-05,
      "loss": 0.7993,
      "step": 13412
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9581087827682495,
      "learning_rate": 1.2049324345909339e-05,
      "loss": 0.8962,
      "step": 13413
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.809830904006958,
      "learning_rate": 1.2048685429511549e-05,
      "loss": 0.8681,
      "step": 13414
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.494099736213684,
      "learning_rate": 1.204804651311376e-05,
      "loss": 0.4346,
      "step": 13415
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.290802001953125,
      "learning_rate": 1.204740759671597e-05,
      "loss": 1.5234,
      "step": 13416
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8030130863189697,
      "learning_rate": 1.2046768680318181e-05,
      "loss": 0.8411,
      "step": 13417
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.4059269428253174,
      "learning_rate": 1.2046129763920391e-05,
      "loss": 0.8392,
      "step": 13418
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9745832681655884,
      "learning_rate": 1.2045490847522603e-05,
      "loss": 1.0206,
      "step": 13419
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7322643995285034,
      "learning_rate": 1.2044851931124813e-05,
      "loss": 0.9532,
      "step": 13420
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5599241256713867,
      "learning_rate": 1.2044213014727024e-05,
      "loss": 1.14,
      "step": 13421
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.2924132347106934,
      "learning_rate": 1.2043574098329234e-05,
      "loss": 0.8996,
      "step": 13422
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.65807580947876,
      "learning_rate": 1.2042935181931446e-05,
      "loss": 1.443,
      "step": 13423
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0622446537017822,
      "learning_rate": 1.2042296265533656e-05,
      "loss": 0.9679,
      "step": 13424
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9525315761566162,
      "learning_rate": 1.2041657349135867e-05,
      "loss": 1.0011,
      "step": 13425
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.39261794090271,
      "learning_rate": 1.2041018432738077e-05,
      "loss": 0.8891,
      "step": 13426
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.607083797454834,
      "learning_rate": 1.2040379516340288e-05,
      "loss": 0.8103,
      "step": 13427
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.734056830406189,
      "learning_rate": 1.2039740599942498e-05,
      "loss": 0.883,
      "step": 13428
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1269617080688477,
      "learning_rate": 1.203910168354471e-05,
      "loss": 1.0121,
      "step": 13429
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5895662307739258,
      "learning_rate": 1.203846276714692e-05,
      "loss": 0.7688,
      "step": 13430
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4622344970703125,
      "learning_rate": 1.2037823850749131e-05,
      "loss": 0.4221,
      "step": 13431
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.270273208618164,
      "learning_rate": 1.2037184934351341e-05,
      "loss": 1.0865,
      "step": 13432
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5402114391326904,
      "learning_rate": 1.2036546017953553e-05,
      "loss": 0.4083,
      "step": 13433
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1178719997406006,
      "learning_rate": 1.203590710155576e-05,
      "loss": 0.9376,
      "step": 13434
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5716607570648193,
      "learning_rate": 1.2035268185157974e-05,
      "loss": 0.8749,
      "step": 13435
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.469465494155884,
      "learning_rate": 1.2034629268760182e-05,
      "loss": 1.0319,
      "step": 13436
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0391976833343506,
      "learning_rate": 1.2033990352362395e-05,
      "loss": 0.8369,
      "step": 13437
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.482926607131958,
      "learning_rate": 1.2033351435964604e-05,
      "loss": 0.9702,
      "step": 13438
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1211023330688477,
      "learning_rate": 1.2032712519566815e-05,
      "loss": 0.9506,
      "step": 13439
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.5889766216278076,
      "learning_rate": 1.2032073603169025e-05,
      "loss": 0.8894,
      "step": 13440
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.3835408687591553,
      "learning_rate": 1.2031434686771237e-05,
      "loss": 1.0058,
      "step": 13441
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0072131156921387,
      "learning_rate": 1.2030795770373446e-05,
      "loss": 0.9586,
      "step": 13442
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7698031663894653,
      "learning_rate": 1.2030156853975658e-05,
      "loss": 0.8959,
      "step": 13443
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.397386074066162,
      "learning_rate": 1.2029517937577868e-05,
      "loss": 0.7454,
      "step": 13444
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.216583251953125,
      "learning_rate": 1.202887902118008e-05,
      "loss": 0.6343,
      "step": 13445
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.8031604290008545,
      "learning_rate": 1.202824010478229e-05,
      "loss": 1.2344,
      "step": 13446
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9278582334518433,
      "learning_rate": 1.20276011883845e-05,
      "loss": 0.9879,
      "step": 13447
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9284114837646484,
      "learning_rate": 1.202696227198671e-05,
      "loss": 0.8518,
      "step": 13448
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5729496479034424,
      "learning_rate": 1.2026323355588922e-05,
      "loss": 0.9544,
      "step": 13449
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.673632264137268,
      "learning_rate": 1.2025684439191132e-05,
      "loss": 0.4081,
      "step": 13450
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3333947658538818,
      "learning_rate": 1.2025045522793344e-05,
      "loss": 0.805,
      "step": 13451
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9338536262512207,
      "learning_rate": 1.2024406606395553e-05,
      "loss": 0.8119,
      "step": 13452
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9567128419876099,
      "learning_rate": 1.2023767689997765e-05,
      "loss": 0.9722,
      "step": 13453
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9047777652740479,
      "learning_rate": 1.2023128773599975e-05,
      "loss": 0.9681,
      "step": 13454
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5913594961166382,
      "learning_rate": 1.2022489857202186e-05,
      "loss": 0.8542,
      "step": 13455
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.80070960521698,
      "learning_rate": 1.2021850940804396e-05,
      "loss": 0.877,
      "step": 13456
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7264090776443481,
      "learning_rate": 1.2021212024406608e-05,
      "loss": 0.9455,
      "step": 13457
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0569870471954346,
      "learning_rate": 1.2020573108008818e-05,
      "loss": 0.9069,
      "step": 13458
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9438456296920776,
      "learning_rate": 1.201993419161103e-05,
      "loss": 0.929,
      "step": 13459
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.6722779273986816,
      "learning_rate": 1.2019295275213239e-05,
      "loss": 1.2379,
      "step": 13460
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9029433727264404,
      "learning_rate": 1.201865635881545e-05,
      "loss": 0.8547,
      "step": 13461
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.492453098297119,
      "learning_rate": 1.201801744241766e-05,
      "loss": 1.9159,
      "step": 13462
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.571981430053711,
      "learning_rate": 1.2017378526019872e-05,
      "loss": 1.0336,
      "step": 13463
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8106789588928223,
      "learning_rate": 1.2016739609622082e-05,
      "loss": 0.9411,
      "step": 13464
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.275355339050293,
      "learning_rate": 1.2016100693224293e-05,
      "loss": 0.9943,
      "step": 13465
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.49776291847229,
      "learning_rate": 1.2015461776826503e-05,
      "loss": 0.9453,
      "step": 13466
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.995330810546875,
      "learning_rate": 1.2014822860428715e-05,
      "loss": 1.2013,
      "step": 13467
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.205831289291382,
      "learning_rate": 1.2014183944030925e-05,
      "loss": 1.027,
      "step": 13468
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.788272500038147,
      "learning_rate": 1.2013545027633136e-05,
      "loss": 0.9759,
      "step": 13469
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7467176914215088,
      "learning_rate": 1.2012906111235344e-05,
      "loss": 0.8754,
      "step": 13470
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2495341300964355,
      "learning_rate": 1.2012267194837558e-05,
      "loss": 0.5389,
      "step": 13471
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5298209190368652,
      "learning_rate": 1.2011628278439766e-05,
      "loss": 0.7764,
      "step": 13472
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6826612949371338,
      "learning_rate": 1.2010989362041979e-05,
      "loss": 0.9651,
      "step": 13473
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.6046667098999023,
      "learning_rate": 1.2010350445644187e-05,
      "loss": 0.9205,
      "step": 13474
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1868956089019775,
      "learning_rate": 1.2009711529246399e-05,
      "loss": 0.9706,
      "step": 13475
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.8396835327148438,
      "learning_rate": 1.2009072612848609e-05,
      "loss": 0.9216,
      "step": 13476
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.3281073570251465,
      "learning_rate": 1.200843369645082e-05,
      "loss": 0.8344,
      "step": 13477
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5605276823043823,
      "learning_rate": 1.200779478005303e-05,
      "loss": 0.7171,
      "step": 13478
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.519312858581543,
      "learning_rate": 1.2007155863655241e-05,
      "loss": 1.3512,
      "step": 13479
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.597890019416809,
      "learning_rate": 1.2006516947257451e-05,
      "loss": 0.8542,
      "step": 13480
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0404438972473145,
      "learning_rate": 1.2005878030859663e-05,
      "loss": 0.8615,
      "step": 13481
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.326284170150757,
      "learning_rate": 1.2005239114461873e-05,
      "loss": 0.9262,
      "step": 13482
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8296728134155273,
      "learning_rate": 1.2004600198064084e-05,
      "loss": 0.879,
      "step": 13483
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.007986545562744,
      "learning_rate": 1.2003961281666294e-05,
      "loss": 1.0388,
      "step": 13484
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.584980845451355,
      "learning_rate": 1.2003322365268506e-05,
      "loss": 0.4922,
      "step": 13485
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0898139476776123,
      "learning_rate": 1.2002683448870716e-05,
      "loss": 0.8977,
      "step": 13486
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.230120897293091,
      "learning_rate": 1.2002044532472927e-05,
      "loss": 0.742,
      "step": 13487
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.43489146232605,
      "learning_rate": 1.2001405616075137e-05,
      "loss": 1.0849,
      "step": 13488
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.985337495803833,
      "learning_rate": 1.2000766699677348e-05,
      "loss": 0.9176,
      "step": 13489
    },
    {
      "epoch": 0.42,
      "grad_norm": 12.61686897277832,
      "learning_rate": 1.2000127783279558e-05,
      "loss": 1.0522,
      "step": 13490
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.4445531368255615,
      "learning_rate": 1.199948886688177e-05,
      "loss": 0.8542,
      "step": 13491
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.753877878189087,
      "learning_rate": 1.199884995048398e-05,
      "loss": 0.883,
      "step": 13492
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7046242952346802,
      "learning_rate": 1.1998211034086191e-05,
      "loss": 0.9808,
      "step": 13493
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9513022899627686,
      "learning_rate": 1.1997572117688401e-05,
      "loss": 0.8968,
      "step": 13494
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9826642274856567,
      "learning_rate": 1.1996933201290613e-05,
      "loss": 0.876,
      "step": 13495
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.459437608718872,
      "learning_rate": 1.1996294284892823e-05,
      "loss": 0.4068,
      "step": 13496
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.108746290206909,
      "learning_rate": 1.1995655368495034e-05,
      "loss": 0.932,
      "step": 13497
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8779544830322266,
      "learning_rate": 1.1995016452097244e-05,
      "loss": 0.9415,
      "step": 13498
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.48337984085083,
      "learning_rate": 1.1994377535699455e-05,
      "loss": 0.8952,
      "step": 13499
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.450739860534668,
      "learning_rate": 1.1993738619301665e-05,
      "loss": 1.0286,
      "step": 13500
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4840174913406372,
      "learning_rate": 1.1993099702903877e-05,
      "loss": 0.8567,
      "step": 13501
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6527183055877686,
      "learning_rate": 1.1992460786506087e-05,
      "loss": 0.8831,
      "step": 13502
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.661992073059082,
      "learning_rate": 1.1991821870108298e-05,
      "loss": 0.7625,
      "step": 13503
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.504840850830078,
      "learning_rate": 1.1991182953710508e-05,
      "loss": 0.9624,
      "step": 13504
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1273059844970703,
      "learning_rate": 1.199054403731272e-05,
      "loss": 0.9221,
      "step": 13505
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.850382685661316,
      "learning_rate": 1.1989905120914928e-05,
      "loss": 0.8771,
      "step": 13506
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.166304111480713,
      "learning_rate": 1.1989266204517141e-05,
      "loss": 1.2517,
      "step": 13507
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.146223306655884,
      "learning_rate": 1.198862728811935e-05,
      "loss": 1.0355,
      "step": 13508
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1098179817199707,
      "learning_rate": 1.1987988371721562e-05,
      "loss": 0.9008,
      "step": 13509
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.859213948249817,
      "learning_rate": 1.198734945532377e-05,
      "loss": 0.789,
      "step": 13510
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.067667007446289,
      "learning_rate": 1.1986710538925982e-05,
      "loss": 1.1473,
      "step": 13511
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7014466524124146,
      "learning_rate": 1.1986071622528192e-05,
      "loss": 0.5001,
      "step": 13512
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.930220365524292,
      "learning_rate": 1.1985432706130404e-05,
      "loss": 0.978,
      "step": 13513
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.408170223236084,
      "learning_rate": 1.1984793789732613e-05,
      "loss": 1.2184,
      "step": 13514
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.805314540863037,
      "learning_rate": 1.1984154873334825e-05,
      "loss": 1.3151,
      "step": 13515
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9290473461151123,
      "learning_rate": 1.1983515956937035e-05,
      "loss": 0.7962,
      "step": 13516
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.5303852558135986,
      "learning_rate": 1.1982877040539246e-05,
      "loss": 0.6691,
      "step": 13517
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1641714572906494,
      "learning_rate": 1.1982238124141456e-05,
      "loss": 1.2158,
      "step": 13518
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8622381687164307,
      "learning_rate": 1.1981599207743668e-05,
      "loss": 0.9544,
      "step": 13519
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7770719528198242,
      "learning_rate": 1.1980960291345878e-05,
      "loss": 0.8876,
      "step": 13520
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6664572954177856,
      "learning_rate": 1.198032137494809e-05,
      "loss": 0.8748,
      "step": 13521
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8185627460479736,
      "learning_rate": 1.1979682458550299e-05,
      "loss": 0.9055,
      "step": 13522
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3395636081695557,
      "learning_rate": 1.197904354215251e-05,
      "loss": 0.3767,
      "step": 13523
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.0906286239624023,
      "learning_rate": 1.197840462575472e-05,
      "loss": 0.9552,
      "step": 13524
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6581655740737915,
      "learning_rate": 1.1977765709356932e-05,
      "loss": 1.1623,
      "step": 13525
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.7904646396636963,
      "learning_rate": 1.1977126792959142e-05,
      "loss": 1.0666,
      "step": 13526
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.079031467437744,
      "learning_rate": 1.1976487876561353e-05,
      "loss": 1.0218,
      "step": 13527
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.824742078781128,
      "learning_rate": 1.1975848960163563e-05,
      "loss": 1.0171,
      "step": 13528
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.302059650421143,
      "learning_rate": 1.1975210043765775e-05,
      "loss": 1.1588,
      "step": 13529
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.6736769676208496,
      "learning_rate": 1.1974571127367985e-05,
      "loss": 1.1386,
      "step": 13530
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0016231536865234,
      "learning_rate": 1.1973932210970196e-05,
      "loss": 0.908,
      "step": 13531
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2625386714935303,
      "learning_rate": 1.1973293294572406e-05,
      "loss": 0.8521,
      "step": 13532
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8958649635314941,
      "learning_rate": 1.1972654378174618e-05,
      "loss": 0.8734,
      "step": 13533
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7446190118789673,
      "learning_rate": 1.1972015461776827e-05,
      "loss": 0.9087,
      "step": 13534
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6155709028244019,
      "learning_rate": 1.1971376545379039e-05,
      "loss": 0.4476,
      "step": 13535
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.360931873321533,
      "learning_rate": 1.1970737628981249e-05,
      "loss": 0.9066,
      "step": 13536
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8939186334609985,
      "learning_rate": 1.197009871258346e-05,
      "loss": 0.9488,
      "step": 13537
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3657381534576416,
      "learning_rate": 1.196945979618567e-05,
      "loss": 0.8741,
      "step": 13538
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0049123764038086,
      "learning_rate": 1.1968820879787882e-05,
      "loss": 0.9057,
      "step": 13539
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7893142700195312,
      "learning_rate": 1.1968181963390092e-05,
      "loss": 0.9405,
      "step": 13540
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8121528625488281,
      "learning_rate": 1.1967543046992303e-05,
      "loss": 0.9451,
      "step": 13541
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.071441650390625,
      "learning_rate": 1.1966904130594511e-05,
      "loss": 0.87,
      "step": 13542
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.8907508850097656,
      "learning_rate": 1.1966265214196725e-05,
      "loss": 1.2459,
      "step": 13543
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9949074983596802,
      "learning_rate": 1.1965626297798933e-05,
      "loss": 0.8172,
      "step": 13544
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.424771785736084,
      "learning_rate": 1.1964987381401146e-05,
      "loss": 1.1616,
      "step": 13545
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.592911720275879,
      "learning_rate": 1.1964348465003354e-05,
      "loss": 0.9651,
      "step": 13546
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7303225994110107,
      "learning_rate": 1.1963709548605567e-05,
      "loss": 0.9405,
      "step": 13547
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.558322787284851,
      "learning_rate": 1.1963070632207776e-05,
      "loss": 0.5454,
      "step": 13548
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.535336017608643,
      "learning_rate": 1.1962431715809987e-05,
      "loss": 0.9135,
      "step": 13549
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8520056009292603,
      "learning_rate": 1.1961792799412197e-05,
      "loss": 0.8628,
      "step": 13550
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6109638214111328,
      "learning_rate": 1.1961153883014408e-05,
      "loss": 0.868,
      "step": 13551
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6053754091262817,
      "learning_rate": 1.1960514966616618e-05,
      "loss": 0.8213,
      "step": 13552
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3702317476272583,
      "learning_rate": 1.195987605021883e-05,
      "loss": 0.3947,
      "step": 13553
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9372395277023315,
      "learning_rate": 1.195923713382104e-05,
      "loss": 0.976,
      "step": 13554
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.957567572593689,
      "learning_rate": 1.1958598217423251e-05,
      "loss": 1.0811,
      "step": 13555
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1644818782806396,
      "learning_rate": 1.1957959301025461e-05,
      "loss": 0.8999,
      "step": 13556
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7412402629852295,
      "learning_rate": 1.1957320384627673e-05,
      "loss": 0.8786,
      "step": 13557
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0033724308013916,
      "learning_rate": 1.1956681468229883e-05,
      "loss": 0.8499,
      "step": 13558
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7609412670135498,
      "learning_rate": 1.1956042551832094e-05,
      "loss": 0.8281,
      "step": 13559
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0004472732543945,
      "learning_rate": 1.1955403635434304e-05,
      "loss": 1.0399,
      "step": 13560
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.720619797706604,
      "learning_rate": 1.1954764719036515e-05,
      "loss": 0.9417,
      "step": 13561
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8731825351715088,
      "learning_rate": 1.1954125802638725e-05,
      "loss": 0.9844,
      "step": 13562
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8459880352020264,
      "learning_rate": 1.1953486886240937e-05,
      "loss": 0.9215,
      "step": 13563
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4943017959594727,
      "learning_rate": 1.1952847969843147e-05,
      "loss": 1.1785,
      "step": 13564
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6736186742782593,
      "learning_rate": 1.1952209053445358e-05,
      "loss": 0.8683,
      "step": 13565
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1340041160583496,
      "learning_rate": 1.1951570137047568e-05,
      "loss": 0.9374,
      "step": 13566
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6813915967941284,
      "learning_rate": 1.195093122064978e-05,
      "loss": 0.8551,
      "step": 13567
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8551089763641357,
      "learning_rate": 1.195029230425199e-05,
      "loss": 0.9533,
      "step": 13568
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.08646559715271,
      "learning_rate": 1.1949653387854201e-05,
      "loss": 1.111,
      "step": 13569
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.779969573020935,
      "learning_rate": 1.1949014471456411e-05,
      "loss": 0.8953,
      "step": 13570
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.5793142318725586,
      "learning_rate": 1.1948375555058622e-05,
      "loss": 0.7565,
      "step": 13571
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7447919845581055,
      "learning_rate": 1.1947736638660832e-05,
      "loss": 0.8676,
      "step": 13572
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.473889350891113,
      "learning_rate": 1.1947097722263044e-05,
      "loss": 1.2499,
      "step": 13573
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.8499398231506348,
      "learning_rate": 1.1946458805865254e-05,
      "loss": 0.7065,
      "step": 13574
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2654929161071777,
      "learning_rate": 1.1945819889467465e-05,
      "loss": 0.5736,
      "step": 13575
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6943938732147217,
      "learning_rate": 1.1945180973069675e-05,
      "loss": 0.9098,
      "step": 13576
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.590793251991272,
      "learning_rate": 1.1944542056671887e-05,
      "loss": 0.9246,
      "step": 13577
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4273276329040527,
      "learning_rate": 1.1943903140274097e-05,
      "loss": 1.0528,
      "step": 13578
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0725085735321045,
      "learning_rate": 1.1943264223876308e-05,
      "loss": 0.9784,
      "step": 13579
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6550930738449097,
      "learning_rate": 1.1942625307478516e-05,
      "loss": 0.8259,
      "step": 13580
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.760146975517273,
      "learning_rate": 1.194198639108073e-05,
      "loss": 0.954,
      "step": 13581
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.645371913909912,
      "learning_rate": 1.1941347474682938e-05,
      "loss": 0.8962,
      "step": 13582
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6689269542694092,
      "learning_rate": 1.1940708558285151e-05,
      "loss": 1.2663,
      "step": 13583
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0269739627838135,
      "learning_rate": 1.1940069641887359e-05,
      "loss": 0.9673,
      "step": 13584
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.4251577854156494,
      "learning_rate": 1.193943072548957e-05,
      "loss": 0.8923,
      "step": 13585
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7696894407272339,
      "learning_rate": 1.193879180909178e-05,
      "loss": 0.8605,
      "step": 13586
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8503966331481934,
      "learning_rate": 1.1938152892693992e-05,
      "loss": 0.9868,
      "step": 13587
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.545638084411621,
      "learning_rate": 1.1937513976296202e-05,
      "loss": 0.8061,
      "step": 13588
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9130463600158691,
      "learning_rate": 1.1936875059898413e-05,
      "loss": 0.9697,
      "step": 13589
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.036701202392578,
      "learning_rate": 1.1936236143500623e-05,
      "loss": 1.4752,
      "step": 13590
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8071554899215698,
      "learning_rate": 1.1935597227102835e-05,
      "loss": 0.8943,
      "step": 13591
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.388301134109497,
      "learning_rate": 1.1934958310705045e-05,
      "loss": 1.1175,
      "step": 13592
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.3022406101226807,
      "learning_rate": 1.1934319394307256e-05,
      "loss": 0.8833,
      "step": 13593
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9559684991836548,
      "learning_rate": 1.1933680477909466e-05,
      "loss": 0.9777,
      "step": 13594
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7049710750579834,
      "learning_rate": 1.1933041561511678e-05,
      "loss": 0.9343,
      "step": 13595
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8888216018676758,
      "learning_rate": 1.1932402645113887e-05,
      "loss": 1.3041,
      "step": 13596
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8591787815093994,
      "learning_rate": 1.1931763728716099e-05,
      "loss": 1.1138,
      "step": 13597
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.180535078048706,
      "learning_rate": 1.1931124812318309e-05,
      "loss": 0.9229,
      "step": 13598
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9343857765197754,
      "learning_rate": 1.193048589592052e-05,
      "loss": 0.8692,
      "step": 13599
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6069045066833496,
      "learning_rate": 1.192984697952273e-05,
      "loss": 0.9219,
      "step": 13600
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6264151334762573,
      "learning_rate": 1.1929208063124942e-05,
      "loss": 0.8441,
      "step": 13601
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2749581336975098,
      "learning_rate": 1.1928569146727152e-05,
      "loss": 1.2187,
      "step": 13602
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9306085109710693,
      "learning_rate": 1.1927930230329363e-05,
      "loss": 0.9965,
      "step": 13603
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.731490969657898,
      "learning_rate": 1.1927291313931573e-05,
      "loss": 0.5093,
      "step": 13604
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.682335376739502,
      "learning_rate": 1.1926652397533785e-05,
      "loss": 1.317,
      "step": 13605
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1728169918060303,
      "learning_rate": 1.1926013481135994e-05,
      "loss": 1.2396,
      "step": 13606
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7035279273986816,
      "learning_rate": 1.1925374564738206e-05,
      "loss": 0.9561,
      "step": 13607
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7473835945129395,
      "learning_rate": 1.1924735648340416e-05,
      "loss": 0.8984,
      "step": 13608
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8860801458358765,
      "learning_rate": 1.1924096731942627e-05,
      "loss": 0.9536,
      "step": 13609
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.754560947418213,
      "learning_rate": 1.1923457815544837e-05,
      "loss": 0.7437,
      "step": 13610
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8137184381484985,
      "learning_rate": 1.1922818899147049e-05,
      "loss": 1.006,
      "step": 13611
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.731947898864746,
      "learning_rate": 1.1922179982749259e-05,
      "loss": 0.9843,
      "step": 13612
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1270077228546143,
      "learning_rate": 1.192154106635147e-05,
      "loss": 1.0651,
      "step": 13613
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.518409013748169,
      "learning_rate": 1.192090214995368e-05,
      "loss": 0.8677,
      "step": 13614
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.794604778289795,
      "learning_rate": 1.1920263233555892e-05,
      "loss": 0.8592,
      "step": 13615
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1603333950042725,
      "learning_rate": 1.19196243171581e-05,
      "loss": 0.9713,
      "step": 13616
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8626161813735962,
      "learning_rate": 1.1918985400760313e-05,
      "loss": 1.2408,
      "step": 13617
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.435241222381592,
      "learning_rate": 1.1918346484362521e-05,
      "loss": 0.9949,
      "step": 13618
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.770060420036316,
      "learning_rate": 1.1917707567964734e-05,
      "loss": 0.949,
      "step": 13619
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.3658459186553955,
      "learning_rate": 1.1917068651566943e-05,
      "loss": 0.906,
      "step": 13620
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0683138370513916,
      "learning_rate": 1.1916429735169154e-05,
      "loss": 0.9087,
      "step": 13621
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6796096563339233,
      "learning_rate": 1.1915790818771364e-05,
      "loss": 0.8801,
      "step": 13622
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7766602039337158,
      "learning_rate": 1.1915151902373575e-05,
      "loss": 0.9225,
      "step": 13623
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2399890422821045,
      "learning_rate": 1.1914512985975785e-05,
      "loss": 0.924,
      "step": 13624
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.45278263092041,
      "learning_rate": 1.1913874069577997e-05,
      "loss": 1.2319,
      "step": 13625
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9922261238098145,
      "learning_rate": 1.1913235153180207e-05,
      "loss": 0.8944,
      "step": 13626
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8206181526184082,
      "learning_rate": 1.1912596236782418e-05,
      "loss": 0.905,
      "step": 13627
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7780020236968994,
      "learning_rate": 1.1911957320384628e-05,
      "loss": 0.8632,
      "step": 13628
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.930635452270508,
      "learning_rate": 1.191131840398684e-05,
      "loss": 1.1704,
      "step": 13629
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6468483209609985,
      "learning_rate": 1.191067948758905e-05,
      "loss": 0.8681,
      "step": 13630
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2622995376586914,
      "learning_rate": 1.1910040571191261e-05,
      "loss": 1.1754,
      "step": 13631
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7166913747787476,
      "learning_rate": 1.1909401654793471e-05,
      "loss": 0.9709,
      "step": 13632
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.038045644760132,
      "learning_rate": 1.1908762738395682e-05,
      "loss": 0.9183,
      "step": 13633
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.693367600440979,
      "learning_rate": 1.1908123821997892e-05,
      "loss": 0.9432,
      "step": 13634
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.41232442855835,
      "learning_rate": 1.1907484905600104e-05,
      "loss": 1.1252,
      "step": 13635
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8752261400222778,
      "learning_rate": 1.1906845989202314e-05,
      "loss": 0.9452,
      "step": 13636
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.156003952026367,
      "learning_rate": 1.1906207072804525e-05,
      "loss": 1.5412,
      "step": 13637
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.13489031791687,
      "learning_rate": 1.1905568156406735e-05,
      "loss": 0.9516,
      "step": 13638
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4678951501846313,
      "learning_rate": 1.1904929240008947e-05,
      "loss": 0.873,
      "step": 13639
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.770953416824341,
      "learning_rate": 1.1904290323611157e-05,
      "loss": 1.1438,
      "step": 13640
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2984049320220947,
      "learning_rate": 1.1903651407213368e-05,
      "loss": 0.9273,
      "step": 13641
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.567406415939331,
      "learning_rate": 1.1903012490815578e-05,
      "loss": 0.9413,
      "step": 13642
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0105347633361816,
      "learning_rate": 1.190237357441779e-05,
      "loss": 0.9326,
      "step": 13643
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5631520748138428,
      "learning_rate": 1.190173465802e-05,
      "loss": 0.9133,
      "step": 13644
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.372920274734497,
      "learning_rate": 1.1901095741622211e-05,
      "loss": 0.5492,
      "step": 13645
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.002699375152588,
      "learning_rate": 1.190045682522442e-05,
      "loss": 0.8996,
      "step": 13646
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.006631851196289,
      "learning_rate": 1.1899817908826632e-05,
      "loss": 0.9656,
      "step": 13647
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5175179243087769,
      "learning_rate": 1.1899178992428842e-05,
      "loss": 0.9357,
      "step": 13648
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.821110725402832,
      "learning_rate": 1.1898540076031054e-05,
      "loss": 0.722,
      "step": 13649
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.6737775802612305,
      "learning_rate": 1.1897901159633264e-05,
      "loss": 1.5704,
      "step": 13650
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.112161874771118,
      "learning_rate": 1.1897262243235475e-05,
      "loss": 0.9304,
      "step": 13651
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7782632112503052,
      "learning_rate": 1.1896623326837683e-05,
      "loss": 0.8554,
      "step": 13652
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6189111471176147,
      "learning_rate": 1.1895984410439896e-05,
      "loss": 0.9802,
      "step": 13653
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7749156951904297,
      "learning_rate": 1.1895345494042105e-05,
      "loss": 0.8828,
      "step": 13654
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9642783403396606,
      "learning_rate": 1.1894706577644318e-05,
      "loss": 0.8642,
      "step": 13655
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0300583839416504,
      "learning_rate": 1.1894067661246526e-05,
      "loss": 0.9408,
      "step": 13656
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5532951354980469,
      "learning_rate": 1.1893428744848738e-05,
      "loss": 0.8957,
      "step": 13657
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.618113040924072,
      "learning_rate": 1.1892789828450947e-05,
      "loss": 1.1779,
      "step": 13658
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.497610569000244,
      "learning_rate": 1.1892150912053159e-05,
      "loss": 1.1783,
      "step": 13659
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4548546075820923,
      "learning_rate": 1.1891511995655369e-05,
      "loss": 0.8455,
      "step": 13660
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6219055652618408,
      "learning_rate": 1.189087307925758e-05,
      "loss": 0.8838,
      "step": 13661
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6839611530303955,
      "learning_rate": 1.189023416285979e-05,
      "loss": 0.8845,
      "step": 13662
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6340255737304688,
      "learning_rate": 1.1889595246462002e-05,
      "loss": 0.7766,
      "step": 13663
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4053127765655518,
      "learning_rate": 1.1888956330064212e-05,
      "loss": 0.889,
      "step": 13664
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8784350156784058,
      "learning_rate": 1.1888317413666423e-05,
      "loss": 0.9063,
      "step": 13665
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7253016233444214,
      "learning_rate": 1.1887678497268633e-05,
      "loss": 0.8941,
      "step": 13666
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9921694993972778,
      "learning_rate": 1.1887039580870845e-05,
      "loss": 1.2433,
      "step": 13667
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8727667331695557,
      "learning_rate": 1.1886400664473054e-05,
      "loss": 0.9336,
      "step": 13668
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9222631454467773,
      "learning_rate": 1.1885761748075266e-05,
      "loss": 1.125,
      "step": 13669
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.368510723114014,
      "learning_rate": 1.1885122831677476e-05,
      "loss": 1.0916,
      "step": 13670
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6104588508605957,
      "learning_rate": 1.1884483915279687e-05,
      "loss": 0.817,
      "step": 13671
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.7917141914367676,
      "learning_rate": 1.1883844998881897e-05,
      "loss": 1.0131,
      "step": 13672
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8690263032913208,
      "learning_rate": 1.1883206082484109e-05,
      "loss": 0.975,
      "step": 13673
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1888599395751953,
      "learning_rate": 1.1882567166086319e-05,
      "loss": 0.9574,
      "step": 13674
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9167336225509644,
      "learning_rate": 1.188192824968853e-05,
      "loss": 1.0555,
      "step": 13675
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.937731146812439,
      "learning_rate": 1.188128933329074e-05,
      "loss": 1.0686,
      "step": 13676
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9190253019332886,
      "learning_rate": 1.1880650416892952e-05,
      "loss": 0.872,
      "step": 13677
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7963380813598633,
      "learning_rate": 1.1880011500495161e-05,
      "loss": 1.0421,
      "step": 13678
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.6085381507873535,
      "learning_rate": 1.1879372584097373e-05,
      "loss": 0.85,
      "step": 13679
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.3766536712646484,
      "learning_rate": 1.1878733667699583e-05,
      "loss": 0.9159,
      "step": 13680
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8593686819076538,
      "learning_rate": 1.1878094751301794e-05,
      "loss": 0.9158,
      "step": 13681
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9382822513580322,
      "learning_rate": 1.1877455834904004e-05,
      "loss": 0.5079,
      "step": 13682
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7658960819244385,
      "learning_rate": 1.1876816918506216e-05,
      "loss": 0.9553,
      "step": 13683
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0489561557769775,
      "learning_rate": 1.1876178002108426e-05,
      "loss": 0.978,
      "step": 13684
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6654647588729858,
      "learning_rate": 1.1875539085710637e-05,
      "loss": 0.4261,
      "step": 13685
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9262160062789917,
      "learning_rate": 1.1874900169312847e-05,
      "loss": 0.7163,
      "step": 13686
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9453026056289673,
      "learning_rate": 1.1874261252915059e-05,
      "loss": 1.0315,
      "step": 13687
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.209312677383423,
      "learning_rate": 1.1873622336517267e-05,
      "loss": 0.6756,
      "step": 13688
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.5912227630615234,
      "learning_rate": 1.187298342011948e-05,
      "loss": 0.6907,
      "step": 13689
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5211787223815918,
      "learning_rate": 1.1872344503721688e-05,
      "loss": 0.9944,
      "step": 13690
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6047955751419067,
      "learning_rate": 1.1871705587323901e-05,
      "loss": 0.7781,
      "step": 13691
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.534972071647644,
      "learning_rate": 1.187106667092611e-05,
      "loss": 0.9445,
      "step": 13692
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5720269680023193,
      "learning_rate": 1.1870427754528323e-05,
      "loss": 0.9622,
      "step": 13693
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7203176021575928,
      "learning_rate": 1.1869788838130531e-05,
      "loss": 0.6072,
      "step": 13694
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.240877628326416,
      "learning_rate": 1.1869149921732743e-05,
      "loss": 0.8613,
      "step": 13695
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8238296508789062,
      "learning_rate": 1.1868511005334952e-05,
      "loss": 0.9089,
      "step": 13696
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7179896831512451,
      "learning_rate": 1.1867872088937164e-05,
      "loss": 1.0103,
      "step": 13697
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7022082805633545,
      "learning_rate": 1.1867233172539374e-05,
      "loss": 0.8679,
      "step": 13698
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.656064510345459,
      "learning_rate": 1.1866594256141585e-05,
      "loss": 0.8083,
      "step": 13699
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.297361373901367,
      "learning_rate": 1.1865955339743795e-05,
      "loss": 1.1471,
      "step": 13700
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.201241970062256,
      "learning_rate": 1.1865316423346007e-05,
      "loss": 0.8587,
      "step": 13701
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3986883163452148,
      "learning_rate": 1.1864677506948217e-05,
      "loss": 0.6858,
      "step": 13702
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.391819477081299,
      "learning_rate": 1.1864038590550428e-05,
      "loss": 1.1536,
      "step": 13703
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7622836828231812,
      "learning_rate": 1.1863399674152638e-05,
      "loss": 0.8185,
      "step": 13704
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.671463966369629,
      "learning_rate": 1.186276075775485e-05,
      "loss": 0.8524,
      "step": 13705
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.3248937129974365,
      "learning_rate": 1.186212184135706e-05,
      "loss": 1.0174,
      "step": 13706
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7115620374679565,
      "learning_rate": 1.1861482924959271e-05,
      "loss": 0.9341,
      "step": 13707
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7201802730560303,
      "learning_rate": 1.186084400856148e-05,
      "loss": 0.8682,
      "step": 13708
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2679219245910645,
      "learning_rate": 1.1860205092163692e-05,
      "loss": 0.8903,
      "step": 13709
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.3703155517578125,
      "learning_rate": 1.1859566175765902e-05,
      "loss": 0.601,
      "step": 13710
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.599900484085083,
      "learning_rate": 1.1858927259368114e-05,
      "loss": 0.8616,
      "step": 13711
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.767488718032837,
      "learning_rate": 1.1858288342970324e-05,
      "loss": 0.8289,
      "step": 13712
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7319588661193848,
      "learning_rate": 1.1857649426572535e-05,
      "loss": 0.6527,
      "step": 13713
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6049950122833252,
      "learning_rate": 1.1857010510174745e-05,
      "loss": 0.9272,
      "step": 13714
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.91093373298645,
      "learning_rate": 1.1856371593776957e-05,
      "loss": 0.7175,
      "step": 13715
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7899237871170044,
      "learning_rate": 1.1855732677379166e-05,
      "loss": 0.8965,
      "step": 13716
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.027702569961548,
      "learning_rate": 1.1855093760981378e-05,
      "loss": 0.8255,
      "step": 13717
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8806962966918945,
      "learning_rate": 1.1854454844583588e-05,
      "loss": 1.4075,
      "step": 13718
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8718540668487549,
      "learning_rate": 1.18538159281858e-05,
      "loss": 1.0976,
      "step": 13719
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8616995811462402,
      "learning_rate": 1.185317701178801e-05,
      "loss": 0.9212,
      "step": 13720
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6746490001678467,
      "learning_rate": 1.185253809539022e-05,
      "loss": 0.8096,
      "step": 13721
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8184216022491455,
      "learning_rate": 1.185189917899243e-05,
      "loss": 1.3576,
      "step": 13722
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.575553059577942,
      "learning_rate": 1.1851260262594639e-05,
      "loss": 0.9153,
      "step": 13723
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.425055742263794,
      "learning_rate": 1.1850621346196852e-05,
      "loss": 0.4067,
      "step": 13724
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.372487783432007,
      "learning_rate": 1.184998242979906e-05,
      "loss": 0.9768,
      "step": 13725
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.863110303878784,
      "learning_rate": 1.1849343513401272e-05,
      "loss": 0.8896,
      "step": 13726
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9347294569015503,
      "learning_rate": 1.1848704597003482e-05,
      "loss": 1.1722,
      "step": 13727
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.732338547706604,
      "learning_rate": 1.1848065680605693e-05,
      "loss": 0.9419,
      "step": 13728
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0156030654907227,
      "learning_rate": 1.1847426764207903e-05,
      "loss": 0.7875,
      "step": 13729
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.028656244277954,
      "learning_rate": 1.1846787847810114e-05,
      "loss": 0.7657,
      "step": 13730
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.1573967933654785,
      "learning_rate": 1.1846148931412324e-05,
      "loss": 1.1097,
      "step": 13731
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1900999546051025,
      "learning_rate": 1.1845510015014536e-05,
      "loss": 0.9692,
      "step": 13732
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9601702690124512,
      "learning_rate": 1.1844871098616746e-05,
      "loss": 0.8409,
      "step": 13733
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.970349311828613,
      "learning_rate": 1.1844232182218957e-05,
      "loss": 1.0897,
      "step": 13734
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.109318971633911,
      "learning_rate": 1.1843593265821167e-05,
      "loss": 0.972,
      "step": 13735
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4883354902267456,
      "learning_rate": 1.1842954349423379e-05,
      "loss": 0.9002,
      "step": 13736
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.4884748458862305,
      "learning_rate": 1.1842315433025589e-05,
      "loss": 1.0707,
      "step": 13737
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8707448244094849,
      "learning_rate": 1.18416765166278e-05,
      "loss": 0.7387,
      "step": 13738
    },
    {
      "epoch": 0.43,
      "grad_norm": 12.530356407165527,
      "learning_rate": 1.184103760023001e-05,
      "loss": 1.1553,
      "step": 13739
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.617353081703186,
      "learning_rate": 1.1840398683832221e-05,
      "loss": 0.8227,
      "step": 13740
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8250818252563477,
      "learning_rate": 1.1839759767434431e-05,
      "loss": 1.2806,
      "step": 13741
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5052456855773926,
      "learning_rate": 1.1839120851036643e-05,
      "loss": 0.4929,
      "step": 13742
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.145236015319824,
      "learning_rate": 1.1838481934638853e-05,
      "loss": 0.9248,
      "step": 13743
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9432601928710938,
      "learning_rate": 1.1837843018241064e-05,
      "loss": 0.9986,
      "step": 13744
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2212822437286377,
      "learning_rate": 1.1837204101843274e-05,
      "loss": 0.8182,
      "step": 13745
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7056070566177368,
      "learning_rate": 1.1836565185445486e-05,
      "loss": 0.8555,
      "step": 13746
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1412265300750732,
      "learning_rate": 1.1835926269047696e-05,
      "loss": 0.6548,
      "step": 13747
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.468868374824524,
      "learning_rate": 1.1835287352649907e-05,
      "loss": 0.4665,
      "step": 13748
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6950334310531616,
      "learning_rate": 1.1834648436252117e-05,
      "loss": 0.8862,
      "step": 13749
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5703277587890625,
      "learning_rate": 1.1834009519854328e-05,
      "loss": 0.9143,
      "step": 13750
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.875580072402954,
      "learning_rate": 1.1833370603456538e-05,
      "loss": 0.9377,
      "step": 13751
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.675289511680603,
      "learning_rate": 1.183273168705875e-05,
      "loss": 0.8013,
      "step": 13752
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6213452816009521,
      "learning_rate": 1.183209277066096e-05,
      "loss": 0.7835,
      "step": 13753
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.214407444000244,
      "learning_rate": 1.1831453854263171e-05,
      "loss": 1.1423,
      "step": 13754
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.544468879699707,
      "learning_rate": 1.183081493786538e-05,
      "loss": 1.0476,
      "step": 13755
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6893815994262695,
      "learning_rate": 1.1830176021467593e-05,
      "loss": 1.3789,
      "step": 13756
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1601555347442627,
      "learning_rate": 1.18295371050698e-05,
      "loss": 0.7981,
      "step": 13757
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6821993589401245,
      "learning_rate": 1.1828898188672014e-05,
      "loss": 0.8988,
      "step": 13758
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4682947397232056,
      "learning_rate": 1.1828259272274222e-05,
      "loss": 0.4043,
      "step": 13759
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.714812159538269,
      "learning_rate": 1.1827620355876435e-05,
      "loss": 0.8188,
      "step": 13760
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2132112979888916,
      "learning_rate": 1.1826981439478644e-05,
      "loss": 1.2102,
      "step": 13761
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0755879878997803,
      "learning_rate": 1.1826342523080855e-05,
      "loss": 0.8731,
      "step": 13762
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6751041412353516,
      "learning_rate": 1.1825703606683065e-05,
      "loss": 0.9353,
      "step": 13763
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.08975887298584,
      "learning_rate": 1.1825064690285277e-05,
      "loss": 0.887,
      "step": 13764
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.198007106781006,
      "learning_rate": 1.1824425773887486e-05,
      "loss": 0.8981,
      "step": 13765
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7321174144744873,
      "learning_rate": 1.1823786857489698e-05,
      "loss": 0.9373,
      "step": 13766
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.768078088760376,
      "learning_rate": 1.1823147941091908e-05,
      "loss": 0.9469,
      "step": 13767
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8718063831329346,
      "learning_rate": 1.182250902469412e-05,
      "loss": 1.0119,
      "step": 13768
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1240391731262207,
      "learning_rate": 1.182187010829633e-05,
      "loss": 1.3859,
      "step": 13769
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.040072202682495,
      "learning_rate": 1.182123119189854e-05,
      "loss": 1.0874,
      "step": 13770
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.492245674133301,
      "learning_rate": 1.182059227550075e-05,
      "loss": 0.9929,
      "step": 13771
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.141782522201538,
      "learning_rate": 1.1819953359102962e-05,
      "loss": 0.8845,
      "step": 13772
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.7171099185943604,
      "learning_rate": 1.1819314442705172e-05,
      "loss": 1.2593,
      "step": 13773
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6725835800170898,
      "learning_rate": 1.1818675526307384e-05,
      "loss": 0.8668,
      "step": 13774
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.9350552558898926,
      "learning_rate": 1.1818036609909593e-05,
      "loss": 0.9141,
      "step": 13775
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.062330484390259,
      "learning_rate": 1.1817397693511805e-05,
      "loss": 0.8188,
      "step": 13776
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.874037981033325,
      "learning_rate": 1.1816758777114015e-05,
      "loss": 0.9477,
      "step": 13777
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.032438278198242,
      "learning_rate": 1.1816119860716226e-05,
      "loss": 0.9047,
      "step": 13778
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0082499980926514,
      "learning_rate": 1.1815480944318436e-05,
      "loss": 0.8559,
      "step": 13779
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.39278244972229,
      "learning_rate": 1.1814842027920648e-05,
      "loss": 0.9323,
      "step": 13780
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7530592679977417,
      "learning_rate": 1.1814203111522858e-05,
      "loss": 0.8334,
      "step": 13781
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8507294654846191,
      "learning_rate": 1.181356419512507e-05,
      "loss": 1.131,
      "step": 13782
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.222756862640381,
      "learning_rate": 1.1812925278727279e-05,
      "loss": 0.8848,
      "step": 13783
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.14551305770874,
      "learning_rate": 1.181228636232949e-05,
      "loss": 1.3127,
      "step": 13784
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5674102306365967,
      "learning_rate": 1.18116474459317e-05,
      "loss": 1.0093,
      "step": 13785
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6933653354644775,
      "learning_rate": 1.1811008529533912e-05,
      "loss": 1.0531,
      "step": 13786
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5276317596435547,
      "learning_rate": 1.1810369613136122e-05,
      "loss": 0.8045,
      "step": 13787
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7444391250610352,
      "learning_rate": 1.1809730696738333e-05,
      "loss": 0.8479,
      "step": 13788
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3797837495803833,
      "learning_rate": 1.1809091780340543e-05,
      "loss": 0.8686,
      "step": 13789
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0874714851379395,
      "learning_rate": 1.1808452863942755e-05,
      "loss": 1.0003,
      "step": 13790
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.7591283321380615,
      "learning_rate": 1.1807813947544965e-05,
      "loss": 0.9954,
      "step": 13791
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.017843008041382,
      "learning_rate": 1.1807175031147176e-05,
      "loss": 0.7639,
      "step": 13792
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.059250593185425,
      "learning_rate": 1.1806536114749384e-05,
      "loss": 0.8836,
      "step": 13793
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6836564540863037,
      "learning_rate": 1.1805897198351598e-05,
      "loss": 0.8685,
      "step": 13794
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5610527992248535,
      "learning_rate": 1.1805258281953806e-05,
      "loss": 1.0115,
      "step": 13795
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.905023455619812,
      "learning_rate": 1.1804619365556019e-05,
      "loss": 0.8812,
      "step": 13796
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2988593578338623,
      "learning_rate": 1.1803980449158227e-05,
      "loss": 0.8407,
      "step": 13797
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.9363577365875244,
      "learning_rate": 1.1803341532760439e-05,
      "loss": 1.0243,
      "step": 13798
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0248618125915527,
      "learning_rate": 1.1802702616362649e-05,
      "loss": 0.9561,
      "step": 13799
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.772025227546692,
      "learning_rate": 1.180206369996486e-05,
      "loss": 0.8329,
      "step": 13800
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.030733108520508,
      "learning_rate": 1.180142478356707e-05,
      "loss": 0.9874,
      "step": 13801
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6920651197433472,
      "learning_rate": 1.1800785867169281e-05,
      "loss": 0.417,
      "step": 13802
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0936336517333984,
      "learning_rate": 1.1800146950771491e-05,
      "loss": 0.9796,
      "step": 13803
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.278414487838745,
      "learning_rate": 1.1799508034373703e-05,
      "loss": 0.9672,
      "step": 13804
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2378649711608887,
      "learning_rate": 1.1798869117975913e-05,
      "loss": 1.2266,
      "step": 13805
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.887873888015747,
      "learning_rate": 1.1798230201578124e-05,
      "loss": 0.9032,
      "step": 13806
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9025287628173828,
      "learning_rate": 1.1797591285180334e-05,
      "loss": 0.8133,
      "step": 13807
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.171210527420044,
      "learning_rate": 1.1796952368782546e-05,
      "loss": 1.0946,
      "step": 13808
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.910062313079834,
      "learning_rate": 1.1796313452384756e-05,
      "loss": 0.8482,
      "step": 13809
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7995071411132812,
      "learning_rate": 1.1795674535986967e-05,
      "loss": 0.8832,
      "step": 13810
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1007251739501953,
      "learning_rate": 1.1795035619589177e-05,
      "loss": 0.9526,
      "step": 13811
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7410027980804443,
      "learning_rate": 1.1794396703191388e-05,
      "loss": 1.0364,
      "step": 13812
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.973930835723877,
      "learning_rate": 1.1793757786793598e-05,
      "loss": 0.8109,
      "step": 13813
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8979419469833374,
      "learning_rate": 1.179311887039581e-05,
      "loss": 0.7152,
      "step": 13814
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7575289011001587,
      "learning_rate": 1.179247995399802e-05,
      "loss": 0.9103,
      "step": 13815
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.3228914737701416,
      "learning_rate": 1.1791841037600231e-05,
      "loss": 1.0673,
      "step": 13816
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8605446815490723,
      "learning_rate": 1.1791202121202441e-05,
      "loss": 0.8885,
      "step": 13817
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.164537191390991,
      "learning_rate": 1.1790563204804653e-05,
      "loss": 0.9611,
      "step": 13818
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5341529846191406,
      "learning_rate": 1.1789924288406863e-05,
      "loss": 0.865,
      "step": 13819
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3844711780548096,
      "learning_rate": 1.1789285372009074e-05,
      "loss": 0.4308,
      "step": 13820
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5554845333099365,
      "learning_rate": 1.1788646455611284e-05,
      "loss": 0.5481,
      "step": 13821
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.952096939086914,
      "learning_rate": 1.1788007539213495e-05,
      "loss": 1.6607,
      "step": 13822
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.131021022796631,
      "learning_rate": 1.1787368622815705e-05,
      "loss": 1.4357,
      "step": 13823
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.466395616531372,
      "learning_rate": 1.1786729706417917e-05,
      "loss": 0.7744,
      "step": 13824
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2966156005859375,
      "learning_rate": 1.1786090790020127e-05,
      "loss": 0.9748,
      "step": 13825
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0534768104553223,
      "learning_rate": 1.1785451873622338e-05,
      "loss": 0.8222,
      "step": 13826
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.4477691650390625,
      "learning_rate": 1.1784812957224548e-05,
      "loss": 0.6404,
      "step": 13827
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.225294589996338,
      "learning_rate": 1.178417404082676e-05,
      "loss": 0.8196,
      "step": 13828
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.850228190422058,
      "learning_rate": 1.1783535124428968e-05,
      "loss": 0.8094,
      "step": 13829
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.3360190391540527,
      "learning_rate": 1.1782896208031181e-05,
      "loss": 1.0294,
      "step": 13830
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.771171808242798,
      "learning_rate": 1.178225729163339e-05,
      "loss": 0.8401,
      "step": 13831
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4731093645095825,
      "learning_rate": 1.1781618375235602e-05,
      "loss": 0.9139,
      "step": 13832
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5597310066223145,
      "learning_rate": 1.178097945883781e-05,
      "loss": 0.9018,
      "step": 13833
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5735118389129639,
      "learning_rate": 1.1780340542440022e-05,
      "loss": 0.8823,
      "step": 13834
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7756894826889038,
      "learning_rate": 1.1779701626042232e-05,
      "loss": 0.8689,
      "step": 13835
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5560407638549805,
      "learning_rate": 1.1779062709644444e-05,
      "loss": 0.8299,
      "step": 13836
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.742562174797058,
      "learning_rate": 1.1778423793246653e-05,
      "loss": 0.8555,
      "step": 13837
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1025664806365967,
      "learning_rate": 1.1777784876848865e-05,
      "loss": 0.8855,
      "step": 13838
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1550533771514893,
      "learning_rate": 1.1777145960451075e-05,
      "loss": 0.9851,
      "step": 13839
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8315843343734741,
      "learning_rate": 1.1776507044053286e-05,
      "loss": 0.9316,
      "step": 13840
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9017869234085083,
      "learning_rate": 1.1775868127655496e-05,
      "loss": 0.9056,
      "step": 13841
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3062562942504883,
      "learning_rate": 1.1775229211257708e-05,
      "loss": 1.0071,
      "step": 13842
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9859747886657715,
      "learning_rate": 1.1774590294859918e-05,
      "loss": 1.0473,
      "step": 13843
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1158180236816406,
      "learning_rate": 1.177395137846213e-05,
      "loss": 0.9467,
      "step": 13844
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.683290481567383,
      "learning_rate": 1.1773312462064339e-05,
      "loss": 1.3518,
      "step": 13845
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.7523574829101562,
      "learning_rate": 1.177267354566655e-05,
      "loss": 0.5889,
      "step": 13846
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2529656887054443,
      "learning_rate": 1.177203462926876e-05,
      "loss": 1.2494,
      "step": 13847
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.05102801322937,
      "learning_rate": 1.1771395712870972e-05,
      "loss": 0.9972,
      "step": 13848
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0628082752227783,
      "learning_rate": 1.1770756796473182e-05,
      "loss": 0.5767,
      "step": 13849
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9766000509262085,
      "learning_rate": 1.1770117880075393e-05,
      "loss": 0.9557,
      "step": 13850
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.676617980003357,
      "learning_rate": 1.1769478963677603e-05,
      "loss": 0.9797,
      "step": 13851
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6795902252197266,
      "learning_rate": 1.1768840047279815e-05,
      "loss": 0.8685,
      "step": 13852
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6816868782043457,
      "learning_rate": 1.1768201130882025e-05,
      "loss": 0.9397,
      "step": 13853
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.6250202655792236,
      "learning_rate": 1.1767562214484236e-05,
      "loss": 1.1252,
      "step": 13854
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.050534963607788,
      "learning_rate": 1.1766923298086446e-05,
      "loss": 1.0665,
      "step": 13855
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.7499382495880127,
      "learning_rate": 1.1766284381688658e-05,
      "loss": 1.0516,
      "step": 13856
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.2662205696105957,
      "learning_rate": 1.1765645465290867e-05,
      "loss": 1.2297,
      "step": 13857
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8914347887039185,
      "learning_rate": 1.1765006548893079e-05,
      "loss": 0.869,
      "step": 13858
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8761579990386963,
      "learning_rate": 1.1764367632495289e-05,
      "loss": 0.9632,
      "step": 13859
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5764042139053345,
      "learning_rate": 1.17637287160975e-05,
      "loss": 0.4644,
      "step": 13860
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8752126693725586,
      "learning_rate": 1.176308979969971e-05,
      "loss": 1.1703,
      "step": 13861
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8642934560775757,
      "learning_rate": 1.1762450883301922e-05,
      "loss": 0.8469,
      "step": 13862
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1746044158935547,
      "learning_rate": 1.1761811966904132e-05,
      "loss": 0.8121,
      "step": 13863
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.4509904384613037,
      "learning_rate": 1.1761173050506343e-05,
      "loss": 0.7232,
      "step": 13864
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7171058654785156,
      "learning_rate": 1.1760534134108551e-05,
      "loss": 0.9514,
      "step": 13865
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9607372283935547,
      "learning_rate": 1.1759895217710765e-05,
      "loss": 0.8978,
      "step": 13866
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.5439155101776123,
      "learning_rate": 1.1759256301312973e-05,
      "loss": 0.8513,
      "step": 13867
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9449400901794434,
      "learning_rate": 1.1758617384915186e-05,
      "loss": 0.8823,
      "step": 13868
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.005662202835083,
      "learning_rate": 1.1757978468517394e-05,
      "loss": 0.9187,
      "step": 13869
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.737639904022217,
      "learning_rate": 1.1757339552119607e-05,
      "loss": 0.7235,
      "step": 13870
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.662513256072998,
      "learning_rate": 1.1756700635721816e-05,
      "loss": 1.041,
      "step": 13871
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.554854154586792,
      "learning_rate": 1.1756061719324027e-05,
      "loss": 1.0045,
      "step": 13872
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7120598554611206,
      "learning_rate": 1.1755422802926237e-05,
      "loss": 0.8403,
      "step": 13873
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4253674745559692,
      "learning_rate": 1.1754783886528448e-05,
      "loss": 0.7531,
      "step": 13874
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.981868267059326,
      "learning_rate": 1.1754144970130658e-05,
      "loss": 1.3984,
      "step": 13875
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4143959283828735,
      "learning_rate": 1.175350605373287e-05,
      "loss": 0.8949,
      "step": 13876
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.4055633544921875,
      "learning_rate": 1.175286713733508e-05,
      "loss": 1.5125,
      "step": 13877
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9069578647613525,
      "learning_rate": 1.1752228220937291e-05,
      "loss": 0.7388,
      "step": 13878
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.16865873336792,
      "learning_rate": 1.1751589304539501e-05,
      "loss": 0.7572,
      "step": 13879
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5590004920959473,
      "learning_rate": 1.1750950388141713e-05,
      "loss": 0.9076,
      "step": 13880
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.720036506652832,
      "learning_rate": 1.1750311471743923e-05,
      "loss": 1.7106,
      "step": 13881
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7893600463867188,
      "learning_rate": 1.1749672555346134e-05,
      "loss": 0.8046,
      "step": 13882
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8166569471359253,
      "learning_rate": 1.1749033638948344e-05,
      "loss": 0.9417,
      "step": 13883
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2130825519561768,
      "learning_rate": 1.1748394722550555e-05,
      "loss": 1.0195,
      "step": 13884
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4031171798706055,
      "learning_rate": 1.1747755806152765e-05,
      "loss": 0.8476,
      "step": 13885
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9852135181427002,
      "learning_rate": 1.1747116889754977e-05,
      "loss": 1.225,
      "step": 13886
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8775429725646973,
      "learning_rate": 1.1746477973357187e-05,
      "loss": 0.8245,
      "step": 13887
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9237560033798218,
      "learning_rate": 1.1745839056959398e-05,
      "loss": 0.9233,
      "step": 13888
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.634459972381592,
      "learning_rate": 1.1745200140561608e-05,
      "loss": 1.589,
      "step": 13889
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0841774940490723,
      "learning_rate": 1.174456122416382e-05,
      "loss": 1.1784,
      "step": 13890
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7770419120788574,
      "learning_rate": 1.174392230776603e-05,
      "loss": 0.849,
      "step": 13891
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9116774797439575,
      "learning_rate": 1.1743283391368241e-05,
      "loss": 0.8073,
      "step": 13892
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8246768712997437,
      "learning_rate": 1.1742644474970451e-05,
      "loss": 1.1139,
      "step": 13893
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6411396265029907,
      "learning_rate": 1.1742005558572662e-05,
      "loss": 0.8403,
      "step": 13894
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.899078369140625,
      "learning_rate": 1.1741366642174872e-05,
      "loss": 0.9004,
      "step": 13895
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4110645055770874,
      "learning_rate": 1.1740727725777084e-05,
      "loss": 0.855,
      "step": 13896
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.535121202468872,
      "learning_rate": 1.1740088809379294e-05,
      "loss": 1.0912,
      "step": 13897
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.7569665908813477,
      "learning_rate": 1.1739449892981505e-05,
      "loss": 1.097,
      "step": 13898
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9962191581726074,
      "learning_rate": 1.1738810976583715e-05,
      "loss": 1.2283,
      "step": 13899
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.039008617401123,
      "learning_rate": 1.1738172060185927e-05,
      "loss": 0.8928,
      "step": 13900
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.25558614730835,
      "learning_rate": 1.1737533143788135e-05,
      "loss": 1.4474,
      "step": 13901
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9410268068313599,
      "learning_rate": 1.1736894227390348e-05,
      "loss": 0.9519,
      "step": 13902
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8639516830444336,
      "learning_rate": 1.1736255310992556e-05,
      "loss": 0.9815,
      "step": 13903
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9584405422210693,
      "learning_rate": 1.173561639459477e-05,
      "loss": 0.9802,
      "step": 13904
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.281782150268555,
      "learning_rate": 1.1734977478196978e-05,
      "loss": 1.4182,
      "step": 13905
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.059693813323975,
      "learning_rate": 1.1734338561799191e-05,
      "loss": 1.02,
      "step": 13906
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5416264533996582,
      "learning_rate": 1.1733699645401399e-05,
      "loss": 0.7632,
      "step": 13907
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.847886562347412,
      "learning_rate": 1.173306072900361e-05,
      "loss": 1.5194,
      "step": 13908
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.466866374015808,
      "learning_rate": 1.173242181260582e-05,
      "loss": 0.9112,
      "step": 13909
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.7376389503479004,
      "learning_rate": 1.1731782896208032e-05,
      "loss": 0.9063,
      "step": 13910
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.688044786453247,
      "learning_rate": 1.1731143979810242e-05,
      "loss": 1.3744,
      "step": 13911
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7275060415267944,
      "learning_rate": 1.1730505063412453e-05,
      "loss": 0.7937,
      "step": 13912
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7250046730041504,
      "learning_rate": 1.1729866147014663e-05,
      "loss": 0.9491,
      "step": 13913
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7020002603530884,
      "learning_rate": 1.1729227230616875e-05,
      "loss": 0.9928,
      "step": 13914
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2331169843673706,
      "learning_rate": 1.1728588314219085e-05,
      "loss": 0.8387,
      "step": 13915
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6542253494262695,
      "learning_rate": 1.1727949397821296e-05,
      "loss": 0.9075,
      "step": 13916
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.143133640289307,
      "learning_rate": 1.1727310481423506e-05,
      "loss": 1.4719,
      "step": 13917
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.9417519569396973,
      "learning_rate": 1.1726671565025718e-05,
      "loss": 1.2037,
      "step": 13918
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.042879819869995,
      "learning_rate": 1.1726032648627927e-05,
      "loss": 0.9466,
      "step": 13919
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4790092706680298,
      "learning_rate": 1.1725393732230139e-05,
      "loss": 0.8524,
      "step": 13920
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7323952913284302,
      "learning_rate": 1.1724754815832349e-05,
      "loss": 0.9112,
      "step": 13921
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9470070600509644,
      "learning_rate": 1.172411589943456e-05,
      "loss": 0.9491,
      "step": 13922
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9413217306137085,
      "learning_rate": 1.172347698303677e-05,
      "loss": 1.3276,
      "step": 13923
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.79378604888916,
      "learning_rate": 1.1722838066638982e-05,
      "loss": 1.0979,
      "step": 13924
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8817845582962036,
      "learning_rate": 1.1722199150241192e-05,
      "loss": 0.9715,
      "step": 13925
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.5195536613464355,
      "learning_rate": 1.1721560233843403e-05,
      "loss": 1.2557,
      "step": 13926
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.525569438934326,
      "learning_rate": 1.1720921317445613e-05,
      "loss": 0.7261,
      "step": 13927
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4766451120376587,
      "learning_rate": 1.1720282401047825e-05,
      "loss": 0.8219,
      "step": 13928
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.870203971862793,
      "learning_rate": 1.1719643484650034e-05,
      "loss": 1.3823,
      "step": 13929
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8117902278900146,
      "learning_rate": 1.1719004568252246e-05,
      "loss": 0.8923,
      "step": 13930
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2261955738067627,
      "learning_rate": 1.1718365651854456e-05,
      "loss": 0.9523,
      "step": 13931
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5909948348999023,
      "learning_rate": 1.1717726735456667e-05,
      "loss": 0.8971,
      "step": 13932
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.167179584503174,
      "learning_rate": 1.1717087819058877e-05,
      "loss": 0.9252,
      "step": 13933
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.744390845298767,
      "learning_rate": 1.1716448902661089e-05,
      "loss": 0.9224,
      "step": 13934
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.24070405960083,
      "learning_rate": 1.1715809986263299e-05,
      "loss": 1.2825,
      "step": 13935
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6039164066314697,
      "learning_rate": 1.171517106986551e-05,
      "loss": 0.9302,
      "step": 13936
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9785033464431763,
      "learning_rate": 1.171453215346772e-05,
      "loss": 1.1888,
      "step": 13937
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4352551698684692,
      "learning_rate": 1.1713893237069932e-05,
      "loss": 0.4098,
      "step": 13938
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.0726542472839355,
      "learning_rate": 1.171325432067214e-05,
      "loss": 1.0248,
      "step": 13939
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8938754796981812,
      "learning_rate": 1.1712615404274353e-05,
      "loss": 0.8464,
      "step": 13940
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.031210422515869,
      "learning_rate": 1.1711976487876561e-05,
      "loss": 0.9014,
      "step": 13941
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.012418031692505,
      "learning_rate": 1.1711337571478774e-05,
      "loss": 0.8876,
      "step": 13942
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8805439472198486,
      "learning_rate": 1.1710698655080983e-05,
      "loss": 1.1732,
      "step": 13943
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1520659923553467,
      "learning_rate": 1.1710059738683194e-05,
      "loss": 0.9429,
      "step": 13944
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0270328521728516,
      "learning_rate": 1.1709420822285404e-05,
      "loss": 1.1485,
      "step": 13945
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5959800481796265,
      "learning_rate": 1.1708781905887615e-05,
      "loss": 0.9212,
      "step": 13946
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6962932348251343,
      "learning_rate": 1.1708142989489825e-05,
      "loss": 0.776,
      "step": 13947
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1545159816741943,
      "learning_rate": 1.1707504073092037e-05,
      "loss": 1.0691,
      "step": 13948
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.229651689529419,
      "learning_rate": 1.1706865156694247e-05,
      "loss": 0.9287,
      "step": 13949
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6717679500579834,
      "learning_rate": 1.1706226240296458e-05,
      "loss": 0.6274,
      "step": 13950
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0556349754333496,
      "learning_rate": 1.1705587323898668e-05,
      "loss": 0.9036,
      "step": 13951
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3613991737365723,
      "learning_rate": 1.170494840750088e-05,
      "loss": 0.9231,
      "step": 13952
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.838050365447998,
      "learning_rate": 1.170430949110309e-05,
      "loss": 1.6356,
      "step": 13953
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.4509048461914062,
      "learning_rate": 1.1703670574705301e-05,
      "loss": 1.2792,
      "step": 13954
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6295390129089355,
      "learning_rate": 1.1703031658307511e-05,
      "loss": 0.8122,
      "step": 13955
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5472321510314941,
      "learning_rate": 1.1702392741909722e-05,
      "loss": 0.5184,
      "step": 13956
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6487444639205933,
      "learning_rate": 1.1701753825511932e-05,
      "loss": 1.1449,
      "step": 13957
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5890607833862305,
      "learning_rate": 1.1701114909114144e-05,
      "loss": 0.5909,
      "step": 13958
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4971158504486084,
      "learning_rate": 1.1700475992716354e-05,
      "loss": 0.8372,
      "step": 13959
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.690138339996338,
      "learning_rate": 1.1699837076318565e-05,
      "loss": 0.8663,
      "step": 13960
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5970826148986816,
      "learning_rate": 1.1699198159920775e-05,
      "loss": 0.9096,
      "step": 13961
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5394872426986694,
      "learning_rate": 1.1698559243522987e-05,
      "loss": 0.8959,
      "step": 13962
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9036837816238403,
      "learning_rate": 1.1697920327125197e-05,
      "loss": 0.8738,
      "step": 13963
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9059333801269531,
      "learning_rate": 1.1697281410727408e-05,
      "loss": 0.8091,
      "step": 13964
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4237509965896606,
      "learning_rate": 1.1696642494329618e-05,
      "loss": 0.7868,
      "step": 13965
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.197498321533203,
      "learning_rate": 1.169600357793183e-05,
      "loss": 0.8564,
      "step": 13966
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.458581805229187,
      "learning_rate": 1.169536466153404e-05,
      "loss": 0.4383,
      "step": 13967
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7553904056549072,
      "learning_rate": 1.1694725745136251e-05,
      "loss": 0.8936,
      "step": 13968
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.042868614196777,
      "learning_rate": 1.169408682873846e-05,
      "loss": 1.2627,
      "step": 13969
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.2107462882995605,
      "learning_rate": 1.1693447912340672e-05,
      "loss": 0.7156,
      "step": 13970
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8800435066223145,
      "learning_rate": 1.1692808995942882e-05,
      "loss": 0.9006,
      "step": 13971
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.584322452545166,
      "learning_rate": 1.1692170079545094e-05,
      "loss": 0.9638,
      "step": 13972
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.585926055908203,
      "learning_rate": 1.1691531163147304e-05,
      "loss": 0.8535,
      "step": 13973
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.7462263107299805,
      "learning_rate": 1.1690892246749515e-05,
      "loss": 0.8145,
      "step": 13974
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7254031896591187,
      "learning_rate": 1.1690253330351723e-05,
      "loss": 0.7541,
      "step": 13975
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9657127857208252,
      "learning_rate": 1.1689614413953936e-05,
      "loss": 0.8924,
      "step": 13976
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.567244052886963,
      "learning_rate": 1.1688975497556145e-05,
      "loss": 1.8051,
      "step": 13977
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6231573820114136,
      "learning_rate": 1.1688336581158358e-05,
      "loss": 0.911,
      "step": 13978
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9942981004714966,
      "learning_rate": 1.1687697664760566e-05,
      "loss": 0.9147,
      "step": 13979
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.9318392276763916,
      "learning_rate": 1.1687058748362778e-05,
      "loss": 1.0212,
      "step": 13980
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9948514699935913,
      "learning_rate": 1.1686419831964987e-05,
      "loss": 0.9426,
      "step": 13981
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7604161500930786,
      "learning_rate": 1.1685780915567199e-05,
      "loss": 0.8906,
      "step": 13982
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8109638690948486,
      "learning_rate": 1.1685141999169409e-05,
      "loss": 0.9357,
      "step": 13983
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8691520690917969,
      "learning_rate": 1.168450308277162e-05,
      "loss": 0.8933,
      "step": 13984
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.382370948791504,
      "learning_rate": 1.168386416637383e-05,
      "loss": 0.387,
      "step": 13985
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9326039552688599,
      "learning_rate": 1.1683225249976042e-05,
      "loss": 0.7556,
      "step": 13986
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.075448513031006,
      "learning_rate": 1.1682586333578252e-05,
      "loss": 0.9426,
      "step": 13987
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4928356409072876,
      "learning_rate": 1.1681947417180463e-05,
      "loss": 0.8769,
      "step": 13988
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3979482650756836,
      "learning_rate": 1.1681308500782673e-05,
      "loss": 0.7822,
      "step": 13989
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.545418739318848,
      "learning_rate": 1.1680669584384885e-05,
      "loss": 1.2468,
      "step": 13990
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.013760566711426,
      "learning_rate": 1.1680030667987094e-05,
      "loss": 0.7227,
      "step": 13991
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1736366748809814,
      "learning_rate": 1.1679391751589306e-05,
      "loss": 0.954,
      "step": 13992
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8690903186798096,
      "learning_rate": 1.1678752835191516e-05,
      "loss": 1.2477,
      "step": 13993
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.692499041557312,
      "learning_rate": 1.1678113918793727e-05,
      "loss": 0.8947,
      "step": 13994
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.3663649559021,
      "learning_rate": 1.1677475002395937e-05,
      "loss": 1.7927,
      "step": 13995
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2852559089660645,
      "learning_rate": 1.1676836085998149e-05,
      "loss": 0.9579,
      "step": 13996
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.741823196411133,
      "learning_rate": 1.1676197169600359e-05,
      "loss": 1.0608,
      "step": 13997
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9955275058746338,
      "learning_rate": 1.167555825320257e-05,
      "loss": 1.0132,
      "step": 13998
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6787598133087158,
      "learning_rate": 1.167491933680478e-05,
      "loss": 0.921,
      "step": 13999
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8941258192062378,
      "learning_rate": 1.1674280420406992e-05,
      "loss": 0.9241,
      "step": 14000
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7999542951583862,
      "learning_rate": 1.1673641504009201e-05,
      "loss": 0.9335,
      "step": 14001
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.543987989425659,
      "learning_rate": 1.1673002587611413e-05,
      "loss": 0.5565,
      "step": 14002
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7611515522003174,
      "learning_rate": 1.1672363671213623e-05,
      "loss": 0.9265,
      "step": 14003
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6325948238372803,
      "learning_rate": 1.1671724754815834e-05,
      "loss": 0.8662,
      "step": 14004
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8627086877822876,
      "learning_rate": 1.1671085838418044e-05,
      "loss": 0.8492,
      "step": 14005
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8413997888565063,
      "learning_rate": 1.1670446922020256e-05,
      "loss": 0.958,
      "step": 14006
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6322360038757324,
      "learning_rate": 1.1669808005622466e-05,
      "loss": 0.9057,
      "step": 14007
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5111619234085083,
      "learning_rate": 1.1669169089224677e-05,
      "loss": 0.8353,
      "step": 14008
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.304326295852661,
      "learning_rate": 1.1668530172826887e-05,
      "loss": 0.9037,
      "step": 14009
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9060167074203491,
      "learning_rate": 1.1667891256429099e-05,
      "loss": 0.8559,
      "step": 14010
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3456168174743652,
      "learning_rate": 1.1667252340031307e-05,
      "loss": 0.9274,
      "step": 14011
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.579153060913086,
      "learning_rate": 1.166661342363352e-05,
      "loss": 1.5973,
      "step": 14012
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.14711594581604,
      "learning_rate": 1.1665974507235728e-05,
      "loss": 0.7376,
      "step": 14013
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8977274894714355,
      "learning_rate": 1.1665335590837941e-05,
      "loss": 0.8265,
      "step": 14014
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8984911441802979,
      "learning_rate": 1.166469667444015e-05,
      "loss": 1.0946,
      "step": 14015
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.4004769325256348,
      "learning_rate": 1.1664057758042363e-05,
      "loss": 0.9461,
      "step": 14016
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.114537477493286,
      "learning_rate": 1.1663418841644571e-05,
      "loss": 1.0134,
      "step": 14017
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.972541332244873,
      "learning_rate": 1.1662779925246783e-05,
      "loss": 0.7833,
      "step": 14018
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.5268261432647705,
      "learning_rate": 1.1662141008848992e-05,
      "loss": 0.9255,
      "step": 14019
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0809929370880127,
      "learning_rate": 1.1661502092451204e-05,
      "loss": 0.9444,
      "step": 14020
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8415287733078003,
      "learning_rate": 1.1660863176053414e-05,
      "loss": 1.0006,
      "step": 14021
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8910006284713745,
      "learning_rate": 1.1660224259655625e-05,
      "loss": 0.9054,
      "step": 14022
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.6411490440368652,
      "learning_rate": 1.1659585343257835e-05,
      "loss": 1.4446,
      "step": 14023
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9727189540863037,
      "learning_rate": 1.1658946426860047e-05,
      "loss": 1.0423,
      "step": 14024
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1184234619140625,
      "learning_rate": 1.1658307510462257e-05,
      "loss": 0.7931,
      "step": 14025
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8256909847259521,
      "learning_rate": 1.1657668594064468e-05,
      "loss": 0.8714,
      "step": 14026
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.3145592212677,
      "learning_rate": 1.1657029677666678e-05,
      "loss": 1.1084,
      "step": 14027
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4381177425384521,
      "learning_rate": 1.165639076126889e-05,
      "loss": 0.8356,
      "step": 14028
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.067337989807129,
      "learning_rate": 1.16557518448711e-05,
      "loss": 1.0784,
      "step": 14029
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.150390386581421,
      "learning_rate": 1.1655112928473311e-05,
      "loss": 0.8719,
      "step": 14030
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.399217128753662,
      "learning_rate": 1.165447401207552e-05,
      "loss": 0.4623,
      "step": 14031
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6711589097976685,
      "learning_rate": 1.1653835095677732e-05,
      "loss": 0.7448,
      "step": 14032
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6937291622161865,
      "learning_rate": 1.1653196179279942e-05,
      "loss": 0.9343,
      "step": 14033
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.827043056488037,
      "learning_rate": 1.1652557262882154e-05,
      "loss": 0.7404,
      "step": 14034
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7779653072357178,
      "learning_rate": 1.1651918346484364e-05,
      "loss": 0.97,
      "step": 14035
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6823809146881104,
      "learning_rate": 1.1651279430086575e-05,
      "loss": 0.861,
      "step": 14036
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5408999919891357,
      "learning_rate": 1.1650640513688785e-05,
      "loss": 0.8437,
      "step": 14037
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.212947368621826,
      "learning_rate": 1.1650001597290996e-05,
      "loss": 0.9263,
      "step": 14038
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.839839220046997,
      "learning_rate": 1.1649362680893206e-05,
      "loss": 1.289,
      "step": 14039
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.948468804359436,
      "learning_rate": 1.1648723764495418e-05,
      "loss": 0.794,
      "step": 14040
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9593523740768433,
      "learning_rate": 1.1648084848097628e-05,
      "loss": 0.9906,
      "step": 14041
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.429216980934143,
      "learning_rate": 1.164744593169984e-05,
      "loss": 0.8114,
      "step": 14042
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0612640380859375,
      "learning_rate": 1.1646807015302049e-05,
      "loss": 0.9702,
      "step": 14043
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9569294452667236,
      "learning_rate": 1.164616809890426e-05,
      "loss": 0.9259,
      "step": 14044
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9820053577423096,
      "learning_rate": 1.164552918250647e-05,
      "loss": 0.9409,
      "step": 14045
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4846696853637695,
      "learning_rate": 1.1644890266108682e-05,
      "loss": 0.9277,
      "step": 14046
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6871004104614258,
      "learning_rate": 1.1644251349710892e-05,
      "loss": 0.8799,
      "step": 14047
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.793910264968872,
      "learning_rate": 1.1643612433313103e-05,
      "loss": 0.898,
      "step": 14048
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7359243631362915,
      "learning_rate": 1.1642973516915312e-05,
      "loss": 1.0164,
      "step": 14049
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6468218564987183,
      "learning_rate": 1.1642334600517525e-05,
      "loss": 0.9127,
      "step": 14050
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8326916694641113,
      "learning_rate": 1.1641695684119733e-05,
      "loss": 0.8847,
      "step": 14051
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5407129526138306,
      "learning_rate": 1.1641056767721946e-05,
      "loss": 0.8901,
      "step": 14052
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.682706356048584,
      "learning_rate": 1.1640417851324154e-05,
      "loss": 0.8778,
      "step": 14053
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.566288709640503,
      "learning_rate": 1.1639778934926366e-05,
      "loss": 0.3687,
      "step": 14054
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.731495976448059,
      "learning_rate": 1.1639140018528576e-05,
      "loss": 0.8452,
      "step": 14055
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.187316656112671,
      "learning_rate": 1.1638501102130787e-05,
      "loss": 1.0186,
      "step": 14056
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5942327976226807,
      "learning_rate": 1.1637862185732997e-05,
      "loss": 0.874,
      "step": 14057
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8969746828079224,
      "learning_rate": 1.1637223269335209e-05,
      "loss": 1.3875,
      "step": 14058
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9035111665725708,
      "learning_rate": 1.1636584352937419e-05,
      "loss": 1.099,
      "step": 14059
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7885288000106812,
      "learning_rate": 1.163594543653963e-05,
      "loss": 0.8483,
      "step": 14060
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5985910892486572,
      "learning_rate": 1.163530652014184e-05,
      "loss": 0.9084,
      "step": 14061
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.987319231033325,
      "learning_rate": 1.1634667603744052e-05,
      "loss": 1.297,
      "step": 14062
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.631334900856018,
      "learning_rate": 1.1634028687346261e-05,
      "loss": 0.9209,
      "step": 14063
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4112746715545654,
      "learning_rate": 1.1633389770948473e-05,
      "loss": 0.9039,
      "step": 14064
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4680373668670654,
      "learning_rate": 1.1632750854550683e-05,
      "loss": 1.0667,
      "step": 14065
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6652233600616455,
      "learning_rate": 1.1632111938152894e-05,
      "loss": 0.8974,
      "step": 14066
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7332000732421875,
      "learning_rate": 1.1631473021755104e-05,
      "loss": 0.8598,
      "step": 14067
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5945135354995728,
      "learning_rate": 1.1630834105357316e-05,
      "loss": 0.9014,
      "step": 14068
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.12283992767334,
      "learning_rate": 1.1630195188959526e-05,
      "loss": 0.9001,
      "step": 14069
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.472543954849243,
      "learning_rate": 1.1629556272561737e-05,
      "loss": 0.9181,
      "step": 14070
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.5718910694122314,
      "learning_rate": 1.1628917356163947e-05,
      "loss": 0.8054,
      "step": 14071
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.020833969116211,
      "learning_rate": 1.1628278439766159e-05,
      "loss": 0.8112,
      "step": 14072
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9734580516815186,
      "learning_rate": 1.1627639523368368e-05,
      "loss": 0.9085,
      "step": 14073
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2389016151428223,
      "learning_rate": 1.162700060697058e-05,
      "loss": 0.9977,
      "step": 14074
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6008868217468262,
      "learning_rate": 1.162636169057279e-05,
      "loss": 0.3819,
      "step": 14075
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.020822048187256,
      "learning_rate": 1.1625722774175001e-05,
      "loss": 0.9363,
      "step": 14076
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.192739486694336,
      "learning_rate": 1.1625083857777211e-05,
      "loss": 0.7624,
      "step": 14077
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4653913974761963,
      "learning_rate": 1.1624444941379423e-05,
      "loss": 1.0151,
      "step": 14078
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0630686283111572,
      "learning_rate": 1.1623806024981633e-05,
      "loss": 0.8918,
      "step": 14079
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2089953422546387,
      "learning_rate": 1.1623167108583844e-05,
      "loss": 0.894,
      "step": 14080
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.121298789978027,
      "learning_rate": 1.1622528192186054e-05,
      "loss": 1.4931,
      "step": 14081
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.791822910308838,
      "learning_rate": 1.1621889275788266e-05,
      "loss": 0.8854,
      "step": 14082
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7069931030273438,
      "learning_rate": 1.1621250359390475e-05,
      "loss": 0.8603,
      "step": 14083
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.727913498878479,
      "learning_rate": 1.1620611442992687e-05,
      "loss": 0.9295,
      "step": 14084
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.113572597503662,
      "learning_rate": 1.1619972526594895e-05,
      "loss": 1.747,
      "step": 14085
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8872509002685547,
      "learning_rate": 1.1619333610197108e-05,
      "loss": 0.9861,
      "step": 14086
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3386223316192627,
      "learning_rate": 1.1618694693799317e-05,
      "loss": 0.8895,
      "step": 14087
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.13350510597229,
      "learning_rate": 1.161805577740153e-05,
      "loss": 0.9962,
      "step": 14088
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6288241147994995,
      "learning_rate": 1.1617416861003738e-05,
      "loss": 0.9048,
      "step": 14089
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.6607143878936768,
      "learning_rate": 1.161677794460595e-05,
      "loss": 0.7523,
      "step": 14090
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.960269808769226,
      "learning_rate": 1.161613902820816e-05,
      "loss": 1.1983,
      "step": 14091
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6686725616455078,
      "learning_rate": 1.1615500111810371e-05,
      "loss": 0.9144,
      "step": 14092
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.849050998687744,
      "learning_rate": 1.161486119541258e-05,
      "loss": 0.9458,
      "step": 14093
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.1657514572143555,
      "learning_rate": 1.1614222279014792e-05,
      "loss": 1.029,
      "step": 14094
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5386583805084229,
      "learning_rate": 1.1613583362617002e-05,
      "loss": 0.8596,
      "step": 14095
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0418319702148438,
      "learning_rate": 1.1612944446219214e-05,
      "loss": 0.8419,
      "step": 14096
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.7884345054626465,
      "learning_rate": 1.1612305529821424e-05,
      "loss": 1.1448,
      "step": 14097
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.28963041305542,
      "learning_rate": 1.1611666613423635e-05,
      "loss": 1.0442,
      "step": 14098
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0858471393585205,
      "learning_rate": 1.1611027697025845e-05,
      "loss": 0.948,
      "step": 14099
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1511707305908203,
      "learning_rate": 1.1610388780628057e-05,
      "loss": 0.8349,
      "step": 14100
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1238527297973633,
      "learning_rate": 1.1609749864230266e-05,
      "loss": 0.9371,
      "step": 14101
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.591615676879883,
      "learning_rate": 1.1609110947832478e-05,
      "loss": 1.1534,
      "step": 14102
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5957684516906738,
      "learning_rate": 1.1608472031434688e-05,
      "loss": 0.79,
      "step": 14103
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5101901292800903,
      "learning_rate": 1.16078331150369e-05,
      "loss": 0.8159,
      "step": 14104
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8509273529052734,
      "learning_rate": 1.160719419863911e-05,
      "loss": 0.8233,
      "step": 14105
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.9241843223571777,
      "learning_rate": 1.160655528224132e-05,
      "loss": 0.8595,
      "step": 14106
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9301255941390991,
      "learning_rate": 1.160591636584353e-05,
      "loss": 0.4047,
      "step": 14107
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.12518310546875,
      "learning_rate": 1.1605277449445742e-05,
      "loss": 1.1483,
      "step": 14108
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0669360160827637,
      "learning_rate": 1.1604638533047952e-05,
      "loss": 0.8794,
      "step": 14109
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7232211828231812,
      "learning_rate": 1.1603999616650162e-05,
      "loss": 0.9078,
      "step": 14110
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0497958660125732,
      "learning_rate": 1.1603360700252373e-05,
      "loss": 1.2401,
      "step": 14111
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.297696113586426,
      "learning_rate": 1.1602721783854583e-05,
      "loss": 0.9825,
      "step": 14112
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4670259952545166,
      "learning_rate": 1.1602082867456795e-05,
      "loss": 0.9171,
      "step": 14113
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8681238889694214,
      "learning_rate": 1.1601443951059005e-05,
      "loss": 0.9057,
      "step": 14114
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.732466220855713,
      "learning_rate": 1.1600805034661216e-05,
      "loss": 0.894,
      "step": 14115
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5834702253341675,
      "learning_rate": 1.1600166118263424e-05,
      "loss": 0.9447,
      "step": 14116
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.136378765106201,
      "learning_rate": 1.1599527201865638e-05,
      "loss": 0.8861,
      "step": 14117
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.248805284500122,
      "learning_rate": 1.1598888285467846e-05,
      "loss": 0.821,
      "step": 14118
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.990899920463562,
      "learning_rate": 1.1598249369070059e-05,
      "loss": 1.3546,
      "step": 14119
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8969489336013794,
      "learning_rate": 1.1597610452672267e-05,
      "loss": 0.959,
      "step": 14120
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.572145700454712,
      "learning_rate": 1.1596971536274479e-05,
      "loss": 0.8887,
      "step": 14121
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.884505033493042,
      "learning_rate": 1.1596332619876689e-05,
      "loss": 1.0208,
      "step": 14122
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0358808040618896,
      "learning_rate": 1.15956937034789e-05,
      "loss": 1.1546,
      "step": 14123
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.844540596008301,
      "learning_rate": 1.159505478708111e-05,
      "loss": 1.5652,
      "step": 14124
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.641390323638916,
      "learning_rate": 1.1594415870683321e-05,
      "loss": 0.9376,
      "step": 14125
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0575296878814697,
      "learning_rate": 1.1593776954285531e-05,
      "loss": 0.8597,
      "step": 14126
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.504034996032715,
      "learning_rate": 1.1593138037887743e-05,
      "loss": 1.0546,
      "step": 14127
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7789011001586914,
      "learning_rate": 1.1592499121489953e-05,
      "loss": 1.2281,
      "step": 14128
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1024856567382812,
      "learning_rate": 1.1591860205092164e-05,
      "loss": 1.0006,
      "step": 14129
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7095398902893066,
      "learning_rate": 1.1591221288694374e-05,
      "loss": 0.867,
      "step": 14130
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.920616865158081,
      "learning_rate": 1.1590582372296586e-05,
      "loss": 0.9216,
      "step": 14131
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.5315375328063965,
      "learning_rate": 1.1589943455898796e-05,
      "loss": 0.8779,
      "step": 14132
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.8916051387786865,
      "learning_rate": 1.1589304539501007e-05,
      "loss": 1.0239,
      "step": 14133
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.014364719390869,
      "learning_rate": 1.1588665623103217e-05,
      "loss": 0.7972,
      "step": 14134
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.324242353439331,
      "learning_rate": 1.1588026706705428e-05,
      "loss": 0.9864,
      "step": 14135
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.023064374923706,
      "learning_rate": 1.1587387790307638e-05,
      "loss": 0.936,
      "step": 14136
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.885164737701416,
      "learning_rate": 1.158674887390985e-05,
      "loss": 0.8949,
      "step": 14137
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6493613719940186,
      "learning_rate": 1.158610995751206e-05,
      "loss": 0.9751,
      "step": 14138
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9350850582122803,
      "learning_rate": 1.1585471041114271e-05,
      "loss": 0.8139,
      "step": 14139
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.091139078140259,
      "learning_rate": 1.1584832124716481e-05,
      "loss": 0.864,
      "step": 14140
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9694143533706665,
      "learning_rate": 1.1584193208318693e-05,
      "loss": 0.9373,
      "step": 14141
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.082097291946411,
      "learning_rate": 1.1583554291920903e-05,
      "loss": 0.8705,
      "step": 14142
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5926874876022339,
      "learning_rate": 1.1582915375523114e-05,
      "loss": 0.9369,
      "step": 14143
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.848447799682617,
      "learning_rate": 1.1582276459125324e-05,
      "loss": 1.1032,
      "step": 14144
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.907926082611084,
      "learning_rate": 1.1581637542727535e-05,
      "loss": 1.1509,
      "step": 14145
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0520236492156982,
      "learning_rate": 1.1580998626329745e-05,
      "loss": 0.951,
      "step": 14146
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9352599382400513,
      "learning_rate": 1.1580359709931957e-05,
      "loss": 0.8698,
      "step": 14147
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9510432481765747,
      "learning_rate": 1.1579720793534167e-05,
      "loss": 0.891,
      "step": 14148
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4815213680267334,
      "learning_rate": 1.1579081877136378e-05,
      "loss": 0.8895,
      "step": 14149
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.8562378883361816,
      "learning_rate": 1.1578442960738588e-05,
      "loss": 0.9349,
      "step": 14150
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.662022590637207,
      "learning_rate": 1.15778040443408e-05,
      "loss": 0.8921,
      "step": 14151
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.34421706199646,
      "learning_rate": 1.1577165127943008e-05,
      "loss": 1.0685,
      "step": 14152
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.769625186920166,
      "learning_rate": 1.1576526211545221e-05,
      "loss": 0.9324,
      "step": 14153
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.880230188369751,
      "learning_rate": 1.157588729514743e-05,
      "loss": 0.8795,
      "step": 14154
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8183385133743286,
      "learning_rate": 1.1575248378749642e-05,
      "loss": 0.8288,
      "step": 14155
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.919558048248291,
      "learning_rate": 1.157460946235185e-05,
      "loss": 0.9531,
      "step": 14156
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1355597972869873,
      "learning_rate": 1.1573970545954062e-05,
      "loss": 1.0561,
      "step": 14157
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9069347381591797,
      "learning_rate": 1.1573331629556272e-05,
      "loss": 0.8775,
      "step": 14158
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2158119678497314,
      "learning_rate": 1.1572692713158484e-05,
      "loss": 0.8505,
      "step": 14159
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1578493118286133,
      "learning_rate": 1.1572053796760693e-05,
      "loss": 0.7769,
      "step": 14160
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.199517250061035,
      "learning_rate": 1.1571414880362905e-05,
      "loss": 0.8406,
      "step": 14161
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.125702381134033,
      "learning_rate": 1.1570775963965115e-05,
      "loss": 0.8716,
      "step": 14162
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.0261311531066895,
      "learning_rate": 1.1570137047567326e-05,
      "loss": 0.8686,
      "step": 14163
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.662651538848877,
      "learning_rate": 1.1569498131169536e-05,
      "loss": 0.8692,
      "step": 14164
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1504807472229004,
      "learning_rate": 1.1568859214771748e-05,
      "loss": 0.739,
      "step": 14165
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.5966527462005615,
      "learning_rate": 1.1568220298373958e-05,
      "loss": 1.0573,
      "step": 14166
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9621981382369995,
      "learning_rate": 1.156758138197617e-05,
      "loss": 0.9177,
      "step": 14167
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.608675718307495,
      "learning_rate": 1.1566942465578379e-05,
      "loss": 0.9617,
      "step": 14168
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5108546018600464,
      "learning_rate": 1.156630354918059e-05,
      "loss": 0.7537,
      "step": 14169
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5641765594482422,
      "learning_rate": 1.15656646327828e-05,
      "loss": 0.9123,
      "step": 14170
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.2504029273986816,
      "learning_rate": 1.1565025716385012e-05,
      "loss": 1.228,
      "step": 14171
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4245524406433105,
      "learning_rate": 1.1564386799987222e-05,
      "loss": 0.4966,
      "step": 14172
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.02897047996521,
      "learning_rate": 1.1563747883589433e-05,
      "loss": 1.0276,
      "step": 14173
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.307694435119629,
      "learning_rate": 1.1563108967191643e-05,
      "loss": 1.1177,
      "step": 14174
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.713049411773682,
      "learning_rate": 1.1562470050793855e-05,
      "loss": 1.5275,
      "step": 14175
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.7057554721832275,
      "learning_rate": 1.1561831134396065e-05,
      "loss": 0.6696,
      "step": 14176
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.285029411315918,
      "learning_rate": 1.1561192217998276e-05,
      "loss": 1.2834,
      "step": 14177
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8115657567977905,
      "learning_rate": 1.1560553301600486e-05,
      "loss": 0.9742,
      "step": 14178
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8128803968429565,
      "learning_rate": 1.1559914385202698e-05,
      "loss": 0.8508,
      "step": 14179
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7600765228271484,
      "learning_rate": 1.1559275468804907e-05,
      "loss": 0.837,
      "step": 14180
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.395592451095581,
      "learning_rate": 1.1558636552407119e-05,
      "loss": 0.9407,
      "step": 14181
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6668773889541626,
      "learning_rate": 1.1557997636009329e-05,
      "loss": 0.8874,
      "step": 14182
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.336747407913208,
      "learning_rate": 1.155735871961154e-05,
      "loss": 1.0485,
      "step": 14183
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.071728229522705,
      "learning_rate": 1.155671980321375e-05,
      "loss": 1.1771,
      "step": 14184
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8700844049453735,
      "learning_rate": 1.1556080886815962e-05,
      "loss": 0.8204,
      "step": 14185
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6427333354949951,
      "learning_rate": 1.1555441970418172e-05,
      "loss": 1.1783,
      "step": 14186
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.265749454498291,
      "learning_rate": 1.1554803054020383e-05,
      "loss": 0.5252,
      "step": 14187
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6601845026016235,
      "learning_rate": 1.1554164137622591e-05,
      "loss": 0.8966,
      "step": 14188
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7195340394973755,
      "learning_rate": 1.1553525221224805e-05,
      "loss": 0.9093,
      "step": 14189
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.3646039962768555,
      "learning_rate": 1.1552886304827013e-05,
      "loss": 1.5518,
      "step": 14190
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.2061750888824463,
      "learning_rate": 1.1552247388429226e-05,
      "loss": 0.9608,
      "step": 14191
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9026575088500977,
      "learning_rate": 1.1551608472031434e-05,
      "loss": 0.8647,
      "step": 14192
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.709105372428894,
      "learning_rate": 1.1550969555633647e-05,
      "loss": 0.8131,
      "step": 14193
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8587874174118042,
      "learning_rate": 1.1550330639235856e-05,
      "loss": 0.9907,
      "step": 14194
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.8958566188812256,
      "learning_rate": 1.1549691722838067e-05,
      "loss": 1.0428,
      "step": 14195
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5059058666229248,
      "learning_rate": 1.1549052806440277e-05,
      "loss": 0.8364,
      "step": 14196
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8593565225601196,
      "learning_rate": 1.1548413890042488e-05,
      "loss": 0.8096,
      "step": 14197
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6103469133377075,
      "learning_rate": 1.1547774973644698e-05,
      "loss": 0.9124,
      "step": 14198
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.54140043258667,
      "learning_rate": 1.154713605724691e-05,
      "loss": 1.0489,
      "step": 14199
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7271722555160522,
      "learning_rate": 1.154649714084912e-05,
      "loss": 0.9441,
      "step": 14200
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.679490566253662,
      "learning_rate": 1.1545858224451331e-05,
      "loss": 0.917,
      "step": 14201
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4018259048461914,
      "learning_rate": 1.1545219308053541e-05,
      "loss": 0.9559,
      "step": 14202
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0214173793792725,
      "learning_rate": 1.1544580391655753e-05,
      "loss": 0.9098,
      "step": 14203
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.835166573524475,
      "learning_rate": 1.1543941475257963e-05,
      "loss": 0.9291,
      "step": 14204
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8514235019683838,
      "learning_rate": 1.1543302558860174e-05,
      "loss": 0.8302,
      "step": 14205
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9505672454833984,
      "learning_rate": 1.1542663642462384e-05,
      "loss": 0.9266,
      "step": 14206
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.05190372467041,
      "learning_rate": 1.1542024726064595e-05,
      "loss": 1.1838,
      "step": 14207
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9071581363677979,
      "learning_rate": 1.1541385809666805e-05,
      "loss": 0.963,
      "step": 14208
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.863920211791992,
      "learning_rate": 1.1540746893269017e-05,
      "loss": 1.3958,
      "step": 14209
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.018486738204956,
      "learning_rate": 1.1540107976871227e-05,
      "loss": 0.8842,
      "step": 14210
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.209924697875977,
      "learning_rate": 1.1539469060473438e-05,
      "loss": 1.2048,
      "step": 14211
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.7387118339538574,
      "learning_rate": 1.1538830144075648e-05,
      "loss": 0.5602,
      "step": 14212
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7474374771118164,
      "learning_rate": 1.153819122767786e-05,
      "loss": 0.8975,
      "step": 14213
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7661688327789307,
      "learning_rate": 1.153755231128007e-05,
      "loss": 0.7796,
      "step": 14214
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5150009393692017,
      "learning_rate": 1.1536913394882281e-05,
      "loss": 0.8285,
      "step": 14215
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.507143497467041,
      "learning_rate": 1.1536274478484491e-05,
      "loss": 0.5548,
      "step": 14216
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1383025646209717,
      "learning_rate": 1.1535635562086702e-05,
      "loss": 1.1571,
      "step": 14217
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.26262092590332,
      "learning_rate": 1.1534996645688912e-05,
      "loss": 1.0256,
      "step": 14218
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9011385440826416,
      "learning_rate": 1.1534357729291124e-05,
      "loss": 0.7206,
      "step": 14219
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.850170612335205,
      "learning_rate": 1.1533718812893334e-05,
      "loss": 0.9921,
      "step": 14220
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.949382781982422,
      "learning_rate": 1.1533079896495545e-05,
      "loss": 1.4745,
      "step": 14221
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0356523990631104,
      "learning_rate": 1.1532440980097755e-05,
      "loss": 0.9402,
      "step": 14222
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7353436946868896,
      "learning_rate": 1.1531802063699967e-05,
      "loss": 0.9534,
      "step": 14223
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.086759090423584,
      "learning_rate": 1.1531163147302175e-05,
      "loss": 0.903,
      "step": 14224
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.142622470855713,
      "learning_rate": 1.1530524230904388e-05,
      "loss": 1.1806,
      "step": 14225
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3743622303009033,
      "learning_rate": 1.1529885314506596e-05,
      "loss": 1.0318,
      "step": 14226
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4584264755249023,
      "learning_rate": 1.152924639810881e-05,
      "loss": 0.9352,
      "step": 14227
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.183130979537964,
      "learning_rate": 1.1528607481711018e-05,
      "loss": 1.2668,
      "step": 14228
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1032333374023438,
      "learning_rate": 1.1527968565313231e-05,
      "loss": 1.176,
      "step": 14229
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0843260288238525,
      "learning_rate": 1.1527329648915439e-05,
      "loss": 0.7957,
      "step": 14230
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.722325086593628,
      "learning_rate": 1.152669073251765e-05,
      "loss": 1.1469,
      "step": 14231
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.70588219165802,
      "learning_rate": 1.152605181611986e-05,
      "loss": 1.0341,
      "step": 14232
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.6367781162261963,
      "learning_rate": 1.1525412899722072e-05,
      "loss": 1.5326,
      "step": 14233
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3751473426818848,
      "learning_rate": 1.1524773983324282e-05,
      "loss": 0.903,
      "step": 14234
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.479907274246216,
      "learning_rate": 1.1524135066926493e-05,
      "loss": 1.0341,
      "step": 14235
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1583471298217773,
      "learning_rate": 1.1523496150528703e-05,
      "loss": 0.9737,
      "step": 14236
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7166780233383179,
      "learning_rate": 1.1522857234130915e-05,
      "loss": 0.8133,
      "step": 14237
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.671036958694458,
      "learning_rate": 1.1522218317733125e-05,
      "loss": 1.034,
      "step": 14238
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.94815731048584,
      "learning_rate": 1.1521579401335336e-05,
      "loss": 0.9187,
      "step": 14239
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5317065715789795,
      "learning_rate": 1.1520940484937546e-05,
      "loss": 0.8481,
      "step": 14240
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.424832344055176,
      "learning_rate": 1.1520301568539758e-05,
      "loss": 1.3288,
      "step": 14241
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9053454399108887,
      "learning_rate": 1.1519662652141967e-05,
      "loss": 0.9289,
      "step": 14242
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9076918363571167,
      "learning_rate": 1.1519023735744179e-05,
      "loss": 0.4038,
      "step": 14243
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.9979724884033203,
      "learning_rate": 1.1518384819346389e-05,
      "loss": 0.8262,
      "step": 14244
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.01011061668396,
      "learning_rate": 1.15177459029486e-05,
      "loss": 1.2864,
      "step": 14245
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7165788412094116,
      "learning_rate": 1.151710698655081e-05,
      "loss": 0.9702,
      "step": 14246
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6125223636627197,
      "learning_rate": 1.1516468070153022e-05,
      "loss": 0.8625,
      "step": 14247
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.759390115737915,
      "learning_rate": 1.1515829153755232e-05,
      "loss": 0.8315,
      "step": 14248
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0343008041381836,
      "learning_rate": 1.1515190237357443e-05,
      "loss": 1.0474,
      "step": 14249
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4515671730041504,
      "learning_rate": 1.1514551320959653e-05,
      "loss": 0.7793,
      "step": 14250
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.187060832977295,
      "learning_rate": 1.1513912404561865e-05,
      "loss": 1.203,
      "step": 14251
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.226508140563965,
      "learning_rate": 1.1513273488164074e-05,
      "loss": 1.1655,
      "step": 14252
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9276111125946045,
      "learning_rate": 1.1512634571766286e-05,
      "loss": 1.148,
      "step": 14253
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.179034471511841,
      "learning_rate": 1.1511995655368496e-05,
      "loss": 0.939,
      "step": 14254
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.6055922508239746,
      "learning_rate": 1.1511356738970707e-05,
      "loss": 1.1014,
      "step": 14255
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.659454822540283,
      "learning_rate": 1.1510717822572917e-05,
      "loss": 0.6643,
      "step": 14256
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.558560848236084,
      "learning_rate": 1.1510078906175129e-05,
      "loss": 1.5869,
      "step": 14257
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.109968662261963,
      "learning_rate": 1.1509439989777339e-05,
      "loss": 1.3142,
      "step": 14258
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0742805004119873,
      "learning_rate": 1.150880107337955e-05,
      "loss": 0.8036,
      "step": 14259
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.6359472274780273,
      "learning_rate": 1.150816215698176e-05,
      "loss": 1.2612,
      "step": 14260
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8176974058151245,
      "learning_rate": 1.1507523240583972e-05,
      "loss": 0.8339,
      "step": 14261
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8593387603759766,
      "learning_rate": 1.150688432418618e-05,
      "loss": 1.004,
      "step": 14262
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.9773306846618652,
      "learning_rate": 1.1506245407788393e-05,
      "loss": 1.2518,
      "step": 14263
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6355018615722656,
      "learning_rate": 1.1505606491390601e-05,
      "loss": 0.4006,
      "step": 14264
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.78300940990448,
      "learning_rate": 1.1504967574992814e-05,
      "loss": 0.936,
      "step": 14265
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.331955432891846,
      "learning_rate": 1.1504328658595023e-05,
      "loss": 1.3464,
      "step": 14266
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4662787914276123,
      "learning_rate": 1.1503689742197234e-05,
      "loss": 1.1863,
      "step": 14267
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.4057745933532715,
      "learning_rate": 1.1503050825799444e-05,
      "loss": 1.3088,
      "step": 14268
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.973193883895874,
      "learning_rate": 1.1502411909401655e-05,
      "loss": 1.1193,
      "step": 14269
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0201125144958496,
      "learning_rate": 1.1501772993003865e-05,
      "loss": 0.9297,
      "step": 14270
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.952980637550354,
      "learning_rate": 1.1501134076606077e-05,
      "loss": 0.8588,
      "step": 14271
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8648961782455444,
      "learning_rate": 1.1500495160208287e-05,
      "loss": 1.2848,
      "step": 14272
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4168083667755127,
      "learning_rate": 1.1499856243810498e-05,
      "loss": 1.0587,
      "step": 14273
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4349758625030518,
      "learning_rate": 1.1499217327412708e-05,
      "loss": 0.7805,
      "step": 14274
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6485681533813477,
      "learning_rate": 1.149857841101492e-05,
      "loss": 0.8923,
      "step": 14275
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.035538911819458,
      "learning_rate": 1.149793949461713e-05,
      "loss": 0.8626,
      "step": 14276
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8434392213821411,
      "learning_rate": 1.1497300578219341e-05,
      "loss": 1.0074,
      "step": 14277
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.1533098220825195,
      "learning_rate": 1.1496661661821551e-05,
      "loss": 0.8723,
      "step": 14278
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7405182123184204,
      "learning_rate": 1.1496022745423762e-05,
      "loss": 0.8569,
      "step": 14279
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.735637903213501,
      "learning_rate": 1.1495383829025972e-05,
      "loss": 0.9829,
      "step": 14280
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.110942840576172,
      "learning_rate": 1.1494744912628184e-05,
      "loss": 1.2999,
      "step": 14281
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.012373924255371,
      "learning_rate": 1.1494105996230394e-05,
      "loss": 1.2606,
      "step": 14282
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9935474395751953,
      "learning_rate": 1.1493467079832605e-05,
      "loss": 0.8529,
      "step": 14283
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9295494556427002,
      "learning_rate": 1.1492828163434815e-05,
      "loss": 0.7837,
      "step": 14284
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.32748556137085,
      "learning_rate": 1.1492189247037027e-05,
      "loss": 1.0413,
      "step": 14285
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.296208381652832,
      "learning_rate": 1.1491550330639237e-05,
      "loss": 0.7207,
      "step": 14286
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3915694952011108,
      "learning_rate": 1.1490911414241448e-05,
      "loss": 0.8574,
      "step": 14287
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6614067554473877,
      "learning_rate": 1.1490272497843658e-05,
      "loss": 0.9246,
      "step": 14288
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.121027708053589,
      "learning_rate": 1.148963358144587e-05,
      "loss": 1.0962,
      "step": 14289
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5159693956375122,
      "learning_rate": 1.148899466504808e-05,
      "loss": 0.9766,
      "step": 14290
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1537179946899414,
      "learning_rate": 1.1488355748650291e-05,
      "loss": 0.9235,
      "step": 14291
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.67948317527771,
      "learning_rate": 1.14877168322525e-05,
      "loss": 0.8793,
      "step": 14292
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1732208728790283,
      "learning_rate": 1.1487077915854712e-05,
      "loss": 0.9533,
      "step": 14293
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.947584629058838,
      "learning_rate": 1.1486438999456922e-05,
      "loss": 1.0389,
      "step": 14294
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.146411418914795,
      "learning_rate": 1.1485800083059134e-05,
      "loss": 0.8826,
      "step": 14295
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.057229518890381,
      "learning_rate": 1.1485161166661344e-05,
      "loss": 0.9894,
      "step": 14296
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9994219541549683,
      "learning_rate": 1.1484522250263555e-05,
      "loss": 0.8973,
      "step": 14297
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4590554237365723,
      "learning_rate": 1.1483883333865763e-05,
      "loss": 0.8166,
      "step": 14298
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.048335552215576,
      "learning_rate": 1.1483244417467976e-05,
      "loss": 1.1511,
      "step": 14299
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4663087129592896,
      "learning_rate": 1.1482605501070185e-05,
      "loss": 0.7709,
      "step": 14300
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.487961769104004,
      "learning_rate": 1.1481966584672398e-05,
      "loss": 1.2866,
      "step": 14301
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.453890562057495,
      "learning_rate": 1.1481327668274606e-05,
      "loss": 1.0193,
      "step": 14302
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1105031967163086,
      "learning_rate": 1.1480688751876818e-05,
      "loss": 0.9114,
      "step": 14303
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1572458744049072,
      "learning_rate": 1.1480049835479027e-05,
      "loss": 1.0099,
      "step": 14304
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7037007808685303,
      "learning_rate": 1.1479410919081239e-05,
      "loss": 0.8691,
      "step": 14305
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6087123155593872,
      "learning_rate": 1.1478772002683449e-05,
      "loss": 0.8764,
      "step": 14306
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.009028911590576,
      "learning_rate": 1.147813308628566e-05,
      "loss": 1.0932,
      "step": 14307
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2811498641967773,
      "learning_rate": 1.147749416988787e-05,
      "loss": 0.9102,
      "step": 14308
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3337630033493042,
      "learning_rate": 1.1476855253490082e-05,
      "loss": 0.4158,
      "step": 14309
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7411415576934814,
      "learning_rate": 1.1476216337092292e-05,
      "loss": 0.8513,
      "step": 14310
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.941143035888672,
      "learning_rate": 1.1475577420694503e-05,
      "loss": 1.0218,
      "step": 14311
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.161735773086548,
      "learning_rate": 1.1474938504296713e-05,
      "loss": 0.7936,
      "step": 14312
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4992750883102417,
      "learning_rate": 1.1474299587898925e-05,
      "loss": 0.8981,
      "step": 14313
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.096416711807251,
      "learning_rate": 1.1473660671501134e-05,
      "loss": 1.0661,
      "step": 14314
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.764204740524292,
      "learning_rate": 1.1473021755103346e-05,
      "loss": 1.1907,
      "step": 14315
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0545527935028076,
      "learning_rate": 1.1472382838705556e-05,
      "loss": 0.8714,
      "step": 14316
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7922929525375366,
      "learning_rate": 1.1471743922307767e-05,
      "loss": 1.0324,
      "step": 14317
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3709255456924438,
      "learning_rate": 1.1471105005909977e-05,
      "loss": 0.4549,
      "step": 14318
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.270787239074707,
      "learning_rate": 1.1470466089512189e-05,
      "loss": 0.9346,
      "step": 14319
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.909545421600342,
      "learning_rate": 1.1469827173114399e-05,
      "loss": 0.5697,
      "step": 14320
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2494218349456787,
      "learning_rate": 1.146918825671661e-05,
      "loss": 0.8506,
      "step": 14321
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2060155868530273,
      "learning_rate": 1.146854934031882e-05,
      "loss": 0.9109,
      "step": 14322
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.48773992061615,
      "learning_rate": 1.1467910423921032e-05,
      "loss": 0.8677,
      "step": 14323
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4935662746429443,
      "learning_rate": 1.1467271507523241e-05,
      "loss": 0.8509,
      "step": 14324
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5671961307525635,
      "learning_rate": 1.1466632591125453e-05,
      "loss": 0.9188,
      "step": 14325
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.0665555000305176,
      "learning_rate": 1.1465993674727663e-05,
      "loss": 0.7333,
      "step": 14326
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8554919958114624,
      "learning_rate": 1.1465354758329874e-05,
      "loss": 0.882,
      "step": 14327
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2279207706451416,
      "learning_rate": 1.1464715841932084e-05,
      "loss": 0.7439,
      "step": 14328
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7312413454055786,
      "learning_rate": 1.1464076925534296e-05,
      "loss": 0.8725,
      "step": 14329
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.575370192527771,
      "learning_rate": 1.1463438009136506e-05,
      "loss": 0.8449,
      "step": 14330
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.745591163635254,
      "learning_rate": 1.1462799092738717e-05,
      "loss": 0.9041,
      "step": 14331
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9351006746292114,
      "learning_rate": 1.1462160176340927e-05,
      "loss": 0.8969,
      "step": 14332
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.098557472229004,
      "learning_rate": 1.1461521259943139e-05,
      "loss": 0.8157,
      "step": 14333
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.300541400909424,
      "learning_rate": 1.1460882343545347e-05,
      "loss": 0.8447,
      "step": 14334
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7345993518829346,
      "learning_rate": 1.146024342714756e-05,
      "loss": 0.9626,
      "step": 14335
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7409974336624146,
      "learning_rate": 1.1459604510749768e-05,
      "loss": 0.8926,
      "step": 14336
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3149611949920654,
      "learning_rate": 1.1458965594351981e-05,
      "loss": 0.8681,
      "step": 14337
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.982199668884277,
      "learning_rate": 1.145832667795419e-05,
      "loss": 1.4811,
      "step": 14338
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8911479711532593,
      "learning_rate": 1.1457687761556403e-05,
      "loss": 0.9005,
      "step": 14339
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.559740662574768,
      "learning_rate": 1.1457048845158611e-05,
      "loss": 0.861,
      "step": 14340
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1995608806610107,
      "learning_rate": 1.1456409928760822e-05,
      "loss": 0.9093,
      "step": 14341
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9506959915161133,
      "learning_rate": 1.1455771012363032e-05,
      "loss": 0.8915,
      "step": 14342
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.6343939304351807,
      "learning_rate": 1.1455132095965244e-05,
      "loss": 0.9237,
      "step": 14343
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.273421287536621,
      "learning_rate": 1.1454493179567454e-05,
      "loss": 1.1912,
      "step": 14344
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.68809175491333,
      "learning_rate": 1.1453854263169665e-05,
      "loss": 0.9255,
      "step": 14345
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.911497712135315,
      "learning_rate": 1.1453215346771875e-05,
      "loss": 0.9347,
      "step": 14346
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2546753883361816,
      "learning_rate": 1.1452576430374087e-05,
      "loss": 0.8256,
      "step": 14347
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.036632776260376,
      "learning_rate": 1.1451937513976297e-05,
      "loss": 0.8444,
      "step": 14348
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7352001667022705,
      "learning_rate": 1.1451298597578508e-05,
      "loss": 0.7388,
      "step": 14349
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5757752656936646,
      "learning_rate": 1.1450659681180718e-05,
      "loss": 0.7779,
      "step": 14350
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.833974838256836,
      "learning_rate": 1.145002076478293e-05,
      "loss": 0.8666,
      "step": 14351
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1718106269836426,
      "learning_rate": 1.144938184838514e-05,
      "loss": 1.0153,
      "step": 14352
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6361762285232544,
      "learning_rate": 1.1448742931987351e-05,
      "loss": 0.967,
      "step": 14353
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8638006448745728,
      "learning_rate": 1.144810401558956e-05,
      "loss": 0.8444,
      "step": 14354
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.682281732559204,
      "learning_rate": 1.1447465099191772e-05,
      "loss": 0.8161,
      "step": 14355
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.61628258228302,
      "learning_rate": 1.1446826182793982e-05,
      "loss": 0.9637,
      "step": 14356
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6227396726608276,
      "learning_rate": 1.1446187266396194e-05,
      "loss": 0.9355,
      "step": 14357
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6392220258712769,
      "learning_rate": 1.1445548349998404e-05,
      "loss": 0.9289,
      "step": 14358
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5149482488632202,
      "learning_rate": 1.1444909433600615e-05,
      "loss": 0.8034,
      "step": 14359
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4967550039291382,
      "learning_rate": 1.1444270517202825e-05,
      "loss": 0.8436,
      "step": 14360
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8831031322479248,
      "learning_rate": 1.1443631600805036e-05,
      "loss": 0.9408,
      "step": 14361
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9124946594238281,
      "learning_rate": 1.1442992684407246e-05,
      "loss": 1.0195,
      "step": 14362
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4212398529052734,
      "learning_rate": 1.1442353768009458e-05,
      "loss": 0.7892,
      "step": 14363
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.6595513820648193,
      "learning_rate": 1.1441714851611668e-05,
      "loss": 1.2613,
      "step": 14364
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8949023485183716,
      "learning_rate": 1.144107593521388e-05,
      "loss": 0.9303,
      "step": 14365
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8650671243667603,
      "learning_rate": 1.1440437018816089e-05,
      "loss": 0.8762,
      "step": 14366
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8384859561920166,
      "learning_rate": 1.14397981024183e-05,
      "loss": 0.9603,
      "step": 14367
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.5453224182128906,
      "learning_rate": 1.143915918602051e-05,
      "loss": 1.6043,
      "step": 14368
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.757382392883301,
      "learning_rate": 1.1438520269622722e-05,
      "loss": 0.9771,
      "step": 14369
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7375811338424683,
      "learning_rate": 1.143788135322493e-05,
      "loss": 0.8745,
      "step": 14370
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0662596225738525,
      "learning_rate": 1.1437242436827143e-05,
      "loss": 1.0434,
      "step": 14371
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1659152507781982,
      "learning_rate": 1.1436603520429352e-05,
      "loss": 0.8726,
      "step": 14372
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9631531238555908,
      "learning_rate": 1.1435964604031565e-05,
      "loss": 0.9544,
      "step": 14373
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8941047191619873,
      "learning_rate": 1.1435325687633773e-05,
      "loss": 0.9521,
      "step": 14374
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6088119745254517,
      "learning_rate": 1.1434686771235986e-05,
      "loss": 0.8535,
      "step": 14375
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8517353534698486,
      "learning_rate": 1.1434047854838194e-05,
      "loss": 0.9046,
      "step": 14376
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2982139587402344,
      "learning_rate": 1.1433408938440406e-05,
      "loss": 1.0156,
      "step": 14377
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.680031180381775,
      "learning_rate": 1.1432770022042616e-05,
      "loss": 0.8546,
      "step": 14378
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.004638433456421,
      "learning_rate": 1.1432131105644827e-05,
      "loss": 0.688,
      "step": 14379
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0083670616149902,
      "learning_rate": 1.1431492189247037e-05,
      "loss": 0.9801,
      "step": 14380
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3907763957977295,
      "learning_rate": 1.1430853272849249e-05,
      "loss": 0.9268,
      "step": 14381
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4390288591384888,
      "learning_rate": 1.1430214356451459e-05,
      "loss": 0.4933,
      "step": 14382
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.302271604537964,
      "learning_rate": 1.142957544005367e-05,
      "loss": 0.7934,
      "step": 14383
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.624780297279358,
      "learning_rate": 1.142893652365588e-05,
      "loss": 1.0407,
      "step": 14384
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5631996393203735,
      "learning_rate": 1.1428297607258092e-05,
      "loss": 0.841,
      "step": 14385
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.405672788619995,
      "learning_rate": 1.1427658690860301e-05,
      "loss": 1.1308,
      "step": 14386
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8513987064361572,
      "learning_rate": 1.1427019774462513e-05,
      "loss": 0.6334,
      "step": 14387
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5932270288467407,
      "learning_rate": 1.1426380858064723e-05,
      "loss": 0.7992,
      "step": 14388
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9583826065063477,
      "learning_rate": 1.1425741941666934e-05,
      "loss": 0.8693,
      "step": 14389
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.652496576309204,
      "learning_rate": 1.1425103025269144e-05,
      "loss": 0.88,
      "step": 14390
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9126790761947632,
      "learning_rate": 1.1424464108871356e-05,
      "loss": 1.1007,
      "step": 14391
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0263476371765137,
      "learning_rate": 1.1423825192473566e-05,
      "loss": 1.0304,
      "step": 14392
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.251922130584717,
      "learning_rate": 1.1423186276075777e-05,
      "loss": 1.3139,
      "step": 14393
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.612858772277832,
      "learning_rate": 1.1422547359677987e-05,
      "loss": 1.1436,
      "step": 14394
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3804209232330322,
      "learning_rate": 1.1421908443280199e-05,
      "loss": 0.8876,
      "step": 14395
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.482163906097412,
      "learning_rate": 1.1421269526882408e-05,
      "loss": 0.8924,
      "step": 14396
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8079897165298462,
      "learning_rate": 1.142063061048462e-05,
      "loss": 1.0079,
      "step": 14397
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3761024475097656,
      "learning_rate": 1.141999169408683e-05,
      "loss": 0.4205,
      "step": 14398
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1016528606414795,
      "learning_rate": 1.1419352777689041e-05,
      "loss": 0.9238,
      "step": 14399
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1416828632354736,
      "learning_rate": 1.1418713861291251e-05,
      "loss": 0.9636,
      "step": 14400
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3701162338256836,
      "learning_rate": 1.1418074944893463e-05,
      "loss": 1.1234,
      "step": 14401
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3589187860488892,
      "learning_rate": 1.1417436028495673e-05,
      "loss": 0.7975,
      "step": 14402
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.045593738555908,
      "learning_rate": 1.1416797112097884e-05,
      "loss": 0.9135,
      "step": 14403
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7310982942581177,
      "learning_rate": 1.1416158195700094e-05,
      "loss": 0.8704,
      "step": 14404
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.023862361907959,
      "learning_rate": 1.1415519279302306e-05,
      "loss": 0.9557,
      "step": 14405
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.513514041900635,
      "learning_rate": 1.1414880362904515e-05,
      "loss": 0.9912,
      "step": 14406
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7237073183059692,
      "learning_rate": 1.1414241446506727e-05,
      "loss": 1.0026,
      "step": 14407
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9919795989990234,
      "learning_rate": 1.1413602530108935e-05,
      "loss": 0.7887,
      "step": 14408
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0853633880615234,
      "learning_rate": 1.1412963613711148e-05,
      "loss": 1.2914,
      "step": 14409
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0191731452941895,
      "learning_rate": 1.1412324697313357e-05,
      "loss": 0.9631,
      "step": 14410
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1346285343170166,
      "learning_rate": 1.141168578091557e-05,
      "loss": 0.9562,
      "step": 14411
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5889360904693604,
      "learning_rate": 1.1411046864517778e-05,
      "loss": 0.8813,
      "step": 14412
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9290066957473755,
      "learning_rate": 1.141040794811999e-05,
      "loss": 1.0177,
      "step": 14413
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.516481876373291,
      "learning_rate": 1.14097690317222e-05,
      "loss": 0.8413,
      "step": 14414
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.047853469848633,
      "learning_rate": 1.1409130115324411e-05,
      "loss": 0.9844,
      "step": 14415
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7473865747451782,
      "learning_rate": 1.140849119892662e-05,
      "loss": 0.9121,
      "step": 14416
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.683763861656189,
      "learning_rate": 1.1407852282528832e-05,
      "loss": 0.9271,
      "step": 14417
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.41874361038208,
      "learning_rate": 1.1407213366131042e-05,
      "loss": 1.5281,
      "step": 14418
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.954866886138916,
      "learning_rate": 1.1406574449733254e-05,
      "loss": 1.158,
      "step": 14419
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.314938545227051,
      "learning_rate": 1.1405935533335464e-05,
      "loss": 1.4475,
      "step": 14420
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.82625150680542,
      "learning_rate": 1.1405296616937675e-05,
      "loss": 0.8124,
      "step": 14421
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.383532166481018,
      "learning_rate": 1.1404657700539885e-05,
      "loss": 0.3869,
      "step": 14422
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5179517269134521,
      "learning_rate": 1.1404018784142097e-05,
      "loss": 0.8858,
      "step": 14423
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.1170077323913574,
      "learning_rate": 1.1403379867744306e-05,
      "loss": 0.8292,
      "step": 14424
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9300155639648438,
      "learning_rate": 1.1402740951346518e-05,
      "loss": 0.8328,
      "step": 14425
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6657460927963257,
      "learning_rate": 1.1402102034948728e-05,
      "loss": 0.9603,
      "step": 14426
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0908236503601074,
      "learning_rate": 1.140146311855094e-05,
      "loss": 1.167,
      "step": 14427
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0665664672851562,
      "learning_rate": 1.140082420215315e-05,
      "loss": 1.1793,
      "step": 14428
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8406974077224731,
      "learning_rate": 1.140018528575536e-05,
      "loss": 0.9336,
      "step": 14429
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.504968523979187,
      "learning_rate": 1.139954636935757e-05,
      "loss": 0.771,
      "step": 14430
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9873275756835938,
      "learning_rate": 1.1398907452959782e-05,
      "loss": 0.6794,
      "step": 14431
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.627642035484314,
      "learning_rate": 1.1398268536561992e-05,
      "loss": 1.269,
      "step": 14432
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.526778221130371,
      "learning_rate": 1.1397629620164204e-05,
      "loss": 1.2424,
      "step": 14433
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.745477557182312,
      "learning_rate": 1.1396990703766413e-05,
      "loss": 0.842,
      "step": 14434
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9039547443389893,
      "learning_rate": 1.1396351787368625e-05,
      "loss": 1.0145,
      "step": 14435
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1143720149993896,
      "learning_rate": 1.1395712870970835e-05,
      "loss": 1.1874,
      "step": 14436
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3011813163757324,
      "learning_rate": 1.1395073954573046e-05,
      "loss": 0.7584,
      "step": 14437
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9432483911514282,
      "learning_rate": 1.1394435038175256e-05,
      "loss": 0.9967,
      "step": 14438
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.581087350845337,
      "learning_rate": 1.1393796121777468e-05,
      "loss": 0.8905,
      "step": 14439
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6385078430175781,
      "learning_rate": 1.1393157205379678e-05,
      "loss": 0.7951,
      "step": 14440
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.739576816558838,
      "learning_rate": 1.1392518288981889e-05,
      "loss": 1.0539,
      "step": 14441
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7012183666229248,
      "learning_rate": 1.1391879372584099e-05,
      "loss": 0.823,
      "step": 14442
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9560245275497437,
      "learning_rate": 1.139124045618631e-05,
      "loss": 1.0022,
      "step": 14443
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2814877033233643,
      "learning_rate": 1.1390601539788519e-05,
      "loss": 0.9813,
      "step": 14444
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.094604253768921,
      "learning_rate": 1.1389962623390732e-05,
      "loss": 0.8487,
      "step": 14445
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7193862199783325,
      "learning_rate": 1.138932370699294e-05,
      "loss": 0.8663,
      "step": 14446
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8379167318344116,
      "learning_rate": 1.1388684790595153e-05,
      "loss": 0.9241,
      "step": 14447
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9238816499710083,
      "learning_rate": 1.1388045874197361e-05,
      "loss": 0.8455,
      "step": 14448
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0766303539276123,
      "learning_rate": 1.1387406957799573e-05,
      "loss": 0.8836,
      "step": 14449
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.02646803855896,
      "learning_rate": 1.1386768041401783e-05,
      "loss": 0.7904,
      "step": 14450
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7861264944076538,
      "learning_rate": 1.1386129125003994e-05,
      "loss": 0.9284,
      "step": 14451
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.110496759414673,
      "learning_rate": 1.1385490208606204e-05,
      "loss": 0.9947,
      "step": 14452
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.840972423553467,
      "learning_rate": 1.1384851292208416e-05,
      "loss": 1.1378,
      "step": 14453
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6907418966293335,
      "learning_rate": 1.1384212375810626e-05,
      "loss": 0.8948,
      "step": 14454
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2838010787963867,
      "learning_rate": 1.1383573459412837e-05,
      "loss": 1.0152,
      "step": 14455
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.90144681930542,
      "learning_rate": 1.1382934543015047e-05,
      "loss": 0.9804,
      "step": 14456
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9847919940948486,
      "learning_rate": 1.1382295626617259e-05,
      "loss": 1.0026,
      "step": 14457
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5094830989837646,
      "learning_rate": 1.1381656710219468e-05,
      "loss": 1.0562,
      "step": 14458
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.498800039291382,
      "learning_rate": 1.138101779382168e-05,
      "loss": 0.8206,
      "step": 14459
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.664181113243103,
      "learning_rate": 1.138037887742389e-05,
      "loss": 0.8814,
      "step": 14460
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8531440496444702,
      "learning_rate": 1.1379739961026101e-05,
      "loss": 1.0135,
      "step": 14461
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.14337420463562,
      "learning_rate": 1.1379101044628311e-05,
      "loss": 0.9754,
      "step": 14462
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5013741254806519,
      "learning_rate": 1.1378462128230523e-05,
      "loss": 0.4052,
      "step": 14463
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7841577529907227,
      "learning_rate": 1.1377823211832733e-05,
      "loss": 0.8419,
      "step": 14464
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.840602159500122,
      "learning_rate": 1.1377184295434944e-05,
      "loss": 0.8077,
      "step": 14465
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.126491069793701,
      "learning_rate": 1.1376545379037154e-05,
      "loss": 1.038,
      "step": 14466
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.101498603820801,
      "learning_rate": 1.1375906462639366e-05,
      "loss": 1.0499,
      "step": 14467
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.6814627647399902,
      "learning_rate": 1.1375267546241575e-05,
      "loss": 1.0926,
      "step": 14468
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8780434131622314,
      "learning_rate": 1.1374628629843787e-05,
      "loss": 0.9071,
      "step": 14469
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.922605276107788,
      "learning_rate": 1.1373989713445997e-05,
      "loss": 0.9066,
      "step": 14470
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5673373937606812,
      "learning_rate": 1.1373350797048208e-05,
      "loss": 0.8376,
      "step": 14471
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9687550067901611,
      "learning_rate": 1.1372711880650418e-05,
      "loss": 0.8383,
      "step": 14472
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.176823616027832,
      "learning_rate": 1.137207296425263e-05,
      "loss": 1.2564,
      "step": 14473
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.655410885810852,
      "learning_rate": 1.137143404785484e-05,
      "loss": 0.9533,
      "step": 14474
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3498153686523438,
      "learning_rate": 1.1370795131457051e-05,
      "loss": 0.9366,
      "step": 14475
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.824563503265381,
      "learning_rate": 1.1370156215059261e-05,
      "loss": 0.9203,
      "step": 14476
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5663254261016846,
      "learning_rate": 1.1369517298661473e-05,
      "loss": 0.8369,
      "step": 14477
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.509224772453308,
      "learning_rate": 1.1368878382263682e-05,
      "loss": 0.4274,
      "step": 14478
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5174376964569092,
      "learning_rate": 1.1368239465865894e-05,
      "loss": 0.9081,
      "step": 14479
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.345888614654541,
      "learning_rate": 1.1367600549468102e-05,
      "loss": 0.8476,
      "step": 14480
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4861845970153809,
      "learning_rate": 1.1366961633070315e-05,
      "loss": 0.8582,
      "step": 14481
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4406911134719849,
      "learning_rate": 1.1366322716672524e-05,
      "loss": 0.4616,
      "step": 14482
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.336345911026001,
      "learning_rate": 1.1365683800274737e-05,
      "loss": 0.4267,
      "step": 14483
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.2043745517730713,
      "learning_rate": 1.1365044883876945e-05,
      "loss": 0.868,
      "step": 14484
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6629127264022827,
      "learning_rate": 1.1364405967479158e-05,
      "loss": 0.9796,
      "step": 14485
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.771707534790039,
      "learning_rate": 1.1363767051081366e-05,
      "loss": 0.6295,
      "step": 14486
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.467679500579834,
      "learning_rate": 1.1363128134683578e-05,
      "loss": 0.9782,
      "step": 14487
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2849793434143066,
      "learning_rate": 1.1362489218285788e-05,
      "loss": 0.7987,
      "step": 14488
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1033740043640137,
      "learning_rate": 1.1361850301888e-05,
      "loss": 0.9054,
      "step": 14489
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5856279134750366,
      "learning_rate": 1.136121138549021e-05,
      "loss": 0.8871,
      "step": 14490
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9801234006881714,
      "learning_rate": 1.136057246909242e-05,
      "loss": 1.2136,
      "step": 14491
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8080813884735107,
      "learning_rate": 1.135993355269463e-05,
      "loss": 0.9082,
      "step": 14492
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4730304479599,
      "learning_rate": 1.1359294636296842e-05,
      "loss": 0.8689,
      "step": 14493
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8114228248596191,
      "learning_rate": 1.1358655719899052e-05,
      "loss": 0.9036,
      "step": 14494
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.7973573207855225,
      "learning_rate": 1.1358016803501262e-05,
      "loss": 1.2741,
      "step": 14495
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.004657506942749,
      "learning_rate": 1.1357377887103473e-05,
      "loss": 0.9616,
      "step": 14496
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.783048391342163,
      "learning_rate": 1.1356738970705683e-05,
      "loss": 1.1604,
      "step": 14497
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.4637646675109863,
      "learning_rate": 1.1356100054307895e-05,
      "loss": 0.8997,
      "step": 14498
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.03108549118042,
      "learning_rate": 1.1355461137910105e-05,
      "loss": 1.553,
      "step": 14499
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.7469074726104736,
      "learning_rate": 1.1354822221512316e-05,
      "loss": 0.8238,
      "step": 14500
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2318637371063232,
      "learning_rate": 1.1354183305114526e-05,
      "loss": 0.8487,
      "step": 14501
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.739471673965454,
      "learning_rate": 1.1353544388716738e-05,
      "loss": 0.9125,
      "step": 14502
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7728663682937622,
      "learning_rate": 1.1352905472318947e-05,
      "loss": 0.8879,
      "step": 14503
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2125930786132812,
      "learning_rate": 1.1352266555921159e-05,
      "loss": 1.1903,
      "step": 14504
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4611119031906128,
      "learning_rate": 1.1351627639523369e-05,
      "loss": 0.8216,
      "step": 14505
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6623436212539673,
      "learning_rate": 1.135098872312558e-05,
      "loss": 0.9549,
      "step": 14506
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8325555324554443,
      "learning_rate": 1.135034980672779e-05,
      "loss": 0.8581,
      "step": 14507
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7285927534103394,
      "learning_rate": 1.1349710890330002e-05,
      "loss": 0.9783,
      "step": 14508
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5759694576263428,
      "learning_rate": 1.1349071973932212e-05,
      "loss": 0.8194,
      "step": 14509
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9499796628952026,
      "learning_rate": 1.1348433057534423e-05,
      "loss": 1.0611,
      "step": 14510
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.49620246887207,
      "learning_rate": 1.1347794141136631e-05,
      "loss": 1.1672,
      "step": 14511
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9915642738342285,
      "learning_rate": 1.1347155224738845e-05,
      "loss": 1.0133,
      "step": 14512
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.038093328475952,
      "learning_rate": 1.1346516308341053e-05,
      "loss": 1.1289,
      "step": 14513
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5842665433883667,
      "learning_rate": 1.1345877391943266e-05,
      "loss": 0.4752,
      "step": 14514
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7912654876708984,
      "learning_rate": 1.1345238475545474e-05,
      "loss": 0.9712,
      "step": 14515
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7358992099761963,
      "learning_rate": 1.1344599559147686e-05,
      "loss": 0.8268,
      "step": 14516
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2854533195495605,
      "learning_rate": 1.1343960642749896e-05,
      "loss": 0.9322,
      "step": 14517
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0159335136413574,
      "learning_rate": 1.1343321726352107e-05,
      "loss": 1.0008,
      "step": 14518
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5824824571609497,
      "learning_rate": 1.1342682809954317e-05,
      "loss": 0.8777,
      "step": 14519
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5841014385223389,
      "learning_rate": 1.1342043893556528e-05,
      "loss": 0.9266,
      "step": 14520
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9955475330352783,
      "learning_rate": 1.1341404977158738e-05,
      "loss": 0.8817,
      "step": 14521
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.058783531188965,
      "learning_rate": 1.134076606076095e-05,
      "loss": 1.4432,
      "step": 14522
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5219990015029907,
      "learning_rate": 1.134012714436316e-05,
      "loss": 0.862,
      "step": 14523
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2108073234558105,
      "learning_rate": 1.1339488227965371e-05,
      "loss": 0.9013,
      "step": 14524
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.291563034057617,
      "learning_rate": 1.1338849311567581e-05,
      "loss": 1.1297,
      "step": 14525
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0816633701324463,
      "learning_rate": 1.1338210395169793e-05,
      "loss": 0.9095,
      "step": 14526
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6354540586471558,
      "learning_rate": 1.1337571478772003e-05,
      "loss": 0.4763,
      "step": 14527
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.00146746635437,
      "learning_rate": 1.1336932562374214e-05,
      "loss": 0.7973,
      "step": 14528
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.140589475631714,
      "learning_rate": 1.1336293645976424e-05,
      "loss": 0.9888,
      "step": 14529
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.681487798690796,
      "learning_rate": 1.1335654729578635e-05,
      "loss": 0.6106,
      "step": 14530
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5617769956588745,
      "learning_rate": 1.1335015813180845e-05,
      "loss": 0.798,
      "step": 14531
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.12109112739563,
      "learning_rate": 1.1334376896783057e-05,
      "loss": 0.9285,
      "step": 14532
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4223953485488892,
      "learning_rate": 1.1333737980385267e-05,
      "loss": 0.8004,
      "step": 14533
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.945120930671692,
      "learning_rate": 1.1333099063987478e-05,
      "loss": 0.7884,
      "step": 14534
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0491952896118164,
      "learning_rate": 1.1332460147589688e-05,
      "loss": 0.9102,
      "step": 14535
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.911277174949646,
      "learning_rate": 1.13318212311919e-05,
      "loss": 0.9893,
      "step": 14536
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5609993934631348,
      "learning_rate": 1.133118231479411e-05,
      "loss": 1.2091,
      "step": 14537
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.305903434753418,
      "learning_rate": 1.1330543398396321e-05,
      "loss": 0.6642,
      "step": 14538
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0045642852783203,
      "learning_rate": 1.1329904481998531e-05,
      "loss": 0.965,
      "step": 14539
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8757588863372803,
      "learning_rate": 1.1329265565600742e-05,
      "loss": 1.0248,
      "step": 14540
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6553421020507812,
      "learning_rate": 1.1328626649202952e-05,
      "loss": 0.8537,
      "step": 14541
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6886144876480103,
      "learning_rate": 1.1327987732805164e-05,
      "loss": 0.9055,
      "step": 14542
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.103529214859009,
      "learning_rate": 1.1327348816407374e-05,
      "loss": 0.6897,
      "step": 14543
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.604273796081543,
      "learning_rate": 1.1326709900009585e-05,
      "loss": 0.8782,
      "step": 14544
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7176473140716553,
      "learning_rate": 1.1326070983611795e-05,
      "loss": 0.9485,
      "step": 14545
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.6481773853302,
      "learning_rate": 1.1325432067214007e-05,
      "loss": 0.8761,
      "step": 14546
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9692517518997192,
      "learning_rate": 1.1324793150816215e-05,
      "loss": 0.9771,
      "step": 14547
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7860769033432007,
      "learning_rate": 1.1324154234418428e-05,
      "loss": 0.9543,
      "step": 14548
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6936901807785034,
      "learning_rate": 1.1323515318020636e-05,
      "loss": 0.8788,
      "step": 14549
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9715397357940674,
      "learning_rate": 1.132287640162285e-05,
      "loss": 0.9796,
      "step": 14550
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6953250169754028,
      "learning_rate": 1.1322237485225058e-05,
      "loss": 1.0207,
      "step": 14551
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.462053656578064,
      "learning_rate": 1.1321598568827271e-05,
      "loss": 0.8259,
      "step": 14552
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.932912826538086,
      "learning_rate": 1.1320959652429479e-05,
      "loss": 0.3565,
      "step": 14553
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0655312538146973,
      "learning_rate": 1.132032073603169e-05,
      "loss": 0.7818,
      "step": 14554
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1034390926361084,
      "learning_rate": 1.13196818196339e-05,
      "loss": 0.9503,
      "step": 14555
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0513343811035156,
      "learning_rate": 1.1319042903236112e-05,
      "loss": 0.923,
      "step": 14556
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.092778205871582,
      "learning_rate": 1.1318403986838322e-05,
      "loss": 0.9369,
      "step": 14557
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.978201985359192,
      "learning_rate": 1.1317765070440533e-05,
      "loss": 0.8414,
      "step": 14558
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3078970909118652,
      "learning_rate": 1.1317126154042743e-05,
      "loss": 1.0008,
      "step": 14559
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.843700647354126,
      "learning_rate": 1.1316487237644955e-05,
      "loss": 0.9967,
      "step": 14560
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5153257846832275,
      "learning_rate": 1.1315848321247165e-05,
      "loss": 0.675,
      "step": 14561
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.0416259765625,
      "learning_rate": 1.1315209404849376e-05,
      "loss": 1.1429,
      "step": 14562
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7867610454559326,
      "learning_rate": 1.1314570488451586e-05,
      "loss": 0.9227,
      "step": 14563
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.535538673400879,
      "learning_rate": 1.1313931572053798e-05,
      "loss": 0.7922,
      "step": 14564
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.227848529815674,
      "learning_rate": 1.1313292655656007e-05,
      "loss": 0.9076,
      "step": 14565
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.008633852005005,
      "learning_rate": 1.1312653739258219e-05,
      "loss": 0.8365,
      "step": 14566
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.360056161880493,
      "learning_rate": 1.1312014822860429e-05,
      "loss": 0.8647,
      "step": 14567
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5045247077941895,
      "learning_rate": 1.131137590646264e-05,
      "loss": 0.7278,
      "step": 14568
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0720129013061523,
      "learning_rate": 1.131073699006485e-05,
      "loss": 0.9809,
      "step": 14569
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.085508823394775,
      "learning_rate": 1.1310098073667062e-05,
      "loss": 1.1969,
      "step": 14570
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.850403070449829,
      "learning_rate": 1.1309459157269272e-05,
      "loss": 0.8489,
      "step": 14571
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4566960334777832,
      "learning_rate": 1.1308820240871483e-05,
      "loss": 0.8218,
      "step": 14572
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.281223773956299,
      "learning_rate": 1.1308181324473693e-05,
      "loss": 1.1307,
      "step": 14573
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9257371425628662,
      "learning_rate": 1.1307542408075905e-05,
      "loss": 1.0023,
      "step": 14574
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.335174083709717,
      "learning_rate": 1.1306903491678114e-05,
      "loss": 1.0856,
      "step": 14575
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9233286380767822,
      "learning_rate": 1.1306264575280326e-05,
      "loss": 0.9963,
      "step": 14576
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8110023736953735,
      "learning_rate": 1.1305625658882536e-05,
      "loss": 0.8739,
      "step": 14577
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.8267314434051514,
      "learning_rate": 1.1304986742484747e-05,
      "loss": 1.0215,
      "step": 14578
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.517886161804199,
      "learning_rate": 1.1304347826086957e-05,
      "loss": 0.7605,
      "step": 14579
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.818647861480713,
      "learning_rate": 1.1303708909689169e-05,
      "loss": 0.8855,
      "step": 14580
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.076429605484009,
      "learning_rate": 1.1303069993291379e-05,
      "loss": 0.7524,
      "step": 14581
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9947614669799805,
      "learning_rate": 1.130243107689359e-05,
      "loss": 1.0462,
      "step": 14582
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5625927448272705,
      "learning_rate": 1.13017921604958e-05,
      "loss": 0.878,
      "step": 14583
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0379743576049805,
      "learning_rate": 1.1301153244098012e-05,
      "loss": 0.9705,
      "step": 14584
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9569470882415771,
      "learning_rate": 1.130051432770022e-05,
      "loss": 0.5734,
      "step": 14585
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.406087636947632,
      "learning_rate": 1.1299875411302433e-05,
      "loss": 1.0312,
      "step": 14586
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.627474308013916,
      "learning_rate": 1.1299236494904641e-05,
      "loss": 0.9099,
      "step": 14587
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.325761795043945,
      "learning_rate": 1.1298597578506854e-05,
      "loss": 1.4036,
      "step": 14588
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.592647075653076,
      "learning_rate": 1.1297958662109063e-05,
      "loss": 0.8344,
      "step": 14589
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6603357791900635,
      "learning_rate": 1.1297319745711274e-05,
      "loss": 0.9863,
      "step": 14590
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0551254749298096,
      "learning_rate": 1.1296680829313484e-05,
      "loss": 1.2744,
      "step": 14591
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1085662841796875,
      "learning_rate": 1.1296041912915695e-05,
      "loss": 0.9726,
      "step": 14592
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5590335130691528,
      "learning_rate": 1.1295402996517905e-05,
      "loss": 0.8959,
      "step": 14593
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9454940557479858,
      "learning_rate": 1.1294764080120117e-05,
      "loss": 1.1091,
      "step": 14594
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.256685733795166,
      "learning_rate": 1.1294125163722327e-05,
      "loss": 1.0499,
      "step": 14595
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5452237129211426,
      "learning_rate": 1.1293486247324538e-05,
      "loss": 0.8824,
      "step": 14596
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6178131103515625,
      "learning_rate": 1.1292847330926748e-05,
      "loss": 0.912,
      "step": 14597
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.832371234893799,
      "learning_rate": 1.129220841452896e-05,
      "loss": 0.6347,
      "step": 14598
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7282625436782837,
      "learning_rate": 1.129156949813117e-05,
      "loss": 1.0187,
      "step": 14599
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5668349266052246,
      "learning_rate": 1.1290930581733381e-05,
      "loss": 0.9566,
      "step": 14600
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5791840553283691,
      "learning_rate": 1.1290291665335591e-05,
      "loss": 0.9521,
      "step": 14601
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.487337350845337,
      "learning_rate": 1.1289652748937802e-05,
      "loss": 0.7976,
      "step": 14602
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4252129793167114,
      "learning_rate": 1.1289013832540012e-05,
      "loss": 0.847,
      "step": 14603
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.9318623542785645,
      "learning_rate": 1.1288374916142224e-05,
      "loss": 1.0487,
      "step": 14604
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.7585761547088623,
      "learning_rate": 1.1287735999744434e-05,
      "loss": 0.9562,
      "step": 14605
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7596766948699951,
      "learning_rate": 1.1287097083346645e-05,
      "loss": 0.9485,
      "step": 14606
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1370248794555664,
      "learning_rate": 1.1286458166948855e-05,
      "loss": 0.9064,
      "step": 14607
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.154780626296997,
      "learning_rate": 1.1285819250551067e-05,
      "loss": 1.1331,
      "step": 14608
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5961438417434692,
      "learning_rate": 1.1285180334153277e-05,
      "loss": 0.9393,
      "step": 14609
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.188124418258667,
      "learning_rate": 1.1284541417755488e-05,
      "loss": 0.8097,
      "step": 14610
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.012237548828125,
      "learning_rate": 1.1283902501357698e-05,
      "loss": 0.8908,
      "step": 14611
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.923656702041626,
      "learning_rate": 1.128326358495991e-05,
      "loss": 1.1757,
      "step": 14612
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7559387683868408,
      "learning_rate": 1.128262466856212e-05,
      "loss": 0.9588,
      "step": 14613
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.6855485439300537,
      "learning_rate": 1.1281985752164331e-05,
      "loss": 1.1066,
      "step": 14614
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5672190189361572,
      "learning_rate": 1.128134683576654e-05,
      "loss": 0.8663,
      "step": 14615
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4422080516815186,
      "learning_rate": 1.1280707919368752e-05,
      "loss": 0.8721,
      "step": 14616
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6035813093185425,
      "learning_rate": 1.1280069002970962e-05,
      "loss": 0.8169,
      "step": 14617
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6449172496795654,
      "learning_rate": 1.1279430086573174e-05,
      "loss": 0.9064,
      "step": 14618
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.826063632965088,
      "learning_rate": 1.1278791170175384e-05,
      "loss": 1.1996,
      "step": 14619
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1497063636779785,
      "learning_rate": 1.1278152253777595e-05,
      "loss": 0.8754,
      "step": 14620
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5769200325012207,
      "learning_rate": 1.1277513337379803e-05,
      "loss": 0.9705,
      "step": 14621
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7606947422027588,
      "learning_rate": 1.1276874420982016e-05,
      "loss": 0.895,
      "step": 14622
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0569963455200195,
      "learning_rate": 1.1276235504584225e-05,
      "loss": 1.0397,
      "step": 14623
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.235833644866943,
      "learning_rate": 1.1275596588186438e-05,
      "loss": 1.2844,
      "step": 14624
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.570463180541992,
      "learning_rate": 1.1274957671788646e-05,
      "loss": 0.944,
      "step": 14625
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1764564514160156,
      "learning_rate": 1.1274318755390858e-05,
      "loss": 0.9798,
      "step": 14626
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7707858085632324,
      "learning_rate": 1.1273679838993067e-05,
      "loss": 0.946,
      "step": 14627
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.325305223464966,
      "learning_rate": 1.1273040922595279e-05,
      "loss": 1.2874,
      "step": 14628
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.125731945037842,
      "learning_rate": 1.1272402006197489e-05,
      "loss": 0.9932,
      "step": 14629
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6798020601272583,
      "learning_rate": 1.12717630897997e-05,
      "loss": 0.9657,
      "step": 14630
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.253004312515259,
      "learning_rate": 1.127112417340191e-05,
      "loss": 0.8401,
      "step": 14631
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0352942943573,
      "learning_rate": 1.1270485257004122e-05,
      "loss": 0.8866,
      "step": 14632
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.885778546333313,
      "learning_rate": 1.1269846340606332e-05,
      "loss": 0.8703,
      "step": 14633
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8616645336151123,
      "learning_rate": 1.1269207424208543e-05,
      "loss": 1.0163,
      "step": 14634
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6856240034103394,
      "learning_rate": 1.1268568507810753e-05,
      "loss": 0.9366,
      "step": 14635
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0546395778656006,
      "learning_rate": 1.1267929591412965e-05,
      "loss": 0.9529,
      "step": 14636
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6358914375305176,
      "learning_rate": 1.1267290675015174e-05,
      "loss": 0.9164,
      "step": 14637
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8567014932632446,
      "learning_rate": 1.1266651758617386e-05,
      "loss": 0.8869,
      "step": 14638
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.44568932056427,
      "learning_rate": 1.1266012842219596e-05,
      "loss": 0.7609,
      "step": 14639
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8816754817962646,
      "learning_rate": 1.1265373925821807e-05,
      "loss": 0.8322,
      "step": 14640
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0619289875030518,
      "learning_rate": 1.1264735009424017e-05,
      "loss": 0.8152,
      "step": 14641
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6713690757751465,
      "learning_rate": 1.1264096093026229e-05,
      "loss": 0.8365,
      "step": 14642
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.713186264038086,
      "learning_rate": 1.1263457176628439e-05,
      "loss": 1.0083,
      "step": 14643
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.139831304550171,
      "learning_rate": 1.126281826023065e-05,
      "loss": 0.7652,
      "step": 14644
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.781732439994812,
      "learning_rate": 1.126217934383286e-05,
      "loss": 0.8608,
      "step": 14645
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9967575073242188,
      "learning_rate": 1.1261540427435072e-05,
      "loss": 0.9619,
      "step": 14646
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.214733123779297,
      "learning_rate": 1.1260901511037281e-05,
      "loss": 1.053,
      "step": 14647
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.173550844192505,
      "learning_rate": 1.1260262594639493e-05,
      "loss": 1.0584,
      "step": 14648
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9300496578216553,
      "learning_rate": 1.1259623678241703e-05,
      "loss": 0.9427,
      "step": 14649
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4238566160202026,
      "learning_rate": 1.1258984761843914e-05,
      "loss": 0.7821,
      "step": 14650
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.683474540710449,
      "learning_rate": 1.1258345845446124e-05,
      "loss": 1.2217,
      "step": 14651
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8593480587005615,
      "learning_rate": 1.1257706929048336e-05,
      "loss": 1.1676,
      "step": 14652
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0260558128356934,
      "learning_rate": 1.1257068012650546e-05,
      "loss": 0.9797,
      "step": 14653
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1155474185943604,
      "learning_rate": 1.1256429096252757e-05,
      "loss": 1.1049,
      "step": 14654
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.325875997543335,
      "learning_rate": 1.1255790179854967e-05,
      "loss": 0.7503,
      "step": 14655
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8664737939834595,
      "learning_rate": 1.1255151263457179e-05,
      "loss": 0.9201,
      "step": 14656
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1643941402435303,
      "learning_rate": 1.1254512347059387e-05,
      "loss": 1.1459,
      "step": 14657
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7935874462127686,
      "learning_rate": 1.12538734306616e-05,
      "loss": 0.882,
      "step": 14658
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.377135753631592,
      "learning_rate": 1.1253234514263808e-05,
      "loss": 0.8012,
      "step": 14659
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8004645109176636,
      "learning_rate": 1.1252595597866021e-05,
      "loss": 0.8137,
      "step": 14660
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.851378917694092,
      "learning_rate": 1.125195668146823e-05,
      "loss": 1.1068,
      "step": 14661
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5523511171340942,
      "learning_rate": 1.1251317765070441e-05,
      "loss": 0.9677,
      "step": 14662
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9618912935256958,
      "learning_rate": 1.1250678848672651e-05,
      "loss": 1.0363,
      "step": 14663
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9443951845169067,
      "learning_rate": 1.1250039932274862e-05,
      "loss": 0.9734,
      "step": 14664
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7905710935592651,
      "learning_rate": 1.1249401015877072e-05,
      "loss": 0.8374,
      "step": 14665
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.8421213626861572,
      "learning_rate": 1.1248762099479284e-05,
      "loss": 1.1532,
      "step": 14666
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6085171699523926,
      "learning_rate": 1.1248123183081494e-05,
      "loss": 0.4257,
      "step": 14667
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3515427112579346,
      "learning_rate": 1.1247484266683705e-05,
      "loss": 0.9583,
      "step": 14668
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6837824583053589,
      "learning_rate": 1.1246845350285915e-05,
      "loss": 0.8455,
      "step": 14669
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.020972728729248,
      "learning_rate": 1.1246206433888127e-05,
      "loss": 0.9151,
      "step": 14670
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9972976446151733,
      "learning_rate": 1.1245567517490337e-05,
      "loss": 0.8673,
      "step": 14671
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.716452956199646,
      "learning_rate": 1.1244928601092548e-05,
      "loss": 0.882,
      "step": 14672
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0591373443603516,
      "learning_rate": 1.1244289684694758e-05,
      "loss": 0.9426,
      "step": 14673
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8718578815460205,
      "learning_rate": 1.124365076829697e-05,
      "loss": 0.9326,
      "step": 14674
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.008195877075195,
      "learning_rate": 1.124301185189918e-05,
      "loss": 0.843,
      "step": 14675
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0519521236419678,
      "learning_rate": 1.1242372935501391e-05,
      "loss": 0.861,
      "step": 14676
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.6643426418304443,
      "learning_rate": 1.12417340191036e-05,
      "loss": 1.0956,
      "step": 14677
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.363074779510498,
      "learning_rate": 1.1241095102705812e-05,
      "loss": 1.2491,
      "step": 14678
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.7935032844543457,
      "learning_rate": 1.1240456186308022e-05,
      "loss": 0.9285,
      "step": 14679
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.168571949005127,
      "learning_rate": 1.1239817269910234e-05,
      "loss": 1.1697,
      "step": 14680
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.647602558135986,
      "learning_rate": 1.1239178353512444e-05,
      "loss": 1.3259,
      "step": 14681
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1028313636779785,
      "learning_rate": 1.1238539437114655e-05,
      "loss": 0.9131,
      "step": 14682
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.695380449295044,
      "learning_rate": 1.1237900520716865e-05,
      "loss": 0.8945,
      "step": 14683
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7743167877197266,
      "learning_rate": 1.1237261604319076e-05,
      "loss": 0.6987,
      "step": 14684
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6927714347839355,
      "learning_rate": 1.1236622687921286e-05,
      "loss": 0.8357,
      "step": 14685
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.335902452468872,
      "learning_rate": 1.1235983771523498e-05,
      "loss": 0.914,
      "step": 14686
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8975739479064941,
      "learning_rate": 1.1235344855125708e-05,
      "loss": 1.0483,
      "step": 14687
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.819143533706665,
      "learning_rate": 1.123470593872792e-05,
      "loss": 0.8839,
      "step": 14688
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.454493761062622,
      "learning_rate": 1.1234067022330129e-05,
      "loss": 0.9122,
      "step": 14689
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6780799627304077,
      "learning_rate": 1.123342810593234e-05,
      "loss": 0.8711,
      "step": 14690
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.504669189453125,
      "learning_rate": 1.123278918953455e-05,
      "loss": 0.7786,
      "step": 14691
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8748923540115356,
      "learning_rate": 1.1232150273136762e-05,
      "loss": 1.0567,
      "step": 14692
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8571733236312866,
      "learning_rate": 1.123151135673897e-05,
      "loss": 0.8221,
      "step": 14693
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9830222129821777,
      "learning_rate": 1.1230872440341183e-05,
      "loss": 0.9089,
      "step": 14694
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9871050119400024,
      "learning_rate": 1.1230233523943392e-05,
      "loss": 0.9288,
      "step": 14695
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.631508469581604,
      "learning_rate": 1.1229594607545605e-05,
      "loss": 0.9365,
      "step": 14696
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.224304676055908,
      "learning_rate": 1.1228955691147813e-05,
      "loss": 1.1053,
      "step": 14697
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7383086681365967,
      "learning_rate": 1.1228316774750026e-05,
      "loss": 0.9273,
      "step": 14698
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.2151272296905518,
      "learning_rate": 1.1227677858352234e-05,
      "loss": 1.0891,
      "step": 14699
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.194650173187256,
      "learning_rate": 1.1227038941954446e-05,
      "loss": 1.2867,
      "step": 14700
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1470510959625244,
      "learning_rate": 1.1226400025556656e-05,
      "loss": 1.2609,
      "step": 14701
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7357642650604248,
      "learning_rate": 1.1225761109158867e-05,
      "loss": 0.875,
      "step": 14702
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.5478057861328125,
      "learning_rate": 1.1225122192761077e-05,
      "loss": 1.2233,
      "step": 14703
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5116676092147827,
      "learning_rate": 1.1224483276363289e-05,
      "loss": 0.9481,
      "step": 14704
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.5813984870910645,
      "learning_rate": 1.1223844359965499e-05,
      "loss": 1.1263,
      "step": 14705
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.74444317817688,
      "learning_rate": 1.122320544356771e-05,
      "loss": 1.5215,
      "step": 14706
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7569552659988403,
      "learning_rate": 1.122256652716992e-05,
      "loss": 0.8165,
      "step": 14707
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8572900295257568,
      "learning_rate": 1.1221927610772132e-05,
      "loss": 0.9198,
      "step": 14708
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9256916046142578,
      "learning_rate": 1.1221288694374341e-05,
      "loss": 0.889,
      "step": 14709
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.680159330368042,
      "learning_rate": 1.1220649777976553e-05,
      "loss": 0.9212,
      "step": 14710
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3338475227355957,
      "learning_rate": 1.1220010861578763e-05,
      "loss": 0.9562,
      "step": 14711
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2028417587280273,
      "learning_rate": 1.1219371945180974e-05,
      "loss": 0.8772,
      "step": 14712
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5647854804992676,
      "learning_rate": 1.1218733028783184e-05,
      "loss": 0.9245,
      "step": 14713
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8211815357208252,
      "learning_rate": 1.1218094112385396e-05,
      "loss": 0.8589,
      "step": 14714
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8748102188110352,
      "learning_rate": 1.1217455195987606e-05,
      "loss": 1.2941,
      "step": 14715
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7168482542037964,
      "learning_rate": 1.1216816279589817e-05,
      "loss": 0.9487,
      "step": 14716
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.019692897796631,
      "learning_rate": 1.1216177363192027e-05,
      "loss": 1.001,
      "step": 14717
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7679617404937744,
      "learning_rate": 1.1215538446794239e-05,
      "loss": 0.9117,
      "step": 14718
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0166521072387695,
      "learning_rate": 1.1214899530396448e-05,
      "loss": 0.9669,
      "step": 14719
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8152408599853516,
      "learning_rate": 1.121426061399866e-05,
      "loss": 0.9094,
      "step": 14720
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3956894874572754,
      "learning_rate": 1.121362169760087e-05,
      "loss": 0.8326,
      "step": 14721
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.231581687927246,
      "learning_rate": 1.1212982781203081e-05,
      "loss": 0.586,
      "step": 14722
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4470937252044678,
      "learning_rate": 1.1212343864805291e-05,
      "loss": 0.8653,
      "step": 14723
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.47442626953125,
      "learning_rate": 1.1211704948407503e-05,
      "loss": 1.3697,
      "step": 14724
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.403209686279297,
      "learning_rate": 1.1211066032009713e-05,
      "loss": 0.9817,
      "step": 14725
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9839838743209839,
      "learning_rate": 1.1210427115611924e-05,
      "loss": 0.8948,
      "step": 14726
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.7322511672973633,
      "learning_rate": 1.1209788199214134e-05,
      "loss": 0.9199,
      "step": 14727
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.790347576141357,
      "learning_rate": 1.1209149282816346e-05,
      "loss": 1.0046,
      "step": 14728
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9304293394088745,
      "learning_rate": 1.1208510366418555e-05,
      "loss": 0.7571,
      "step": 14729
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9368923902511597,
      "learning_rate": 1.1207871450020767e-05,
      "loss": 0.8499,
      "step": 14730
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.048206329345703,
      "learning_rate": 1.1207232533622975e-05,
      "loss": 0.8906,
      "step": 14731
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6355141401290894,
      "learning_rate": 1.1206593617225188e-05,
      "loss": 0.6865,
      "step": 14732
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1948447227478027,
      "learning_rate": 1.1205954700827397e-05,
      "loss": 1.2375,
      "step": 14733
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.633169174194336,
      "learning_rate": 1.120531578442961e-05,
      "loss": 0.8797,
      "step": 14734
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6296703815460205,
      "learning_rate": 1.1204676868031818e-05,
      "loss": 0.865,
      "step": 14735
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0196568965911865,
      "learning_rate": 1.120403795163403e-05,
      "loss": 0.8707,
      "step": 14736
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.103761672973633,
      "learning_rate": 1.120339903523624e-05,
      "loss": 0.9477,
      "step": 14737
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4865901470184326,
      "learning_rate": 1.1202760118838451e-05,
      "loss": 0.4518,
      "step": 14738
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.9494247436523438,
      "learning_rate": 1.120212120244066e-05,
      "loss": 1.2601,
      "step": 14739
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7110087871551514,
      "learning_rate": 1.1201482286042872e-05,
      "loss": 1.2185,
      "step": 14740
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.877915620803833,
      "learning_rate": 1.1200843369645082e-05,
      "loss": 1.0513,
      "step": 14741
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.745836853981018,
      "learning_rate": 1.1200204453247294e-05,
      "loss": 0.8831,
      "step": 14742
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9319075345993042,
      "learning_rate": 1.1199565536849504e-05,
      "loss": 1.0629,
      "step": 14743
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.178870916366577,
      "learning_rate": 1.1198926620451715e-05,
      "loss": 0.9466,
      "step": 14744
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5049463510513306,
      "learning_rate": 1.1198287704053925e-05,
      "loss": 0.8437,
      "step": 14745
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.863327980041504,
      "learning_rate": 1.1197648787656136e-05,
      "loss": 1.1028,
      "step": 14746
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4524668455123901,
      "learning_rate": 1.1197009871258346e-05,
      "loss": 0.8356,
      "step": 14747
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6413735151290894,
      "learning_rate": 1.1196370954860558e-05,
      "loss": 0.7917,
      "step": 14748
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5730962753295898,
      "learning_rate": 1.1195732038462768e-05,
      "loss": 0.7955,
      "step": 14749
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9581210613250732,
      "learning_rate": 1.119509312206498e-05,
      "loss": 0.8593,
      "step": 14750
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.933924674987793,
      "learning_rate": 1.1194454205667189e-05,
      "loss": 0.8201,
      "step": 14751
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.616275429725647,
      "learning_rate": 1.11938152892694e-05,
      "loss": 0.9167,
      "step": 14752
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2251791954040527,
      "learning_rate": 1.119317637287161e-05,
      "loss": 0.7247,
      "step": 14753
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.690064549446106,
      "learning_rate": 1.1192537456473822e-05,
      "loss": 0.8622,
      "step": 14754
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.4009294509887695,
      "learning_rate": 1.1191898540076032e-05,
      "loss": 1.3133,
      "step": 14755
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.6710751056671143,
      "learning_rate": 1.1191259623678243e-05,
      "loss": 1.0794,
      "step": 14756
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9745501279830933,
      "learning_rate": 1.1190620707280453e-05,
      "loss": 1.2214,
      "step": 14757
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9193357229232788,
      "learning_rate": 1.1189981790882665e-05,
      "loss": 0.8833,
      "step": 14758
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8869335651397705,
      "learning_rate": 1.1189342874484875e-05,
      "loss": 0.9343,
      "step": 14759
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9292099475860596,
      "learning_rate": 1.1188703958087086e-05,
      "loss": 1.0258,
      "step": 14760
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.8542962074279785,
      "learning_rate": 1.1188065041689296e-05,
      "loss": 1.0178,
      "step": 14761
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7331452369689941,
      "learning_rate": 1.1187426125291508e-05,
      "loss": 1.0003,
      "step": 14762
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.017324447631836,
      "learning_rate": 1.1186787208893718e-05,
      "loss": 0.8827,
      "step": 14763
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9355452060699463,
      "learning_rate": 1.1186148292495929e-05,
      "loss": 0.8409,
      "step": 14764
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.151169538497925,
      "learning_rate": 1.1185509376098139e-05,
      "loss": 0.6764,
      "step": 14765
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7414088249206543,
      "learning_rate": 1.118487045970035e-05,
      "loss": 1.0035,
      "step": 14766
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.227074384689331,
      "learning_rate": 1.1184231543302559e-05,
      "loss": 0.9662,
      "step": 14767
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6118299961090088,
      "learning_rate": 1.1183592626904772e-05,
      "loss": 0.9425,
      "step": 14768
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3011090755462646,
      "learning_rate": 1.118295371050698e-05,
      "loss": 0.9749,
      "step": 14769
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6464040279388428,
      "learning_rate": 1.1182314794109193e-05,
      "loss": 0.8438,
      "step": 14770
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.732141137123108,
      "learning_rate": 1.1181675877711401e-05,
      "loss": 0.8314,
      "step": 14771
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7182321548461914,
      "learning_rate": 1.1181036961313613e-05,
      "loss": 0.4197,
      "step": 14772
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8188371658325195,
      "learning_rate": 1.1180398044915823e-05,
      "loss": 0.9321,
      "step": 14773
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8192451000213623,
      "learning_rate": 1.1179759128518034e-05,
      "loss": 0.8686,
      "step": 14774
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8887416124343872,
      "learning_rate": 1.1179120212120244e-05,
      "loss": 0.8303,
      "step": 14775
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5404456853866577,
      "learning_rate": 1.1178481295722456e-05,
      "loss": 0.9058,
      "step": 14776
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0557239055633545,
      "learning_rate": 1.1177842379324666e-05,
      "loss": 0.9313,
      "step": 14777
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5376766920089722,
      "learning_rate": 1.1177203462926877e-05,
      "loss": 0.8378,
      "step": 14778
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4590564966201782,
      "learning_rate": 1.1176564546529087e-05,
      "loss": 0.8686,
      "step": 14779
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.769518256187439,
      "learning_rate": 1.1175925630131299e-05,
      "loss": 0.8566,
      "step": 14780
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.476032018661499,
      "learning_rate": 1.1175286713733508e-05,
      "loss": 0.8218,
      "step": 14781
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9590997695922852,
      "learning_rate": 1.117464779733572e-05,
      "loss": 1.2846,
      "step": 14782
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7522494792938232,
      "learning_rate": 1.117400888093793e-05,
      "loss": 0.8931,
      "step": 14783
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5028295516967773,
      "learning_rate": 1.1173369964540141e-05,
      "loss": 0.5105,
      "step": 14784
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7350586652755737,
      "learning_rate": 1.1172731048142351e-05,
      "loss": 0.8229,
      "step": 14785
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.289361000061035,
      "learning_rate": 1.1172092131744563e-05,
      "loss": 0.8849,
      "step": 14786
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0866615772247314,
      "learning_rate": 1.1171453215346773e-05,
      "loss": 0.916,
      "step": 14787
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.254560470581055,
      "learning_rate": 1.1170814298948984e-05,
      "loss": 0.9076,
      "step": 14788
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2371580600738525,
      "learning_rate": 1.1170175382551194e-05,
      "loss": 0.9572,
      "step": 14789
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.791022777557373,
      "learning_rate": 1.1169536466153406e-05,
      "loss": 0.8467,
      "step": 14790
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.371715545654297,
      "learning_rate": 1.1168897549755615e-05,
      "loss": 0.7569,
      "step": 14791
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9093023538589478,
      "learning_rate": 1.1168258633357827e-05,
      "loss": 0.8859,
      "step": 14792
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.661908745765686,
      "learning_rate": 1.1167619716960037e-05,
      "loss": 0.8385,
      "step": 14793
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.7231392860412598,
      "learning_rate": 1.1166980800562248e-05,
      "loss": 0.9968,
      "step": 14794
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8050007820129395,
      "learning_rate": 1.1166341884164458e-05,
      "loss": 1.1262,
      "step": 14795
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.57760488986969,
      "learning_rate": 1.116570296776667e-05,
      "loss": 0.9563,
      "step": 14796
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.598093032836914,
      "learning_rate": 1.116506405136888e-05,
      "loss": 0.4004,
      "step": 14797
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.143729209899902,
      "learning_rate": 1.1164425134971091e-05,
      "loss": 1.2337,
      "step": 14798
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.767454981803894,
      "learning_rate": 1.1163786218573301e-05,
      "loss": 0.9632,
      "step": 14799
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9892503023147583,
      "learning_rate": 1.1163147302175513e-05,
      "loss": 1.0294,
      "step": 14800
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0406835079193115,
      "learning_rate": 1.1162508385777722e-05,
      "loss": 0.9109,
      "step": 14801
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.500396251678467,
      "learning_rate": 1.1161869469379934e-05,
      "loss": 1.3056,
      "step": 14802
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.884966254234314,
      "learning_rate": 1.1161230552982142e-05,
      "loss": 0.8749,
      "step": 14803
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0474653244018555,
      "learning_rate": 1.1160591636584355e-05,
      "loss": 0.9203,
      "step": 14804
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9894468784332275,
      "learning_rate": 1.1159952720186564e-05,
      "loss": 0.9878,
      "step": 14805
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.989534378051758,
      "learning_rate": 1.1159313803788777e-05,
      "loss": 0.7493,
      "step": 14806
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.5007550716400146,
      "learning_rate": 1.1158674887390985e-05,
      "loss": 1.1005,
      "step": 14807
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6819871664047241,
      "learning_rate": 1.1158035970993198e-05,
      "loss": 0.8328,
      "step": 14808
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7604403495788574,
      "learning_rate": 1.1157397054595406e-05,
      "loss": 0.9616,
      "step": 14809
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.674495220184326,
      "learning_rate": 1.1156758138197618e-05,
      "loss": 0.6792,
      "step": 14810
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.029890537261963,
      "learning_rate": 1.1156119221799828e-05,
      "loss": 0.9109,
      "step": 14811
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8299400806427002,
      "learning_rate": 1.115548030540204e-05,
      "loss": 0.9751,
      "step": 14812
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3770880699157715,
      "learning_rate": 1.115484138900425e-05,
      "loss": 0.8474,
      "step": 14813
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9353041648864746,
      "learning_rate": 1.115420247260646e-05,
      "loss": 0.916,
      "step": 14814
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3951060771942139,
      "learning_rate": 1.115356355620867e-05,
      "loss": 0.8192,
      "step": 14815
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.623290777206421,
      "learning_rate": 1.1152924639810882e-05,
      "loss": 1.2384,
      "step": 14816
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7733166217803955,
      "learning_rate": 1.1152285723413092e-05,
      "loss": 1.0766,
      "step": 14817
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9897392988204956,
      "learning_rate": 1.1151646807015304e-05,
      "loss": 0.634,
      "step": 14818
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7000948190689087,
      "learning_rate": 1.1151007890617513e-05,
      "loss": 0.8774,
      "step": 14819
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4699885845184326,
      "learning_rate": 1.1150368974219725e-05,
      "loss": 0.6278,
      "step": 14820
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5830093622207642,
      "learning_rate": 1.1149730057821935e-05,
      "loss": 0.9609,
      "step": 14821
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1708273887634277,
      "learning_rate": 1.1149091141424146e-05,
      "loss": 0.7773,
      "step": 14822
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.279329776763916,
      "learning_rate": 1.1148452225026356e-05,
      "loss": 1.099,
      "step": 14823
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.333172559738159,
      "learning_rate": 1.1147813308628568e-05,
      "loss": 0.5085,
      "step": 14824
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.289353609085083,
      "learning_rate": 1.1147174392230778e-05,
      "loss": 0.793,
      "step": 14825
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.66726016998291,
      "learning_rate": 1.1146535475832989e-05,
      "loss": 1.045,
      "step": 14826
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7349004745483398,
      "learning_rate": 1.1145896559435199e-05,
      "loss": 0.9494,
      "step": 14827
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7198424339294434,
      "learning_rate": 1.114525764303741e-05,
      "loss": 0.8803,
      "step": 14828
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5484706163406372,
      "learning_rate": 1.114461872663962e-05,
      "loss": 0.9746,
      "step": 14829
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2703473567962646,
      "learning_rate": 1.1143979810241832e-05,
      "loss": 1.0842,
      "step": 14830
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6105327606201172,
      "learning_rate": 1.1143340893844042e-05,
      "loss": 0.8381,
      "step": 14831
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.349316120147705,
      "learning_rate": 1.1142701977446253e-05,
      "loss": 1.4134,
      "step": 14832
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5004656314849854,
      "learning_rate": 1.1142063061048463e-05,
      "loss": 0.7179,
      "step": 14833
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8488746881484985,
      "learning_rate": 1.1141424144650675e-05,
      "loss": 0.8709,
      "step": 14834
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6375274658203125,
      "learning_rate": 1.1140785228252885e-05,
      "loss": 0.9525,
      "step": 14835
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.501035213470459,
      "learning_rate": 1.1140146311855096e-05,
      "loss": 0.865,
      "step": 14836
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1841671466827393,
      "learning_rate": 1.1139507395457306e-05,
      "loss": 0.8038,
      "step": 14837
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.905085563659668,
      "learning_rate": 1.1138868479059518e-05,
      "loss": 1.0952,
      "step": 14838
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5510531663894653,
      "learning_rate": 1.1138229562661726e-05,
      "loss": 0.8949,
      "step": 14839
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6629959344863892,
      "learning_rate": 1.1137590646263939e-05,
      "loss": 0.7214,
      "step": 14840
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.755138874053955,
      "learning_rate": 1.1136951729866147e-05,
      "loss": 0.9393,
      "step": 14841
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6351898908615112,
      "learning_rate": 1.113631281346836e-05,
      "loss": 0.8229,
      "step": 14842
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.358704090118408,
      "learning_rate": 1.1135673897070568e-05,
      "loss": 0.9457,
      "step": 14843
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7393927574157715,
      "learning_rate": 1.1135034980672782e-05,
      "loss": 0.843,
      "step": 14844
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.7434117794036865,
      "learning_rate": 1.113439606427499e-05,
      "loss": 0.6423,
      "step": 14845
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8943703174591064,
      "learning_rate": 1.1133757147877201e-05,
      "loss": 1.0124,
      "step": 14846
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.371927499771118,
      "learning_rate": 1.1133118231479411e-05,
      "loss": 1.1379,
      "step": 14847
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.6178224086761475,
      "learning_rate": 1.1132479315081623e-05,
      "loss": 0.5248,
      "step": 14848
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6994194984436035,
      "learning_rate": 1.1131840398683833e-05,
      "loss": 0.9003,
      "step": 14849
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.8100130558013916,
      "learning_rate": 1.1131201482286044e-05,
      "loss": 0.867,
      "step": 14850
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.246913433074951,
      "learning_rate": 1.1130562565888254e-05,
      "loss": 1.0497,
      "step": 14851
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7023861408233643,
      "learning_rate": 1.1129923649490466e-05,
      "loss": 0.8232,
      "step": 14852
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.5654516220092773,
      "learning_rate": 1.1129284733092675e-05,
      "loss": 0.9761,
      "step": 14853
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.9764368534088135,
      "learning_rate": 1.1128645816694887e-05,
      "loss": 0.9326,
      "step": 14854
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5351738929748535,
      "learning_rate": 1.1128006900297097e-05,
      "loss": 0.9353,
      "step": 14855
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.398736000061035,
      "learning_rate": 1.1127367983899308e-05,
      "loss": 0.9443,
      "step": 14856
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.3409950733184814,
      "learning_rate": 1.1126729067501518e-05,
      "loss": 1.281,
      "step": 14857
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.115365743637085,
      "learning_rate": 1.112609015110373e-05,
      "loss": 0.9821,
      "step": 14858
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.789199948310852,
      "learning_rate": 1.112545123470594e-05,
      "loss": 0.7427,
      "step": 14859
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.5972530841827393,
      "learning_rate": 1.1124812318308151e-05,
      "loss": 0.8267,
      "step": 14860
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.6113483905792236,
      "learning_rate": 1.1124173401910361e-05,
      "loss": 0.9066,
      "step": 14861
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.051959037780762,
      "learning_rate": 1.1123534485512573e-05,
      "loss": 1.0643,
      "step": 14862
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7155499458312988,
      "learning_rate": 1.1122895569114782e-05,
      "loss": 0.8869,
      "step": 14863
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7604314088821411,
      "learning_rate": 1.1122256652716994e-05,
      "loss": 1.0026,
      "step": 14864
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9005277156829834,
      "learning_rate": 1.1121617736319204e-05,
      "loss": 0.9981,
      "step": 14865
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.835659146308899,
      "learning_rate": 1.1120978819921415e-05,
      "loss": 0.8941,
      "step": 14866
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7030251026153564,
      "learning_rate": 1.1120339903523625e-05,
      "loss": 0.8213,
      "step": 14867
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5184953212738037,
      "learning_rate": 1.1119700987125837e-05,
      "loss": 0.8257,
      "step": 14868
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4178266525268555,
      "learning_rate": 1.1119062070728047e-05,
      "loss": 0.7272,
      "step": 14869
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.89469313621521,
      "learning_rate": 1.1118423154330258e-05,
      "loss": 1.1223,
      "step": 14870
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.699334621429443,
      "learning_rate": 1.1117784237932468e-05,
      "loss": 1.5877,
      "step": 14871
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.097471237182617,
      "learning_rate": 1.111714532153468e-05,
      "loss": 0.8814,
      "step": 14872
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3914077281951904,
      "learning_rate": 1.111650640513689e-05,
      "loss": 0.8645,
      "step": 14873
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9305890798568726,
      "learning_rate": 1.1115867488739101e-05,
      "loss": 0.8503,
      "step": 14874
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5056254863739014,
      "learning_rate": 1.1115228572341311e-05,
      "loss": 0.3642,
      "step": 14875
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.167294502258301,
      "learning_rate": 1.1114589655943522e-05,
      "loss": 0.6941,
      "step": 14876
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.6908414363861084,
      "learning_rate": 1.111395073954573e-05,
      "loss": 1.2727,
      "step": 14877
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.891597032546997,
      "learning_rate": 1.1113311823147944e-05,
      "loss": 0.918,
      "step": 14878
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6845561265945435,
      "learning_rate": 1.1112672906750152e-05,
      "loss": 0.4034,
      "step": 14879
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1034469604492188,
      "learning_rate": 1.1112033990352365e-05,
      "loss": 0.7697,
      "step": 14880
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.654207944869995,
      "learning_rate": 1.1111395073954573e-05,
      "loss": 1.123,
      "step": 14881
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0191779136657715,
      "learning_rate": 1.1110756157556783e-05,
      "loss": 0.954,
      "step": 14882
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.308616042137146,
      "learning_rate": 1.1110117241158995e-05,
      "loss": 0.7703,
      "step": 14883
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.86291766166687,
      "learning_rate": 1.1109478324761205e-05,
      "loss": 0.8084,
      "step": 14884
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.765329837799072,
      "learning_rate": 1.1108839408363416e-05,
      "loss": 1.4741,
      "step": 14885
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.69496750831604,
      "learning_rate": 1.1108200491965626e-05,
      "loss": 0.9,
      "step": 14886
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.547786235809326,
      "learning_rate": 1.1107561575567838e-05,
      "loss": 1.3749,
      "step": 14887
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8492960929870605,
      "learning_rate": 1.1106922659170047e-05,
      "loss": 0.7596,
      "step": 14888
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5718499422073364,
      "learning_rate": 1.1106283742772259e-05,
      "loss": 0.8492,
      "step": 14889
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8894169330596924,
      "learning_rate": 1.1105644826374469e-05,
      "loss": 0.8634,
      "step": 14890
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6066718101501465,
      "learning_rate": 1.110500590997668e-05,
      "loss": 0.8839,
      "step": 14891
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.423676013946533,
      "learning_rate": 1.110436699357889e-05,
      "loss": 1.7409,
      "step": 14892
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5413743257522583,
      "learning_rate": 1.1103728077181102e-05,
      "loss": 0.8532,
      "step": 14893
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9868628978729248,
      "learning_rate": 1.1103089160783312e-05,
      "loss": 1.1089,
      "step": 14894
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.265925168991089,
      "learning_rate": 1.1102450244385523e-05,
      "loss": 0.9493,
      "step": 14895
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.848145008087158,
      "learning_rate": 1.1101811327987733e-05,
      "loss": 1.0709,
      "step": 14896
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.315913200378418,
      "learning_rate": 1.1101172411589945e-05,
      "loss": 0.5915,
      "step": 14897
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.901548147201538,
      "learning_rate": 1.1100533495192154e-05,
      "loss": 0.9063,
      "step": 14898
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5810110569000244,
      "learning_rate": 1.1099894578794366e-05,
      "loss": 0.9713,
      "step": 14899
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.6456127166748047,
      "learning_rate": 1.1099255662396576e-05,
      "loss": 0.8751,
      "step": 14900
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.120418071746826,
      "learning_rate": 1.1098616745998787e-05,
      "loss": 1.1814,
      "step": 14901
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.160885810852051,
      "learning_rate": 1.1097977829600997e-05,
      "loss": 1.2713,
      "step": 14902
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5190157890319824,
      "learning_rate": 1.1097338913203209e-05,
      "loss": 0.4454,
      "step": 14903
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.191889524459839,
      "learning_rate": 1.1096699996805419e-05,
      "loss": 0.911,
      "step": 14904
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2039406299591064,
      "learning_rate": 1.109606108040763e-05,
      "loss": 0.8524,
      "step": 14905
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9160767793655396,
      "learning_rate": 1.109542216400984e-05,
      "loss": 0.4942,
      "step": 14906
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6551920175552368,
      "learning_rate": 1.1094783247612052e-05,
      "loss": 0.441,
      "step": 14907
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4531424045562744,
      "learning_rate": 1.109414433121426e-05,
      "loss": 1.0271,
      "step": 14908
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6466879844665527,
      "learning_rate": 1.1093505414816473e-05,
      "loss": 0.8162,
      "step": 14909
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8947477340698242,
      "learning_rate": 1.1092866498418681e-05,
      "loss": 0.9166,
      "step": 14910
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.70304536819458,
      "learning_rate": 1.1092227582020894e-05,
      "loss": 0.8539,
      "step": 14911
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4226040840148926,
      "learning_rate": 1.1091588665623103e-05,
      "loss": 1.166,
      "step": 14912
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5359381437301636,
      "learning_rate": 1.1090949749225314e-05,
      "loss": 0.783,
      "step": 14913
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7443057298660278,
      "learning_rate": 1.1090310832827524e-05,
      "loss": 0.9482,
      "step": 14914
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7160866260528564,
      "learning_rate": 1.1089671916429735e-05,
      "loss": 0.8891,
      "step": 14915
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.284555912017822,
      "learning_rate": 1.1089033000031945e-05,
      "loss": 1.5754,
      "step": 14916
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.097978353500366,
      "learning_rate": 1.1088394083634157e-05,
      "loss": 0.9452,
      "step": 14917
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.317667245864868,
      "learning_rate": 1.1087755167236367e-05,
      "loss": 0.865,
      "step": 14918
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7685260772705078,
      "learning_rate": 1.1087116250838578e-05,
      "loss": 0.8537,
      "step": 14919
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3069514036178589,
      "learning_rate": 1.1086477334440788e-05,
      "loss": 0.8077,
      "step": 14920
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6163487434387207,
      "learning_rate": 1.1085838418043e-05,
      "loss": 0.9448,
      "step": 14921
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6980290412902832,
      "learning_rate": 1.108519950164521e-05,
      "loss": 0.9509,
      "step": 14922
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6537092924118042,
      "learning_rate": 1.1084560585247421e-05,
      "loss": 0.3381,
      "step": 14923
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4168022871017456,
      "learning_rate": 1.1083921668849631e-05,
      "loss": 0.7817,
      "step": 14924
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.656923770904541,
      "learning_rate": 1.1083282752451842e-05,
      "loss": 0.9063,
      "step": 14925
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5341403484344482,
      "learning_rate": 1.1082643836054052e-05,
      "loss": 1.0383,
      "step": 14926
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.343835353851318,
      "learning_rate": 1.1082004919656264e-05,
      "loss": 0.8907,
      "step": 14927
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7244757413864136,
      "learning_rate": 1.1081366003258474e-05,
      "loss": 0.8002,
      "step": 14928
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.648165225982666,
      "learning_rate": 1.1080727086860685e-05,
      "loss": 0.8088,
      "step": 14929
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3066463470458984,
      "learning_rate": 1.1080088170462895e-05,
      "loss": 1.0616,
      "step": 14930
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5864651203155518,
      "learning_rate": 1.1079449254065107e-05,
      "loss": 1.0705,
      "step": 14931
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8252696990966797,
      "learning_rate": 1.1078810337667317e-05,
      "loss": 0.8454,
      "step": 14932
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0660481452941895,
      "learning_rate": 1.1078171421269528e-05,
      "loss": 0.8421,
      "step": 14933
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.593433141708374,
      "learning_rate": 1.1077532504871738e-05,
      "loss": 0.9129,
      "step": 14934
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7107102870941162,
      "learning_rate": 1.107689358847395e-05,
      "loss": 0.9388,
      "step": 14935
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.893219470977783,
      "learning_rate": 1.107625467207616e-05,
      "loss": 0.9128,
      "step": 14936
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0180349349975586,
      "learning_rate": 1.1075615755678371e-05,
      "loss": 0.7342,
      "step": 14937
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9808976650238037,
      "learning_rate": 1.107497683928058e-05,
      "loss": 0.9257,
      "step": 14938
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.7508127689361572,
      "learning_rate": 1.1074337922882792e-05,
      "loss": 0.7387,
      "step": 14939
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.604971170425415,
      "learning_rate": 1.1073699006485002e-05,
      "loss": 0.8864,
      "step": 14940
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6863986253738403,
      "learning_rate": 1.1073060090087214e-05,
      "loss": 0.8896,
      "step": 14941
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4042165279388428,
      "learning_rate": 1.1072421173689424e-05,
      "loss": 0.8252,
      "step": 14942
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.084073781967163,
      "learning_rate": 1.1071782257291635e-05,
      "loss": 0.9379,
      "step": 14943
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.033588171005249,
      "learning_rate": 1.1071143340893843e-05,
      "loss": 0.8379,
      "step": 14944
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6759519577026367,
      "learning_rate": 1.1070504424496056e-05,
      "loss": 0.8484,
      "step": 14945
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.464035749435425,
      "learning_rate": 1.1069865508098265e-05,
      "loss": 1.2283,
      "step": 14946
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.899539947509766,
      "learning_rate": 1.1069226591700478e-05,
      "loss": 1.3523,
      "step": 14947
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.004976272583008,
      "learning_rate": 1.1068587675302686e-05,
      "loss": 0.8164,
      "step": 14948
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4302217960357666,
      "learning_rate": 1.1067948758904898e-05,
      "loss": 0.7994,
      "step": 14949
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7263489961624146,
      "learning_rate": 1.1067309842507107e-05,
      "loss": 0.8885,
      "step": 14950
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5090787410736084,
      "learning_rate": 1.1066670926109319e-05,
      "loss": 0.9415,
      "step": 14951
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.830298900604248,
      "learning_rate": 1.1066032009711529e-05,
      "loss": 0.947,
      "step": 14952
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0414175987243652,
      "learning_rate": 1.106539309331374e-05,
      "loss": 0.9953,
      "step": 14953
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9951225519180298,
      "learning_rate": 1.106475417691595e-05,
      "loss": 0.8244,
      "step": 14954
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7320671081542969,
      "learning_rate": 1.1064115260518162e-05,
      "loss": 0.4516,
      "step": 14955
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.038180351257324,
      "learning_rate": 1.1063476344120372e-05,
      "loss": 0.8691,
      "step": 14956
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8485251665115356,
      "learning_rate": 1.1062837427722583e-05,
      "loss": 1.3901,
      "step": 14957
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4131836891174316,
      "learning_rate": 1.1062198511324793e-05,
      "loss": 1.1027,
      "step": 14958
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8135809898376465,
      "learning_rate": 1.1061559594927005e-05,
      "loss": 0.8583,
      "step": 14959
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1783933639526367,
      "learning_rate": 1.1060920678529214e-05,
      "loss": 0.8536,
      "step": 14960
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6563317775726318,
      "learning_rate": 1.1060281762131426e-05,
      "loss": 0.8939,
      "step": 14961
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0288453102111816,
      "learning_rate": 1.1059642845733636e-05,
      "loss": 1.0573,
      "step": 14962
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7708185911178589,
      "learning_rate": 1.1059003929335847e-05,
      "loss": 0.8473,
      "step": 14963
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9565234184265137,
      "learning_rate": 1.1058365012938057e-05,
      "loss": 0.9264,
      "step": 14964
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.776879906654358,
      "learning_rate": 1.1057726096540269e-05,
      "loss": 0.9659,
      "step": 14965
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4858880043029785,
      "learning_rate": 1.1057087180142479e-05,
      "loss": 0.6561,
      "step": 14966
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.557224750518799,
      "learning_rate": 1.105644826374469e-05,
      "loss": 0.7441,
      "step": 14967
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.07039475440979,
      "learning_rate": 1.10558093473469e-05,
      "loss": 0.8622,
      "step": 14968
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.214263677597046,
      "learning_rate": 1.1055170430949112e-05,
      "loss": 0.9766,
      "step": 14969
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8369718790054321,
      "learning_rate": 1.1054531514551321e-05,
      "loss": 0.8336,
      "step": 14970
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2927284240722656,
      "learning_rate": 1.1053892598153533e-05,
      "loss": 0.9987,
      "step": 14971
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7618359327316284,
      "learning_rate": 1.1053253681755743e-05,
      "loss": 0.9599,
      "step": 14972
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5511707067489624,
      "learning_rate": 1.1052614765357954e-05,
      "loss": 0.8847,
      "step": 14973
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.762981414794922,
      "learning_rate": 1.1051975848960164e-05,
      "loss": 0.8557,
      "step": 14974
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8299267292022705,
      "learning_rate": 1.1051336932562376e-05,
      "loss": 0.903,
      "step": 14975
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.607839107513428,
      "learning_rate": 1.1050698016164586e-05,
      "loss": 1.6879,
      "step": 14976
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8353183269500732,
      "learning_rate": 1.1050059099766797e-05,
      "loss": 0.9272,
      "step": 14977
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6747076511383057,
      "learning_rate": 1.1049420183369007e-05,
      "loss": 0.9873,
      "step": 14978
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.2462661266326904,
      "learning_rate": 1.1048781266971219e-05,
      "loss": 0.8404,
      "step": 14979
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7595670223236084,
      "learning_rate": 1.1048142350573427e-05,
      "loss": 0.8659,
      "step": 14980
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8550055027008057,
      "learning_rate": 1.104750343417564e-05,
      "loss": 0.8936,
      "step": 14981
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.953689455986023,
      "learning_rate": 1.1046864517777848e-05,
      "loss": 0.9327,
      "step": 14982
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4945929050445557,
      "learning_rate": 1.1046225601380061e-05,
      "loss": 0.8966,
      "step": 14983
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9916917085647583,
      "learning_rate": 1.104558668498227e-05,
      "loss": 0.8676,
      "step": 14984
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0339536666870117,
      "learning_rate": 1.1044947768584481e-05,
      "loss": 1.0312,
      "step": 14985
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9939326047897339,
      "learning_rate": 1.1044308852186691e-05,
      "loss": 0.8847,
      "step": 14986
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9818334579467773,
      "learning_rate": 1.1043669935788902e-05,
      "loss": 0.9506,
      "step": 14987
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5088905096054077,
      "learning_rate": 1.1043031019391112e-05,
      "loss": 0.4836,
      "step": 14988
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.610277771949768,
      "learning_rate": 1.1042392102993324e-05,
      "loss": 0.903,
      "step": 14989
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.553204894065857,
      "learning_rate": 1.1041753186595534e-05,
      "loss": 0.9494,
      "step": 14990
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0056779384613037,
      "learning_rate": 1.1041114270197745e-05,
      "loss": 0.9167,
      "step": 14991
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.879468321800232,
      "learning_rate": 1.1040475353799955e-05,
      "loss": 1.0135,
      "step": 14992
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.768765926361084,
      "learning_rate": 1.1039836437402167e-05,
      "loss": 0.5529,
      "step": 14993
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8134722709655762,
      "learning_rate": 1.1039197521004377e-05,
      "loss": 0.8707,
      "step": 14994
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7810862064361572,
      "learning_rate": 1.1038558604606588e-05,
      "loss": 0.9049,
      "step": 14995
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.3363430500030518,
      "learning_rate": 1.1037919688208798e-05,
      "loss": 0.7827,
      "step": 14996
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6342805624008179,
      "learning_rate": 1.103728077181101e-05,
      "loss": 0.9244,
      "step": 14997
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5351099967956543,
      "learning_rate": 1.103664185541322e-05,
      "loss": 0.9332,
      "step": 14998
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8987163305282593,
      "learning_rate": 1.1036002939015431e-05,
      "loss": 1.0164,
      "step": 14999
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.466477394104004,
      "learning_rate": 1.103536402261764e-05,
      "loss": 1.0238,
      "step": 15000
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.888444423675537,
      "learning_rate": 1.1034725106219852e-05,
      "loss": 0.9018,
      "step": 15001
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6418412923812866,
      "learning_rate": 1.1034086189822062e-05,
      "loss": 0.8508,
      "step": 15002
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9239577054977417,
      "learning_rate": 1.1033447273424274e-05,
      "loss": 0.839,
      "step": 15003
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.33194899559021,
      "learning_rate": 1.1032808357026484e-05,
      "loss": 0.9419,
      "step": 15004
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.012007713317871,
      "learning_rate": 1.1032169440628695e-05,
      "loss": 0.8819,
      "step": 15005
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7439484596252441,
      "learning_rate": 1.1031530524230905e-05,
      "loss": 0.989,
      "step": 15006
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5285441875457764,
      "learning_rate": 1.1030891607833116e-05,
      "loss": 0.3927,
      "step": 15007
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.260289192199707,
      "learning_rate": 1.1030252691435326e-05,
      "loss": 1.0502,
      "step": 15008
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9233609437942505,
      "learning_rate": 1.1029613775037538e-05,
      "loss": 0.8172,
      "step": 15009
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7818273305892944,
      "learning_rate": 1.1028974858639748e-05,
      "loss": 0.9196,
      "step": 15010
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.201897144317627,
      "learning_rate": 1.102833594224196e-05,
      "loss": 0.9309,
      "step": 15011
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.727931022644043,
      "learning_rate": 1.1027697025844169e-05,
      "loss": 0.8287,
      "step": 15012
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6093063354492188,
      "learning_rate": 1.102705810944638e-05,
      "loss": 0.9807,
      "step": 15013
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.8329663276672363,
      "learning_rate": 1.102641919304859e-05,
      "loss": 0.8496,
      "step": 15014
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.249758243560791,
      "learning_rate": 1.1025780276650802e-05,
      "loss": 0.7885,
      "step": 15015
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.596586227416992,
      "learning_rate": 1.102514136025301e-05,
      "loss": 1.6963,
      "step": 15016
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.652463674545288,
      "learning_rate": 1.1024502443855223e-05,
      "loss": 0.815,
      "step": 15017
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.677971839904785,
      "learning_rate": 1.1023863527457432e-05,
      "loss": 1.1723,
      "step": 15018
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.679974317550659,
      "learning_rate": 1.1023224611059645e-05,
      "loss": 0.9998,
      "step": 15019
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.327275276184082,
      "learning_rate": 1.1022585694661853e-05,
      "loss": 1.2142,
      "step": 15020
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.3307154178619385,
      "learning_rate": 1.1021946778264066e-05,
      "loss": 1.0658,
      "step": 15021
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.25112247467041,
      "learning_rate": 1.1021307861866274e-05,
      "loss": 1.0436,
      "step": 15022
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3333816528320312,
      "learning_rate": 1.1020668945468486e-05,
      "loss": 1.018,
      "step": 15023
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.567647933959961,
      "learning_rate": 1.1020030029070696e-05,
      "loss": 1.2111,
      "step": 15024
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8234312534332275,
      "learning_rate": 1.1019391112672907e-05,
      "loss": 1.214,
      "step": 15025
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7227988243103027,
      "learning_rate": 1.1018752196275117e-05,
      "loss": 0.8533,
      "step": 15026
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.592087984085083,
      "learning_rate": 1.1018113279877329e-05,
      "loss": 0.9077,
      "step": 15027
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.16823410987854,
      "learning_rate": 1.1017474363479539e-05,
      "loss": 0.9325,
      "step": 15028
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9493803977966309,
      "learning_rate": 1.101683544708175e-05,
      "loss": 0.9506,
      "step": 15029
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.3906924724578857,
      "learning_rate": 1.101619653068396e-05,
      "loss": 1.0589,
      "step": 15030
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6122360229492188,
      "learning_rate": 1.1015557614286172e-05,
      "loss": 0.8324,
      "step": 15031
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7635685205459595,
      "learning_rate": 1.1014918697888381e-05,
      "loss": 0.9105,
      "step": 15032
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6919344663619995,
      "learning_rate": 1.1014279781490593e-05,
      "loss": 0.9577,
      "step": 15033
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.955310583114624,
      "learning_rate": 1.1013640865092803e-05,
      "loss": 0.8403,
      "step": 15034
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4898290634155273,
      "learning_rate": 1.1013001948695014e-05,
      "loss": 0.9249,
      "step": 15035
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8364931344985962,
      "learning_rate": 1.1012363032297224e-05,
      "loss": 0.8336,
      "step": 15036
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6654448509216309,
      "learning_rate": 1.1011724115899436e-05,
      "loss": 0.9646,
      "step": 15037
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5686038732528687,
      "learning_rate": 1.1011085199501646e-05,
      "loss": 0.8707,
      "step": 15038
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8202130794525146,
      "learning_rate": 1.1010446283103857e-05,
      "loss": 1.1922,
      "step": 15039
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.477694034576416,
      "learning_rate": 1.1009807366706067e-05,
      "loss": 1.0115,
      "step": 15040
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9732381105422974,
      "learning_rate": 1.1009168450308279e-05,
      "loss": 0.9735,
      "step": 15041
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.508915901184082,
      "learning_rate": 1.1008529533910488e-05,
      "loss": 0.9524,
      "step": 15042
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6114908456802368,
      "learning_rate": 1.10078906175127e-05,
      "loss": 0.8579,
      "step": 15043
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.884021520614624,
      "learning_rate": 1.100725170111491e-05,
      "loss": 0.9392,
      "step": 15044
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.441851019859314,
      "learning_rate": 1.1006612784717121e-05,
      "loss": 0.8044,
      "step": 15045
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.881251335144043,
      "learning_rate": 1.1005973868319331e-05,
      "loss": 1.4709,
      "step": 15046
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5871832370758057,
      "learning_rate": 1.1005334951921543e-05,
      "loss": 0.9023,
      "step": 15047
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.416928768157959,
      "learning_rate": 1.1004696035523753e-05,
      "loss": 0.9738,
      "step": 15048
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.034601211547852,
      "learning_rate": 1.1004057119125964e-05,
      "loss": 1.5252,
      "step": 15049
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6368598937988281,
      "learning_rate": 1.1003418202728174e-05,
      "loss": 0.9699,
      "step": 15050
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7426594495773315,
      "learning_rate": 1.1002779286330386e-05,
      "loss": 0.951,
      "step": 15051
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.383115768432617,
      "learning_rate": 1.1002140369932595e-05,
      "loss": 0.9581,
      "step": 15052
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7129855155944824,
      "learning_rate": 1.1001501453534807e-05,
      "loss": 0.8776,
      "step": 15053
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.059551477432251,
      "learning_rate": 1.1000862537137015e-05,
      "loss": 0.7711,
      "step": 15054
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6110732555389404,
      "learning_rate": 1.1000223620739228e-05,
      "loss": 1.069,
      "step": 15055
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8084018230438232,
      "learning_rate": 1.0999584704341437e-05,
      "loss": 0.8222,
      "step": 15056
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0893752574920654,
      "learning_rate": 1.099894578794365e-05,
      "loss": 0.9081,
      "step": 15057
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8897645473480225,
      "learning_rate": 1.0998306871545858e-05,
      "loss": 0.9363,
      "step": 15058
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.676192283630371,
      "learning_rate": 1.099766795514807e-05,
      "loss": 0.835,
      "step": 15059
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7741031646728516,
      "learning_rate": 1.099702903875028e-05,
      "loss": 0.8665,
      "step": 15060
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8291984796524048,
      "learning_rate": 1.0996390122352491e-05,
      "loss": 0.9861,
      "step": 15061
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.661102294921875,
      "learning_rate": 1.09957512059547e-05,
      "loss": 0.864,
      "step": 15062
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.5572800636291504,
      "learning_rate": 1.0995112289556912e-05,
      "loss": 1.6755,
      "step": 15063
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0494041442871094,
      "learning_rate": 1.0994473373159122e-05,
      "loss": 1.0533,
      "step": 15064
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5155997276306152,
      "learning_rate": 1.0993834456761334e-05,
      "loss": 0.7867,
      "step": 15065
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8427791595458984,
      "learning_rate": 1.0993195540363544e-05,
      "loss": 0.7976,
      "step": 15066
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1561896800994873,
      "learning_rate": 1.0992556623965755e-05,
      "loss": 0.9956,
      "step": 15067
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.835458278656006,
      "learning_rate": 1.0991917707567965e-05,
      "loss": 1.4844,
      "step": 15068
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.229623794555664,
      "learning_rate": 1.0991278791170176e-05,
      "loss": 0.833,
      "step": 15069
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.9184048175811768,
      "learning_rate": 1.0990639874772386e-05,
      "loss": 1.0689,
      "step": 15070
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.244734287261963,
      "learning_rate": 1.0990000958374598e-05,
      "loss": 1.0157,
      "step": 15071
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5182467699050903,
      "learning_rate": 1.0989362041976808e-05,
      "loss": 0.868,
      "step": 15072
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.52461576461792,
      "learning_rate": 1.098872312557902e-05,
      "loss": 0.4393,
      "step": 15073
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.911532402038574,
      "learning_rate": 1.0988084209181229e-05,
      "loss": 1.2393,
      "step": 15074
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.289017915725708,
      "learning_rate": 1.098744529278344e-05,
      "loss": 0.8902,
      "step": 15075
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.5594046115875244,
      "learning_rate": 1.098680637638565e-05,
      "loss": 1.0219,
      "step": 15076
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.8857011795043945,
      "learning_rate": 1.0986167459987862e-05,
      "loss": 0.9245,
      "step": 15077
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8815077543258667,
      "learning_rate": 1.0985528543590072e-05,
      "loss": 1.0382,
      "step": 15078
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.67828106880188,
      "learning_rate": 1.0984889627192283e-05,
      "loss": 1.1754,
      "step": 15079
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7558835744857788,
      "learning_rate": 1.0984250710794493e-05,
      "loss": 0.8643,
      "step": 15080
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6759910583496094,
      "learning_rate": 1.0983611794396705e-05,
      "loss": 0.9043,
      "step": 15081
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.7202506065368652,
      "learning_rate": 1.0982972877998915e-05,
      "loss": 0.9372,
      "step": 15082
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.743813395500183,
      "learning_rate": 1.0982333961601126e-05,
      "loss": 0.8027,
      "step": 15083
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9703644514083862,
      "learning_rate": 1.0981695045203336e-05,
      "loss": 0.8347,
      "step": 15084
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.424002170562744,
      "learning_rate": 1.0981056128805548e-05,
      "loss": 1.6907,
      "step": 15085
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6393592357635498,
      "learning_rate": 1.0980417212407758e-05,
      "loss": 0.6479,
      "step": 15086
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5193259716033936,
      "learning_rate": 1.0979778296009969e-05,
      "loss": 0.8298,
      "step": 15087
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.4958770275115967,
      "learning_rate": 1.0979139379612179e-05,
      "loss": 1.0004,
      "step": 15088
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8675109148025513,
      "learning_rate": 1.097850046321439e-05,
      "loss": 0.9001,
      "step": 15089
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1764674186706543,
      "learning_rate": 1.0977861546816599e-05,
      "loss": 0.951,
      "step": 15090
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.420705318450928,
      "learning_rate": 1.0977222630418812e-05,
      "loss": 1.4798,
      "step": 15091
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7519421577453613,
      "learning_rate": 1.097658371402102e-05,
      "loss": 0.9887,
      "step": 15092
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.80423104763031,
      "learning_rate": 1.0975944797623233e-05,
      "loss": 1.0023,
      "step": 15093
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.438753843307495,
      "learning_rate": 1.0975305881225441e-05,
      "loss": 0.7773,
      "step": 15094
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0781655311584473,
      "learning_rate": 1.0974666964827653e-05,
      "loss": 0.8873,
      "step": 15095
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5860886573791504,
      "learning_rate": 1.0974028048429863e-05,
      "loss": 0.7999,
      "step": 15096
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9557650089263916,
      "learning_rate": 1.0973389132032074e-05,
      "loss": 0.8157,
      "step": 15097
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.728875756263733,
      "learning_rate": 1.0972750215634284e-05,
      "loss": 1.1584,
      "step": 15098
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.651261806488037,
      "learning_rate": 1.0972111299236496e-05,
      "loss": 0.9061,
      "step": 15099
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.131950855255127,
      "learning_rate": 1.0971472382838706e-05,
      "loss": 1.0921,
      "step": 15100
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4507020711898804,
      "learning_rate": 1.0970833466440917e-05,
      "loss": 0.7727,
      "step": 15101
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.894423246383667,
      "learning_rate": 1.0970194550043127e-05,
      "loss": 0.7638,
      "step": 15102
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0072553157806396,
      "learning_rate": 1.0969555633645339e-05,
      "loss": 0.9013,
      "step": 15103
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.807479739189148,
      "learning_rate": 1.0968916717247548e-05,
      "loss": 0.7291,
      "step": 15104
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7917226552963257,
      "learning_rate": 1.096827780084976e-05,
      "loss": 0.901,
      "step": 15105
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2855513095855713,
      "learning_rate": 1.096763888445197e-05,
      "loss": 1.0897,
      "step": 15106
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.21824049949646,
      "learning_rate": 1.0966999968054181e-05,
      "loss": 1.116,
      "step": 15107
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5446124076843262,
      "learning_rate": 1.0966361051656391e-05,
      "loss": 0.8692,
      "step": 15108
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8732500076293945,
      "learning_rate": 1.0965722135258603e-05,
      "loss": 0.867,
      "step": 15109
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7129353284835815,
      "learning_rate": 1.0965083218860813e-05,
      "loss": 0.8957,
      "step": 15110
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0036025047302246,
      "learning_rate": 1.0964444302463024e-05,
      "loss": 1.0475,
      "step": 15111
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4428825378417969,
      "learning_rate": 1.0963805386065234e-05,
      "loss": 0.8438,
      "step": 15112
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8437048196792603,
      "learning_rate": 1.0963166469667446e-05,
      "loss": 0.8869,
      "step": 15113
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5462738275527954,
      "learning_rate": 1.0962527553269655e-05,
      "loss": 0.8187,
      "step": 15114
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6598889827728271,
      "learning_rate": 1.0961888636871867e-05,
      "loss": 0.8825,
      "step": 15115
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.114480972290039,
      "learning_rate": 1.0961249720474077e-05,
      "loss": 1.0829,
      "step": 15116
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7210228443145752,
      "learning_rate": 1.0960610804076288e-05,
      "loss": 0.9613,
      "step": 15117
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.81987726688385,
      "learning_rate": 1.0959971887678498e-05,
      "loss": 0.8264,
      "step": 15118
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.783879280090332,
      "learning_rate": 1.095933297128071e-05,
      "loss": 0.9176,
      "step": 15119
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7491124868392944,
      "learning_rate": 1.095869405488292e-05,
      "loss": 0.9543,
      "step": 15120
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.054356575012207,
      "learning_rate": 1.0958055138485131e-05,
      "loss": 0.9067,
      "step": 15121
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3173065185546875,
      "learning_rate": 1.0957416222087341e-05,
      "loss": 0.9427,
      "step": 15122
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.251378536224365,
      "learning_rate": 1.0956777305689553e-05,
      "loss": 1.5217,
      "step": 15123
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.773960828781128,
      "learning_rate": 1.0956138389291762e-05,
      "loss": 0.9587,
      "step": 15124
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1798343658447266,
      "learning_rate": 1.0955499472893974e-05,
      "loss": 0.9666,
      "step": 15125
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7039731740951538,
      "learning_rate": 1.0954860556496182e-05,
      "loss": 0.9238,
      "step": 15126
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.464437484741211,
      "learning_rate": 1.0954221640098395e-05,
      "loss": 0.7721,
      "step": 15127
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9720360040664673,
      "learning_rate": 1.0953582723700604e-05,
      "loss": 0.9626,
      "step": 15128
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8827910423278809,
      "learning_rate": 1.0952943807302817e-05,
      "loss": 1.0996,
      "step": 15129
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.789675235748291,
      "learning_rate": 1.0952304890905025e-05,
      "loss": 0.936,
      "step": 15130
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0524942874908447,
      "learning_rate": 1.0951665974507236e-05,
      "loss": 0.9458,
      "step": 15131
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.244631052017212,
      "learning_rate": 1.0951027058109446e-05,
      "loss": 0.6068,
      "step": 15132
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.251704216003418,
      "learning_rate": 1.0950388141711658e-05,
      "loss": 1.0298,
      "step": 15133
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.215008020401001,
      "learning_rate": 1.0949749225313868e-05,
      "loss": 1.0443,
      "step": 15134
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7843379974365234,
      "learning_rate": 1.094911030891608e-05,
      "loss": 0.711,
      "step": 15135
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4370392560958862,
      "learning_rate": 1.0948471392518289e-05,
      "loss": 0.8299,
      "step": 15136
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8965226411819458,
      "learning_rate": 1.09478324761205e-05,
      "loss": 0.8344,
      "step": 15137
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3942558765411377,
      "learning_rate": 1.094719355972271e-05,
      "loss": 0.7604,
      "step": 15138
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6604437828063965,
      "learning_rate": 1.0946554643324922e-05,
      "loss": 0.9427,
      "step": 15139
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1651575565338135,
      "learning_rate": 1.0945915726927132e-05,
      "loss": 0.7741,
      "step": 15140
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5398181676864624,
      "learning_rate": 1.0945276810529343e-05,
      "loss": 0.8444,
      "step": 15141
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6239497661590576,
      "learning_rate": 1.0944637894131553e-05,
      "loss": 0.8818,
      "step": 15142
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.103285312652588,
      "learning_rate": 1.0943998977733765e-05,
      "loss": 0.9998,
      "step": 15143
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.970735788345337,
      "learning_rate": 1.0943360061335975e-05,
      "loss": 0.547,
      "step": 15144
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.490485906600952,
      "learning_rate": 1.0942721144938186e-05,
      "loss": 1.0479,
      "step": 15145
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7816941738128662,
      "learning_rate": 1.0942082228540396e-05,
      "loss": 0.873,
      "step": 15146
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7200051546096802,
      "learning_rate": 1.0941443312142608e-05,
      "loss": 0.9246,
      "step": 15147
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8803144693374634,
      "learning_rate": 1.0940804395744818e-05,
      "loss": 0.938,
      "step": 15148
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6985336542129517,
      "learning_rate": 1.0940165479347029e-05,
      "loss": 0.8868,
      "step": 15149
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.409743547439575,
      "learning_rate": 1.0939526562949239e-05,
      "loss": 0.8112,
      "step": 15150
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.8580243587493896,
      "learning_rate": 1.093888764655145e-05,
      "loss": 0.9538,
      "step": 15151
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7702982425689697,
      "learning_rate": 1.093824873015366e-05,
      "loss": 0.9032,
      "step": 15152
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.7594211101531982,
      "learning_rate": 1.0937609813755872e-05,
      "loss": 0.851,
      "step": 15153
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2861390113830566,
      "learning_rate": 1.0936970897358082e-05,
      "loss": 0.9609,
      "step": 15154
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.163910150527954,
      "learning_rate": 1.0936331980960293e-05,
      "loss": 0.9541,
      "step": 15155
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9585682153701782,
      "learning_rate": 1.0935693064562503e-05,
      "loss": 1.0266,
      "step": 15156
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9071894884109497,
      "learning_rate": 1.0935054148164715e-05,
      "loss": 0.5141,
      "step": 15157
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.952662229537964,
      "learning_rate": 1.0934415231766925e-05,
      "loss": 0.9955,
      "step": 15158
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.2797179222106934,
      "learning_rate": 1.0933776315369136e-05,
      "loss": 0.8245,
      "step": 15159
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.625852108001709,
      "learning_rate": 1.0933137398971346e-05,
      "loss": 0.8892,
      "step": 15160
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8105655908584595,
      "learning_rate": 1.0932498482573557e-05,
      "loss": 0.5763,
      "step": 15161
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.5510101318359375,
      "learning_rate": 1.0931859566175766e-05,
      "loss": 0.8262,
      "step": 15162
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7383977174758911,
      "learning_rate": 1.0931220649777979e-05,
      "loss": 0.9002,
      "step": 15163
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.3818793296813965,
      "learning_rate": 1.0930581733380187e-05,
      "loss": 0.5972,
      "step": 15164
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0795345306396484,
      "learning_rate": 1.09299428169824e-05,
      "loss": 1.0891,
      "step": 15165
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7489484548568726,
      "learning_rate": 1.0929303900584608e-05,
      "loss": 1.1031,
      "step": 15166
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.7203779220581055,
      "learning_rate": 1.0928664984186822e-05,
      "loss": 1.1249,
      "step": 15167
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6839513778686523,
      "learning_rate": 1.092802606778903e-05,
      "loss": 0.8274,
      "step": 15168
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7156615257263184,
      "learning_rate": 1.0927387151391241e-05,
      "loss": 0.9567,
      "step": 15169
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9076275825500488,
      "learning_rate": 1.0926748234993451e-05,
      "loss": 1.1948,
      "step": 15170
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0824906826019287,
      "learning_rate": 1.0926109318595663e-05,
      "loss": 1.0159,
      "step": 15171
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6701021194458008,
      "learning_rate": 1.0925470402197873e-05,
      "loss": 0.8816,
      "step": 15172
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.545312404632568,
      "learning_rate": 1.0924831485800084e-05,
      "loss": 1.238,
      "step": 15173
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.4732489585876465,
      "learning_rate": 1.0924192569402294e-05,
      "loss": 0.932,
      "step": 15174
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4476863145828247,
      "learning_rate": 1.0923553653004506e-05,
      "loss": 0.8724,
      "step": 15175
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6370446681976318,
      "learning_rate": 1.0922914736606715e-05,
      "loss": 0.9011,
      "step": 15176
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4731370210647583,
      "learning_rate": 1.0922275820208927e-05,
      "loss": 0.4265,
      "step": 15177
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.5879387855529785,
      "learning_rate": 1.0921636903811137e-05,
      "loss": 1.1091,
      "step": 15178
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9047737121582031,
      "learning_rate": 1.0920997987413348e-05,
      "loss": 0.9313,
      "step": 15179
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.1226515769958496,
      "learning_rate": 1.0920359071015558e-05,
      "loss": 0.7613,
      "step": 15180
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.501229763031006,
      "learning_rate": 1.091972015461777e-05,
      "loss": 1.4311,
      "step": 15181
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8000397682189941,
      "learning_rate": 1.091908123821998e-05,
      "loss": 0.7673,
      "step": 15182
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3085159063339233,
      "learning_rate": 1.0918442321822191e-05,
      "loss": 0.4743,
      "step": 15183
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4936314821243286,
      "learning_rate": 1.0917803405424401e-05,
      "loss": 0.8812,
      "step": 15184
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5571473836898804,
      "learning_rate": 1.0917164489026613e-05,
      "loss": 0.435,
      "step": 15185
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2445731163024902,
      "learning_rate": 1.0916525572628822e-05,
      "loss": 0.9949,
      "step": 15186
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.453153133392334,
      "learning_rate": 1.0915886656231034e-05,
      "loss": 0.939,
      "step": 15187
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4993199110031128,
      "learning_rate": 1.0915247739833244e-05,
      "loss": 0.3829,
      "step": 15188
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.711503028869629,
      "learning_rate": 1.0914608823435455e-05,
      "loss": 0.8881,
      "step": 15189
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.294675827026367,
      "learning_rate": 1.0913969907037665e-05,
      "loss": 1.036,
      "step": 15190
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.967695713043213,
      "learning_rate": 1.0913330990639877e-05,
      "loss": 0.8823,
      "step": 15191
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6320858001708984,
      "learning_rate": 1.0912692074242087e-05,
      "loss": 0.9219,
      "step": 15192
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.607433557510376,
      "learning_rate": 1.0912053157844298e-05,
      "loss": 0.9016,
      "step": 15193
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7903732061386108,
      "learning_rate": 1.0911414241446508e-05,
      "loss": 0.9298,
      "step": 15194
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3983705043792725,
      "learning_rate": 1.091077532504872e-05,
      "loss": 0.6988,
      "step": 15195
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9429876804351807,
      "learning_rate": 1.091013640865093e-05,
      "loss": 0.708,
      "step": 15196
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9268332719802856,
      "learning_rate": 1.0909497492253141e-05,
      "loss": 1.3347,
      "step": 15197
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4875075817108154,
      "learning_rate": 1.0908858575855351e-05,
      "loss": 0.9523,
      "step": 15198
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2809767723083496,
      "learning_rate": 1.0908219659457562e-05,
      "loss": 1.0618,
      "step": 15199
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6842838525772095,
      "learning_rate": 1.090758074305977e-05,
      "loss": 0.8377,
      "step": 15200
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.734391450881958,
      "learning_rate": 1.0906941826661984e-05,
      "loss": 0.9649,
      "step": 15201
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9644962549209595,
      "learning_rate": 1.0906302910264192e-05,
      "loss": 0.8624,
      "step": 15202
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8563748598098755,
      "learning_rate": 1.0905663993866405e-05,
      "loss": 0.8574,
      "step": 15203
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9477282762527466,
      "learning_rate": 1.0905025077468613e-05,
      "loss": 0.8927,
      "step": 15204
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6004761457443237,
      "learning_rate": 1.0904386161070825e-05,
      "loss": 0.8074,
      "step": 15205
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7463304996490479,
      "learning_rate": 1.0903747244673035e-05,
      "loss": 0.7034,
      "step": 15206
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.818060278892517,
      "learning_rate": 1.0903108328275246e-05,
      "loss": 0.9077,
      "step": 15207
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.642536997795105,
      "learning_rate": 1.0902469411877456e-05,
      "loss": 0.86,
      "step": 15208
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5270274877548218,
      "learning_rate": 1.0901830495479668e-05,
      "loss": 0.8118,
      "step": 15209
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9827293157577515,
      "learning_rate": 1.0901191579081878e-05,
      "loss": 0.8084,
      "step": 15210
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5628820657730103,
      "learning_rate": 1.0900552662684089e-05,
      "loss": 0.92,
      "step": 15211
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8634709119796753,
      "learning_rate": 1.0899913746286299e-05,
      "loss": 0.8377,
      "step": 15212
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.218357563018799,
      "learning_rate": 1.089927482988851e-05,
      "loss": 0.8321,
      "step": 15213
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9078643321990967,
      "learning_rate": 1.089863591349072e-05,
      "loss": 0.8741,
      "step": 15214
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.090322256088257,
      "learning_rate": 1.0897996997092932e-05,
      "loss": 0.8345,
      "step": 15215
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0007479190826416,
      "learning_rate": 1.0897358080695142e-05,
      "loss": 0.9534,
      "step": 15216
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.8217413425445557,
      "learning_rate": 1.0896719164297353e-05,
      "loss": 1.0438,
      "step": 15217
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.309648513793945,
      "learning_rate": 1.0896080247899563e-05,
      "loss": 1.5492,
      "step": 15218
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6621896028518677,
      "learning_rate": 1.0895441331501775e-05,
      "loss": 0.813,
      "step": 15219
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6155339479446411,
      "learning_rate": 1.0894802415103985e-05,
      "loss": 0.8082,
      "step": 15220
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.582648754119873,
      "learning_rate": 1.0894163498706196e-05,
      "loss": 1.1648,
      "step": 15221
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6035637855529785,
      "learning_rate": 1.0893524582308406e-05,
      "loss": 1.003,
      "step": 15222
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.0361998081207275,
      "learning_rate": 1.0892885665910618e-05,
      "loss": 1.0441,
      "step": 15223
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.463155746459961,
      "learning_rate": 1.0892246749512827e-05,
      "loss": 0.7029,
      "step": 15224
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.630763530731201,
      "learning_rate": 1.0891607833115039e-05,
      "loss": 1.0064,
      "step": 15225
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9930403232574463,
      "learning_rate": 1.0890968916717249e-05,
      "loss": 1.0428,
      "step": 15226
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8574501276016235,
      "learning_rate": 1.089033000031946e-05,
      "loss": 0.9217,
      "step": 15227
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0747971534729004,
      "learning_rate": 1.088969108392167e-05,
      "loss": 1.0294,
      "step": 15228
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.130906105041504,
      "learning_rate": 1.0889052167523882e-05,
      "loss": 1.0012,
      "step": 15229
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.8554911613464355,
      "learning_rate": 1.0888413251126092e-05,
      "loss": 0.8277,
      "step": 15230
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8688926696777344,
      "learning_rate": 1.0887774334728303e-05,
      "loss": 0.8764,
      "step": 15231
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.607423305511475,
      "learning_rate": 1.0887135418330513e-05,
      "loss": 1.5676,
      "step": 15232
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9702531099319458,
      "learning_rate": 1.0886496501932725e-05,
      "loss": 1.1488,
      "step": 15233
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3682587146759033,
      "learning_rate": 1.0885857585534934e-05,
      "loss": 1.016,
      "step": 15234
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.441472053527832,
      "learning_rate": 1.0885218669137146e-05,
      "loss": 1.1654,
      "step": 15235
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8016350269317627,
      "learning_rate": 1.0884579752739354e-05,
      "loss": 0.974,
      "step": 15236
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6763581037521362,
      "learning_rate": 1.0883940836341567e-05,
      "loss": 0.4625,
      "step": 15237
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8708595037460327,
      "learning_rate": 1.0883301919943775e-05,
      "loss": 0.9223,
      "step": 15238
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6812247037887573,
      "learning_rate": 1.0882663003545989e-05,
      "loss": 0.8502,
      "step": 15239
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8716737031936646,
      "learning_rate": 1.0882024087148197e-05,
      "loss": 0.9299,
      "step": 15240
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.58233642578125,
      "learning_rate": 1.0881385170750408e-05,
      "loss": 0.7257,
      "step": 15241
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5021108388900757,
      "learning_rate": 1.0880746254352618e-05,
      "loss": 0.8895,
      "step": 15242
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5699411630630493,
      "learning_rate": 1.088010733795483e-05,
      "loss": 0.9562,
      "step": 15243
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.72645902633667,
      "learning_rate": 1.087946842155704e-05,
      "loss": 0.9032,
      "step": 15244
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4354382753372192,
      "learning_rate": 1.0878829505159251e-05,
      "loss": 0.8862,
      "step": 15245
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.770448684692383,
      "learning_rate": 1.0878190588761461e-05,
      "loss": 1.5709,
      "step": 15246
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.8952088356018066,
      "learning_rate": 1.0877551672363673e-05,
      "loss": 0.9495,
      "step": 15247
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7179981470108032,
      "learning_rate": 1.0876912755965882e-05,
      "loss": 0.9422,
      "step": 15248
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9816049337387085,
      "learning_rate": 1.0876273839568094e-05,
      "loss": 0.9877,
      "step": 15249
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.99911367893219,
      "learning_rate": 1.0875634923170304e-05,
      "loss": 1.0001,
      "step": 15250
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.245485544204712,
      "learning_rate": 1.0874996006772515e-05,
      "loss": 1.0887,
      "step": 15251
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6176129579544067,
      "learning_rate": 1.0874357090374725e-05,
      "loss": 1.1816,
      "step": 15252
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6400773525238037,
      "learning_rate": 1.0873718173976937e-05,
      "loss": 0.9512,
      "step": 15253
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4998849630355835,
      "learning_rate": 1.0873079257579147e-05,
      "loss": 0.8194,
      "step": 15254
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.534560203552246,
      "learning_rate": 1.0872440341181358e-05,
      "loss": 0.87,
      "step": 15255
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6465615034103394,
      "learning_rate": 1.0871801424783568e-05,
      "loss": 0.752,
      "step": 15256
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6836440563201904,
      "learning_rate": 1.087116250838578e-05,
      "loss": 0.8952,
      "step": 15257
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.081606388092041,
      "learning_rate": 1.087052359198799e-05,
      "loss": 0.935,
      "step": 15258
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4621970653533936,
      "learning_rate": 1.0869884675590201e-05,
      "loss": 0.4176,
      "step": 15259
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0275814533233643,
      "learning_rate": 1.0869245759192411e-05,
      "loss": 0.8044,
      "step": 15260
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2501134872436523,
      "learning_rate": 1.0868606842794622e-05,
      "loss": 0.8815,
      "step": 15261
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7525423765182495,
      "learning_rate": 1.0867967926396832e-05,
      "loss": 0.8618,
      "step": 15262
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.506343364715576,
      "learning_rate": 1.0867329009999044e-05,
      "loss": 0.8279,
      "step": 15263
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.29890513420105,
      "learning_rate": 1.0866690093601254e-05,
      "loss": 0.802,
      "step": 15264
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6514613628387451,
      "learning_rate": 1.0866051177203465e-05,
      "loss": 0.7905,
      "step": 15265
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.116502285003662,
      "learning_rate": 1.0865412260805675e-05,
      "loss": 1.5491,
      "step": 15266
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4876704216003418,
      "learning_rate": 1.0864773344407887e-05,
      "loss": 0.735,
      "step": 15267
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9488211870193481,
      "learning_rate": 1.0864134428010096e-05,
      "loss": 0.9074,
      "step": 15268
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.200322389602661,
      "learning_rate": 1.0863495511612305e-05,
      "loss": 1.0585,
      "step": 15269
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.369082450866699,
      "learning_rate": 1.0862856595214518e-05,
      "loss": 0.9973,
      "step": 15270
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.4004340171813965,
      "learning_rate": 1.0862217678816726e-05,
      "loss": 1.0368,
      "step": 15271
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5438634157180786,
      "learning_rate": 1.0861578762418938e-05,
      "loss": 0.9321,
      "step": 15272
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8721956014633179,
      "learning_rate": 1.0860939846021147e-05,
      "loss": 0.8511,
      "step": 15273
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9311431646347046,
      "learning_rate": 1.0860300929623359e-05,
      "loss": 0.9154,
      "step": 15274
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.924070119857788,
      "learning_rate": 1.0859662013225569e-05,
      "loss": 0.9041,
      "step": 15275
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8547699451446533,
      "learning_rate": 1.085902309682778e-05,
      "loss": 1.1531,
      "step": 15276
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.068922758102417,
      "learning_rate": 1.085838418042999e-05,
      "loss": 1.0508,
      "step": 15277
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9017664194107056,
      "learning_rate": 1.0857745264032202e-05,
      "loss": 0.8764,
      "step": 15278
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4937891960144043,
      "learning_rate": 1.0857106347634412e-05,
      "loss": 0.762,
      "step": 15279
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0742363929748535,
      "learning_rate": 1.0856467431236623e-05,
      "loss": 1.2927,
      "step": 15280
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.996907114982605,
      "learning_rate": 1.0855828514838833e-05,
      "loss": 1.1704,
      "step": 15281
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8358640670776367,
      "learning_rate": 1.0855189598441045e-05,
      "loss": 0.948,
      "step": 15282
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7393133640289307,
      "learning_rate": 1.0854550682043254e-05,
      "loss": 1.0027,
      "step": 15283
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.494305372238159,
      "learning_rate": 1.0853911765645466e-05,
      "loss": 0.8416,
      "step": 15284
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9623949527740479,
      "learning_rate": 1.0853272849247676e-05,
      "loss": 0.9219,
      "step": 15285
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.830801248550415,
      "learning_rate": 1.0852633932849887e-05,
      "loss": 0.9317,
      "step": 15286
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.505314826965332,
      "learning_rate": 1.0851995016452097e-05,
      "loss": 0.9575,
      "step": 15287
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.5721664428710938,
      "learning_rate": 1.0851356100054309e-05,
      "loss": 0.9034,
      "step": 15288
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.9676663875579834,
      "learning_rate": 1.0850717183656519e-05,
      "loss": 1.1138,
      "step": 15289
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5108190774917603,
      "learning_rate": 1.085007826725873e-05,
      "loss": 0.8373,
      "step": 15290
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.750206708908081,
      "learning_rate": 1.084943935086094e-05,
      "loss": 0.8357,
      "step": 15291
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5726337432861328,
      "learning_rate": 1.0848800434463152e-05,
      "loss": 0.8987,
      "step": 15292
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7798813581466675,
      "learning_rate": 1.0848161518065361e-05,
      "loss": 0.9124,
      "step": 15293
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.316833972930908,
      "learning_rate": 1.0847522601667573e-05,
      "loss": 1.3393,
      "step": 15294
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9595155715942383,
      "learning_rate": 1.0846883685269783e-05,
      "loss": 0.8363,
      "step": 15295
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4862364530563354,
      "learning_rate": 1.0846244768871994e-05,
      "loss": 0.8952,
      "step": 15296
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6357818841934204,
      "learning_rate": 1.0845605852474204e-05,
      "loss": 0.932,
      "step": 15297
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6380678415298462,
      "learning_rate": 1.0844966936076416e-05,
      "loss": 0.8955,
      "step": 15298
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.4558680057525635,
      "learning_rate": 1.0844328019678626e-05,
      "loss": 1.0908,
      "step": 15299
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.5661048889160156,
      "learning_rate": 1.0843689103280837e-05,
      "loss": 0.7142,
      "step": 15300
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9420281648635864,
      "learning_rate": 1.0843050186883047e-05,
      "loss": 0.8419,
      "step": 15301
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.116114377975464,
      "learning_rate": 1.0842411270485259e-05,
      "loss": 1.0817,
      "step": 15302
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.768420934677124,
      "learning_rate": 1.0841772354087467e-05,
      "loss": 0.8371,
      "step": 15303
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8385599851608276,
      "learning_rate": 1.084113343768968e-05,
      "loss": 0.7722,
      "step": 15304
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.880514621734619,
      "learning_rate": 1.0840494521291888e-05,
      "loss": 1.1658,
      "step": 15305
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.967300295829773,
      "learning_rate": 1.0839855604894101e-05,
      "loss": 1.0358,
      "step": 15306
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8121451139450073,
      "learning_rate": 1.083921668849631e-05,
      "loss": 0.8331,
      "step": 15307
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.991487503051758,
      "learning_rate": 1.0838577772098521e-05,
      "loss": 1.6968,
      "step": 15308
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8273411989212036,
      "learning_rate": 1.0837938855700731e-05,
      "loss": 0.8689,
      "step": 15309
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7972711324691772,
      "learning_rate": 1.0837299939302942e-05,
      "loss": 0.9554,
      "step": 15310
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6080217361450195,
      "learning_rate": 1.0836661022905152e-05,
      "loss": 0.8449,
      "step": 15311
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.833467721939087,
      "learning_rate": 1.0836022106507364e-05,
      "loss": 0.921,
      "step": 15312
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0011961460113525,
      "learning_rate": 1.0835383190109574e-05,
      "loss": 0.8808,
      "step": 15313
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6245496273040771,
      "learning_rate": 1.0834744273711785e-05,
      "loss": 0.8032,
      "step": 15314
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7783547639846802,
      "learning_rate": 1.0834105357313995e-05,
      "loss": 0.8827,
      "step": 15315
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.778397083282471,
      "learning_rate": 1.0833466440916207e-05,
      "loss": 1.4677,
      "step": 15316
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.033317804336548,
      "learning_rate": 1.0832827524518417e-05,
      "loss": 1.0218,
      "step": 15317
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.857325792312622,
      "learning_rate": 1.0832188608120628e-05,
      "loss": 1.3676,
      "step": 15318
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.94100022315979,
      "learning_rate": 1.0831549691722838e-05,
      "loss": 0.6297,
      "step": 15319
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.092445135116577,
      "learning_rate": 1.083091077532505e-05,
      "loss": 0.9449,
      "step": 15320
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.538062572479248,
      "learning_rate": 1.083027185892726e-05,
      "loss": 1.2146,
      "step": 15321
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9157146215438843,
      "learning_rate": 1.0829632942529471e-05,
      "loss": 0.4263,
      "step": 15322
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6790854930877686,
      "learning_rate": 1.082899402613168e-05,
      "loss": 0.9622,
      "step": 15323
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2246387004852295,
      "learning_rate": 1.0828355109733892e-05,
      "loss": 1.0529,
      "step": 15324
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.634133219718933,
      "learning_rate": 1.0827716193336102e-05,
      "loss": 0.8025,
      "step": 15325
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.941712737083435,
      "learning_rate": 1.0827077276938314e-05,
      "loss": 1.1375,
      "step": 15326
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9886205196380615,
      "learning_rate": 1.0826438360540524e-05,
      "loss": 0.8809,
      "step": 15327
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4878392219543457,
      "learning_rate": 1.0825799444142735e-05,
      "loss": 0.4039,
      "step": 15328
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0543277263641357,
      "learning_rate": 1.0825160527744945e-05,
      "loss": 1.002,
      "step": 15329
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.2328665256500244,
      "learning_rate": 1.0824521611347156e-05,
      "loss": 0.8563,
      "step": 15330
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2564377784729004,
      "learning_rate": 1.0823882694949366e-05,
      "loss": 0.9225,
      "step": 15331
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2654311656951904,
      "learning_rate": 1.0823243778551578e-05,
      "loss": 0.8384,
      "step": 15332
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.55110764503479,
      "learning_rate": 1.0822604862153788e-05,
      "loss": 0.4492,
      "step": 15333
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7535145282745361,
      "learning_rate": 1.0821965945756e-05,
      "loss": 0.9589,
      "step": 15334
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0719118118286133,
      "learning_rate": 1.0821327029358209e-05,
      "loss": 0.6351,
      "step": 15335
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.9266815185546875,
      "learning_rate": 1.082068811296042e-05,
      "loss": 0.942,
      "step": 15336
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.9796204566955566,
      "learning_rate": 1.082004919656263e-05,
      "loss": 0.9378,
      "step": 15337
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8596934080123901,
      "learning_rate": 1.0819410280164842e-05,
      "loss": 1.0117,
      "step": 15338
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.309117317199707,
      "learning_rate": 1.081877136376705e-05,
      "loss": 0.501,
      "step": 15339
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3792933225631714,
      "learning_rate": 1.0818132447369263e-05,
      "loss": 0.8839,
      "step": 15340
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.022829055786133,
      "learning_rate": 1.0817493530971472e-05,
      "loss": 1.2058,
      "step": 15341
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5180413722991943,
      "learning_rate": 1.0816854614573685e-05,
      "loss": 0.8383,
      "step": 15342
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.818432331085205,
      "learning_rate": 1.0816215698175893e-05,
      "loss": 0.8497,
      "step": 15343
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.174104690551758,
      "learning_rate": 1.0815576781778106e-05,
      "loss": 0.7289,
      "step": 15344
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6749367713928223,
      "learning_rate": 1.0814937865380314e-05,
      "loss": 1.2584,
      "step": 15345
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3361215591430664,
      "learning_rate": 1.0814298948982526e-05,
      "loss": 0.8877,
      "step": 15346
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5113914012908936,
      "learning_rate": 1.0813660032584736e-05,
      "loss": 0.6625,
      "step": 15347
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.421379566192627,
      "learning_rate": 1.0813021116186947e-05,
      "loss": 0.8595,
      "step": 15348
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8030900955200195,
      "learning_rate": 1.0812382199789157e-05,
      "loss": 0.9331,
      "step": 15349
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8451091051101685,
      "learning_rate": 1.0811743283391369e-05,
      "loss": 1.1411,
      "step": 15350
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.793175458908081,
      "learning_rate": 1.0811104366993579e-05,
      "loss": 0.8179,
      "step": 15351
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6804850101470947,
      "learning_rate": 1.081046545059579e-05,
      "loss": 0.8656,
      "step": 15352
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.78829824924469,
      "learning_rate": 1.0809826534198e-05,
      "loss": 0.8961,
      "step": 15353
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5916111469268799,
      "learning_rate": 1.0809187617800212e-05,
      "loss": 0.788,
      "step": 15354
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2152161598205566,
      "learning_rate": 1.0808548701402421e-05,
      "loss": 1.0068,
      "step": 15355
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4367762804031372,
      "learning_rate": 1.0807909785004633e-05,
      "loss": 0.8073,
      "step": 15356
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.502152919769287,
      "learning_rate": 1.0807270868606843e-05,
      "loss": 0.383,
      "step": 15357
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8071210384368896,
      "learning_rate": 1.0806631952209054e-05,
      "loss": 0.9933,
      "step": 15358
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0659396648406982,
      "learning_rate": 1.0805993035811264e-05,
      "loss": 0.9903,
      "step": 15359
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5508251190185547,
      "learning_rate": 1.0805354119413476e-05,
      "loss": 0.8975,
      "step": 15360
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.0752854347229,
      "learning_rate": 1.0804715203015686e-05,
      "loss": 1.473,
      "step": 15361
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4632188081741333,
      "learning_rate": 1.0804076286617897e-05,
      "loss": 0.8368,
      "step": 15362
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5477567911148071,
      "learning_rate": 1.0803437370220107e-05,
      "loss": 0.83,
      "step": 15363
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0186619758605957,
      "learning_rate": 1.0802798453822319e-05,
      "loss": 0.7802,
      "step": 15364
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.204000473022461,
      "learning_rate": 1.0802159537424528e-05,
      "loss": 1.6293,
      "step": 15365
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9044710397720337,
      "learning_rate": 1.080152062102674e-05,
      "loss": 0.9271,
      "step": 15366
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7195965051651,
      "learning_rate": 1.080088170462895e-05,
      "loss": 1.027,
      "step": 15367
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.8477134704589844,
      "learning_rate": 1.0800242788231161e-05,
      "loss": 0.9076,
      "step": 15368
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.141571521759033,
      "learning_rate": 1.0799603871833371e-05,
      "loss": 0.9349,
      "step": 15369
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.557055711746216,
      "learning_rate": 1.0798964955435583e-05,
      "loss": 0.8797,
      "step": 15370
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3559272289276123,
      "learning_rate": 1.0798326039037793e-05,
      "loss": 0.9247,
      "step": 15371
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.411838531494141,
      "learning_rate": 1.0797687122640004e-05,
      "loss": 1.3505,
      "step": 15372
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5473275184631348,
      "learning_rate": 1.0797048206242214e-05,
      "loss": 1.1325,
      "step": 15373
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6101036071777344,
      "learning_rate": 1.0796409289844426e-05,
      "loss": 0.9981,
      "step": 15374
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.703345537185669,
      "learning_rate": 1.0795770373446635e-05,
      "loss": 0.9431,
      "step": 15375
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9830141067504883,
      "learning_rate": 1.0795131457048847e-05,
      "loss": 1.225,
      "step": 15376
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7205703258514404,
      "learning_rate": 1.0794492540651055e-05,
      "loss": 1.0321,
      "step": 15377
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.055809020996094,
      "learning_rate": 1.0793853624253268e-05,
      "loss": 1.9156,
      "step": 15378
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6635608673095703,
      "learning_rate": 1.0793214707855477e-05,
      "loss": 0.8954,
      "step": 15379
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6539785861968994,
      "learning_rate": 1.079257579145769e-05,
      "loss": 0.8527,
      "step": 15380
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.841089129447937,
      "learning_rate": 1.0791936875059898e-05,
      "loss": 0.8966,
      "step": 15381
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3962836265563965,
      "learning_rate": 1.079129795866211e-05,
      "loss": 0.9395,
      "step": 15382
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.507093071937561,
      "learning_rate": 1.079065904226432e-05,
      "loss": 0.3392,
      "step": 15383
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7001373767852783,
      "learning_rate": 1.0790020125866531e-05,
      "loss": 0.8628,
      "step": 15384
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9221347570419312,
      "learning_rate": 1.078938120946874e-05,
      "loss": 0.9159,
      "step": 15385
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.025974750518799,
      "learning_rate": 1.0788742293070952e-05,
      "loss": 1.0163,
      "step": 15386
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.349536895751953,
      "learning_rate": 1.0788103376673162e-05,
      "loss": 0.7299,
      "step": 15387
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.266732931137085,
      "learning_rate": 1.0787464460275374e-05,
      "loss": 0.9114,
      "step": 15388
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5480263233184814,
      "learning_rate": 1.0786825543877584e-05,
      "loss": 0.9567,
      "step": 15389
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8153492212295532,
      "learning_rate": 1.0786186627479795e-05,
      "loss": 0.8611,
      "step": 15390
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5623674392700195,
      "learning_rate": 1.0785547711082005e-05,
      "loss": 0.6966,
      "step": 15391
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8668556213378906,
      "learning_rate": 1.0784908794684216e-05,
      "loss": 1.1067,
      "step": 15392
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7055233716964722,
      "learning_rate": 1.0784269878286426e-05,
      "loss": 0.8817,
      "step": 15393
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4722094535827637,
      "learning_rate": 1.0783630961888638e-05,
      "loss": 1.1913,
      "step": 15394
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.384330987930298,
      "learning_rate": 1.0782992045490848e-05,
      "loss": 1.0278,
      "step": 15395
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3620989322662354,
      "learning_rate": 1.078235312909306e-05,
      "loss": 1.1463,
      "step": 15396
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9519530534744263,
      "learning_rate": 1.0781714212695269e-05,
      "loss": 0.9462,
      "step": 15397
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5746002197265625,
      "learning_rate": 1.078107529629748e-05,
      "loss": 0.8925,
      "step": 15398
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.329143762588501,
      "learning_rate": 1.078043637989969e-05,
      "loss": 0.9231,
      "step": 15399
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.326098918914795,
      "learning_rate": 1.0779797463501902e-05,
      "loss": 0.7995,
      "step": 15400
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.297523498535156,
      "learning_rate": 1.0779158547104112e-05,
      "loss": 1.0431,
      "step": 15401
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5718376636505127,
      "learning_rate": 1.0778519630706323e-05,
      "loss": 0.5874,
      "step": 15402
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6032828092575073,
      "learning_rate": 1.0777880714308533e-05,
      "loss": 0.5509,
      "step": 15403
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4245145320892334,
      "learning_rate": 1.0777241797910745e-05,
      "loss": 0.8699,
      "step": 15404
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1587512493133545,
      "learning_rate": 1.0776602881512955e-05,
      "loss": 1.1912,
      "step": 15405
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.710422396659851,
      "learning_rate": 1.0775963965115166e-05,
      "loss": 0.86,
      "step": 15406
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6999772787094116,
      "learning_rate": 1.0775325048717376e-05,
      "loss": 0.9026,
      "step": 15407
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.8383140563964844,
      "learning_rate": 1.0774686132319588e-05,
      "loss": 0.9381,
      "step": 15408
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7142518758773804,
      "learning_rate": 1.0774047215921798e-05,
      "loss": 0.8311,
      "step": 15409
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.58930504322052,
      "learning_rate": 1.0773408299524009e-05,
      "loss": 0.9691,
      "step": 15410
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9908732175827026,
      "learning_rate": 1.0772769383126219e-05,
      "loss": 0.9817,
      "step": 15411
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8434178829193115,
      "learning_rate": 1.077213046672843e-05,
      "loss": 1.1728,
      "step": 15412
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.665718674659729,
      "learning_rate": 1.0771491550330639e-05,
      "loss": 0.9682,
      "step": 15413
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9614315032958984,
      "learning_rate": 1.0770852633932852e-05,
      "loss": 0.8429,
      "step": 15414
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0331647396087646,
      "learning_rate": 1.077021371753506e-05,
      "loss": 1.0139,
      "step": 15415
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4928789138793945,
      "learning_rate": 1.0769574801137273e-05,
      "loss": 0.8811,
      "step": 15416
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5611279010772705,
      "learning_rate": 1.0768935884739481e-05,
      "loss": 0.8585,
      "step": 15417
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.957399845123291,
      "learning_rate": 1.0768296968341693e-05,
      "loss": 0.8618,
      "step": 15418
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3355294466018677,
      "learning_rate": 1.0767658051943903e-05,
      "loss": 0.4609,
      "step": 15419
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.2098917961120605,
      "learning_rate": 1.0767019135546114e-05,
      "loss": 1.2653,
      "step": 15420
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0119781494140625,
      "learning_rate": 1.0766380219148324e-05,
      "loss": 1.1916,
      "step": 15421
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.297300100326538,
      "learning_rate": 1.0765741302750536e-05,
      "loss": 0.7629,
      "step": 15422
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4629476070404053,
      "learning_rate": 1.0765102386352746e-05,
      "loss": 0.9455,
      "step": 15423
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7695086002349854,
      "learning_rate": 1.0764463469954957e-05,
      "loss": 0.9776,
      "step": 15424
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.890403985977173,
      "learning_rate": 1.0763824553557167e-05,
      "loss": 0.8693,
      "step": 15425
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.194206714630127,
      "learning_rate": 1.0763185637159379e-05,
      "loss": 0.834,
      "step": 15426
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.15684175491333,
      "learning_rate": 1.0762546720761588e-05,
      "loss": 1.1134,
      "step": 15427
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.983040452003479,
      "learning_rate": 1.07619078043638e-05,
      "loss": 0.8601,
      "step": 15428
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.446469783782959,
      "learning_rate": 1.076126888796601e-05,
      "loss": 1.3607,
      "step": 15429
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8906453847885132,
      "learning_rate": 1.0760629971568221e-05,
      "loss": 0.8814,
      "step": 15430
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7922793626785278,
      "learning_rate": 1.0759991055170431e-05,
      "loss": 0.8259,
      "step": 15431
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6386168003082275,
      "learning_rate": 1.0759352138772643e-05,
      "loss": 0.8927,
      "step": 15432
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4552931785583496,
      "learning_rate": 1.0758713222374853e-05,
      "loss": 0.9792,
      "step": 15433
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.810139775276184,
      "learning_rate": 1.0758074305977064e-05,
      "loss": 0.8167,
      "step": 15434
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6508183479309082,
      "learning_rate": 1.0757435389579274e-05,
      "loss": 0.8708,
      "step": 15435
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.604609727859497,
      "learning_rate": 1.0756796473181486e-05,
      "loss": 0.7936,
      "step": 15436
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.7939910888671875,
      "learning_rate": 1.0756157556783695e-05,
      "loss": 1.3195,
      "step": 15437
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.757285714149475,
      "learning_rate": 1.0755518640385907e-05,
      "loss": 0.8402,
      "step": 15438
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0750787258148193,
      "learning_rate": 1.0754879723988117e-05,
      "loss": 0.9729,
      "step": 15439
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8778568506240845,
      "learning_rate": 1.0754240807590328e-05,
      "loss": 0.9839,
      "step": 15440
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6808890104293823,
      "learning_rate": 1.0753601891192538e-05,
      "loss": 0.8932,
      "step": 15441
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2977020740509033,
      "learning_rate": 1.075296297479475e-05,
      "loss": 1.0099,
      "step": 15442
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9181698560714722,
      "learning_rate": 1.075232405839696e-05,
      "loss": 0.8808,
      "step": 15443
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6940433979034424,
      "learning_rate": 1.0751685141999171e-05,
      "loss": 1.0211,
      "step": 15444
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.159306287765503,
      "learning_rate": 1.0751046225601381e-05,
      "loss": 1.1842,
      "step": 15445
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.337682008743286,
      "learning_rate": 1.0750407309203593e-05,
      "loss": 0.7663,
      "step": 15446
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.0936007499694824,
      "learning_rate": 1.0749768392805802e-05,
      "loss": 1.1543,
      "step": 15447
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7867083549499512,
      "learning_rate": 1.0749129476408014e-05,
      "loss": 0.7501,
      "step": 15448
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6038992404937744,
      "learning_rate": 1.0748490560010222e-05,
      "loss": 0.4873,
      "step": 15449
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.049636363983154,
      "learning_rate": 1.0747851643612435e-05,
      "loss": 0.9829,
      "step": 15450
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7555184364318848,
      "learning_rate": 1.0747212727214644e-05,
      "loss": 0.6528,
      "step": 15451
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8790326118469238,
      "learning_rate": 1.0746573810816857e-05,
      "loss": 0.8814,
      "step": 15452
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8755723237991333,
      "learning_rate": 1.0745934894419065e-05,
      "loss": 0.9379,
      "step": 15453
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5976758003234863,
      "learning_rate": 1.0745295978021276e-05,
      "loss": 0.8665,
      "step": 15454
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6387650966644287,
      "learning_rate": 1.0744657061623486e-05,
      "loss": 0.9229,
      "step": 15455
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.580902099609375,
      "learning_rate": 1.0744018145225698e-05,
      "loss": 0.891,
      "step": 15456
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6313447952270508,
      "learning_rate": 1.0743379228827908e-05,
      "loss": 0.8705,
      "step": 15457
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2990305423736572,
      "learning_rate": 1.074274031243012e-05,
      "loss": 1.0001,
      "step": 15458
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6071321964263916,
      "learning_rate": 1.0742101396032329e-05,
      "loss": 0.9191,
      "step": 15459
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0708556175231934,
      "learning_rate": 1.074146247963454e-05,
      "loss": 0.8743,
      "step": 15460
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8418025970458984,
      "learning_rate": 1.074082356323675e-05,
      "loss": 0.9414,
      "step": 15461
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9910502433776855,
      "learning_rate": 1.0740184646838962e-05,
      "loss": 0.829,
      "step": 15462
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6724284887313843,
      "learning_rate": 1.0739545730441172e-05,
      "loss": 0.9205,
      "step": 15463
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0400733947753906,
      "learning_rate": 1.0738906814043383e-05,
      "loss": 1.1117,
      "step": 15464
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5946028232574463,
      "learning_rate": 1.0738267897645593e-05,
      "loss": 0.5474,
      "step": 15465
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2003133296966553,
      "learning_rate": 1.0737628981247805e-05,
      "loss": 1.1885,
      "step": 15466
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5883417129516602,
      "learning_rate": 1.0736990064850015e-05,
      "loss": 0.786,
      "step": 15467
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7204506397247314,
      "learning_rate": 1.0736351148452226e-05,
      "loss": 0.9298,
      "step": 15468
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6979339122772217,
      "learning_rate": 1.0735712232054436e-05,
      "loss": 1.1878,
      "step": 15469
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.811036229133606,
      "learning_rate": 1.0735073315656648e-05,
      "loss": 0.7616,
      "step": 15470
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7837269306182861,
      "learning_rate": 1.0734434399258858e-05,
      "loss": 0.9155,
      "step": 15471
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7175670862197876,
      "learning_rate": 1.0733795482861069e-05,
      "loss": 0.8772,
      "step": 15472
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.064204454421997,
      "learning_rate": 1.0733156566463279e-05,
      "loss": 0.8888,
      "step": 15473
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9580187797546387,
      "learning_rate": 1.073251765006549e-05,
      "loss": 1.1683,
      "step": 15474
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2824522256851196,
      "learning_rate": 1.07318787336677e-05,
      "loss": 0.7661,
      "step": 15475
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5980007648468018,
      "learning_rate": 1.0731239817269912e-05,
      "loss": 0.9505,
      "step": 15476
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.241982936859131,
      "learning_rate": 1.0730600900872122e-05,
      "loss": 0.8066,
      "step": 15477
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3317290544509888,
      "learning_rate": 1.0729961984474333e-05,
      "loss": 0.408,
      "step": 15478
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.650938868522644,
      "learning_rate": 1.0729323068076543e-05,
      "loss": 0.8287,
      "step": 15479
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5491001605987549,
      "learning_rate": 1.0728684151678755e-05,
      "loss": 0.9128,
      "step": 15480
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4476884603500366,
      "learning_rate": 1.0728045235280965e-05,
      "loss": 0.4702,
      "step": 15481
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2650651931762695,
      "learning_rate": 1.0727406318883176e-05,
      "loss": 1.1598,
      "step": 15482
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.738656759262085,
      "learning_rate": 1.0726767402485386e-05,
      "loss": 0.8522,
      "step": 15483
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.146697998046875,
      "learning_rate": 1.0726128486087597e-05,
      "loss": 1.0529,
      "step": 15484
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8286927938461304,
      "learning_rate": 1.0725489569689806e-05,
      "loss": 0.8704,
      "step": 15485
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.672294020652771,
      "learning_rate": 1.0724850653292019e-05,
      "loss": 0.8333,
      "step": 15486
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.533431887626648,
      "learning_rate": 1.0724211736894227e-05,
      "loss": 0.909,
      "step": 15487
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9242750406265259,
      "learning_rate": 1.072357282049644e-05,
      "loss": 0.8303,
      "step": 15488
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9520236253738403,
      "learning_rate": 1.0722933904098648e-05,
      "loss": 0.7923,
      "step": 15489
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2083802223205566,
      "learning_rate": 1.0722294987700862e-05,
      "loss": 1.1679,
      "step": 15490
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.817554950714111,
      "learning_rate": 1.072165607130307e-05,
      "loss": 1.3483,
      "step": 15491
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7705581188201904,
      "learning_rate": 1.0721017154905281e-05,
      "loss": 0.8704,
      "step": 15492
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.8147337436676025,
      "learning_rate": 1.0720378238507491e-05,
      "loss": 1.228,
      "step": 15493
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.012960910797119,
      "learning_rate": 1.0719739322109703e-05,
      "loss": 0.9664,
      "step": 15494
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5831507444381714,
      "learning_rate": 1.0719100405711913e-05,
      "loss": 0.738,
      "step": 15495
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.7655160427093506,
      "learning_rate": 1.0718461489314124e-05,
      "loss": 1.5499,
      "step": 15496
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.855668067932129,
      "learning_rate": 1.0717822572916334e-05,
      "loss": 0.589,
      "step": 15497
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9828764200210571,
      "learning_rate": 1.0717183656518546e-05,
      "loss": 1.0262,
      "step": 15498
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0859551429748535,
      "learning_rate": 1.0716544740120755e-05,
      "loss": 0.9315,
      "step": 15499
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5684398412704468,
      "learning_rate": 1.0715905823722967e-05,
      "loss": 0.8196,
      "step": 15500
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.862244725227356,
      "learning_rate": 1.0715266907325177e-05,
      "loss": 0.8537,
      "step": 15501
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8526561260223389,
      "learning_rate": 1.0714627990927388e-05,
      "loss": 0.9403,
      "step": 15502
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.208780288696289,
      "learning_rate": 1.0713989074529598e-05,
      "loss": 1.1206,
      "step": 15503
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9757212400436401,
      "learning_rate": 1.071335015813181e-05,
      "loss": 0.8715,
      "step": 15504
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.889214515686035,
      "learning_rate": 1.071271124173402e-05,
      "loss": 0.8061,
      "step": 15505
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.965493083000183,
      "learning_rate": 1.0712072325336231e-05,
      "loss": 1.101,
      "step": 15506
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.988736867904663,
      "learning_rate": 1.0711433408938441e-05,
      "loss": 0.8915,
      "step": 15507
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.107100009918213,
      "learning_rate": 1.0710794492540653e-05,
      "loss": 1.4167,
      "step": 15508
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.023365020751953,
      "learning_rate": 1.0710155576142862e-05,
      "loss": 1.1929,
      "step": 15509
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.661095142364502,
      "learning_rate": 1.0709516659745074e-05,
      "loss": 0.9771,
      "step": 15510
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.024462938308716,
      "learning_rate": 1.0708877743347284e-05,
      "loss": 1.1993,
      "step": 15511
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.355304002761841,
      "learning_rate": 1.0708238826949495e-05,
      "loss": 0.9847,
      "step": 15512
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6845412254333496,
      "learning_rate": 1.0707599910551705e-05,
      "loss": 1.0631,
      "step": 15513
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7078471183776855,
      "learning_rate": 1.0706960994153917e-05,
      "loss": 0.7992,
      "step": 15514
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5861618518829346,
      "learning_rate": 1.0706322077756127e-05,
      "loss": 0.7329,
      "step": 15515
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.259239673614502,
      "learning_rate": 1.0705683161358338e-05,
      "loss": 0.7847,
      "step": 15516
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.742384433746338,
      "learning_rate": 1.0705044244960548e-05,
      "loss": 0.9163,
      "step": 15517
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.8266966342926025,
      "learning_rate": 1.070440532856276e-05,
      "loss": 1.1871,
      "step": 15518
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3255059719085693,
      "learning_rate": 1.070376641216497e-05,
      "loss": 0.6983,
      "step": 15519
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.651373386383057,
      "learning_rate": 1.0703127495767181e-05,
      "loss": 1.1722,
      "step": 15520
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8025054931640625,
      "learning_rate": 1.070248857936939e-05,
      "loss": 1.0156,
      "step": 15521
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.5977935791015625,
      "learning_rate": 1.0701849662971602e-05,
      "loss": 1.2137,
      "step": 15522
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.1289167404174805,
      "learning_rate": 1.070121074657381e-05,
      "loss": 1.5002,
      "step": 15523
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.149840831756592,
      "learning_rate": 1.0700571830176024e-05,
      "loss": 1.45,
      "step": 15524
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6661934852600098,
      "learning_rate": 1.0699932913778232e-05,
      "loss": 0.9713,
      "step": 15525
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0682342052459717,
      "learning_rate": 1.0699293997380445e-05,
      "loss": 0.8288,
      "step": 15526
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8186416625976562,
      "learning_rate": 1.0698655080982653e-05,
      "loss": 0.8746,
      "step": 15527
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5916504859924316,
      "learning_rate": 1.0698016164584865e-05,
      "loss": 0.8087,
      "step": 15528
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6071398258209229,
      "learning_rate": 1.0697377248187075e-05,
      "loss": 0.7969,
      "step": 15529
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.8564887046813965,
      "learning_rate": 1.0696738331789286e-05,
      "loss": 1.2503,
      "step": 15530
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.548309564590454,
      "learning_rate": 1.0696099415391496e-05,
      "loss": 0.8955,
      "step": 15531
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8267072439193726,
      "learning_rate": 1.0695460498993708e-05,
      "loss": 1.0797,
      "step": 15532
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.1816418170928955,
      "learning_rate": 1.0694821582595918e-05,
      "loss": 0.9325,
      "step": 15533
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.301398754119873,
      "learning_rate": 1.0694182666198129e-05,
      "loss": 1.3178,
      "step": 15534
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1306240558624268,
      "learning_rate": 1.0693543749800339e-05,
      "loss": 1.1764,
      "step": 15535
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4264745712280273,
      "learning_rate": 1.069290483340255e-05,
      "loss": 0.8521,
      "step": 15536
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7927955389022827,
      "learning_rate": 1.069226591700476e-05,
      "loss": 1.0382,
      "step": 15537
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4630107879638672,
      "learning_rate": 1.0691627000606972e-05,
      "loss": 0.8584,
      "step": 15538
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0216903686523438,
      "learning_rate": 1.0690988084209182e-05,
      "loss": 0.893,
      "step": 15539
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.663682699203491,
      "learning_rate": 1.0690349167811393e-05,
      "loss": 0.909,
      "step": 15540
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4570152759552002,
      "learning_rate": 1.0689710251413603e-05,
      "loss": 0.7402,
      "step": 15541
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.706491470336914,
      "learning_rate": 1.0689071335015815e-05,
      "loss": 0.8483,
      "step": 15542
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.268603801727295,
      "learning_rate": 1.0688432418618025e-05,
      "loss": 1.4491,
      "step": 15543
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.543657898902893,
      "learning_rate": 1.0687793502220236e-05,
      "loss": 0.9867,
      "step": 15544
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7000997066497803,
      "learning_rate": 1.0687154585822446e-05,
      "loss": 0.9018,
      "step": 15545
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5709614753723145,
      "learning_rate": 1.0686515669424657e-05,
      "loss": 0.9294,
      "step": 15546
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6756207942962646,
      "learning_rate": 1.0685876753026867e-05,
      "loss": 0.8905,
      "step": 15547
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.144641637802124,
      "learning_rate": 1.0685237836629079e-05,
      "loss": 0.9899,
      "step": 15548
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.440438747406006,
      "learning_rate": 1.0684598920231289e-05,
      "loss": 1.7206,
      "step": 15549
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.046586275100708,
      "learning_rate": 1.06839600038335e-05,
      "loss": 0.8246,
      "step": 15550
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2807135581970215,
      "learning_rate": 1.068332108743571e-05,
      "loss": 1.0834,
      "step": 15551
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1681602001190186,
      "learning_rate": 1.0682682171037922e-05,
      "loss": 0.8686,
      "step": 15552
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9499726295471191,
      "learning_rate": 1.0682043254640132e-05,
      "loss": 0.9278,
      "step": 15553
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9035371541976929,
      "learning_rate": 1.0681404338242343e-05,
      "loss": 0.941,
      "step": 15554
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.8876657485961914,
      "learning_rate": 1.0680765421844553e-05,
      "loss": 0.9131,
      "step": 15555
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4530692100524902,
      "learning_rate": 1.0680126505446764e-05,
      "loss": 0.9962,
      "step": 15556
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4664634466171265,
      "learning_rate": 1.0679487589048974e-05,
      "loss": 0.8397,
      "step": 15557
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8378835916519165,
      "learning_rate": 1.0678848672651186e-05,
      "loss": 0.8114,
      "step": 15558
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9604688882827759,
      "learning_rate": 1.0678209756253394e-05,
      "loss": 0.9461,
      "step": 15559
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5508952140808105,
      "learning_rate": 1.0677570839855607e-05,
      "loss": 1.0348,
      "step": 15560
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.319638729095459,
      "learning_rate": 1.0676931923457815e-05,
      "loss": 1.4168,
      "step": 15561
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.871969223022461,
      "learning_rate": 1.0676293007060029e-05,
      "loss": 0.9187,
      "step": 15562
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1399986743927,
      "learning_rate": 1.0675654090662237e-05,
      "loss": 0.5258,
      "step": 15563
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.568007230758667,
      "learning_rate": 1.0675015174264448e-05,
      "loss": 0.8575,
      "step": 15564
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.162312030792236,
      "learning_rate": 1.0674376257866658e-05,
      "loss": 1.3605,
      "step": 15565
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7714930772781372,
      "learning_rate": 1.067373734146887e-05,
      "loss": 0.8612,
      "step": 15566
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.955855369567871,
      "learning_rate": 1.067309842507108e-05,
      "loss": 0.6133,
      "step": 15567
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5664085149765015,
      "learning_rate": 1.0672459508673291e-05,
      "loss": 0.9206,
      "step": 15568
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0271337032318115,
      "learning_rate": 1.0671820592275501e-05,
      "loss": 0.9717,
      "step": 15569
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7071813344955444,
      "learning_rate": 1.0671181675877713e-05,
      "loss": 0.8489,
      "step": 15570
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7465046644210815,
      "learning_rate": 1.0670542759479922e-05,
      "loss": 0.9617,
      "step": 15571
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8710240125656128,
      "learning_rate": 1.0669903843082134e-05,
      "loss": 0.836,
      "step": 15572
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.650234699249268,
      "learning_rate": 1.0669264926684344e-05,
      "loss": 1.3813,
      "step": 15573
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9113954305648804,
      "learning_rate": 1.0668626010286555e-05,
      "loss": 0.9233,
      "step": 15574
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.746502161026001,
      "learning_rate": 1.0667987093888765e-05,
      "loss": 0.8376,
      "step": 15575
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6046929359436035,
      "learning_rate": 1.0667348177490977e-05,
      "loss": 0.848,
      "step": 15576
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.158308029174805,
      "learning_rate": 1.0666709261093187e-05,
      "loss": 1.1553,
      "step": 15577
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7926857471466064,
      "learning_rate": 1.0666070344695398e-05,
      "loss": 0.9784,
      "step": 15578
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8624976873397827,
      "learning_rate": 1.0665431428297608e-05,
      "loss": 0.4937,
      "step": 15579
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9681658744812012,
      "learning_rate": 1.066479251189982e-05,
      "loss": 0.9184,
      "step": 15580
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9209121465682983,
      "learning_rate": 1.066415359550203e-05,
      "loss": 0.9174,
      "step": 15581
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.308366060256958,
      "learning_rate": 1.0663514679104241e-05,
      "loss": 1.1247,
      "step": 15582
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0713627338409424,
      "learning_rate": 1.0662875762706451e-05,
      "loss": 0.7541,
      "step": 15583
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.724015712738037,
      "learning_rate": 1.0662236846308662e-05,
      "loss": 0.8721,
      "step": 15584
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8767707347869873,
      "learning_rate": 1.0661597929910872e-05,
      "loss": 0.9751,
      "step": 15585
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8427960872650146,
      "learning_rate": 1.0660959013513084e-05,
      "loss": 0.8743,
      "step": 15586
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5389199256896973,
      "learning_rate": 1.0660320097115294e-05,
      "loss": 0.7097,
      "step": 15587
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.602581977844238,
      "learning_rate": 1.0659681180717505e-05,
      "loss": 0.9988,
      "step": 15588
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7273125648498535,
      "learning_rate": 1.0659042264319715e-05,
      "loss": 0.9558,
      "step": 15589
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.434253692626953,
      "learning_rate": 1.0658403347921927e-05,
      "loss": 0.9019,
      "step": 15590
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.297579765319824,
      "learning_rate": 1.0657764431524136e-05,
      "loss": 1.6161,
      "step": 15591
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.2683393955230713,
      "learning_rate": 1.0657125515126348e-05,
      "loss": 0.8156,
      "step": 15592
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5514943599700928,
      "learning_rate": 1.0656486598728558e-05,
      "loss": 0.8755,
      "step": 15593
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6969242095947266,
      "learning_rate": 1.065584768233077e-05,
      "loss": 1.0821,
      "step": 15594
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.427126407623291,
      "learning_rate": 1.0655208765932978e-05,
      "loss": 1.0456,
      "step": 15595
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4870755672454834,
      "learning_rate": 1.065456984953519e-05,
      "loss": 0.488,
      "step": 15596
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6605569124221802,
      "learning_rate": 1.0653930933137399e-05,
      "loss": 0.8327,
      "step": 15597
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7637044191360474,
      "learning_rate": 1.0653292016739612e-05,
      "loss": 0.8945,
      "step": 15598
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6734027862548828,
      "learning_rate": 1.065265310034182e-05,
      "loss": 0.9937,
      "step": 15599
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.191033840179443,
      "learning_rate": 1.0652014183944032e-05,
      "loss": 1.137,
      "step": 15600
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8176097869873047,
      "learning_rate": 1.0651375267546242e-05,
      "loss": 1.2401,
      "step": 15601
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9645020961761475,
      "learning_rate": 1.0650736351148453e-05,
      "loss": 0.8837,
      "step": 15602
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6920411586761475,
      "learning_rate": 1.0650097434750663e-05,
      "loss": 0.9674,
      "step": 15603
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6360316276550293,
      "learning_rate": 1.0649458518352875e-05,
      "loss": 0.862,
      "step": 15604
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3346304893493652,
      "learning_rate": 1.0648819601955085e-05,
      "loss": 0.8505,
      "step": 15605
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7125815153121948,
      "learning_rate": 1.0648180685557296e-05,
      "loss": 0.842,
      "step": 15606
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.940998911857605,
      "learning_rate": 1.0647541769159506e-05,
      "loss": 0.9026,
      "step": 15607
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3845930099487305,
      "learning_rate": 1.0646902852761718e-05,
      "loss": 0.8961,
      "step": 15608
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7540876865386963,
      "learning_rate": 1.0646263936363927e-05,
      "loss": 0.6023,
      "step": 15609
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7925059795379639,
      "learning_rate": 1.0645625019966139e-05,
      "loss": 0.8071,
      "step": 15610
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.224100112915039,
      "learning_rate": 1.0644986103568349e-05,
      "loss": 0.8854,
      "step": 15611
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7887552976608276,
      "learning_rate": 1.064434718717056e-05,
      "loss": 0.9908,
      "step": 15612
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.8903884887695312,
      "learning_rate": 1.064370827077277e-05,
      "loss": 1.0827,
      "step": 15613
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6859582662582397,
      "learning_rate": 1.0643069354374982e-05,
      "loss": 1.0872,
      "step": 15614
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3627235889434814,
      "learning_rate": 1.0642430437977192e-05,
      "loss": 0.9241,
      "step": 15615
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5680992603302,
      "learning_rate": 1.0641791521579403e-05,
      "loss": 1.0459,
      "step": 15616
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9455289840698242,
      "learning_rate": 1.0641152605181613e-05,
      "loss": 1.0503,
      "step": 15617
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8672858476638794,
      "learning_rate": 1.0640513688783825e-05,
      "loss": 0.7548,
      "step": 15618
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.821359395980835,
      "learning_rate": 1.0639874772386034e-05,
      "loss": 0.8849,
      "step": 15619
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9544709920883179,
      "learning_rate": 1.0639235855988246e-05,
      "loss": 0.7754,
      "step": 15620
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4831119775772095,
      "learning_rate": 1.0638596939590456e-05,
      "loss": 0.9208,
      "step": 15621
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.041907787322998,
      "learning_rate": 1.0637958023192667e-05,
      "loss": 1.2039,
      "step": 15622
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3945960998535156,
      "learning_rate": 1.0637319106794877e-05,
      "loss": 0.8425,
      "step": 15623
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6616935729980469,
      "learning_rate": 1.0636680190397089e-05,
      "loss": 0.8495,
      "step": 15624
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.8401641845703125,
      "learning_rate": 1.0636041273999299e-05,
      "loss": 0.8186,
      "step": 15625
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.690674901008606,
      "learning_rate": 1.063540235760151e-05,
      "loss": 0.911,
      "step": 15626
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6315542459487915,
      "learning_rate": 1.063476344120372e-05,
      "loss": 0.8974,
      "step": 15627
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6342068910598755,
      "learning_rate": 1.0634124524805932e-05,
      "loss": 0.9134,
      "step": 15628
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4724167585372925,
      "learning_rate": 1.0633485608408141e-05,
      "loss": 0.9431,
      "step": 15629
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.2021052837371826,
      "learning_rate": 1.0632846692010353e-05,
      "loss": 0.8873,
      "step": 15630
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.642035961151123,
      "learning_rate": 1.0632207775612561e-05,
      "loss": 1.6911,
      "step": 15631
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.855381965637207,
      "learning_rate": 1.0631568859214774e-05,
      "loss": 1.0087,
      "step": 15632
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.449810028076172,
      "learning_rate": 1.0630929942816982e-05,
      "loss": 0.8705,
      "step": 15633
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6852017641067505,
      "learning_rate": 1.0630291026419196e-05,
      "loss": 0.9702,
      "step": 15634
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.671389579772949,
      "learning_rate": 1.0629652110021404e-05,
      "loss": 1.4216,
      "step": 15635
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8378700017929077,
      "learning_rate": 1.0629013193623617e-05,
      "loss": 0.9521,
      "step": 15636
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0891268253326416,
      "learning_rate": 1.0628374277225825e-05,
      "loss": 0.9539,
      "step": 15637
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9157767295837402,
      "learning_rate": 1.0627735360828037e-05,
      "loss": 0.948,
      "step": 15638
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.440461158752441,
      "learning_rate": 1.0627096444430247e-05,
      "loss": 1.0098,
      "step": 15639
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6458353996276855,
      "learning_rate": 1.0626457528032458e-05,
      "loss": 0.9309,
      "step": 15640
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.531479001045227,
      "learning_rate": 1.0625818611634668e-05,
      "loss": 0.9455,
      "step": 15641
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7383852005004883,
      "learning_rate": 1.062517969523688e-05,
      "loss": 0.9191,
      "step": 15642
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.715598225593567,
      "learning_rate": 1.062454077883909e-05,
      "loss": 0.9451,
      "step": 15643
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4678332805633545,
      "learning_rate": 1.0623901862441301e-05,
      "loss": 0.9548,
      "step": 15644
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.475589394569397,
      "learning_rate": 1.0623262946043511e-05,
      "loss": 0.804,
      "step": 15645
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.515360951423645,
      "learning_rate": 1.0622624029645722e-05,
      "loss": 0.8842,
      "step": 15646
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.673762321472168,
      "learning_rate": 1.0621985113247932e-05,
      "loss": 0.8158,
      "step": 15647
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.0975427627563477,
      "learning_rate": 1.0621346196850144e-05,
      "loss": 0.8575,
      "step": 15648
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.968693733215332,
      "learning_rate": 1.0620707280452354e-05,
      "loss": 1.0884,
      "step": 15649
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9394837617874146,
      "learning_rate": 1.0620068364054565e-05,
      "loss": 1.0229,
      "step": 15650
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.628408670425415,
      "learning_rate": 1.0619429447656775e-05,
      "loss": 0.8227,
      "step": 15651
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0674779415130615,
      "learning_rate": 1.0618790531258987e-05,
      "loss": 0.9195,
      "step": 15652
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2090256214141846,
      "learning_rate": 1.0618151614861196e-05,
      "loss": 0.7313,
      "step": 15653
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.336671352386475,
      "learning_rate": 1.0617512698463408e-05,
      "loss": 1.5363,
      "step": 15654
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7476545572280884,
      "learning_rate": 1.0616873782065618e-05,
      "loss": 0.889,
      "step": 15655
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.639238119125366,
      "learning_rate": 1.0616234865667828e-05,
      "loss": 0.8066,
      "step": 15656
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.59722101688385,
      "learning_rate": 1.061559594927004e-05,
      "loss": 0.9837,
      "step": 15657
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7537826299667358,
      "learning_rate": 1.0614957032872249e-05,
      "loss": 0.9469,
      "step": 15658
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7527196407318115,
      "learning_rate": 1.061431811647446e-05,
      "loss": 0.9558,
      "step": 15659
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0521652698516846,
      "learning_rate": 1.061367920007667e-05,
      "loss": 0.9647,
      "step": 15660
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8848495483398438,
      "learning_rate": 1.0613040283678882e-05,
      "loss": 0.8886,
      "step": 15661
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7550592422485352,
      "learning_rate": 1.061240136728109e-05,
      "loss": 0.9406,
      "step": 15662
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.745155096054077,
      "learning_rate": 1.0611762450883303e-05,
      "loss": 0.8126,
      "step": 15663
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9223086833953857,
      "learning_rate": 1.0611123534485512e-05,
      "loss": 0.9192,
      "step": 15664
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.499269485473633,
      "learning_rate": 1.0610484618087725e-05,
      "loss": 1.3572,
      "step": 15665
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6590241193771362,
      "learning_rate": 1.0609845701689933e-05,
      "loss": 0.8931,
      "step": 15666
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.3107011318206787,
      "learning_rate": 1.0609206785292146e-05,
      "loss": 0.7927,
      "step": 15667
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9077218770980835,
      "learning_rate": 1.0608567868894354e-05,
      "loss": 0.8994,
      "step": 15668
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.9800310134887695,
      "learning_rate": 1.0607928952496566e-05,
      "loss": 1.0746,
      "step": 15669
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.779935598373413,
      "learning_rate": 1.0607290036098776e-05,
      "loss": 0.9273,
      "step": 15670
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6392402648925781,
      "learning_rate": 1.0606651119700987e-05,
      "loss": 0.7868,
      "step": 15671
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7771716117858887,
      "learning_rate": 1.0606012203303197e-05,
      "loss": 0.9544,
      "step": 15672
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.837809443473816,
      "learning_rate": 1.0605373286905409e-05,
      "loss": 0.8919,
      "step": 15673
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.733880877494812,
      "learning_rate": 1.0604734370507619e-05,
      "loss": 0.9634,
      "step": 15674
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.812289714813232,
      "learning_rate": 1.060409545410983e-05,
      "loss": 1.0133,
      "step": 15675
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.787891149520874,
      "learning_rate": 1.060345653771204e-05,
      "loss": 1.255,
      "step": 15676
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.53777015209198,
      "learning_rate": 1.0602817621314252e-05,
      "loss": 0.806,
      "step": 15677
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.6692538261413574,
      "learning_rate": 1.0602178704916461e-05,
      "loss": 1.0104,
      "step": 15678
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.525943398475647,
      "learning_rate": 1.0601539788518673e-05,
      "loss": 0.8078,
      "step": 15679
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.587552785873413,
      "learning_rate": 1.0600900872120883e-05,
      "loss": 0.8725,
      "step": 15680
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.28412389755249,
      "learning_rate": 1.0600261955723094e-05,
      "loss": 1.4028,
      "step": 15681
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7469239234924316,
      "learning_rate": 1.0599623039325304e-05,
      "loss": 0.849,
      "step": 15682
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0339279174804688,
      "learning_rate": 1.0598984122927516e-05,
      "loss": 0.5618,
      "step": 15683
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.601198673248291,
      "learning_rate": 1.0598345206529726e-05,
      "loss": 0.9081,
      "step": 15684
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6705636978149414,
      "learning_rate": 1.0597706290131937e-05,
      "loss": 0.4081,
      "step": 15685
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.068995952606201,
      "learning_rate": 1.0597067373734147e-05,
      "loss": 0.9629,
      "step": 15686
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6476492881774902,
      "learning_rate": 1.0596428457336359e-05,
      "loss": 0.8531,
      "step": 15687
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.9630231857299805,
      "learning_rate": 1.0595789540938568e-05,
      "loss": 1.1271,
      "step": 15688
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5200310945510864,
      "learning_rate": 1.059515062454078e-05,
      "loss": 0.8823,
      "step": 15689
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8506956100463867,
      "learning_rate": 1.059451170814299e-05,
      "loss": 0.8701,
      "step": 15690
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5845866203308105,
      "learning_rate": 1.0593872791745201e-05,
      "loss": 0.8569,
      "step": 15691
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.52993631362915,
      "learning_rate": 1.0593233875347411e-05,
      "loss": 1.4912,
      "step": 15692
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6559642553329468,
      "learning_rate": 1.0592594958949623e-05,
      "loss": 0.8306,
      "step": 15693
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2872977256774902,
      "learning_rate": 1.0591956042551833e-05,
      "loss": 0.8586,
      "step": 15694
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.652493476867676,
      "learning_rate": 1.0591317126154044e-05,
      "loss": 1.121,
      "step": 15695
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8694144487380981,
      "learning_rate": 1.0590678209756254e-05,
      "loss": 0.873,
      "step": 15696
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.130417585372925,
      "learning_rate": 1.0590039293358466e-05,
      "loss": 1.213,
      "step": 15697
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5522334575653076,
      "learning_rate": 1.0589400376960675e-05,
      "loss": 0.8992,
      "step": 15698
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.074970245361328,
      "learning_rate": 1.0588761460562887e-05,
      "loss": 1.7402,
      "step": 15699
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7204686403274536,
      "learning_rate": 1.0588122544165095e-05,
      "loss": 0.5166,
      "step": 15700
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.600903034210205,
      "learning_rate": 1.0587483627767308e-05,
      "loss": 0.8869,
      "step": 15701
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.89633047580719,
      "learning_rate": 1.0586844711369517e-05,
      "loss": 0.8855,
      "step": 15702
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.117774248123169,
      "learning_rate": 1.058620579497173e-05,
      "loss": 0.931,
      "step": 15703
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3264154195785522,
      "learning_rate": 1.0585566878573938e-05,
      "loss": 0.7191,
      "step": 15704
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6313897371292114,
      "learning_rate": 1.058492796217615e-05,
      "loss": 0.5031,
      "step": 15705
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3759448528289795,
      "learning_rate": 1.058428904577836e-05,
      "loss": 0.7415,
      "step": 15706
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9677940607070923,
      "learning_rate": 1.0583650129380571e-05,
      "loss": 0.9318,
      "step": 15707
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.091311454772949,
      "learning_rate": 1.058301121298278e-05,
      "loss": 0.9953,
      "step": 15708
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3589749336242676,
      "learning_rate": 1.0582372296584992e-05,
      "loss": 0.8507,
      "step": 15709
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.5387701988220215,
      "learning_rate": 1.0581733380187202e-05,
      "loss": 0.9752,
      "step": 15710
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.414273738861084,
      "learning_rate": 1.0581094463789414e-05,
      "loss": 0.8611,
      "step": 15711
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6821959018707275,
      "learning_rate": 1.0580455547391624e-05,
      "loss": 0.8009,
      "step": 15712
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7951124906539917,
      "learning_rate": 1.0579816630993835e-05,
      "loss": 0.9854,
      "step": 15713
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8596642017364502,
      "learning_rate": 1.0579177714596045e-05,
      "loss": 0.8591,
      "step": 15714
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.640484094619751,
      "learning_rate": 1.0578538798198256e-05,
      "loss": 0.8611,
      "step": 15715
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7115156650543213,
      "learning_rate": 1.0577899881800466e-05,
      "loss": 1.6281,
      "step": 15716
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4388364553451538,
      "learning_rate": 1.0577260965402678e-05,
      "loss": 0.8171,
      "step": 15717
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3469337224960327,
      "learning_rate": 1.0576622049004888e-05,
      "loss": 0.8421,
      "step": 15718
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.624595284461975,
      "learning_rate": 1.05759831326071e-05,
      "loss": 0.8112,
      "step": 15719
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4217920303344727,
      "learning_rate": 1.0575344216209309e-05,
      "loss": 0.8903,
      "step": 15720
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.6314902305603027,
      "learning_rate": 1.057470529981152e-05,
      "loss": 1.134,
      "step": 15721
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2037250995635986,
      "learning_rate": 1.057406638341373e-05,
      "loss": 0.8947,
      "step": 15722
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0644822120666504,
      "learning_rate": 1.0573427467015942e-05,
      "loss": 1.021,
      "step": 15723
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7391653060913086,
      "learning_rate": 1.0572788550618152e-05,
      "loss": 0.6231,
      "step": 15724
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8530491590499878,
      "learning_rate": 1.0572149634220363e-05,
      "loss": 0.8989,
      "step": 15725
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.606156826019287,
      "learning_rate": 1.0571510717822573e-05,
      "loss": 1.3294,
      "step": 15726
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7092704772949219,
      "learning_rate": 1.0570871801424785e-05,
      "loss": 0.887,
      "step": 15727
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4392497539520264,
      "learning_rate": 1.0570232885026995e-05,
      "loss": 0.8288,
      "step": 15728
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8682407140731812,
      "learning_rate": 1.0569593968629206e-05,
      "loss": 0.9906,
      "step": 15729
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3535354137420654,
      "learning_rate": 1.0568955052231416e-05,
      "loss": 0.8802,
      "step": 15730
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6346794366836548,
      "learning_rate": 1.0568316135833628e-05,
      "loss": 0.9587,
      "step": 15731
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7149288654327393,
      "learning_rate": 1.0567677219435838e-05,
      "loss": 0.914,
      "step": 15732
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.935129165649414,
      "learning_rate": 1.0567038303038049e-05,
      "loss": 1.0017,
      "step": 15733
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7994495630264282,
      "learning_rate": 1.0566399386640259e-05,
      "loss": 0.8934,
      "step": 15734
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5902544260025024,
      "learning_rate": 1.056576047024247e-05,
      "loss": 0.9377,
      "step": 15735
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.667419910430908,
      "learning_rate": 1.0565121553844679e-05,
      "loss": 0.8229,
      "step": 15736
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.449582099914551,
      "learning_rate": 1.0564482637446892e-05,
      "loss": 1.0435,
      "step": 15737
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8151808977127075,
      "learning_rate": 1.05638437210491e-05,
      "loss": 1.1917,
      "step": 15738
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7251989841461182,
      "learning_rate": 1.0563204804651313e-05,
      "loss": 0.8664,
      "step": 15739
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.13303279876709,
      "learning_rate": 1.0562565888253521e-05,
      "loss": 1.0838,
      "step": 15740
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.490204930305481,
      "learning_rate": 1.0561926971855733e-05,
      "loss": 0.8622,
      "step": 15741
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.658206820487976,
      "learning_rate": 1.0561288055457943e-05,
      "loss": 0.9039,
      "step": 15742
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.2022149562835693,
      "learning_rate": 1.0560649139060154e-05,
      "loss": 0.5562,
      "step": 15743
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.049121379852295,
      "learning_rate": 1.0560010222662364e-05,
      "loss": 0.9239,
      "step": 15744
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.65526819229126,
      "learning_rate": 1.0559371306264576e-05,
      "loss": 1.2653,
      "step": 15745
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8305141925811768,
      "learning_rate": 1.0558732389866786e-05,
      "loss": 1.0199,
      "step": 15746
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.8530192375183105,
      "learning_rate": 1.0558093473468997e-05,
      "loss": 1.1971,
      "step": 15747
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9083852767944336,
      "learning_rate": 1.0557454557071207e-05,
      "loss": 0.9334,
      "step": 15748
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8969544172286987,
      "learning_rate": 1.0556815640673419e-05,
      "loss": 0.8872,
      "step": 15749
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7179877758026123,
      "learning_rate": 1.0556176724275628e-05,
      "loss": 0.9204,
      "step": 15750
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7587498426437378,
      "learning_rate": 1.055553780787784e-05,
      "loss": 0.8773,
      "step": 15751
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1035075187683105,
      "learning_rate": 1.055489889148005e-05,
      "loss": 0.751,
      "step": 15752
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7459992170333862,
      "learning_rate": 1.0554259975082261e-05,
      "loss": 0.9083,
      "step": 15753
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0327632427215576,
      "learning_rate": 1.0553621058684471e-05,
      "loss": 0.9238,
      "step": 15754
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5525023937225342,
      "learning_rate": 1.0552982142286683e-05,
      "loss": 0.8645,
      "step": 15755
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9947527647018433,
      "learning_rate": 1.0552343225888893e-05,
      "loss": 0.9423,
      "step": 15756
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.721438407897949,
      "learning_rate": 1.0551704309491104e-05,
      "loss": 1.3253,
      "step": 15757
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.9646928310394287,
      "learning_rate": 1.0551065393093314e-05,
      "loss": 1.0713,
      "step": 15758
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.818374514579773,
      "learning_rate": 1.0550426476695526e-05,
      "loss": 0.937,
      "step": 15759
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7646156549453735,
      "learning_rate": 1.0549787560297735e-05,
      "loss": 0.8876,
      "step": 15760
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.906148076057434,
      "learning_rate": 1.0549148643899947e-05,
      "loss": 0.9738,
      "step": 15761
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8587334156036377,
      "learning_rate": 1.0548509727502157e-05,
      "loss": 0.9376,
      "step": 15762
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7000929117202759,
      "learning_rate": 1.0547870811104368e-05,
      "loss": 0.9245,
      "step": 15763
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2849957942962646,
      "learning_rate": 1.0547231894706578e-05,
      "loss": 0.8592,
      "step": 15764
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7173333168029785,
      "learning_rate": 1.054659297830879e-05,
      "loss": 0.7845,
      "step": 15765
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.18031120300293,
      "learning_rate": 1.0545954061911e-05,
      "loss": 1.5435,
      "step": 15766
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.921043872833252,
      "learning_rate": 1.0545315145513211e-05,
      "loss": 0.9105,
      "step": 15767
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.8110337257385254,
      "learning_rate": 1.0544676229115421e-05,
      "loss": 0.8938,
      "step": 15768
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.102001190185547,
      "learning_rate": 1.0544037312717633e-05,
      "loss": 0.9895,
      "step": 15769
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6509828567504883,
      "learning_rate": 1.0543398396319842e-05,
      "loss": 1.0081,
      "step": 15770
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.85945463180542,
      "learning_rate": 1.0542759479922054e-05,
      "loss": 1.1847,
      "step": 15771
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7762036323547363,
      "learning_rate": 1.0542120563524262e-05,
      "loss": 0.9921,
      "step": 15772
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7371970415115356,
      "learning_rate": 1.0541481647126475e-05,
      "loss": 1.0011,
      "step": 15773
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3938817977905273,
      "learning_rate": 1.0540842730728684e-05,
      "loss": 1.0196,
      "step": 15774
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5942857265472412,
      "learning_rate": 1.0540203814330897e-05,
      "loss": 0.866,
      "step": 15775
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0708510875701904,
      "learning_rate": 1.0539564897933105e-05,
      "loss": 0.9235,
      "step": 15776
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.941880226135254,
      "learning_rate": 1.0538925981535316e-05,
      "loss": 1.2202,
      "step": 15777
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2662525177001953,
      "learning_rate": 1.0538287065137526e-05,
      "loss": 0.8822,
      "step": 15778
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2932080030441284,
      "learning_rate": 1.0537648148739738e-05,
      "loss": 0.506,
      "step": 15779
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.846485137939453,
      "learning_rate": 1.0537009232341948e-05,
      "loss": 1.1425,
      "step": 15780
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6036534309387207,
      "learning_rate": 1.053637031594416e-05,
      "loss": 0.8719,
      "step": 15781
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3434078693389893,
      "learning_rate": 1.0535731399546369e-05,
      "loss": 0.8809,
      "step": 15782
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.065012454986572,
      "learning_rate": 1.053509248314858e-05,
      "loss": 1.3594,
      "step": 15783
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.9279613494873047,
      "learning_rate": 1.053445356675079e-05,
      "loss": 1.1058,
      "step": 15784
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.75624680519104,
      "learning_rate": 1.0533814650353002e-05,
      "loss": 0.8781,
      "step": 15785
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.710477352142334,
      "learning_rate": 1.0533175733955212e-05,
      "loss": 0.8595,
      "step": 15786
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.244162082672119,
      "learning_rate": 1.0532536817557423e-05,
      "loss": 0.8656,
      "step": 15787
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8953771591186523,
      "learning_rate": 1.0531897901159633e-05,
      "loss": 0.935,
      "step": 15788
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.756988048553467,
      "learning_rate": 1.0531258984761845e-05,
      "loss": 0.7159,
      "step": 15789
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0485379695892334,
      "learning_rate": 1.0530620068364055e-05,
      "loss": 0.8261,
      "step": 15790
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2343037128448486,
      "learning_rate": 1.0529981151966266e-05,
      "loss": 1.0591,
      "step": 15791
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.338740348815918,
      "learning_rate": 1.0529342235568476e-05,
      "loss": 0.8757,
      "step": 15792
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7427302598953247,
      "learning_rate": 1.0528703319170688e-05,
      "loss": 0.8941,
      "step": 15793
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6252679824829102,
      "learning_rate": 1.0528064402772898e-05,
      "loss": 0.4099,
      "step": 15794
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.387984037399292,
      "learning_rate": 1.0527425486375109e-05,
      "loss": 1.3214,
      "step": 15795
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.319849967956543,
      "learning_rate": 1.0526786569977319e-05,
      "loss": 0.7642,
      "step": 15796
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8226896524429321,
      "learning_rate": 1.052614765357953e-05,
      "loss": 0.9575,
      "step": 15797
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.1918678283691406,
      "learning_rate": 1.052550873718174e-05,
      "loss": 0.7374,
      "step": 15798
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6559585332870483,
      "learning_rate": 1.0524869820783952e-05,
      "loss": 0.9556,
      "step": 15799
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9919542074203491,
      "learning_rate": 1.0524230904386162e-05,
      "loss": 1.2135,
      "step": 15800
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3942478895187378,
      "learning_rate": 1.0523591987988373e-05,
      "loss": 0.8742,
      "step": 15801
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.2891390323638916,
      "learning_rate": 1.0522953071590583e-05,
      "loss": 1.209,
      "step": 15802
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.297217845916748,
      "learning_rate": 1.0522314155192795e-05,
      "loss": 0.7144,
      "step": 15803
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7082525491714478,
      "learning_rate": 1.0521675238795005e-05,
      "loss": 1.0286,
      "step": 15804
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.663902997970581,
      "learning_rate": 1.0521036322397216e-05,
      "loss": 0.8327,
      "step": 15805
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1884231567382812,
      "learning_rate": 1.0520397405999426e-05,
      "loss": 0.9514,
      "step": 15806
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.788827896118164,
      "learning_rate": 1.0519758489601637e-05,
      "loss": 1.6077,
      "step": 15807
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5195348262786865,
      "learning_rate": 1.0519119573203846e-05,
      "loss": 0.9132,
      "step": 15808
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.6972432136535645,
      "learning_rate": 1.0518480656806059e-05,
      "loss": 1.4072,
      "step": 15809
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.166110515594482,
      "learning_rate": 1.0517841740408267e-05,
      "loss": 1.2745,
      "step": 15810
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8936015367507935,
      "learning_rate": 1.051720282401048e-05,
      "loss": 0.8051,
      "step": 15811
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.6511690616607666,
      "learning_rate": 1.0516563907612688e-05,
      "loss": 1.4368,
      "step": 15812
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0651607513427734,
      "learning_rate": 1.0515924991214902e-05,
      "loss": 0.8666,
      "step": 15813
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.4688947200775146,
      "learning_rate": 1.051528607481711e-05,
      "loss": 0.8818,
      "step": 15814
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0020461082458496,
      "learning_rate": 1.0514647158419321e-05,
      "loss": 0.915,
      "step": 15815
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6087496280670166,
      "learning_rate": 1.0514008242021531e-05,
      "loss": 0.4222,
      "step": 15816
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.483318567276001,
      "learning_rate": 1.0513369325623743e-05,
      "loss": 0.4296,
      "step": 15817
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1432721614837646,
      "learning_rate": 1.0512730409225953e-05,
      "loss": 0.9572,
      "step": 15818
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.382080316543579,
      "learning_rate": 1.0512091492828164e-05,
      "loss": 1.0882,
      "step": 15819
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.418147325515747,
      "learning_rate": 1.0511452576430374e-05,
      "loss": 0.6037,
      "step": 15820
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.09364652633667,
      "learning_rate": 1.0510813660032586e-05,
      "loss": 0.956,
      "step": 15821
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.634666919708252,
      "learning_rate": 1.0510174743634795e-05,
      "loss": 1.0864,
      "step": 15822
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.064807176589966,
      "learning_rate": 1.0509535827237007e-05,
      "loss": 1.1298,
      "step": 15823
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.548513650894165,
      "learning_rate": 1.0508896910839217e-05,
      "loss": 1.0916,
      "step": 15824
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.117934465408325,
      "learning_rate": 1.0508257994441428e-05,
      "loss": 0.8941,
      "step": 15825
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.4650471210479736,
      "learning_rate": 1.0507619078043638e-05,
      "loss": 1.1576,
      "step": 15826
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7969517707824707,
      "learning_rate": 1.050698016164585e-05,
      "loss": 0.9203,
      "step": 15827
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7189435958862305,
      "learning_rate": 1.050634124524806e-05,
      "loss": 1.0248,
      "step": 15828
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2946009635925293,
      "learning_rate": 1.0505702328850271e-05,
      "loss": 0.812,
      "step": 15829
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7762562036514282,
      "learning_rate": 1.0505063412452481e-05,
      "loss": 0.797,
      "step": 15830
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9542406797409058,
      "learning_rate": 1.0504424496054693e-05,
      "loss": 0.9931,
      "step": 15831
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.5256054401397705,
      "learning_rate": 1.0503785579656902e-05,
      "loss": 0.75,
      "step": 15832
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2025279998779297,
      "learning_rate": 1.0503146663259114e-05,
      "loss": 0.5969,
      "step": 15833
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.848322868347168,
      "learning_rate": 1.0502507746861324e-05,
      "loss": 0.9818,
      "step": 15834
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.051947832107544,
      "learning_rate": 1.0501868830463535e-05,
      "loss": 0.8976,
      "step": 15835
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.650015950202942,
      "learning_rate": 1.0501229914065745e-05,
      "loss": 0.8428,
      "step": 15836
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4442896842956543,
      "learning_rate": 1.0500590997667957e-05,
      "loss": 0.7825,
      "step": 15837
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5861306190490723,
      "learning_rate": 1.0499952081270167e-05,
      "loss": 0.8551,
      "step": 15838
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.007260322570801,
      "learning_rate": 1.0499313164872378e-05,
      "loss": 0.8492,
      "step": 15839
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.683675765991211,
      "learning_rate": 1.0498674248474588e-05,
      "loss": 0.7853,
      "step": 15840
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7670785188674927,
      "learning_rate": 1.04980353320768e-05,
      "loss": 1.1416,
      "step": 15841
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.2505245208740234,
      "learning_rate": 1.049739641567901e-05,
      "loss": 1.0457,
      "step": 15842
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6829406023025513,
      "learning_rate": 1.0496757499281221e-05,
      "loss": 0.8933,
      "step": 15843
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7484489679336548,
      "learning_rate": 1.049611858288343e-05,
      "loss": 0.8588,
      "step": 15844
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2773470878601074,
      "learning_rate": 1.0495479666485642e-05,
      "loss": 0.939,
      "step": 15845
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.4310555458068848,
      "learning_rate": 1.049484075008785e-05,
      "loss": 0.8733,
      "step": 15846
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6422266960144043,
      "learning_rate": 1.0494201833690064e-05,
      "loss": 0.973,
      "step": 15847
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5752607583999634,
      "learning_rate": 1.0493562917292272e-05,
      "loss": 0.8254,
      "step": 15848
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8665317296981812,
      "learning_rate": 1.0492924000894485e-05,
      "loss": 0.9129,
      "step": 15849
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.182713508605957,
      "learning_rate": 1.0492285084496693e-05,
      "loss": 1.036,
      "step": 15850
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7430129051208496,
      "learning_rate": 1.0491646168098905e-05,
      "loss": 0.8738,
      "step": 15851
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.04807448387146,
      "learning_rate": 1.0491007251701115e-05,
      "loss": 0.8863,
      "step": 15852
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2714059352874756,
      "learning_rate": 1.0490368335303326e-05,
      "loss": 0.8109,
      "step": 15853
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.4119977951049805,
      "learning_rate": 1.0489729418905536e-05,
      "loss": 1.1331,
      "step": 15854
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.634931206703186,
      "learning_rate": 1.0489090502507748e-05,
      "loss": 0.9155,
      "step": 15855
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.030768871307373,
      "learning_rate": 1.0488451586109958e-05,
      "loss": 1.213,
      "step": 15856
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1705663204193115,
      "learning_rate": 1.0487812669712169e-05,
      "loss": 0.7561,
      "step": 15857
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.347426414489746,
      "learning_rate": 1.0487173753314379e-05,
      "loss": 1.1822,
      "step": 15858
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.061022996902466,
      "learning_rate": 1.048653483691659e-05,
      "loss": 1.0915,
      "step": 15859
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0434370040893555,
      "learning_rate": 1.04858959205188e-05,
      "loss": 1.0485,
      "step": 15860
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6435296535491943,
      "learning_rate": 1.0485257004121012e-05,
      "loss": 0.8145,
      "step": 15861
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9918807744979858,
      "learning_rate": 1.0484618087723222e-05,
      "loss": 0.8877,
      "step": 15862
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6424429416656494,
      "learning_rate": 1.0483979171325433e-05,
      "loss": 0.9194,
      "step": 15863
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.4007887840271,
      "learning_rate": 1.0483340254927643e-05,
      "loss": 0.9469,
      "step": 15864
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6695889234542847,
      "learning_rate": 1.0482701338529855e-05,
      "loss": 0.8555,
      "step": 15865
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0607118606567383,
      "learning_rate": 1.0482062422132065e-05,
      "loss": 0.9751,
      "step": 15866
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0084621906280518,
      "learning_rate": 1.0481423505734276e-05,
      "loss": 0.9652,
      "step": 15867
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8173084259033203,
      "learning_rate": 1.0480784589336486e-05,
      "loss": 0.9519,
      "step": 15868
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8249564170837402,
      "learning_rate": 1.0480145672938697e-05,
      "loss": 0.9988,
      "step": 15869
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.413621187210083,
      "learning_rate": 1.0479506756540907e-05,
      "loss": 1.0338,
      "step": 15870
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.115269422531128,
      "learning_rate": 1.0478867840143119e-05,
      "loss": 0.9337,
      "step": 15871
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8471044301986694,
      "learning_rate": 1.0478228923745329e-05,
      "loss": 0.9616,
      "step": 15872
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.546351432800293,
      "learning_rate": 1.047759000734754e-05,
      "loss": 0.9238,
      "step": 15873
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.304847002029419,
      "learning_rate": 1.047695109094975e-05,
      "loss": 0.6566,
      "step": 15874
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.217427730560303,
      "learning_rate": 1.0476312174551962e-05,
      "loss": 1.4612,
      "step": 15875
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.034126043319702,
      "learning_rate": 1.0475673258154172e-05,
      "loss": 0.9726,
      "step": 15876
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.924450397491455,
      "learning_rate": 1.0475034341756383e-05,
      "loss": 0.8486,
      "step": 15877
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9373894929885864,
      "learning_rate": 1.0474395425358593e-05,
      "loss": 0.9935,
      "step": 15878
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.7604193687438965,
      "learning_rate": 1.0473756508960804e-05,
      "loss": 1.3372,
      "step": 15879
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8985686302185059,
      "learning_rate": 1.0473117592563014e-05,
      "loss": 0.912,
      "step": 15880
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0816686153411865,
      "learning_rate": 1.0472478676165226e-05,
      "loss": 0.9446,
      "step": 15881
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.581464409828186,
      "learning_rate": 1.0471839759767434e-05,
      "loss": 0.4404,
      "step": 15882
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.175419569015503,
      "learning_rate": 1.0471200843369647e-05,
      "loss": 0.8699,
      "step": 15883
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.117431640625,
      "learning_rate": 1.0470561926971855e-05,
      "loss": 0.9912,
      "step": 15884
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.081998109817505,
      "learning_rate": 1.0469923010574069e-05,
      "loss": 0.9297,
      "step": 15885
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.737621545791626,
      "learning_rate": 1.0469284094176277e-05,
      "loss": 0.8536,
      "step": 15886
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.402014970779419,
      "learning_rate": 1.0468645177778488e-05,
      "loss": 1.2216,
      "step": 15887
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.10982608795166,
      "learning_rate": 1.0468006261380698e-05,
      "loss": 0.876,
      "step": 15888
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5859253406524658,
      "learning_rate": 1.046736734498291e-05,
      "loss": 0.9129,
      "step": 15889
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7757664918899536,
      "learning_rate": 1.046672842858512e-05,
      "loss": 0.8254,
      "step": 15890
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.960658311843872,
      "learning_rate": 1.0466089512187331e-05,
      "loss": 0.7435,
      "step": 15891
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3938283920288086,
      "learning_rate": 1.0465450595789541e-05,
      "loss": 0.9366,
      "step": 15892
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.5372486114501953,
      "learning_rate": 1.0464811679391753e-05,
      "loss": 0.9829,
      "step": 15893
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8425499200820923,
      "learning_rate": 1.0464172762993962e-05,
      "loss": 0.7184,
      "step": 15894
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.695311188697815,
      "learning_rate": 1.0463533846596174e-05,
      "loss": 0.8509,
      "step": 15895
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.62751042842865,
      "learning_rate": 1.0462894930198384e-05,
      "loss": 0.8822,
      "step": 15896
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3740196228027344,
      "learning_rate": 1.0462256013800595e-05,
      "loss": 0.9216,
      "step": 15897
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9511544704437256,
      "learning_rate": 1.0461617097402805e-05,
      "loss": 0.6386,
      "step": 15898
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.004957914352417,
      "learning_rate": 1.0460978181005017e-05,
      "loss": 0.843,
      "step": 15899
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9155128002166748,
      "learning_rate": 1.0460339264607227e-05,
      "loss": 0.9289,
      "step": 15900
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4896520376205444,
      "learning_rate": 1.0459700348209438e-05,
      "loss": 0.7604,
      "step": 15901
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.856331467628479,
      "learning_rate": 1.0459061431811648e-05,
      "loss": 0.8918,
      "step": 15902
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.932479977607727,
      "learning_rate": 1.045842251541386e-05,
      "loss": 0.8241,
      "step": 15903
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7804380655288696,
      "learning_rate": 1.045778359901607e-05,
      "loss": 0.8537,
      "step": 15904
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5066542625427246,
      "learning_rate": 1.0457144682618281e-05,
      "loss": 0.8371,
      "step": 15905
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2928667068481445,
      "learning_rate": 1.0456505766220491e-05,
      "loss": 0.9622,
      "step": 15906
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.301698923110962,
      "learning_rate": 1.0455866849822702e-05,
      "loss": 0.9452,
      "step": 15907
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.332697629928589,
      "learning_rate": 1.0455227933424912e-05,
      "loss": 0.812,
      "step": 15908
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.4834694862365723,
      "learning_rate": 1.0454589017027124e-05,
      "loss": 0.8256,
      "step": 15909
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.121434688568115,
      "learning_rate": 1.0453950100629334e-05,
      "loss": 1.2985,
      "step": 15910
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6453840732574463,
      "learning_rate": 1.0453311184231545e-05,
      "loss": 0.8737,
      "step": 15911
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.315694570541382,
      "learning_rate": 1.0452672267833755e-05,
      "loss": 1.2223,
      "step": 15912
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.1371541023254395,
      "learning_rate": 1.0452033351435967e-05,
      "loss": 1.5525,
      "step": 15913
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9263300895690918,
      "learning_rate": 1.0451394435038176e-05,
      "loss": 0.9717,
      "step": 15914
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.827826976776123,
      "learning_rate": 1.0450755518640388e-05,
      "loss": 1.0803,
      "step": 15915
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.662481307983398,
      "learning_rate": 1.0450116602242598e-05,
      "loss": 1.0753,
      "step": 15916
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4595340490341187,
      "learning_rate": 1.044947768584481e-05,
      "loss": 0.7523,
      "step": 15917
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.4304721355438232,
      "learning_rate": 1.0448838769447018e-05,
      "loss": 0.8566,
      "step": 15918
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.050596237182617,
      "learning_rate": 1.044819985304923e-05,
      "loss": 0.9696,
      "step": 15919
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.8796355724334717,
      "learning_rate": 1.0447560936651439e-05,
      "loss": 0.7819,
      "step": 15920
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.27470064163208,
      "learning_rate": 1.0446922020253652e-05,
      "loss": 1.0528,
      "step": 15921
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1216912269592285,
      "learning_rate": 1.044628310385586e-05,
      "loss": 0.9061,
      "step": 15922
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5733932256698608,
      "learning_rate": 1.0445644187458072e-05,
      "loss": 0.9317,
      "step": 15923
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8084354400634766,
      "learning_rate": 1.0445005271060282e-05,
      "loss": 0.9545,
      "step": 15924
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8363804817199707,
      "learning_rate": 1.0444366354662493e-05,
      "loss": 0.8781,
      "step": 15925
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.356132984161377,
      "learning_rate": 1.0443727438264703e-05,
      "loss": 0.57,
      "step": 15926
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9204562902450562,
      "learning_rate": 1.0443088521866915e-05,
      "loss": 0.8697,
      "step": 15927
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9291538000106812,
      "learning_rate": 1.0442449605469125e-05,
      "loss": 1.2061,
      "step": 15928
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.289245843887329,
      "learning_rate": 1.0441810689071336e-05,
      "loss": 1.0935,
      "step": 15929
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.5157954692840576,
      "learning_rate": 1.0441171772673546e-05,
      "loss": 1.1397,
      "step": 15930
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7011210918426514,
      "learning_rate": 1.0440532856275758e-05,
      "loss": 0.8857,
      "step": 15931
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2997223138809204,
      "learning_rate": 1.0439893939877967e-05,
      "loss": 0.4259,
      "step": 15932
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6696721315383911,
      "learning_rate": 1.0439255023480179e-05,
      "loss": 0.8732,
      "step": 15933
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.597161889076233,
      "learning_rate": 1.0438616107082389e-05,
      "loss": 0.8839,
      "step": 15934
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.279277801513672,
      "learning_rate": 1.04379771906846e-05,
      "loss": 0.8869,
      "step": 15935
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7168524265289307,
      "learning_rate": 1.043733827428681e-05,
      "loss": 1.0572,
      "step": 15936
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9203213453292847,
      "learning_rate": 1.0436699357889022e-05,
      "loss": 0.4118,
      "step": 15937
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5901771783828735,
      "learning_rate": 1.0436060441491232e-05,
      "loss": 0.9478,
      "step": 15938
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.2941479682922363,
      "learning_rate": 1.0435421525093443e-05,
      "loss": 0.8705,
      "step": 15939
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6594412326812744,
      "learning_rate": 1.0434782608695653e-05,
      "loss": 0.7625,
      "step": 15940
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5459074974060059,
      "learning_rate": 1.0434143692297865e-05,
      "loss": 0.824,
      "step": 15941
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.203639507293701,
      "learning_rate": 1.0433504775900074e-05,
      "loss": 0.9319,
      "step": 15942
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7700163125991821,
      "learning_rate": 1.0432865859502286e-05,
      "loss": 0.9213,
      "step": 15943
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7209452390670776,
      "learning_rate": 1.0432226943104496e-05,
      "loss": 0.9279,
      "step": 15944
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.149653196334839,
      "learning_rate": 1.0431588026706707e-05,
      "loss": 0.9446,
      "step": 15945
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4134577512741089,
      "learning_rate": 1.0430949110308917e-05,
      "loss": 0.7301,
      "step": 15946
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7598689794540405,
      "learning_rate": 1.0430310193911129e-05,
      "loss": 0.9056,
      "step": 15947
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.8609371185302734,
      "learning_rate": 1.0429671277513339e-05,
      "loss": 0.7833,
      "step": 15948
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.135341167449951,
      "learning_rate": 1.042903236111555e-05,
      "loss": 0.8146,
      "step": 15949
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.101022481918335,
      "learning_rate": 1.042839344471776e-05,
      "loss": 1.0428,
      "step": 15950
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.2606937885284424,
      "learning_rate": 1.0427754528319972e-05,
      "loss": 0.8453,
      "step": 15951
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6277590990066528,
      "learning_rate": 1.0427115611922181e-05,
      "loss": 0.9199,
      "step": 15952
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.5775914192199707,
      "learning_rate": 1.0426476695524393e-05,
      "loss": 0.9475,
      "step": 15953
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.5321526527404785,
      "learning_rate": 1.0425837779126601e-05,
      "loss": 1.1334,
      "step": 15954
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9212106466293335,
      "learning_rate": 1.0425198862728814e-05,
      "loss": 0.8266,
      "step": 15955
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.4338033199310303,
      "learning_rate": 1.0424559946331022e-05,
      "loss": 0.8105,
      "step": 15956
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8004248142242432,
      "learning_rate": 1.0423921029933236e-05,
      "loss": 0.849,
      "step": 15957
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.297708511352539,
      "learning_rate": 1.0423282113535444e-05,
      "loss": 0.9046,
      "step": 15958
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.060852527618408,
      "learning_rate": 1.0422643197137657e-05,
      "loss": 1.1898,
      "step": 15959
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.236130714416504,
      "learning_rate": 1.0422004280739865e-05,
      "loss": 0.8706,
      "step": 15960
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.468384027481079,
      "learning_rate": 1.0421365364342077e-05,
      "loss": 0.7272,
      "step": 15961
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.4300124645233154,
      "learning_rate": 1.0420726447944287e-05,
      "loss": 0.6458,
      "step": 15962
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0712928771972656,
      "learning_rate": 1.0420087531546498e-05,
      "loss": 0.8666,
      "step": 15963
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0244100093841553,
      "learning_rate": 1.0419448615148708e-05,
      "loss": 0.9416,
      "step": 15964
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1378438472747803,
      "learning_rate": 1.041880969875092e-05,
      "loss": 1.0523,
      "step": 15965
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4751533269882202,
      "learning_rate": 1.041817078235313e-05,
      "loss": 0.7777,
      "step": 15966
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9052801132202148,
      "learning_rate": 1.0417531865955341e-05,
      "loss": 0.8379,
      "step": 15967
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4676817655563354,
      "learning_rate": 1.0416892949557551e-05,
      "loss": 0.4269,
      "step": 15968
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.084911823272705,
      "learning_rate": 1.0416254033159762e-05,
      "loss": 0.9759,
      "step": 15969
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.3726067543029785,
      "learning_rate": 1.0415615116761972e-05,
      "loss": 0.8816,
      "step": 15970
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.4078707695007324,
      "learning_rate": 1.0414976200364184e-05,
      "loss": 0.8119,
      "step": 15971
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4899756908416748,
      "learning_rate": 1.0414337283966394e-05,
      "loss": 0.8162,
      "step": 15972
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.106440305709839,
      "learning_rate": 1.0413698367568605e-05,
      "loss": 0.8558,
      "step": 15973
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.468971848487854,
      "learning_rate": 1.0413059451170815e-05,
      "loss": 0.7984,
      "step": 15974
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7137333154678345,
      "learning_rate": 1.0412420534773027e-05,
      "loss": 0.9644,
      "step": 15975
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.345327138900757,
      "learning_rate": 1.0411781618375236e-05,
      "loss": 1.1111,
      "step": 15976
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2888636589050293,
      "learning_rate": 1.0411142701977448e-05,
      "loss": 0.9954,
      "step": 15977
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2311441898345947,
      "learning_rate": 1.0410503785579658e-05,
      "loss": 1.0561,
      "step": 15978
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8573338985443115,
      "learning_rate": 1.040986486918187e-05,
      "loss": 0.9965,
      "step": 15979
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7839775085449219,
      "learning_rate": 1.040922595278408e-05,
      "loss": 0.8865,
      "step": 15980
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.234326124191284,
      "learning_rate": 1.040858703638629e-05,
      "loss": 0.6713,
      "step": 15981
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9143449068069458,
      "learning_rate": 1.04079481199885e-05,
      "loss": 0.9286,
      "step": 15982
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8306080102920532,
      "learning_rate": 1.0407309203590712e-05,
      "loss": 1.0057,
      "step": 15983
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.856332778930664,
      "learning_rate": 1.0406670287192922e-05,
      "loss": 0.4128,
      "step": 15984
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2798314094543457,
      "learning_rate": 1.0406031370795134e-05,
      "loss": 0.8903,
      "step": 15985
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5554801225662231,
      "learning_rate": 1.0405392454397343e-05,
      "loss": 0.8006,
      "step": 15986
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6961439847946167,
      "learning_rate": 1.0404753537999555e-05,
      "loss": 0.3942,
      "step": 15987
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.644327163696289,
      "learning_rate": 1.0404114621601765e-05,
      "loss": 0.4899,
      "step": 15988
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5820270776748657,
      "learning_rate": 1.0403475705203976e-05,
      "loss": 0.8626,
      "step": 15989
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.715928077697754,
      "learning_rate": 1.0402836788806186e-05,
      "loss": 0.7489,
      "step": 15990
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.912402868270874,
      "learning_rate": 1.0402197872408398e-05,
      "loss": 1.0812,
      "step": 15991
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.614242076873779,
      "learning_rate": 1.0401558956010606e-05,
      "loss": 1.1925,
      "step": 15992
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.054967164993286,
      "learning_rate": 1.040092003961282e-05,
      "loss": 0.9245,
      "step": 15993
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.078894853591919,
      "learning_rate": 1.0400281123215027e-05,
      "loss": 0.9568,
      "step": 15994
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5718841552734375,
      "learning_rate": 1.039964220681724e-05,
      "loss": 0.8878,
      "step": 15995
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5640606880187988,
      "learning_rate": 1.0399003290419449e-05,
      "loss": 0.4054,
      "step": 15996
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.459918975830078,
      "learning_rate": 1.039836437402166e-05,
      "loss": 1.0066,
      "step": 15997
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.9278793334960938,
      "learning_rate": 1.039772545762387e-05,
      "loss": 1.0821,
      "step": 15998
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4427810907363892,
      "learning_rate": 1.0397086541226082e-05,
      "loss": 0.9225,
      "step": 15999
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9772394895553589,
      "learning_rate": 1.0396447624828292e-05,
      "loss": 0.7754,
      "step": 16000
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0648934841156006,
      "learning_rate": 1.0395808708430503e-05,
      "loss": 0.843,
      "step": 16001
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2180910110473633,
      "learning_rate": 1.0395169792032713e-05,
      "loss": 0.6797,
      "step": 16002
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7962102890014648,
      "learning_rate": 1.0394530875634925e-05,
      "loss": 0.8999,
      "step": 16003
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.830141067504883,
      "learning_rate": 1.0393891959237134e-05,
      "loss": 1.0685,
      "step": 16004
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6599379777908325,
      "learning_rate": 1.0393253042839346e-05,
      "loss": 0.7025,
      "step": 16005
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0994839668273926,
      "learning_rate": 1.0392614126441556e-05,
      "loss": 1.1711,
      "step": 16006
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5969328880310059,
      "learning_rate": 1.0391975210043767e-05,
      "loss": 0.9111,
      "step": 16007
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.704310417175293,
      "learning_rate": 1.0391336293645977e-05,
      "loss": 0.7051,
      "step": 16008
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.765647053718567,
      "learning_rate": 1.0390697377248189e-05,
      "loss": 0.8397,
      "step": 16009
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3706086874008179,
      "learning_rate": 1.0390058460850399e-05,
      "loss": 0.8166,
      "step": 16010
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.768192768096924,
      "learning_rate": 1.038941954445261e-05,
      "loss": 0.6646,
      "step": 16011
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0290544033050537,
      "learning_rate": 1.038878062805482e-05,
      "loss": 0.5119,
      "step": 16012
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5544757843017578,
      "learning_rate": 1.0388141711657032e-05,
      "loss": 0.7871,
      "step": 16013
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5533612966537476,
      "learning_rate": 1.0387502795259241e-05,
      "loss": 0.7608,
      "step": 16014
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9257640838623047,
      "learning_rate": 1.0386863878861453e-05,
      "loss": 0.9131,
      "step": 16015
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5826845169067383,
      "learning_rate": 1.0386224962463663e-05,
      "loss": 1.1011,
      "step": 16016
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8702486753463745,
      "learning_rate": 1.0385586046065874e-05,
      "loss": 0.8734,
      "step": 16017
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8172669410705566,
      "learning_rate": 1.0384947129668084e-05,
      "loss": 0.9046,
      "step": 16018
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.874448299407959,
      "learning_rate": 1.0384308213270296e-05,
      "loss": 0.8655,
      "step": 16019
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4912630319595337,
      "learning_rate": 1.0383669296872506e-05,
      "loss": 0.8263,
      "step": 16020
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.775946021080017,
      "learning_rate": 1.0383030380474717e-05,
      "loss": 0.9453,
      "step": 16021
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.1077520847320557,
      "learning_rate": 1.0382391464076927e-05,
      "loss": 1.0796,
      "step": 16022
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.561634063720703,
      "learning_rate": 1.0381752547679139e-05,
      "loss": 0.7292,
      "step": 16023
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9683783054351807,
      "learning_rate": 1.0381113631281348e-05,
      "loss": 0.9715,
      "step": 16024
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0266849994659424,
      "learning_rate": 1.038047471488356e-05,
      "loss": 0.6461,
      "step": 16025
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9237247705459595,
      "learning_rate": 1.037983579848577e-05,
      "loss": 0.8253,
      "step": 16026
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.060451030731201,
      "learning_rate": 1.0379196882087981e-05,
      "loss": 1.4167,
      "step": 16027
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.756182074546814,
      "learning_rate": 1.037855796569019e-05,
      "loss": 0.8106,
      "step": 16028
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2176787853240967,
      "learning_rate": 1.0377919049292403e-05,
      "loss": 0.9612,
      "step": 16029
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1019856929779053,
      "learning_rate": 1.0377280132894611e-05,
      "loss": 1.0294,
      "step": 16030
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.048560619354248,
      "learning_rate": 1.0376641216496824e-05,
      "loss": 1.1817,
      "step": 16031
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.286198139190674,
      "learning_rate": 1.0376002300099032e-05,
      "loss": 0.9989,
      "step": 16032
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7017326354980469,
      "learning_rate": 1.0375363383701244e-05,
      "loss": 0.7629,
      "step": 16033
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5088770389556885,
      "learning_rate": 1.0374724467303454e-05,
      "loss": 0.8383,
      "step": 16034
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2093088626861572,
      "learning_rate": 1.0374085550905665e-05,
      "loss": 0.9475,
      "step": 16035
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.337847948074341,
      "learning_rate": 1.0373446634507875e-05,
      "loss": 1.1001,
      "step": 16036
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.918608546257019,
      "learning_rate": 1.0372807718110087e-05,
      "loss": 0.9238,
      "step": 16037
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0247855186462402,
      "learning_rate": 1.0372168801712296e-05,
      "loss": 0.8794,
      "step": 16038
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6528750658035278,
      "learning_rate": 1.0371529885314508e-05,
      "loss": 0.9199,
      "step": 16039
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.90348482131958,
      "learning_rate": 1.0370890968916718e-05,
      "loss": 0.8466,
      "step": 16040
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6101899147033691,
      "learning_rate": 1.0370252052518928e-05,
      "loss": 0.8038,
      "step": 16041
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9091860055923462,
      "learning_rate": 1.036961313612114e-05,
      "loss": 0.9904,
      "step": 16042
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.934849262237549,
      "learning_rate": 1.0368974219723349e-05,
      "loss": 1.5281,
      "step": 16043
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0388171672821045,
      "learning_rate": 1.036833530332556e-05,
      "loss": 1.0203,
      "step": 16044
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8129022121429443,
      "learning_rate": 1.036769638692777e-05,
      "loss": 0.8951,
      "step": 16045
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7858392000198364,
      "learning_rate": 1.0367057470529982e-05,
      "loss": 0.9091,
      "step": 16046
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.8000741004943848,
      "learning_rate": 1.0366418554132192e-05,
      "loss": 0.9739,
      "step": 16047
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5254361629486084,
      "learning_rate": 1.0365779637734403e-05,
      "loss": 1.0834,
      "step": 16048
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.076462507247925,
      "learning_rate": 1.0365140721336613e-05,
      "loss": 0.8283,
      "step": 16049
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.795054316520691,
      "learning_rate": 1.0364501804938825e-05,
      "loss": 0.8706,
      "step": 16050
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9222320318222046,
      "learning_rate": 1.0363862888541035e-05,
      "loss": 0.7688,
      "step": 16051
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8721847534179688,
      "learning_rate": 1.0363223972143246e-05,
      "loss": 0.9149,
      "step": 16052
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.884237289428711,
      "learning_rate": 1.0362585055745456e-05,
      "loss": 0.9416,
      "step": 16053
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.4551150798797607,
      "learning_rate": 1.0361946139347668e-05,
      "loss": 1.0137,
      "step": 16054
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5204600095748901,
      "learning_rate": 1.0361307222949878e-05,
      "loss": 0.7856,
      "step": 16055
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8303672075271606,
      "learning_rate": 1.0360668306552089e-05,
      "loss": 0.8343,
      "step": 16056
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.738276720046997,
      "learning_rate": 1.0360029390154299e-05,
      "loss": 0.4521,
      "step": 16057
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.8202524185180664,
      "learning_rate": 1.035939047375651e-05,
      "loss": 0.8388,
      "step": 16058
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.719810128211975,
      "learning_rate": 1.0358751557358719e-05,
      "loss": 0.8394,
      "step": 16059
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6115771532058716,
      "learning_rate": 1.0358112640960932e-05,
      "loss": 0.4336,
      "step": 16060
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7925090789794922,
      "learning_rate": 1.035747372456314e-05,
      "loss": 1.1194,
      "step": 16061
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7343940734863281,
      "learning_rate": 1.0356834808165353e-05,
      "loss": 0.7822,
      "step": 16062
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.545721411705017,
      "learning_rate": 1.0356195891767561e-05,
      "loss": 0.9306,
      "step": 16063
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.929212212562561,
      "learning_rate": 1.0355556975369773e-05,
      "loss": 0.9505,
      "step": 16064
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.899153470993042,
      "learning_rate": 1.0354918058971983e-05,
      "loss": 1.1631,
      "step": 16065
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.655228614807129,
      "learning_rate": 1.0354279142574194e-05,
      "loss": 0.8492,
      "step": 16066
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7927398681640625,
      "learning_rate": 1.0353640226176404e-05,
      "loss": 0.8228,
      "step": 16067
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6990476846694946,
      "learning_rate": 1.0353001309778616e-05,
      "loss": 0.9563,
      "step": 16068
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7565605640411377,
      "learning_rate": 1.0352362393380826e-05,
      "loss": 0.7962,
      "step": 16069
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7143750190734863,
      "learning_rate": 1.0351723476983037e-05,
      "loss": 0.8825,
      "step": 16070
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2293920516967773,
      "learning_rate": 1.0351084560585247e-05,
      "loss": 1.2042,
      "step": 16071
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4366604089736938,
      "learning_rate": 1.0350445644187459e-05,
      "loss": 0.8736,
      "step": 16072
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9020702838897705,
      "learning_rate": 1.0349806727789668e-05,
      "loss": 0.7219,
      "step": 16073
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6265424489974976,
      "learning_rate": 1.034916781139188e-05,
      "loss": 0.9408,
      "step": 16074
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7792251110076904,
      "learning_rate": 1.034852889499409e-05,
      "loss": 0.9236,
      "step": 16075
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7027292251586914,
      "learning_rate": 1.0347889978596301e-05,
      "loss": 0.8249,
      "step": 16076
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9382438659667969,
      "learning_rate": 1.0347251062198511e-05,
      "loss": 0.9423,
      "step": 16077
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.669238567352295,
      "learning_rate": 1.0346612145800723e-05,
      "loss": 0.7642,
      "step": 16078
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.553519368171692,
      "learning_rate": 1.0345973229402933e-05,
      "loss": 0.8805,
      "step": 16079
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7711187601089478,
      "learning_rate": 1.0345334313005144e-05,
      "loss": 1.0197,
      "step": 16080
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.408734679222107,
      "learning_rate": 1.0344695396607354e-05,
      "loss": 0.8802,
      "step": 16081
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.008169412612915,
      "learning_rate": 1.0344056480209566e-05,
      "loss": 0.8163,
      "step": 16082
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5625616312026978,
      "learning_rate": 1.0343417563811775e-05,
      "loss": 0.9566,
      "step": 16083
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7803395986557007,
      "learning_rate": 1.0342778647413987e-05,
      "loss": 0.9067,
      "step": 16084
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.188960552215576,
      "learning_rate": 1.0342139731016197e-05,
      "loss": 1.0328,
      "step": 16085
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.072723388671875,
      "learning_rate": 1.0341500814618408e-05,
      "loss": 1.1108,
      "step": 16086
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5225577354431152,
      "learning_rate": 1.0340861898220618e-05,
      "loss": 0.8719,
      "step": 16087
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8395023345947266,
      "learning_rate": 1.034022298182283e-05,
      "loss": 0.9555,
      "step": 16088
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.164093971252441,
      "learning_rate": 1.033958406542504e-05,
      "loss": 1.5654,
      "step": 16089
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5623884201049805,
      "learning_rate": 1.0338945149027251e-05,
      "loss": 1.0829,
      "step": 16090
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3593547344207764,
      "learning_rate": 1.0338306232629461e-05,
      "loss": 0.4961,
      "step": 16091
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5300391912460327,
      "learning_rate": 1.0337667316231673e-05,
      "loss": 0.8335,
      "step": 16092
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7329251766204834,
      "learning_rate": 1.0337028399833882e-05,
      "loss": 0.9276,
      "step": 16093
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5341976881027222,
      "learning_rate": 1.0336389483436094e-05,
      "loss": 0.8624,
      "step": 16094
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9002442359924316,
      "learning_rate": 1.0335750567038302e-05,
      "loss": 1.0861,
      "step": 16095
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.72536563873291,
      "learning_rate": 1.0335111650640515e-05,
      "loss": 1.1035,
      "step": 16096
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3480703830718994,
      "learning_rate": 1.0334472734242724e-05,
      "loss": 0.9002,
      "step": 16097
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6180380582809448,
      "learning_rate": 1.0333833817844937e-05,
      "loss": 0.9097,
      "step": 16098
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.220446825027466,
      "learning_rate": 1.0333194901447145e-05,
      "loss": 1.1285,
      "step": 16099
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3503766059875488,
      "learning_rate": 1.0332555985049356e-05,
      "loss": 0.8625,
      "step": 16100
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3931732177734375,
      "learning_rate": 1.0331917068651566e-05,
      "loss": 0.8285,
      "step": 16101
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1286535263061523,
      "learning_rate": 1.0331278152253778e-05,
      "loss": 1.2028,
      "step": 16102
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.708229899406433,
      "learning_rate": 1.0330639235855988e-05,
      "loss": 0.9248,
      "step": 16103
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6146436929702759,
      "learning_rate": 1.03300003194582e-05,
      "loss": 0.8629,
      "step": 16104
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4146863222122192,
      "learning_rate": 1.0329361403060409e-05,
      "loss": 0.8569,
      "step": 16105
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5969605445861816,
      "learning_rate": 1.032872248666262e-05,
      "loss": 0.8968,
      "step": 16106
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.079622268676758,
      "learning_rate": 1.032808357026483e-05,
      "loss": 1.132,
      "step": 16107
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3043203353881836,
      "learning_rate": 1.0327444653867042e-05,
      "loss": 0.9423,
      "step": 16108
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5599371194839478,
      "learning_rate": 1.0326805737469252e-05,
      "loss": 0.841,
      "step": 16109
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.132693767547607,
      "learning_rate": 1.0326166821071463e-05,
      "loss": 0.9997,
      "step": 16110
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4336374998092651,
      "learning_rate": 1.0325527904673673e-05,
      "loss": 0.869,
      "step": 16111
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.714463472366333,
      "learning_rate": 1.0324888988275885e-05,
      "loss": 1.0866,
      "step": 16112
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.664786458015442,
      "learning_rate": 1.0324250071878095e-05,
      "loss": 0.9382,
      "step": 16113
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4746195077896118,
      "learning_rate": 1.0323611155480306e-05,
      "loss": 0.512,
      "step": 16114
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.480642318725586,
      "learning_rate": 1.0322972239082516e-05,
      "loss": 1.1612,
      "step": 16115
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7290195226669312,
      "learning_rate": 1.0322333322684728e-05,
      "loss": 0.7446,
      "step": 16116
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7273566722869873,
      "learning_rate": 1.0321694406286938e-05,
      "loss": 0.9937,
      "step": 16117
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.324889659881592,
      "learning_rate": 1.0321055489889149e-05,
      "loss": 0.592,
      "step": 16118
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.614828109741211,
      "learning_rate": 1.0320416573491359e-05,
      "loss": 1.4568,
      "step": 16119
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.023219585418701,
      "learning_rate": 1.031977765709357e-05,
      "loss": 0.777,
      "step": 16120
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.226431369781494,
      "learning_rate": 1.031913874069578e-05,
      "loss": 0.9729,
      "step": 16121
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6448386907577515,
      "learning_rate": 1.0318499824297992e-05,
      "loss": 0.8879,
      "step": 16122
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7773785591125488,
      "learning_rate": 1.0317860907900202e-05,
      "loss": 0.844,
      "step": 16123
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7624181509017944,
      "learning_rate": 1.0317221991502413e-05,
      "loss": 0.8956,
      "step": 16124
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6317365169525146,
      "learning_rate": 1.0316583075104623e-05,
      "loss": 0.8291,
      "step": 16125
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2489964962005615,
      "learning_rate": 1.0315944158706835e-05,
      "loss": 0.9439,
      "step": 16126
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.24658727645874,
      "learning_rate": 1.0315305242309045e-05,
      "loss": 1.1284,
      "step": 16127
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6676377058029175,
      "learning_rate": 1.0314666325911256e-05,
      "loss": 0.8427,
      "step": 16128
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8441466093063354,
      "learning_rate": 1.0314027409513466e-05,
      "loss": 0.9134,
      "step": 16129
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.61809504032135,
      "learning_rate": 1.0313388493115677e-05,
      "loss": 0.9488,
      "step": 16130
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.773361086845398,
      "learning_rate": 1.0312749576717886e-05,
      "loss": 0.902,
      "step": 16131
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8419533967971802,
      "learning_rate": 1.0312110660320099e-05,
      "loss": 0.8166,
      "step": 16132
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.673449993133545,
      "learning_rate": 1.0311471743922307e-05,
      "loss": 0.7169,
      "step": 16133
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3555922508239746,
      "learning_rate": 1.031083282752452e-05,
      "loss": 0.9634,
      "step": 16134
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.6308531761169434,
      "learning_rate": 1.0310193911126728e-05,
      "loss": 0.9277,
      "step": 16135
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6493629217147827,
      "learning_rate": 1.0309554994728942e-05,
      "loss": 0.9185,
      "step": 16136
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8274117708206177,
      "learning_rate": 1.030891607833115e-05,
      "loss": 0.9033,
      "step": 16137
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.636014699935913,
      "learning_rate": 1.0308277161933361e-05,
      "loss": 0.4193,
      "step": 16138
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2994633913040161,
      "learning_rate": 1.0307638245535571e-05,
      "loss": 0.4613,
      "step": 16139
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6970069408416748,
      "learning_rate": 1.0306999329137783e-05,
      "loss": 0.9232,
      "step": 16140
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.929161787033081,
      "learning_rate": 1.0306360412739993e-05,
      "loss": 0.9399,
      "step": 16141
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8557353019714355,
      "learning_rate": 1.0305721496342204e-05,
      "loss": 0.9403,
      "step": 16142
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1170544624328613,
      "learning_rate": 1.0305082579944414e-05,
      "loss": 0.899,
      "step": 16143
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5228229761123657,
      "learning_rate": 1.0304443663546626e-05,
      "loss": 0.8625,
      "step": 16144
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.133044958114624,
      "learning_rate": 1.0303804747148835e-05,
      "loss": 0.9646,
      "step": 16145
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4717422723770142,
      "learning_rate": 1.0303165830751047e-05,
      "loss": 0.8115,
      "step": 16146
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8382759094238281,
      "learning_rate": 1.0302526914353257e-05,
      "loss": 0.8142,
      "step": 16147
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.094728469848633,
      "learning_rate": 1.0301887997955468e-05,
      "loss": 1.2016,
      "step": 16148
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5535268783569336,
      "learning_rate": 1.0301249081557678e-05,
      "loss": 0.8386,
      "step": 16149
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2209603786468506,
      "learning_rate": 1.030061016515989e-05,
      "loss": 0.4472,
      "step": 16150
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9232953786849976,
      "learning_rate": 1.02999712487621e-05,
      "loss": 0.958,
      "step": 16151
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8148256540298462,
      "learning_rate": 1.0299332332364311e-05,
      "loss": 0.9169,
      "step": 16152
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.3259851932525635,
      "learning_rate": 1.0298693415966521e-05,
      "loss": 0.95,
      "step": 16153
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.865348219871521,
      "learning_rate": 1.0298054499568733e-05,
      "loss": 0.8413,
      "step": 16154
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.065847873687744,
      "learning_rate": 1.0297415583170942e-05,
      "loss": 1.5506,
      "step": 16155
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2527525424957275,
      "learning_rate": 1.0296776666773154e-05,
      "loss": 1.1715,
      "step": 16156
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.254884719848633,
      "learning_rate": 1.0296137750375364e-05,
      "loss": 0.919,
      "step": 16157
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1200203895568848,
      "learning_rate": 1.0295498833977575e-05,
      "loss": 1.2231,
      "step": 16158
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.9883885383605957,
      "learning_rate": 1.0294859917579785e-05,
      "loss": 1.2693,
      "step": 16159
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8137212991714478,
      "learning_rate": 1.0294221001181997e-05,
      "loss": 0.875,
      "step": 16160
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5352963209152222,
      "learning_rate": 1.0293582084784207e-05,
      "loss": 0.9147,
      "step": 16161
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9236505031585693,
      "learning_rate": 1.0292943168386418e-05,
      "loss": 0.756,
      "step": 16162
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7325210571289062,
      "learning_rate": 1.0292304251988628e-05,
      "loss": 0.9411,
      "step": 16163
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.894836664199829,
      "learning_rate": 1.029166533559084e-05,
      "loss": 0.8003,
      "step": 16164
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.636762022972107,
      "learning_rate": 1.029102641919305e-05,
      "loss": 0.772,
      "step": 16165
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.291501998901367,
      "learning_rate": 1.0290387502795261e-05,
      "loss": 1.1318,
      "step": 16166
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.6925907135009766,
      "learning_rate": 1.0289748586397469e-05,
      "loss": 0.8715,
      "step": 16167
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.65644371509552,
      "learning_rate": 1.0289109669999682e-05,
      "loss": 0.342,
      "step": 16168
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.147506237030029,
      "learning_rate": 1.028847075360189e-05,
      "loss": 1.2546,
      "step": 16169
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6174863576889038,
      "learning_rate": 1.0287831837204104e-05,
      "loss": 0.9021,
      "step": 16170
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7321851253509521,
      "learning_rate": 1.0287192920806312e-05,
      "loss": 0.8746,
      "step": 16171
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.589606761932373,
      "learning_rate": 1.0286554004408525e-05,
      "loss": 0.758,
      "step": 16172
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.657422423362732,
      "learning_rate": 1.0285915088010733e-05,
      "loss": 0.918,
      "step": 16173
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.874232292175293,
      "learning_rate": 1.0285276171612945e-05,
      "loss": 0.8098,
      "step": 16174
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5695799589157104,
      "learning_rate": 1.0284637255215155e-05,
      "loss": 0.8324,
      "step": 16175
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.944443702697754,
      "learning_rate": 1.0283998338817366e-05,
      "loss": 0.9277,
      "step": 16176
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8444948196411133,
      "learning_rate": 1.0283359422419576e-05,
      "loss": 0.9672,
      "step": 16177
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.510258436203003,
      "learning_rate": 1.0282720506021788e-05,
      "loss": 0.8333,
      "step": 16178
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.281068801879883,
      "learning_rate": 1.0282081589623998e-05,
      "loss": 0.8694,
      "step": 16179
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.1321187019348145,
      "learning_rate": 1.0281442673226209e-05,
      "loss": 1.0545,
      "step": 16180
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.832567811012268,
      "learning_rate": 1.0280803756828419e-05,
      "loss": 0.8713,
      "step": 16181
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.204542875289917,
      "learning_rate": 1.028016484043063e-05,
      "loss": 1.0203,
      "step": 16182
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8803508281707764,
      "learning_rate": 1.027952592403284e-05,
      "loss": 1.0042,
      "step": 16183
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7131757736206055,
      "learning_rate": 1.0278887007635052e-05,
      "loss": 0.8594,
      "step": 16184
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4642870426177979,
      "learning_rate": 1.0278248091237262e-05,
      "loss": 0.3948,
      "step": 16185
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.111348628997803,
      "learning_rate": 1.0277609174839473e-05,
      "loss": 0.8895,
      "step": 16186
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7213338613510132,
      "learning_rate": 1.0276970258441683e-05,
      "loss": 1.0048,
      "step": 16187
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5927737951278687,
      "learning_rate": 1.0276331342043895e-05,
      "loss": 0.5573,
      "step": 16188
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3265037536621094,
      "learning_rate": 1.0275692425646105e-05,
      "loss": 0.9257,
      "step": 16189
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7177013158798218,
      "learning_rate": 1.0275053509248316e-05,
      "loss": 0.8272,
      "step": 16190
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.828871011734009,
      "learning_rate": 1.0274414592850526e-05,
      "loss": 1.1801,
      "step": 16191
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5458760261535645,
      "learning_rate": 1.0273775676452737e-05,
      "loss": 0.8872,
      "step": 16192
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6163225173950195,
      "learning_rate": 1.0273136760054947e-05,
      "loss": 0.7278,
      "step": 16193
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9168726205825806,
      "learning_rate": 1.0272497843657159e-05,
      "loss": 0.9621,
      "step": 16194
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.002629518508911,
      "learning_rate": 1.0271858927259369e-05,
      "loss": 0.761,
      "step": 16195
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.324812889099121,
      "learning_rate": 1.027122001086158e-05,
      "loss": 0.8439,
      "step": 16196
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9657472372055054,
      "learning_rate": 1.027058109446379e-05,
      "loss": 1.0111,
      "step": 16197
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8693181276321411,
      "learning_rate": 1.0269942178066002e-05,
      "loss": 1.2215,
      "step": 16198
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6966423988342285,
      "learning_rate": 1.0269303261668212e-05,
      "loss": 0.8764,
      "step": 16199
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7968571186065674,
      "learning_rate": 1.0268664345270423e-05,
      "loss": 0.7814,
      "step": 16200
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8356035947799683,
      "learning_rate": 1.0268025428872633e-05,
      "loss": 0.8789,
      "step": 16201
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.101304054260254,
      "learning_rate": 1.0267386512474844e-05,
      "loss": 0.5219,
      "step": 16202
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0217771530151367,
      "learning_rate": 1.0266747596077054e-05,
      "loss": 0.8719,
      "step": 16203
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.131610870361328,
      "learning_rate": 1.0266108679679266e-05,
      "loss": 1.2364,
      "step": 16204
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.807644248008728,
      "learning_rate": 1.0265469763281474e-05,
      "loss": 0.8941,
      "step": 16205
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.075397491455078,
      "learning_rate": 1.0264830846883687e-05,
      "loss": 1.0395,
      "step": 16206
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7872915267944336,
      "learning_rate": 1.0264191930485895e-05,
      "loss": 0.9712,
      "step": 16207
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8750213384628296,
      "learning_rate": 1.0263553014088109e-05,
      "loss": 0.8799,
      "step": 16208
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0140020847320557,
      "learning_rate": 1.0262914097690317e-05,
      "loss": 1.1689,
      "step": 16209
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0411455631256104,
      "learning_rate": 1.0262275181292528e-05,
      "loss": 1.185,
      "step": 16210
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7403455972671509,
      "learning_rate": 1.0261636264894738e-05,
      "loss": 0.8978,
      "step": 16211
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6014013290405273,
      "learning_rate": 1.026099734849695e-05,
      "loss": 1.0231,
      "step": 16212
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.440783739089966,
      "learning_rate": 1.026035843209916e-05,
      "loss": 0.9362,
      "step": 16213
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6713026762008667,
      "learning_rate": 1.0259719515701371e-05,
      "loss": 0.9578,
      "step": 16214
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2751972675323486,
      "learning_rate": 1.0259080599303581e-05,
      "loss": 1.0069,
      "step": 16215
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9582014083862305,
      "learning_rate": 1.0258441682905793e-05,
      "loss": 0.9112,
      "step": 16216
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.981816291809082,
      "learning_rate": 1.0257802766508002e-05,
      "loss": 1.1285,
      "step": 16217
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1460587978363037,
      "learning_rate": 1.0257163850110214e-05,
      "loss": 0.9939,
      "step": 16218
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6895043849945068,
      "learning_rate": 1.0256524933712424e-05,
      "loss": 0.9262,
      "step": 16219
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.749913215637207,
      "learning_rate": 1.0255886017314635e-05,
      "loss": 0.8243,
      "step": 16220
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.8813931941986084,
      "learning_rate": 1.0255247100916845e-05,
      "loss": 0.945,
      "step": 16221
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1995584964752197,
      "learning_rate": 1.0254608184519057e-05,
      "loss": 1.0632,
      "step": 16222
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5822436809539795,
      "learning_rate": 1.0253969268121267e-05,
      "loss": 0.7965,
      "step": 16223
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8496817350387573,
      "learning_rate": 1.0253330351723478e-05,
      "loss": 1.213,
      "step": 16224
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3989269733428955,
      "learning_rate": 1.0252691435325688e-05,
      "loss": 0.8403,
      "step": 16225
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.846636176109314,
      "learning_rate": 1.02520525189279e-05,
      "loss": 0.8377,
      "step": 16226
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1336331367492676,
      "learning_rate": 1.025141360253011e-05,
      "loss": 0.7779,
      "step": 16227
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6931895017623901,
      "learning_rate": 1.0250774686132321e-05,
      "loss": 1.0081,
      "step": 16228
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8535596132278442,
      "learning_rate": 1.025013576973453e-05,
      "loss": 0.9996,
      "step": 16229
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8628848791122437,
      "learning_rate": 1.0249496853336742e-05,
      "loss": 0.9351,
      "step": 16230
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0615360736846924,
      "learning_rate": 1.0248857936938952e-05,
      "loss": 1.1496,
      "step": 16231
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5561833381652832,
      "learning_rate": 1.0248219020541164e-05,
      "loss": 0.9148,
      "step": 16232
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.040229082107544,
      "learning_rate": 1.0247580104143374e-05,
      "loss": 0.8901,
      "step": 16233
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4611507654190063,
      "learning_rate": 1.0246941187745585e-05,
      "loss": 0.7414,
      "step": 16234
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.9842936992645264,
      "learning_rate": 1.0246302271347795e-05,
      "loss": 0.6962,
      "step": 16235
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.585317850112915,
      "learning_rate": 1.0245663354950007e-05,
      "loss": 0.8825,
      "step": 16236
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5742225646972656,
      "learning_rate": 1.0245024438552216e-05,
      "loss": 0.8247,
      "step": 16237
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6011903285980225,
      "learning_rate": 1.0244385522154428e-05,
      "loss": 1.0205,
      "step": 16238
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.918959379196167,
      "learning_rate": 1.0243746605756638e-05,
      "loss": 0.9351,
      "step": 16239
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.819401502609253,
      "learning_rate": 1.024310768935885e-05,
      "loss": 0.8339,
      "step": 16240
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5456095933914185,
      "learning_rate": 1.0242468772961058e-05,
      "loss": 0.4418,
      "step": 16241
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9019840955734253,
      "learning_rate": 1.024182985656327e-05,
      "loss": 0.8585,
      "step": 16242
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7661603689193726,
      "learning_rate": 1.0241190940165479e-05,
      "loss": 1.0343,
      "step": 16243
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8837430477142334,
      "learning_rate": 1.0240552023767692e-05,
      "loss": 0.942,
      "step": 16244
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8468207120895386,
      "learning_rate": 1.02399131073699e-05,
      "loss": 0.9588,
      "step": 16245
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9440542459487915,
      "learning_rate": 1.0239274190972112e-05,
      "loss": 1.0579,
      "step": 16246
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3972500562667847,
      "learning_rate": 1.0238635274574322e-05,
      "loss": 0.8628,
      "step": 16247
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.4287517070770264,
      "learning_rate": 1.0237996358176533e-05,
      "loss": 0.602,
      "step": 16248
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.440672755241394,
      "learning_rate": 1.0237357441778743e-05,
      "loss": 0.7863,
      "step": 16249
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6303412914276123,
      "learning_rate": 1.0236718525380955e-05,
      "loss": 0.8642,
      "step": 16250
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7121270895004272,
      "learning_rate": 1.0236079608983165e-05,
      "loss": 0.9978,
      "step": 16251
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1715691089630127,
      "learning_rate": 1.0235440692585376e-05,
      "loss": 0.9804,
      "step": 16252
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.322749614715576,
      "learning_rate": 1.0234801776187586e-05,
      "loss": 0.7912,
      "step": 16253
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.116818428039551,
      "learning_rate": 1.0234162859789797e-05,
      "loss": 0.9801,
      "step": 16254
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4980589151382446,
      "learning_rate": 1.0233523943392007e-05,
      "loss": 0.8409,
      "step": 16255
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.695030689239502,
      "learning_rate": 1.0232885026994219e-05,
      "loss": 0.7999,
      "step": 16256
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.705676555633545,
      "learning_rate": 1.0232246110596429e-05,
      "loss": 0.8759,
      "step": 16257
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4439232349395752,
      "learning_rate": 1.023160719419864e-05,
      "loss": 0.4993,
      "step": 16258
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9814081192016602,
      "learning_rate": 1.023096827780085e-05,
      "loss": 0.9247,
      "step": 16259
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4962997436523438,
      "learning_rate": 1.0230329361403062e-05,
      "loss": 0.8168,
      "step": 16260
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8056918382644653,
      "learning_rate": 1.0229690445005272e-05,
      "loss": 0.9007,
      "step": 16261
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3090085983276367,
      "learning_rate": 1.0229051528607483e-05,
      "loss": 0.9235,
      "step": 16262
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.045214891433716,
      "learning_rate": 1.0228412612209693e-05,
      "loss": 1.0955,
      "step": 16263
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5887473821640015,
      "learning_rate": 1.0227773695811904e-05,
      "loss": 0.818,
      "step": 16264
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.122020959854126,
      "learning_rate": 1.0227134779414114e-05,
      "loss": 0.8798,
      "step": 16265
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6142058372497559,
      "learning_rate": 1.0226495863016326e-05,
      "loss": 0.7945,
      "step": 16266
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0623388290405273,
      "learning_rate": 1.0225856946618536e-05,
      "loss": 0.8519,
      "step": 16267
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6394429206848145,
      "learning_rate": 1.0225218030220747e-05,
      "loss": 0.8012,
      "step": 16268
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7932333946228027,
      "learning_rate": 1.0224579113822957e-05,
      "loss": 0.7912,
      "step": 16269
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4252454042434692,
      "learning_rate": 1.0223940197425169e-05,
      "loss": 0.8467,
      "step": 16270
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2933051586151123,
      "learning_rate": 1.0223301281027379e-05,
      "loss": 1.0054,
      "step": 16271
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.789285659790039,
      "learning_rate": 1.022266236462959e-05,
      "loss": 1.1823,
      "step": 16272
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4932053089141846,
      "learning_rate": 1.02220234482318e-05,
      "loss": 0.7917,
      "step": 16273
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6443750858306885,
      "learning_rate": 1.0221384531834011e-05,
      "loss": 0.8996,
      "step": 16274
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.015016555786133,
      "learning_rate": 1.0220745615436221e-05,
      "loss": 0.9338,
      "step": 16275
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6717395782470703,
      "learning_rate": 1.0220106699038433e-05,
      "loss": 0.9081,
      "step": 16276
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.526015043258667,
      "learning_rate": 1.0219467782640641e-05,
      "loss": 0.8824,
      "step": 16277
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7588759660720825,
      "learning_rate": 1.0218828866242854e-05,
      "loss": 0.7596,
      "step": 16278
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5883792638778687,
      "learning_rate": 1.0218189949845062e-05,
      "loss": 0.876,
      "step": 16279
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.4013309478759766,
      "learning_rate": 1.0217551033447276e-05,
      "loss": 1.079,
      "step": 16280
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.4766013622283936,
      "learning_rate": 1.0216912117049484e-05,
      "loss": 0.9358,
      "step": 16281
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7814133167266846,
      "learning_rate": 1.0216273200651697e-05,
      "loss": 0.8052,
      "step": 16282
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5242739915847778,
      "learning_rate": 1.0215634284253905e-05,
      "loss": 0.8771,
      "step": 16283
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.843133807182312,
      "learning_rate": 1.0214995367856117e-05,
      "loss": 0.8662,
      "step": 16284
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1378695964813232,
      "learning_rate": 1.0214356451458327e-05,
      "loss": 0.736,
      "step": 16285
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0268940925598145,
      "learning_rate": 1.0213717535060538e-05,
      "loss": 0.9162,
      "step": 16286
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5340067148208618,
      "learning_rate": 1.0213078618662748e-05,
      "loss": 0.8611,
      "step": 16287
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.307175159454346,
      "learning_rate": 1.021243970226496e-05,
      "loss": 1.4342,
      "step": 16288
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8144134283065796,
      "learning_rate": 1.021180078586717e-05,
      "loss": 0.9223,
      "step": 16289
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.705733299255371,
      "learning_rate": 1.0211161869469381e-05,
      "loss": 0.7681,
      "step": 16290
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.280595064163208,
      "learning_rate": 1.0210522953071591e-05,
      "loss": 0.7544,
      "step": 16291
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1035454273223877,
      "learning_rate": 1.0209884036673802e-05,
      "loss": 0.9014,
      "step": 16292
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1938095092773438,
      "learning_rate": 1.0209245120276012e-05,
      "loss": 0.9161,
      "step": 16293
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1999638080596924,
      "learning_rate": 1.0208606203878224e-05,
      "loss": 0.9354,
      "step": 16294
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0021588802337646,
      "learning_rate": 1.0207967287480434e-05,
      "loss": 0.9272,
      "step": 16295
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.472827434539795,
      "learning_rate": 1.0207328371082645e-05,
      "loss": 1.6219,
      "step": 16296
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5668962001800537,
      "learning_rate": 1.0206689454684855e-05,
      "loss": 0.8843,
      "step": 16297
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7643107175827026,
      "learning_rate": 1.0206050538287067e-05,
      "loss": 0.9237,
      "step": 16298
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6210672855377197,
      "learning_rate": 1.0205411621889276e-05,
      "loss": 0.8064,
      "step": 16299
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5809887647628784,
      "learning_rate": 1.0204772705491488e-05,
      "loss": 0.3606,
      "step": 16300
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.1169376373291016,
      "learning_rate": 1.0204133789093698e-05,
      "loss": 0.8346,
      "step": 16301
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8875386714935303,
      "learning_rate": 1.020349487269591e-05,
      "loss": 0.8544,
      "step": 16302
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5610073804855347,
      "learning_rate": 1.020285595629812e-05,
      "loss": 0.7889,
      "step": 16303
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9606764316558838,
      "learning_rate": 1.020221703990033e-05,
      "loss": 0.9737,
      "step": 16304
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.025963544845581,
      "learning_rate": 1.020157812350254e-05,
      "loss": 0.8298,
      "step": 16305
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.420974850654602,
      "learning_rate": 1.0200939207104752e-05,
      "loss": 0.4275,
      "step": 16306
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4038519859313965,
      "learning_rate": 1.0200300290706962e-05,
      "loss": 0.9302,
      "step": 16307
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4555182456970215,
      "learning_rate": 1.0199661374309174e-05,
      "loss": 0.6908,
      "step": 16308
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.300985097885132,
      "learning_rate": 1.0199022457911383e-05,
      "loss": 0.7515,
      "step": 16309
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.031907320022583,
      "learning_rate": 1.0198383541513595e-05,
      "loss": 0.9722,
      "step": 16310
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6481223106384277,
      "learning_rate": 1.0197744625115805e-05,
      "loss": 0.7648,
      "step": 16311
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8890941143035889,
      "learning_rate": 1.0197105708718016e-05,
      "loss": 0.8787,
      "step": 16312
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7678816318511963,
      "learning_rate": 1.0196466792320226e-05,
      "loss": 0.8095,
      "step": 16313
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.242349863052368,
      "learning_rate": 1.0195827875922438e-05,
      "loss": 1.2115,
      "step": 16314
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.913406252861023,
      "learning_rate": 1.0195188959524646e-05,
      "loss": 0.97,
      "step": 16315
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9203650951385498,
      "learning_rate": 1.019455004312686e-05,
      "loss": 0.8204,
      "step": 16316
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4745324850082397,
      "learning_rate": 1.0193911126729067e-05,
      "loss": 0.8336,
      "step": 16317
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.511872410774231,
      "learning_rate": 1.019327221033128e-05,
      "loss": 0.8473,
      "step": 16318
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0242905616760254,
      "learning_rate": 1.0192633293933489e-05,
      "loss": 1.2135,
      "step": 16319
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.124345302581787,
      "learning_rate": 1.01919943775357e-05,
      "loss": 1.4468,
      "step": 16320
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.3638370037078857,
      "learning_rate": 1.019135546113791e-05,
      "loss": 1.3013,
      "step": 16321
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.414558172225952,
      "learning_rate": 1.0190716544740122e-05,
      "loss": 0.9783,
      "step": 16322
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6742992401123047,
      "learning_rate": 1.0190077628342332e-05,
      "loss": 0.8162,
      "step": 16323
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.611647367477417,
      "learning_rate": 1.0189438711944543e-05,
      "loss": 0.8473,
      "step": 16324
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5557894706726074,
      "learning_rate": 1.0188799795546753e-05,
      "loss": 1.2965,
      "step": 16325
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1470441818237305,
      "learning_rate": 1.0188160879148965e-05,
      "loss": 0.97,
      "step": 16326
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.755431056022644,
      "learning_rate": 1.0187521962751174e-05,
      "loss": 0.8288,
      "step": 16327
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.68414306640625,
      "learning_rate": 1.0186883046353386e-05,
      "loss": 0.8064,
      "step": 16328
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7930985689163208,
      "learning_rate": 1.0186244129955596e-05,
      "loss": 0.8727,
      "step": 16329
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6064212322235107,
      "learning_rate": 1.0185605213557807e-05,
      "loss": 0.8223,
      "step": 16330
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.610001564025879,
      "learning_rate": 1.0184966297160017e-05,
      "loss": 0.9235,
      "step": 16331
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.547760009765625,
      "learning_rate": 1.0184327380762229e-05,
      "loss": 0.9336,
      "step": 16332
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5309486389160156,
      "learning_rate": 1.0183688464364439e-05,
      "loss": 0.8447,
      "step": 16333
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9840787649154663,
      "learning_rate": 1.018304954796665e-05,
      "loss": 0.9727,
      "step": 16334
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0445971488952637,
      "learning_rate": 1.018241063156886e-05,
      "loss": 0.9963,
      "step": 16335
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1782357692718506,
      "learning_rate": 1.0181771715171072e-05,
      "loss": 1.018,
      "step": 16336
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.025618076324463,
      "learning_rate": 1.0181132798773281e-05,
      "loss": 1.1997,
      "step": 16337
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5394786596298218,
      "learning_rate": 1.0180493882375493e-05,
      "loss": 0.8635,
      "step": 16338
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6910436153411865,
      "learning_rate": 1.0179854965977703e-05,
      "loss": 0.9244,
      "step": 16339
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.321908473968506,
      "learning_rate": 1.0179216049579914e-05,
      "loss": 0.6893,
      "step": 16340
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.175206184387207,
      "learning_rate": 1.0178577133182124e-05,
      "loss": 0.9771,
      "step": 16341
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.995656967163086,
      "learning_rate": 1.0177938216784336e-05,
      "loss": 0.9676,
      "step": 16342
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7723777294158936,
      "learning_rate": 1.0177299300386546e-05,
      "loss": 0.6219,
      "step": 16343
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.330610990524292,
      "learning_rate": 1.0176660383988757e-05,
      "loss": 0.7591,
      "step": 16344
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8173401355743408,
      "learning_rate": 1.0176021467590967e-05,
      "loss": 0.9258,
      "step": 16345
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9017879962921143,
      "learning_rate": 1.0175382551193179e-05,
      "loss": 0.8279,
      "step": 16346
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0796966552734375,
      "learning_rate": 1.0174743634795388e-05,
      "loss": 1.0513,
      "step": 16347
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9160479307174683,
      "learning_rate": 1.01741047183976e-05,
      "loss": 0.663,
      "step": 16348
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6566486358642578,
      "learning_rate": 1.017346580199981e-05,
      "loss": 0.9308,
      "step": 16349
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.7525227069854736,
      "learning_rate": 1.0172826885602021e-05,
      "loss": 0.9405,
      "step": 16350
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5442485809326172,
      "learning_rate": 1.017218796920423e-05,
      "loss": 0.7809,
      "step": 16351
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6001008749008179,
      "learning_rate": 1.0171549052806443e-05,
      "loss": 0.8655,
      "step": 16352
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8622398376464844,
      "learning_rate": 1.0170910136408651e-05,
      "loss": 0.8315,
      "step": 16353
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9778568744659424,
      "learning_rate": 1.0170271220010864e-05,
      "loss": 0.924,
      "step": 16354
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0603835582733154,
      "learning_rate": 1.0169632303613072e-05,
      "loss": 0.8775,
      "step": 16355
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7063060998916626,
      "learning_rate": 1.0168993387215284e-05,
      "loss": 0.9088,
      "step": 16356
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7515019178390503,
      "learning_rate": 1.0168354470817494e-05,
      "loss": 0.8487,
      "step": 16357
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.4266462326049805,
      "learning_rate": 1.0167715554419705e-05,
      "loss": 1.2793,
      "step": 16358
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3930160999298096,
      "learning_rate": 1.0167076638021915e-05,
      "loss": 1.2647,
      "step": 16359
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6123961210250854,
      "learning_rate": 1.0166437721624127e-05,
      "loss": 0.8223,
      "step": 16360
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5616499185562134,
      "learning_rate": 1.0165798805226336e-05,
      "loss": 0.7503,
      "step": 16361
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6443655490875244,
      "learning_rate": 1.0165159888828548e-05,
      "loss": 0.8575,
      "step": 16362
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5576651096343994,
      "learning_rate": 1.0164520972430758e-05,
      "loss": 0.8212,
      "step": 16363
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5757427215576172,
      "learning_rate": 1.016388205603297e-05,
      "loss": 0.9035,
      "step": 16364
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.9845423698425293,
      "learning_rate": 1.016324313963518e-05,
      "loss": 1.1745,
      "step": 16365
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.44497549533844,
      "learning_rate": 1.016260422323739e-05,
      "loss": 0.4383,
      "step": 16366
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.45851469039917,
      "learning_rate": 1.01619653068396e-05,
      "loss": 1.4092,
      "step": 16367
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.744841456413269,
      "learning_rate": 1.0161326390441812e-05,
      "loss": 0.8541,
      "step": 16368
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4422457218170166,
      "learning_rate": 1.0160687474044022e-05,
      "loss": 0.9832,
      "step": 16369
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9818239212036133,
      "learning_rate": 1.0160048557646234e-05,
      "loss": 0.9135,
      "step": 16370
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.475238800048828,
      "learning_rate": 1.0159409641248443e-05,
      "loss": 1.1898,
      "step": 16371
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9168528318405151,
      "learning_rate": 1.0158770724850655e-05,
      "loss": 0.7694,
      "step": 16372
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1843314170837402,
      "learning_rate": 1.0158131808452865e-05,
      "loss": 1.0028,
      "step": 16373
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.9917633533477783,
      "learning_rate": 1.0157492892055076e-05,
      "loss": 1.1147,
      "step": 16374
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9046614170074463,
      "learning_rate": 1.0156853975657286e-05,
      "loss": 0.8901,
      "step": 16375
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5730260610580444,
      "learning_rate": 1.0156215059259498e-05,
      "loss": 0.886,
      "step": 16376
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.88971745967865,
      "learning_rate": 1.0155576142861708e-05,
      "loss": 0.876,
      "step": 16377
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.81175696849823,
      "learning_rate": 1.015493722646392e-05,
      "loss": 1.0169,
      "step": 16378
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0857155323028564,
      "learning_rate": 1.0154298310066129e-05,
      "loss": 0.9647,
      "step": 16379
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1508407592773438,
      "learning_rate": 1.015365939366834e-05,
      "loss": 0.9723,
      "step": 16380
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4594861268997192,
      "learning_rate": 1.015302047727055e-05,
      "loss": 0.5932,
      "step": 16381
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.745579719543457,
      "learning_rate": 1.0152381560872762e-05,
      "loss": 1.0849,
      "step": 16382
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8263169527053833,
      "learning_rate": 1.0151742644474972e-05,
      "loss": 0.865,
      "step": 16383
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.667130947113037,
      "learning_rate": 1.0151103728077183e-05,
      "loss": 0.7415,
      "step": 16384
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.439204216003418,
      "learning_rate": 1.0150464811679393e-05,
      "loss": 1.844,
      "step": 16385
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6393100023269653,
      "learning_rate": 1.0149825895281605e-05,
      "loss": 0.8071,
      "step": 16386
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.935608148574829,
      "learning_rate": 1.0149186978883813e-05,
      "loss": 1.098,
      "step": 16387
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0324342250823975,
      "learning_rate": 1.0148548062486026e-05,
      "loss": 0.867,
      "step": 16388
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9783891439437866,
      "learning_rate": 1.0147909146088234e-05,
      "loss": 1.2071,
      "step": 16389
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6058343648910522,
      "learning_rate": 1.0147270229690448e-05,
      "loss": 0.882,
      "step": 16390
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9739643335342407,
      "learning_rate": 1.0146631313292656e-05,
      "loss": 0.9368,
      "step": 16391
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.707792043685913,
      "learning_rate": 1.0145992396894867e-05,
      "loss": 0.8822,
      "step": 16392
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9193493127822876,
      "learning_rate": 1.0145353480497077e-05,
      "loss": 0.8128,
      "step": 16393
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0193817615509033,
      "learning_rate": 1.0144714564099289e-05,
      "loss": 1.3211,
      "step": 16394
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4857780933380127,
      "learning_rate": 1.0144075647701499e-05,
      "loss": 1.0133,
      "step": 16395
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0962905883789062,
      "learning_rate": 1.014343673130371e-05,
      "loss": 1.2267,
      "step": 16396
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.71980357170105,
      "learning_rate": 1.014279781490592e-05,
      "loss": 0.6675,
      "step": 16397
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.009990692138672,
      "learning_rate": 1.0142158898508132e-05,
      "loss": 0.9495,
      "step": 16398
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6389575004577637,
      "learning_rate": 1.0141519982110341e-05,
      "loss": 1.0134,
      "step": 16399
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.732679605484009,
      "learning_rate": 1.0140881065712553e-05,
      "loss": 0.8925,
      "step": 16400
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8054291009902954,
      "learning_rate": 1.0140242149314763e-05,
      "loss": 1.0015,
      "step": 16401
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1137585639953613,
      "learning_rate": 1.0139603232916974e-05,
      "loss": 0.7721,
      "step": 16402
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7236651182174683,
      "learning_rate": 1.0138964316519184e-05,
      "loss": 0.8307,
      "step": 16403
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6344164609909058,
      "learning_rate": 1.0138325400121396e-05,
      "loss": 0.8832,
      "step": 16404
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6508924961090088,
      "learning_rate": 1.0137686483723606e-05,
      "loss": 0.8465,
      "step": 16405
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.312448501586914,
      "learning_rate": 1.0137047567325817e-05,
      "loss": 0.8026,
      "step": 16406
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9800589084625244,
      "learning_rate": 1.0136408650928027e-05,
      "loss": 0.93,
      "step": 16407
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.8931374549865723,
      "learning_rate": 1.0135769734530239e-05,
      "loss": 0.707,
      "step": 16408
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6859021186828613,
      "learning_rate": 1.0135130818132448e-05,
      "loss": 0.385,
      "step": 16409
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.308502197265625,
      "learning_rate": 1.013449190173466e-05,
      "loss": 1.0851,
      "step": 16410
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4154365062713623,
      "learning_rate": 1.013385298533687e-05,
      "loss": 1.0408,
      "step": 16411
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4833399057388306,
      "learning_rate": 1.0133214068939081e-05,
      "loss": 0.869,
      "step": 16412
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7394797801971436,
      "learning_rate": 1.0132575152541291e-05,
      "loss": 1.017,
      "step": 16413
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.890750527381897,
      "learning_rate": 1.0131936236143503e-05,
      "loss": 0.9686,
      "step": 16414
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.164141893386841,
      "learning_rate": 1.0131297319745713e-05,
      "loss": 0.8625,
      "step": 16415
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7246272563934326,
      "learning_rate": 1.0130658403347924e-05,
      "loss": 0.7768,
      "step": 16416
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7497223615646362,
      "learning_rate": 1.0130019486950134e-05,
      "loss": 0.7496,
      "step": 16417
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.884021282196045,
      "learning_rate": 1.0129380570552346e-05,
      "loss": 0.9031,
      "step": 16418
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1202166080474854,
      "learning_rate": 1.0128741654154555e-05,
      "loss": 1.0692,
      "step": 16419
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3467354774475098,
      "learning_rate": 1.0128102737756767e-05,
      "loss": 0.9401,
      "step": 16420
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7261403799057007,
      "learning_rate": 1.0127463821358977e-05,
      "loss": 1.1674,
      "step": 16421
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8776880502700806,
      "learning_rate": 1.0126824904961188e-05,
      "loss": 0.8123,
      "step": 16422
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6192078590393066,
      "learning_rate": 1.0126185988563396e-05,
      "loss": 0.9308,
      "step": 16423
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.2563412189483643,
      "learning_rate": 1.012554707216561e-05,
      "loss": 1.0358,
      "step": 16424
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.592153310775757,
      "learning_rate": 1.0124908155767818e-05,
      "loss": 0.6546,
      "step": 16425
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4688117504119873,
      "learning_rate": 1.0124269239370031e-05,
      "loss": 0.7541,
      "step": 16426
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5351091623306274,
      "learning_rate": 1.012363032297224e-05,
      "loss": 0.7558,
      "step": 16427
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7413350343704224,
      "learning_rate": 1.0122991406574449e-05,
      "loss": 0.8504,
      "step": 16428
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5547385215759277,
      "learning_rate": 1.012235249017666e-05,
      "loss": 0.8665,
      "step": 16429
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8766200542449951,
      "learning_rate": 1.012171357377887e-05,
      "loss": 0.9304,
      "step": 16430
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9186197519302368,
      "learning_rate": 1.0121074657381082e-05,
      "loss": 0.5067,
      "step": 16431
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.027674913406372,
      "learning_rate": 1.0120435740983292e-05,
      "loss": 0.6645,
      "step": 16432
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.092843770980835,
      "learning_rate": 1.0119796824585503e-05,
      "loss": 0.9046,
      "step": 16433
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8670510053634644,
      "learning_rate": 1.0119157908187713e-05,
      "loss": 0.8892,
      "step": 16434
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.8984556198120117,
      "learning_rate": 1.0118518991789925e-05,
      "loss": 0.9862,
      "step": 16435
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4620264768600464,
      "learning_rate": 1.0117880075392135e-05,
      "loss": 0.7833,
      "step": 16436
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.962871551513672,
      "learning_rate": 1.0117241158994346e-05,
      "loss": 0.7609,
      "step": 16437
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6077871322631836,
      "learning_rate": 1.0116602242596556e-05,
      "loss": 0.9337,
      "step": 16438
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.2545008659362793,
      "learning_rate": 1.0115963326198768e-05,
      "loss": 0.669,
      "step": 16439
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.020664691925049,
      "learning_rate": 1.0115324409800978e-05,
      "loss": 0.9002,
      "step": 16440
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5600088834762573,
      "learning_rate": 1.0114685493403189e-05,
      "loss": 0.8371,
      "step": 16441
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9770879745483398,
      "learning_rate": 1.0114046577005399e-05,
      "loss": 0.9114,
      "step": 16442
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.668863296508789,
      "learning_rate": 1.011340766060761e-05,
      "loss": 1.1854,
      "step": 16443
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1397836208343506,
      "learning_rate": 1.011276874420982e-05,
      "loss": 0.7742,
      "step": 16444
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.816738605499268,
      "learning_rate": 1.0112129827812032e-05,
      "loss": 1.1054,
      "step": 16445
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5329670906066895,
      "learning_rate": 1.0111490911414242e-05,
      "loss": 0.9378,
      "step": 16446
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.678905963897705,
      "learning_rate": 1.0110851995016453e-05,
      "loss": 0.8932,
      "step": 16447
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.49591064453125,
      "learning_rate": 1.0110213078618663e-05,
      "loss": 0.9031,
      "step": 16448
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6452629566192627,
      "learning_rate": 1.0109574162220875e-05,
      "loss": 0.9222,
      "step": 16449
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6764118671417236,
      "learning_rate": 1.0108935245823085e-05,
      "loss": 0.7879,
      "step": 16450
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.424657106399536,
      "learning_rate": 1.0108296329425296e-05,
      "loss": 0.9634,
      "step": 16451
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5233668088912964,
      "learning_rate": 1.0107657413027506e-05,
      "loss": 0.9188,
      "step": 16452
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6895887851715088,
      "learning_rate": 1.0107018496629717e-05,
      "loss": 0.7733,
      "step": 16453
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4460115432739258,
      "learning_rate": 1.0106379580231926e-05,
      "loss": 0.4267,
      "step": 16454
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7468079328536987,
      "learning_rate": 1.0105740663834139e-05,
      "loss": 0.8978,
      "step": 16455
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.45845365524292,
      "learning_rate": 1.0105101747436347e-05,
      "loss": 0.6636,
      "step": 16456
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7267441749572754,
      "learning_rate": 1.010446283103856e-05,
      "loss": 0.6787,
      "step": 16457
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5670818090438843,
      "learning_rate": 1.0103823914640768e-05,
      "loss": 0.8647,
      "step": 16458
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.918135404586792,
      "learning_rate": 1.0103184998242982e-05,
      "loss": 0.9244,
      "step": 16459
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0306434631347656,
      "learning_rate": 1.010254608184519e-05,
      "loss": 1.2913,
      "step": 16460
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.899666428565979,
      "learning_rate": 1.0101907165447401e-05,
      "loss": 0.9913,
      "step": 16461
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1362195014953613,
      "learning_rate": 1.0101268249049611e-05,
      "loss": 0.8843,
      "step": 16462
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1070144176483154,
      "learning_rate": 1.0100629332651823e-05,
      "loss": 0.8416,
      "step": 16463
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.7033653259277344,
      "learning_rate": 1.0099990416254033e-05,
      "loss": 0.9627,
      "step": 16464
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.162333965301514,
      "learning_rate": 1.0099351499856244e-05,
      "loss": 1.0945,
      "step": 16465
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.128125190734863,
      "learning_rate": 1.0098712583458454e-05,
      "loss": 1.8844,
      "step": 16466
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7599235773086548,
      "learning_rate": 1.0098073667060666e-05,
      "loss": 0.8729,
      "step": 16467
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8449634313583374,
      "learning_rate": 1.0097434750662875e-05,
      "loss": 0.8764,
      "step": 16468
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.7002007961273193,
      "learning_rate": 1.0096795834265087e-05,
      "loss": 1.0017,
      "step": 16469
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.208543539047241,
      "learning_rate": 1.0096156917867297e-05,
      "loss": 0.8924,
      "step": 16470
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.500261068344116,
      "learning_rate": 1.0095518001469508e-05,
      "loss": 0.8668,
      "step": 16471
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7128300666809082,
      "learning_rate": 1.0094879085071718e-05,
      "loss": 0.8525,
      "step": 16472
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8748823404312134,
      "learning_rate": 1.009424016867393e-05,
      "loss": 0.8182,
      "step": 16473
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7401078939437866,
      "learning_rate": 1.009360125227614e-05,
      "loss": 0.8878,
      "step": 16474
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0050244331359863,
      "learning_rate": 1.0092962335878351e-05,
      "loss": 1.0377,
      "step": 16475
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7445790767669678,
      "learning_rate": 1.0092323419480561e-05,
      "loss": 0.8961,
      "step": 16476
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1955318450927734,
      "learning_rate": 1.0091684503082773e-05,
      "loss": 0.5819,
      "step": 16477
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6113102436065674,
      "learning_rate": 1.0091045586684982e-05,
      "loss": 0.8946,
      "step": 16478
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7410491704940796,
      "learning_rate": 1.0090406670287194e-05,
      "loss": 0.8992,
      "step": 16479
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3590316772460938,
      "learning_rate": 1.0089767753889404e-05,
      "loss": 0.8813,
      "step": 16480
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.391918659210205,
      "learning_rate": 1.0089128837491615e-05,
      "loss": 0.9194,
      "step": 16481
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8094816207885742,
      "learning_rate": 1.0088489921093825e-05,
      "loss": 0.8674,
      "step": 16482
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8818289041519165,
      "learning_rate": 1.0087851004696037e-05,
      "loss": 0.9247,
      "step": 16483
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.81552255153656,
      "learning_rate": 1.0087212088298247e-05,
      "loss": 0.9162,
      "step": 16484
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0310444831848145,
      "learning_rate": 1.0086573171900458e-05,
      "loss": 0.8274,
      "step": 16485
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0386054515838623,
      "learning_rate": 1.0085934255502668e-05,
      "loss": 0.9458,
      "step": 16486
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9908488988876343,
      "learning_rate": 1.008529533910488e-05,
      "loss": 0.899,
      "step": 16487
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.654197335243225,
      "learning_rate": 1.008465642270709e-05,
      "loss": 0.896,
      "step": 16488
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7440561056137085,
      "learning_rate": 1.0084017506309301e-05,
      "loss": 0.7845,
      "step": 16489
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5233317613601685,
      "learning_rate": 1.0083378589911509e-05,
      "loss": 0.9008,
      "step": 16490
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4120720624923706,
      "learning_rate": 1.0082739673513722e-05,
      "loss": 0.7606,
      "step": 16491
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.761475086212158,
      "learning_rate": 1.008210075711593e-05,
      "loss": 0.8636,
      "step": 16492
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9288649559020996,
      "learning_rate": 1.0081461840718144e-05,
      "loss": 0.5543,
      "step": 16493
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.163132667541504,
      "learning_rate": 1.0080822924320352e-05,
      "loss": 1.0042,
      "step": 16494
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6777281761169434,
      "learning_rate": 1.0080184007922565e-05,
      "loss": 0.8812,
      "step": 16495
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8550101518630981,
      "learning_rate": 1.0079545091524773e-05,
      "loss": 0.8479,
      "step": 16496
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.4042420387268066,
      "learning_rate": 1.0078906175126985e-05,
      "loss": 1.051,
      "step": 16497
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6131259202957153,
      "learning_rate": 1.0078267258729195e-05,
      "loss": 0.8224,
      "step": 16498
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.091876268386841,
      "learning_rate": 1.0077628342331406e-05,
      "loss": 0.7315,
      "step": 16499
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.066373348236084,
      "learning_rate": 1.0076989425933616e-05,
      "loss": 1.265,
      "step": 16500
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.1189682483673096,
      "learning_rate": 1.0076350509535828e-05,
      "loss": 1.426,
      "step": 16501
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.2673184871673584,
      "learning_rate": 1.0075711593138038e-05,
      "loss": 0.9693,
      "step": 16502
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.2384414672851562,
      "learning_rate": 1.0075072676740249e-05,
      "loss": 0.8908,
      "step": 16503
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.29440975189209,
      "learning_rate": 1.0074433760342459e-05,
      "loss": 1.0385,
      "step": 16504
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6672593355178833,
      "learning_rate": 1.007379484394467e-05,
      "loss": 0.9033,
      "step": 16505
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.974098563194275,
      "learning_rate": 1.007315592754688e-05,
      "loss": 0.8471,
      "step": 16506
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0835540294647217,
      "learning_rate": 1.0072517011149092e-05,
      "loss": 1.0369,
      "step": 16507
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7611083984375,
      "learning_rate": 1.0071878094751302e-05,
      "loss": 0.9416,
      "step": 16508
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8960365056991577,
      "learning_rate": 1.0071239178353513e-05,
      "loss": 1.004,
      "step": 16509
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.003610849380493,
      "learning_rate": 1.0070600261955723e-05,
      "loss": 1.0751,
      "step": 16510
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5705643892288208,
      "learning_rate": 1.0069961345557935e-05,
      "loss": 0.8981,
      "step": 16511
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.982414722442627,
      "learning_rate": 1.0069322429160145e-05,
      "loss": 1.5506,
      "step": 16512
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.016465663909912,
      "learning_rate": 1.0068683512762356e-05,
      "loss": 0.9235,
      "step": 16513
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6464641094207764,
      "learning_rate": 1.0068044596364566e-05,
      "loss": 0.3569,
      "step": 16514
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3300049304962158,
      "learning_rate": 1.0067405679966777e-05,
      "loss": 0.7138,
      "step": 16515
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.722810983657837,
      "learning_rate": 1.0066766763568987e-05,
      "loss": 0.7304,
      "step": 16516
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8322323560714722,
      "learning_rate": 1.0066127847171199e-05,
      "loss": 0.9096,
      "step": 16517
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0798192024230957,
      "learning_rate": 1.0065488930773409e-05,
      "loss": 0.9251,
      "step": 16518
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.205514669418335,
      "learning_rate": 1.006485001437562e-05,
      "loss": 0.9675,
      "step": 16519
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1584696769714355,
      "learning_rate": 1.006421109797783e-05,
      "loss": 1.218,
      "step": 16520
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7010360956192017,
      "learning_rate": 1.0063572181580042e-05,
      "loss": 0.9049,
      "step": 16521
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.961987257003784,
      "learning_rate": 1.0062933265182252e-05,
      "loss": 1.2038,
      "step": 16522
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7938692569732666,
      "learning_rate": 1.0062294348784463e-05,
      "loss": 0.8638,
      "step": 16523
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.2883548736572266,
      "learning_rate": 1.0061655432386673e-05,
      "loss": 1.02,
      "step": 16524
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.878519296646118,
      "learning_rate": 1.0061016515988884e-05,
      "loss": 1.023,
      "step": 16525
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.910670518875122,
      "learning_rate": 1.0060377599591094e-05,
      "loss": 0.8439,
      "step": 16526
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.876247525215149,
      "learning_rate": 1.0059738683193306e-05,
      "loss": 0.8286,
      "step": 16527
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9738590717315674,
      "learning_rate": 1.0059099766795514e-05,
      "loss": 1.1706,
      "step": 16528
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7741143703460693,
      "learning_rate": 1.0058460850397727e-05,
      "loss": 0.8775,
      "step": 16529
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1558051109313965,
      "learning_rate": 1.0057821933999935e-05,
      "loss": 0.98,
      "step": 16530
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9828972816467285,
      "learning_rate": 1.0057183017602149e-05,
      "loss": 0.8335,
      "step": 16531
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.842209815979004,
      "learning_rate": 1.0056544101204357e-05,
      "loss": 0.873,
      "step": 16532
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.8226754665374756,
      "learning_rate": 1.0055905184806568e-05,
      "loss": 0.92,
      "step": 16533
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.035918951034546,
      "learning_rate": 1.0055266268408778e-05,
      "loss": 0.9964,
      "step": 16534
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0809621810913086,
      "learning_rate": 1.005462735201099e-05,
      "loss": 1.3255,
      "step": 16535
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.070434093475342,
      "learning_rate": 1.00539884356132e-05,
      "loss": 0.9308,
      "step": 16536
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.825420618057251,
      "learning_rate": 1.0053349519215411e-05,
      "loss": 0.9692,
      "step": 16537
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6795517206192017,
      "learning_rate": 1.0052710602817621e-05,
      "loss": 0.9072,
      "step": 16538
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8935554027557373,
      "learning_rate": 1.0052071686419833e-05,
      "loss": 0.9323,
      "step": 16539
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1353654861450195,
      "learning_rate": 1.0051432770022042e-05,
      "loss": 0.8381,
      "step": 16540
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.326871156692505,
      "learning_rate": 1.0050793853624254e-05,
      "loss": 0.9339,
      "step": 16541
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.795457363128662,
      "learning_rate": 1.0050154937226464e-05,
      "loss": 1.1212,
      "step": 16542
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.711379051208496,
      "learning_rate": 1.0049516020828675e-05,
      "loss": 0.8498,
      "step": 16543
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7611474990844727,
      "learning_rate": 1.0048877104430885e-05,
      "loss": 0.8836,
      "step": 16544
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.400939464569092,
      "learning_rate": 1.0048238188033097e-05,
      "loss": 1.4009,
      "step": 16545
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4673291444778442,
      "learning_rate": 1.0047599271635307e-05,
      "loss": 0.843,
      "step": 16546
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6558688879013062,
      "learning_rate": 1.0046960355237518e-05,
      "loss": 0.9488,
      "step": 16547
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4664974212646484,
      "learning_rate": 1.0046321438839728e-05,
      "loss": 0.8347,
      "step": 16548
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9275405406951904,
      "learning_rate": 1.004568252244194e-05,
      "loss": 1.2471,
      "step": 16549
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.8548758029937744,
      "learning_rate": 1.004504360604415e-05,
      "loss": 1.1218,
      "step": 16550
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1058104038238525,
      "learning_rate": 1.0044404689646361e-05,
      "loss": 0.7927,
      "step": 16551
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9643062353134155,
      "learning_rate": 1.004376577324857e-05,
      "loss": 0.7404,
      "step": 16552
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6248366832733154,
      "learning_rate": 1.0043126856850782e-05,
      "loss": 1.0704,
      "step": 16553
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.077262878417969,
      "learning_rate": 1.0042487940452992e-05,
      "loss": 1.2437,
      "step": 16554
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6272982358932495,
      "learning_rate": 1.0041849024055204e-05,
      "loss": 0.8073,
      "step": 16555
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7181440591812134,
      "learning_rate": 1.0041210107657414e-05,
      "loss": 0.7223,
      "step": 16556
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.7445926666259766,
      "learning_rate": 1.0040571191259625e-05,
      "loss": 0.8979,
      "step": 16557
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.003767967224121,
      "learning_rate": 1.0039932274861835e-05,
      "loss": 1.1188,
      "step": 16558
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9485247135162354,
      "learning_rate": 1.0039293358464047e-05,
      "loss": 1.0647,
      "step": 16559
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6170060634613037,
      "learning_rate": 1.0038654442066256e-05,
      "loss": 0.884,
      "step": 16560
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.888611316680908,
      "learning_rate": 1.0038015525668468e-05,
      "loss": 0.5668,
      "step": 16561
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.999571681022644,
      "learning_rate": 1.0037376609270678e-05,
      "loss": 1.0043,
      "step": 16562
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7996567487716675,
      "learning_rate": 1.003673769287289e-05,
      "loss": 0.7474,
      "step": 16563
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.527602672576904,
      "learning_rate": 1.0036098776475098e-05,
      "loss": 1.3237,
      "step": 16564
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.1626296043396,
      "learning_rate": 1.003545986007731e-05,
      "loss": 1.2944,
      "step": 16565
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8661788702011108,
      "learning_rate": 1.0034820943679519e-05,
      "loss": 0.8681,
      "step": 16566
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.714879035949707,
      "learning_rate": 1.0034182027281732e-05,
      "loss": 1.2975,
      "step": 16567
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6323678493499756,
      "learning_rate": 1.003354311088394e-05,
      "loss": 1.1125,
      "step": 16568
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0308279991149902,
      "learning_rate": 1.0032904194486152e-05,
      "loss": 0.987,
      "step": 16569
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.557833433151245,
      "learning_rate": 1.0032265278088362e-05,
      "loss": 1.0332,
      "step": 16570
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.586686372756958,
      "learning_rate": 1.0031626361690573e-05,
      "loss": 0.4914,
      "step": 16571
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.548126220703125,
      "learning_rate": 1.0030987445292783e-05,
      "loss": 1.4981,
      "step": 16572
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0678389072418213,
      "learning_rate": 1.0030348528894995e-05,
      "loss": 0.8699,
      "step": 16573
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.3737387657165527,
      "learning_rate": 1.0029709612497205e-05,
      "loss": 1.2323,
      "step": 16574
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7868828773498535,
      "learning_rate": 1.0029070696099416e-05,
      "loss": 1.0989,
      "step": 16575
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.2541391849517822,
      "learning_rate": 1.0028431779701626e-05,
      "loss": 0.9044,
      "step": 16576
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6098532676696777,
      "learning_rate": 1.0027792863303837e-05,
      "loss": 1.0649,
      "step": 16577
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9917258024215698,
      "learning_rate": 1.0027153946906047e-05,
      "loss": 0.9829,
      "step": 16578
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.803524136543274,
      "learning_rate": 1.0026515030508259e-05,
      "loss": 0.9425,
      "step": 16579
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4372998476028442,
      "learning_rate": 1.0025876114110469e-05,
      "loss": 0.4236,
      "step": 16580
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8084540367126465,
      "learning_rate": 1.002523719771268e-05,
      "loss": 1.0447,
      "step": 16581
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.361547946929932,
      "learning_rate": 1.002459828131489e-05,
      "loss": 1.5321,
      "step": 16582
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5218160152435303,
      "learning_rate": 1.0023959364917102e-05,
      "loss": 1.3419,
      "step": 16583
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.8469560146331787,
      "learning_rate": 1.0023320448519312e-05,
      "loss": 0.9234,
      "step": 16584
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.661088228225708,
      "learning_rate": 1.0022681532121523e-05,
      "loss": 1.0032,
      "step": 16585
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.3729963302612305,
      "learning_rate": 1.0022042615723733e-05,
      "loss": 0.9654,
      "step": 16586
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.54850435256958,
      "learning_rate": 1.0021403699325944e-05,
      "loss": 0.9228,
      "step": 16587
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.657927989959717,
      "learning_rate": 1.0020764782928154e-05,
      "loss": 1.7552,
      "step": 16588
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.252582550048828,
      "learning_rate": 1.0020125866530366e-05,
      "loss": 0.9889,
      "step": 16589
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.4462392330169678,
      "learning_rate": 1.0019486950132576e-05,
      "loss": 0.988,
      "step": 16590
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6436880826950073,
      "learning_rate": 1.0018848033734787e-05,
      "loss": 0.793,
      "step": 16591
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.199252605438232,
      "learning_rate": 1.0018209117336997e-05,
      "loss": 1.2427,
      "step": 16592
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0192556381225586,
      "learning_rate": 1.0017570200939209e-05,
      "loss": 0.7913,
      "step": 16593
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9507001638412476,
      "learning_rate": 1.0016931284541419e-05,
      "loss": 0.986,
      "step": 16594
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.5795979499816895,
      "learning_rate": 1.001629236814363e-05,
      "loss": 1.4732,
      "step": 16595
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7207194566726685,
      "learning_rate": 1.001565345174584e-05,
      "loss": 0.8701,
      "step": 16596
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8554924726486206,
      "learning_rate": 1.0015014535348051e-05,
      "loss": 0.9005,
      "step": 16597
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4050383567810059,
      "learning_rate": 1.0014375618950261e-05,
      "loss": 0.8032,
      "step": 16598
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.683227062225342,
      "learning_rate": 1.0013736702552473e-05,
      "loss": 1.4273,
      "step": 16599
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7414302825927734,
      "learning_rate": 1.0013097786154681e-05,
      "loss": 0.8481,
      "step": 16600
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.031111717224121,
      "learning_rate": 1.0012458869756894e-05,
      "loss": 0.9928,
      "step": 16601
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3768224716186523,
      "learning_rate": 1.0011819953359102e-05,
      "loss": 0.9388,
      "step": 16602
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8204843997955322,
      "learning_rate": 1.0011181036961316e-05,
      "loss": 0.8642,
      "step": 16603
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7737443447113037,
      "learning_rate": 1.0010542120563524e-05,
      "loss": 0.9645,
      "step": 16604
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6667544841766357,
      "learning_rate": 1.0009903204165737e-05,
      "loss": 0.8193,
      "step": 16605
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.750872611999512,
      "learning_rate": 1.0009264287767945e-05,
      "loss": 1.3366,
      "step": 16606
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7237061262130737,
      "learning_rate": 1.0008625371370157e-05,
      "loss": 1.0032,
      "step": 16607
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.4156198501586914,
      "learning_rate": 1.0007986454972367e-05,
      "loss": 1.139,
      "step": 16608
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1778478622436523,
      "learning_rate": 1.0007347538574578e-05,
      "loss": 0.7224,
      "step": 16609
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9232501983642578,
      "learning_rate": 1.0006708622176788e-05,
      "loss": 0.7052,
      "step": 16610
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4918462038040161,
      "learning_rate": 1.0006069705779e-05,
      "loss": 0.932,
      "step": 16611
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.1093060970306396,
      "learning_rate": 1.000543078938121e-05,
      "loss": 0.8774,
      "step": 16612
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.292722463607788,
      "learning_rate": 1.0004791872983421e-05,
      "loss": 0.8222,
      "step": 16613
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6400713920593262,
      "learning_rate": 1.000415295658563e-05,
      "loss": 0.918,
      "step": 16614
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7018650770187378,
      "learning_rate": 1.0003514040187842e-05,
      "loss": 0.6544,
      "step": 16615
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8722903728485107,
      "learning_rate": 1.0002875123790052e-05,
      "loss": 0.8183,
      "step": 16616
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7295829057693481,
      "learning_rate": 1.0002236207392264e-05,
      "loss": 1.1306,
      "step": 16617
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.1276702880859375,
      "learning_rate": 1.0001597290994474e-05,
      "loss": 1.1874,
      "step": 16618
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.747588634490967,
      "learning_rate": 1.0000958374596685e-05,
      "loss": 1.8053,
      "step": 16619
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9378454685211182,
      "learning_rate": 1.0000319458198895e-05,
      "loss": 0.898,
      "step": 16620
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7895361185073853,
      "learning_rate": 9.999680541801105e-06,
      "loss": 1.0184,
      "step": 16621
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.9962797164917,
      "learning_rate": 9.999041625403316e-06,
      "loss": 0.774,
      "step": 16622
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.653526544570923,
      "learning_rate": 9.998402709005526e-06,
      "loss": 0.9899,
      "step": 16623
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1845626831054688,
      "learning_rate": 9.997763792607738e-06,
      "loss": 0.976,
      "step": 16624
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.5959792137146,
      "learning_rate": 9.997124876209948e-06,
      "loss": 1.547,
      "step": 16625
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7086013555526733,
      "learning_rate": 9.99648595981216e-06,
      "loss": 0.4714,
      "step": 16626
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7743103504180908,
      "learning_rate": 9.995847043414369e-06,
      "loss": 0.9163,
      "step": 16627
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7244707345962524,
      "learning_rate": 9.99520812701658e-06,
      "loss": 0.9335,
      "step": 16628
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4283941984176636,
      "learning_rate": 9.99456921061879e-06,
      "loss": 0.9121,
      "step": 16629
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.988511562347412,
      "learning_rate": 9.993930294221002e-06,
      "loss": 0.5744,
      "step": 16630
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9533861875534058,
      "learning_rate": 9.993291377823212e-06,
      "loss": 0.9101,
      "step": 16631
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1457087993621826,
      "learning_rate": 9.992652461425423e-06,
      "loss": 0.8414,
      "step": 16632
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5325045585632324,
      "learning_rate": 9.992013545027633e-06,
      "loss": 0.9157,
      "step": 16633
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6298092603683472,
      "learning_rate": 9.991374628629845e-06,
      "loss": 0.8504,
      "step": 16634
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8198819160461426,
      "learning_rate": 9.990735712232055e-06,
      "loss": 0.8019,
      "step": 16635
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8439751863479614,
      "learning_rate": 9.990096795834265e-06,
      "loss": 0.852,
      "step": 16636
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.566521644592285,
      "learning_rate": 9.989457879436476e-06,
      "loss": 1.0713,
      "step": 16637
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.940549373626709,
      "learning_rate": 9.988818963038686e-06,
      "loss": 0.9137,
      "step": 16638
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.471174716949463,
      "learning_rate": 9.988180046640897e-06,
      "loss": 0.7174,
      "step": 16639
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8692543506622314,
      "learning_rate": 9.987541130243107e-06,
      "loss": 0.9617,
      "step": 16640
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.1046836376190186,
      "learning_rate": 9.986902213845319e-06,
      "loss": 0.828,
      "step": 16641
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7220473289489746,
      "learning_rate": 9.986263297447529e-06,
      "loss": 1.0049,
      "step": 16642
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9285707473754883,
      "learning_rate": 9.98562438104974e-06,
      "loss": 0.9905,
      "step": 16643
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8371047973632812,
      "learning_rate": 9.98498546465195e-06,
      "loss": 0.8449,
      "step": 16644
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.154649019241333,
      "learning_rate": 9.984346548254162e-06,
      "loss": 0.8435,
      "step": 16645
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.038991928100586,
      "learning_rate": 9.983707631856372e-06,
      "loss": 0.9057,
      "step": 16646
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6155487298965454,
      "learning_rate": 9.983068715458583e-06,
      "loss": 0.9139,
      "step": 16647
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9291563034057617,
      "learning_rate": 9.982429799060793e-06,
      "loss": 1.0121,
      "step": 16648
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.649143934249878,
      "learning_rate": 9.981790882663004e-06,
      "loss": 0.8428,
      "step": 16649
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9504213333129883,
      "learning_rate": 9.981151966265214e-06,
      "loss": 1.0734,
      "step": 16650
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7415703535079956,
      "learning_rate": 9.980513049867426e-06,
      "loss": 0.9157,
      "step": 16651
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8611938953399658,
      "learning_rate": 9.979874133469636e-06,
      "loss": 0.8232,
      "step": 16652
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0229361057281494,
      "learning_rate": 9.979235217071847e-06,
      "loss": 0.9686,
      "step": 16653
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3325773477554321,
      "learning_rate": 9.978596300674057e-06,
      "loss": 0.7642,
      "step": 16654
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.703012228012085,
      "learning_rate": 9.977957384276267e-06,
      "loss": 0.818,
      "step": 16655
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6549237966537476,
      "learning_rate": 9.977318467878479e-06,
      "loss": 0.8882,
      "step": 16656
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.498119592666626,
      "learning_rate": 9.976679551480688e-06,
      "loss": 0.8414,
      "step": 16657
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.623396396636963,
      "learning_rate": 9.9760406350829e-06,
      "loss": 0.8774,
      "step": 16658
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.041656494140625,
      "learning_rate": 9.97540171868511e-06,
      "loss": 1.0929,
      "step": 16659
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.0679943561553955,
      "learning_rate": 9.974762802287321e-06,
      "loss": 0.8686,
      "step": 16660
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.656957983970642,
      "learning_rate": 9.974123885889531e-06,
      "loss": 0.7848,
      "step": 16661
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7125482559204102,
      "learning_rate": 9.973484969491743e-06,
      "loss": 0.8575,
      "step": 16662
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6013104915618896,
      "learning_rate": 9.972846053093953e-06,
      "loss": 0.8971,
      "step": 16663
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6972883939743042,
      "learning_rate": 9.972207136696164e-06,
      "loss": 0.9407,
      "step": 16664
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8925585746765137,
      "learning_rate": 9.971568220298374e-06,
      "loss": 1.2172,
      "step": 16665
    },
    {
      "epoch": 0.52,
      "grad_norm": 20.11966323852539,
      "learning_rate": 9.970929303900586e-06,
      "loss": 1.5115,
      "step": 16666
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9999934434890747,
      "learning_rate": 9.970290387502795e-06,
      "loss": 0.8489,
      "step": 16667
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8194799423217773,
      "learning_rate": 9.969651471105007e-06,
      "loss": 0.8155,
      "step": 16668
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6787115335464478,
      "learning_rate": 9.969012554707217e-06,
      "loss": 0.3356,
      "step": 16669
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.077153444290161,
      "learning_rate": 9.968373638309428e-06,
      "loss": 0.9923,
      "step": 16670
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.800002098083496,
      "learning_rate": 9.967734721911638e-06,
      "loss": 0.9402,
      "step": 16671
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.7953908443450928,
      "learning_rate": 9.96709580551385e-06,
      "loss": 0.8542,
      "step": 16672
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.10990047454834,
      "learning_rate": 9.96645688911606e-06,
      "loss": 0.9419,
      "step": 16673
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.021657943725586,
      "learning_rate": 9.96581797271827e-06,
      "loss": 1.0619,
      "step": 16674
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.968957543373108,
      "learning_rate": 9.965179056320481e-06,
      "loss": 0.8798,
      "step": 16675
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5105162858963013,
      "learning_rate": 9.964540139922691e-06,
      "loss": 0.8342,
      "step": 16676
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7071268558502197,
      "learning_rate": 9.963901223524902e-06,
      "loss": 0.8535,
      "step": 16677
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1724159717559814,
      "learning_rate": 9.963262307127112e-06,
      "loss": 0.9412,
      "step": 16678
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.708560585975647,
      "learning_rate": 9.962623390729324e-06,
      "loss": 0.7652,
      "step": 16679
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9714000225067139,
      "learning_rate": 9.961984474331534e-06,
      "loss": 1.0536,
      "step": 16680
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.026148796081543,
      "learning_rate": 9.961345557933745e-06,
      "loss": 0.8664,
      "step": 16681
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7350705862045288,
      "learning_rate": 9.960706641535955e-06,
      "loss": 0.8701,
      "step": 16682
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1662182807922363,
      "learning_rate": 9.960067725138167e-06,
      "loss": 0.9205,
      "step": 16683
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6208857297897339,
      "learning_rate": 9.959428808740376e-06,
      "loss": 0.9295,
      "step": 16684
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0250399112701416,
      "learning_rate": 9.958789892342588e-06,
      "loss": 0.9529,
      "step": 16685
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9108433723449707,
      "learning_rate": 9.958150975944798e-06,
      "loss": 0.8446,
      "step": 16686
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7959470748901367,
      "learning_rate": 9.95751205954701e-06,
      "loss": 0.7804,
      "step": 16687
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6243590116500854,
      "learning_rate": 9.95687314314922e-06,
      "loss": 0.9218,
      "step": 16688
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5597056150436401,
      "learning_rate": 9.95623422675143e-06,
      "loss": 0.9174,
      "step": 16689
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8726308345794678,
      "learning_rate": 9.95559531035364e-06,
      "loss": 0.8334,
      "step": 16690
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.988932728767395,
      "learning_rate": 9.95495639395585e-06,
      "loss": 0.9381,
      "step": 16691
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2163405418395996,
      "learning_rate": 9.954317477558062e-06,
      "loss": 0.812,
      "step": 16692
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7952669858932495,
      "learning_rate": 9.953678561160272e-06,
      "loss": 0.8526,
      "step": 16693
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6707549095153809,
      "learning_rate": 9.953039644762483e-06,
      "loss": 0.845,
      "step": 16694
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5086586475372314,
      "learning_rate": 9.952400728364693e-06,
      "loss": 0.8467,
      "step": 16695
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5966905355453491,
      "learning_rate": 9.951761811966905e-06,
      "loss": 0.8123,
      "step": 16696
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6399195194244385,
      "learning_rate": 9.951122895569115e-06,
      "loss": 0.4059,
      "step": 16697
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0611937046051025,
      "learning_rate": 9.950483979171326e-06,
      "loss": 0.8476,
      "step": 16698
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8948348760604858,
      "learning_rate": 9.949845062773536e-06,
      "loss": 0.8428,
      "step": 16699
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.3535730838775635,
      "learning_rate": 9.949206146375748e-06,
      "loss": 1.0224,
      "step": 16700
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9044232368469238,
      "learning_rate": 9.948567229977958e-06,
      "loss": 0.9855,
      "step": 16701
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.178497791290283,
      "learning_rate": 9.947928313580169e-06,
      "loss": 1.0291,
      "step": 16702
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1173107624053955,
      "learning_rate": 9.947289397182379e-06,
      "loss": 0.9782,
      "step": 16703
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0653605461120605,
      "learning_rate": 9.94665048078459e-06,
      "loss": 0.9941,
      "step": 16704
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.627476453781128,
      "learning_rate": 9.9460115643868e-06,
      "loss": 0.789,
      "step": 16705
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.854099154472351,
      "learning_rate": 9.945372647989012e-06,
      "loss": 1.2005,
      "step": 16706
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.896788477897644,
      "learning_rate": 9.944733731591222e-06,
      "loss": 0.9478,
      "step": 16707
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.398771047592163,
      "learning_rate": 9.944094815193433e-06,
      "loss": 0.8582,
      "step": 16708
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.4538798332214355,
      "learning_rate": 9.943455898795643e-06,
      "loss": 0.7982,
      "step": 16709
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5896886587142944,
      "learning_rate": 9.942816982397853e-06,
      "loss": 0.8864,
      "step": 16710
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.248030424118042,
      "learning_rate": 9.942178066000065e-06,
      "loss": 1.0138,
      "step": 16711
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2797775268554688,
      "learning_rate": 9.941539149602274e-06,
      "loss": 0.8589,
      "step": 16712
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9695096015930176,
      "learning_rate": 9.940900233204486e-06,
      "loss": 0.9971,
      "step": 16713
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.559549570083618,
      "learning_rate": 9.940261316806696e-06,
      "loss": 0.7566,
      "step": 16714
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5761927366256714,
      "learning_rate": 9.939622400408907e-06,
      "loss": 0.8219,
      "step": 16715
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5634695291519165,
      "learning_rate": 9.938983484011117e-06,
      "loss": 0.8918,
      "step": 16716
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5361541509628296,
      "learning_rate": 9.938344567613329e-06,
      "loss": 0.7766,
      "step": 16717
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7338902950286865,
      "learning_rate": 9.937705651215539e-06,
      "loss": 0.9468,
      "step": 16718
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9879406690597534,
      "learning_rate": 9.93706673481775e-06,
      "loss": 0.9612,
      "step": 16719
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8548288345336914,
      "learning_rate": 9.93642781841996e-06,
      "loss": 0.9207,
      "step": 16720
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.778057336807251,
      "learning_rate": 9.935788902022172e-06,
      "loss": 0.9104,
      "step": 16721
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8638861179351807,
      "learning_rate": 9.935149985624381e-06,
      "loss": 0.9463,
      "step": 16722
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.861238956451416,
      "learning_rate": 9.934511069226593e-06,
      "loss": 0.9044,
      "step": 16723
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8582090139389038,
      "learning_rate": 9.933872152828803e-06,
      "loss": 0.9674,
      "step": 16724
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.869406819343567,
      "learning_rate": 9.933233236431014e-06,
      "loss": 0.9684,
      "step": 16725
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.198570728302002,
      "learning_rate": 9.932594320033224e-06,
      "loss": 0.996,
      "step": 16726
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0161280632019043,
      "learning_rate": 9.931955403635436e-06,
      "loss": 1.2174,
      "step": 16727
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7647346258163452,
      "learning_rate": 9.931316487237646e-06,
      "loss": 0.9009,
      "step": 16728
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8685230016708374,
      "learning_rate": 9.930677570839855e-06,
      "loss": 0.9079,
      "step": 16729
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4841856956481934,
      "learning_rate": 9.930038654442067e-06,
      "loss": 0.896,
      "step": 16730
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8176137208938599,
      "learning_rate": 9.929399738044277e-06,
      "loss": 1.2095,
      "step": 16731
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7565929889678955,
      "learning_rate": 9.928760821646488e-06,
      "loss": 0.7577,
      "step": 16732
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1415646076202393,
      "learning_rate": 9.928121905248698e-06,
      "loss": 0.8227,
      "step": 16733
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.094942569732666,
      "learning_rate": 9.92748298885091e-06,
      "loss": 0.9392,
      "step": 16734
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.750183582305908,
      "learning_rate": 9.92684407245312e-06,
      "loss": 1.1335,
      "step": 16735
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2474169731140137,
      "learning_rate": 9.926205156055331e-06,
      "loss": 1.0071,
      "step": 16736
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.8367342948913574,
      "learning_rate": 9.925566239657541e-06,
      "loss": 0.6488,
      "step": 16737
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.679647445678711,
      "learning_rate": 9.924927323259753e-06,
      "loss": 0.9346,
      "step": 16738
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.163261651992798,
      "learning_rate": 9.924288406861962e-06,
      "loss": 0.7334,
      "step": 16739
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6896038055419922,
      "learning_rate": 9.923649490464174e-06,
      "loss": 0.7933,
      "step": 16740
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6901648044586182,
      "learning_rate": 9.923010574066384e-06,
      "loss": 0.8379,
      "step": 16741
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.197520732879639,
      "learning_rate": 9.922371657668595e-06,
      "loss": 1.4168,
      "step": 16742
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.698529839515686,
      "learning_rate": 9.921732741270805e-06,
      "loss": 0.887,
      "step": 16743
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8526155948638916,
      "learning_rate": 9.921093824873017e-06,
      "loss": 0.9065,
      "step": 16744
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.588717937469482,
      "learning_rate": 9.920454908475227e-06,
      "loss": 1.4359,
      "step": 16745
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9277808666229248,
      "learning_rate": 9.919815992077436e-06,
      "loss": 0.8911,
      "step": 16746
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.287997007369995,
      "learning_rate": 9.919177075679648e-06,
      "loss": 0.9522,
      "step": 16747
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.693922996520996,
      "learning_rate": 9.918538159281858e-06,
      "loss": 0.8263,
      "step": 16748
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.932241916656494,
      "learning_rate": 9.91789924288407e-06,
      "loss": 2.2328,
      "step": 16749
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.181522846221924,
      "learning_rate": 9.91726032648628e-06,
      "loss": 0.9236,
      "step": 16750
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.991714358329773,
      "learning_rate": 9.91662141008849e-06,
      "loss": 0.9539,
      "step": 16751
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8239426612854004,
      "learning_rate": 9.9159824936907e-06,
      "loss": 0.9434,
      "step": 16752
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.277270793914795,
      "learning_rate": 9.915343577292912e-06,
      "loss": 0.6905,
      "step": 16753
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5310124158859253,
      "learning_rate": 9.914704660895122e-06,
      "loss": 0.8647,
      "step": 16754
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8641303777694702,
      "learning_rate": 9.914065744497334e-06,
      "loss": 0.8901,
      "step": 16755
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.055311679840088,
      "learning_rate": 9.913426828099543e-06,
      "loss": 0.7486,
      "step": 16756
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1606459617614746,
      "learning_rate": 9.912787911701755e-06,
      "loss": 1.1791,
      "step": 16757
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.22672176361084,
      "learning_rate": 9.912148995303965e-06,
      "loss": 1.0578,
      "step": 16758
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8548470735549927,
      "learning_rate": 9.911510078906176e-06,
      "loss": 0.8217,
      "step": 16759
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.864869475364685,
      "learning_rate": 9.910871162508386e-06,
      "loss": 0.9491,
      "step": 16760
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5872366428375244,
      "learning_rate": 9.910232246110598e-06,
      "loss": 0.7919,
      "step": 16761
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.660077452659607,
      "learning_rate": 9.909593329712808e-06,
      "loss": 0.8617,
      "step": 16762
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2289412021636963,
      "learning_rate": 9.90895441331502e-06,
      "loss": 0.8083,
      "step": 16763
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5322158336639404,
      "learning_rate": 9.908315496917229e-06,
      "loss": 0.8337,
      "step": 16764
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.663503646850586,
      "learning_rate": 9.907676580519439e-06,
      "loss": 0.8743,
      "step": 16765
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.3410351276397705,
      "learning_rate": 9.90703766412165e-06,
      "loss": 0.5635,
      "step": 16766
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1105504035949707,
      "learning_rate": 9.90639874772386e-06,
      "loss": 0.9491,
      "step": 16767
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6553242206573486,
      "learning_rate": 9.905759831326072e-06,
      "loss": 0.4409,
      "step": 16768
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6159840822219849,
      "learning_rate": 9.905120914928282e-06,
      "loss": 0.8443,
      "step": 16769
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7001633644104004,
      "learning_rate": 9.904481998530493e-06,
      "loss": 0.8887,
      "step": 16770
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0284149646759033,
      "learning_rate": 9.903843082132703e-06,
      "loss": 1.057,
      "step": 16771
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0958776473999023,
      "learning_rate": 9.903204165734915e-06,
      "loss": 0.908,
      "step": 16772
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6587624549865723,
      "learning_rate": 9.902565249337125e-06,
      "loss": 0.4024,
      "step": 16773
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5477497577667236,
      "learning_rate": 9.901926332939336e-06,
      "loss": 0.9519,
      "step": 16774
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.630600452423096,
      "learning_rate": 9.901287416541546e-06,
      "loss": 1.3378,
      "step": 16775
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4383412599563599,
      "learning_rate": 9.900648500143757e-06,
      "loss": 0.703,
      "step": 16776
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.545950412750244,
      "learning_rate": 9.900009583745967e-06,
      "loss": 2.0238,
      "step": 16777
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4862091541290283,
      "learning_rate": 9.899370667348179e-06,
      "loss": 0.8556,
      "step": 16778
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.4009804725646973,
      "learning_rate": 9.898731750950389e-06,
      "loss": 0.4955,
      "step": 16779
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8841179609298706,
      "learning_rate": 9.8980928345526e-06,
      "loss": 0.9195,
      "step": 16780
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.3081581592559814,
      "learning_rate": 9.89745391815481e-06,
      "loss": 0.9337,
      "step": 16781
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0342936515808105,
      "learning_rate": 9.89681500175702e-06,
      "loss": 0.9494,
      "step": 16782
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.6485750675201416,
      "learning_rate": 9.896176085359232e-06,
      "loss": 0.9941,
      "step": 16783
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.4521257877349854,
      "learning_rate": 9.895537168961441e-06,
      "loss": 1.0996,
      "step": 16784
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.847248077392578,
      "learning_rate": 9.894898252563653e-06,
      "loss": 1.2151,
      "step": 16785
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6947377920150757,
      "learning_rate": 9.894259336165863e-06,
      "loss": 0.9108,
      "step": 16786
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.7086679935455322,
      "learning_rate": 9.893620419768074e-06,
      "loss": 0.8631,
      "step": 16787
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0662107467651367,
      "learning_rate": 9.892981503370284e-06,
      "loss": 1.0187,
      "step": 16788
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.168320655822754,
      "learning_rate": 9.892342586972496e-06,
      "loss": 1.2645,
      "step": 16789
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.319010019302368,
      "learning_rate": 9.891703670574706e-06,
      "loss": 0.8751,
      "step": 16790
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5537631511688232,
      "learning_rate": 9.891064754176917e-06,
      "loss": 0.8034,
      "step": 16791
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.501796245574951,
      "learning_rate": 9.890425837779127e-06,
      "loss": 1.0949,
      "step": 16792
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.3697917461395264,
      "learning_rate": 9.889786921381339e-06,
      "loss": 0.9041,
      "step": 16793
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.284043550491333,
      "learning_rate": 9.889148004983548e-06,
      "loss": 1.2852,
      "step": 16794
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.738071084022522,
      "learning_rate": 9.88850908858576e-06,
      "loss": 0.9025,
      "step": 16795
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.534567952156067,
      "learning_rate": 9.88787017218797e-06,
      "loss": 0.8372,
      "step": 16796
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2054598331451416,
      "learning_rate": 9.887231255790181e-06,
      "loss": 0.7965,
      "step": 16797
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7714725732803345,
      "learning_rate": 9.886592339392391e-06,
      "loss": 0.9664,
      "step": 16798
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7024149894714355,
      "learning_rate": 9.885953422994603e-06,
      "loss": 0.8797,
      "step": 16799
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9252188205718994,
      "learning_rate": 9.885314506596813e-06,
      "loss": 0.8892,
      "step": 16800
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5346992015838623,
      "learning_rate": 9.884675590199022e-06,
      "loss": 0.8085,
      "step": 16801
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.105466365814209,
      "learning_rate": 9.884036673801234e-06,
      "loss": 0.9767,
      "step": 16802
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.506505250930786,
      "learning_rate": 9.883397757403444e-06,
      "loss": 0.8804,
      "step": 16803
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4726040363311768,
      "learning_rate": 9.882758841005655e-06,
      "loss": 0.8859,
      "step": 16804
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8934544324874878,
      "learning_rate": 9.882119924607865e-06,
      "loss": 1.2687,
      "step": 16805
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5234485864639282,
      "learning_rate": 9.881481008210077e-06,
      "loss": 0.8465,
      "step": 16806
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.900915265083313,
      "learning_rate": 9.880842091812287e-06,
      "loss": 0.9777,
      "step": 16807
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9416406154632568,
      "learning_rate": 9.880203175414498e-06,
      "loss": 0.956,
      "step": 16808
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9709035158157349,
      "learning_rate": 9.879564259016708e-06,
      "loss": 0.8919,
      "step": 16809
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.853498935699463,
      "learning_rate": 9.87892534261892e-06,
      "loss": 0.9098,
      "step": 16810
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.028221368789673,
      "learning_rate": 9.87828642622113e-06,
      "loss": 1.0283,
      "step": 16811
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8076279163360596,
      "learning_rate": 9.877647509823341e-06,
      "loss": 0.8625,
      "step": 16812
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9413704872131348,
      "learning_rate": 9.87700859342555e-06,
      "loss": 0.9823,
      "step": 16813
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6863845586776733,
      "learning_rate": 9.876369677027762e-06,
      "loss": 0.8753,
      "step": 16814
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6786068677902222,
      "learning_rate": 9.875730760629972e-06,
      "loss": 0.7657,
      "step": 16815
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2021636962890625,
      "learning_rate": 9.875091844232184e-06,
      "loss": 1.001,
      "step": 16816
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.959918975830078,
      "learning_rate": 9.874452927834394e-06,
      "loss": 0.8195,
      "step": 16817
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8179408311843872,
      "learning_rate": 9.873814011436605e-06,
      "loss": 0.8784,
      "step": 16818
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8861887454986572,
      "learning_rate": 9.873175095038815e-06,
      "loss": 1.1213,
      "step": 16819
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3896113634109497,
      "learning_rate": 9.872536178641025e-06,
      "loss": 0.7029,
      "step": 16820
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0737478733062744,
      "learning_rate": 9.871897262243236e-06,
      "loss": 0.9978,
      "step": 16821
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9959070682525635,
      "learning_rate": 9.871258345845446e-06,
      "loss": 1.069,
      "step": 16822
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.3860056400299072,
      "learning_rate": 9.870619429447658e-06,
      "loss": 0.8459,
      "step": 16823
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.809131145477295,
      "learning_rate": 9.869980513049868e-06,
      "loss": 0.8816,
      "step": 16824
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8690338134765625,
      "learning_rate": 9.86934159665208e-06,
      "loss": 0.9048,
      "step": 16825
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8281797170639038,
      "learning_rate": 9.868702680254289e-06,
      "loss": 1.1782,
      "step": 16826
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5212453603744507,
      "learning_rate": 9.8680637638565e-06,
      "loss": 0.3713,
      "step": 16827
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5888671875,
      "learning_rate": 9.86742484745871e-06,
      "loss": 0.8286,
      "step": 16828
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4270793199539185,
      "learning_rate": 9.866785931060922e-06,
      "loss": 0.4481,
      "step": 16829
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6798741817474365,
      "learning_rate": 9.866147014663132e-06,
      "loss": 0.9233,
      "step": 16830
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6671279668807983,
      "learning_rate": 9.865508098265343e-06,
      "loss": 0.9296,
      "step": 16831
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.849392056465149,
      "learning_rate": 9.864869181867553e-06,
      "loss": 0.8683,
      "step": 16832
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5046859979629517,
      "learning_rate": 9.864230265469765e-06,
      "loss": 0.9034,
      "step": 16833
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6667497158050537,
      "learning_rate": 9.863591349071975e-06,
      "loss": 0.9027,
      "step": 16834
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9511314630508423,
      "learning_rate": 9.862952432674186e-06,
      "loss": 1.324,
      "step": 16835
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.183047294616699,
      "learning_rate": 9.862313516276396e-06,
      "loss": 0.8827,
      "step": 16836
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.5273284912109375,
      "learning_rate": 9.861674599878606e-06,
      "loss": 1.5535,
      "step": 16837
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.603658676147461,
      "learning_rate": 9.861035683480817e-06,
      "loss": 1.003,
      "step": 16838
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.3545632362365723,
      "learning_rate": 9.860396767083027e-06,
      "loss": 0.74,
      "step": 16839
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5758640766143799,
      "learning_rate": 9.859757850685239e-06,
      "loss": 0.7704,
      "step": 16840
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9718478918075562,
      "learning_rate": 9.859118934287449e-06,
      "loss": 0.8418,
      "step": 16841
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.286140203475952,
      "learning_rate": 9.85848001788966e-06,
      "loss": 0.7448,
      "step": 16842
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9534575939178467,
      "learning_rate": 9.85784110149187e-06,
      "loss": 0.9631,
      "step": 16843
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6740726232528687,
      "learning_rate": 9.857202185094082e-06,
      "loss": 1.2016,
      "step": 16844
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.85282301902771,
      "learning_rate": 9.856563268696292e-06,
      "loss": 0.8792,
      "step": 16845
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0171918869018555,
      "learning_rate": 9.855924352298503e-06,
      "loss": 0.9244,
      "step": 16846
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6515157222747803,
      "learning_rate": 9.855285435900713e-06,
      "loss": 0.9515,
      "step": 16847
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2598557472229004,
      "learning_rate": 9.854646519502924e-06,
      "loss": 0.8837,
      "step": 16848
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.379939317703247,
      "learning_rate": 9.854007603105134e-06,
      "loss": 0.7247,
      "step": 16849
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5796493291854858,
      "learning_rate": 9.853368686707346e-06,
      "loss": 0.8506,
      "step": 16850
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.922680616378784,
      "learning_rate": 9.852729770309556e-06,
      "loss": 1.2891,
      "step": 16851
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.46388840675354,
      "learning_rate": 9.852090853911767e-06,
      "loss": 0.9285,
      "step": 16852
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.297347068786621,
      "learning_rate": 9.851451937513977e-06,
      "loss": 1.0894,
      "step": 16853
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7504074573516846,
      "learning_rate": 9.850813021116189e-06,
      "loss": 1.0183,
      "step": 16854
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6237561702728271,
      "learning_rate": 9.850174104718399e-06,
      "loss": 0.7819,
      "step": 16855
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.327932357788086,
      "learning_rate": 9.849535188320608e-06,
      "loss": 0.8497,
      "step": 16856
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.2398908138275146,
      "learning_rate": 9.84889627192282e-06,
      "loss": 0.6529,
      "step": 16857
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.187358856201172,
      "learning_rate": 9.84825735552503e-06,
      "loss": 0.7207,
      "step": 16858
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9918134212493896,
      "learning_rate": 9.847618439127241e-06,
      "loss": 0.9708,
      "step": 16859
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6423369646072388,
      "learning_rate": 9.846979522729451e-06,
      "loss": 0.8992,
      "step": 16860
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.931809425354004,
      "learning_rate": 9.846340606331663e-06,
      "loss": 0.9641,
      "step": 16861
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.702955722808838,
      "learning_rate": 9.845701689933873e-06,
      "loss": 0.8717,
      "step": 16862
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0842843055725098,
      "learning_rate": 9.845062773536084e-06,
      "loss": 0.7546,
      "step": 16863
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4755783081054688,
      "learning_rate": 9.844423857138294e-06,
      "loss": 0.3875,
      "step": 16864
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.295382499694824,
      "learning_rate": 9.843784940740506e-06,
      "loss": 1.2646,
      "step": 16865
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7094124555587769,
      "learning_rate": 9.843146024342715e-06,
      "loss": 0.8691,
      "step": 16866
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.4459311962127686,
      "learning_rate": 9.842507107944927e-06,
      "loss": 0.9143,
      "step": 16867
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8369488716125488,
      "learning_rate": 9.841868191547137e-06,
      "loss": 0.9588,
      "step": 16868
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.119236707687378,
      "learning_rate": 9.841229275149348e-06,
      "loss": 0.468,
      "step": 16869
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.052137851715088,
      "learning_rate": 9.840590358751558e-06,
      "loss": 0.9132,
      "step": 16870
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5130188465118408,
      "learning_rate": 9.83995144235377e-06,
      "loss": 0.8376,
      "step": 16871
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5580754280090332,
      "learning_rate": 9.83931252595598e-06,
      "loss": 0.8649,
      "step": 16872
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.836009502410889,
      "learning_rate": 9.838673609558191e-06,
      "loss": 1.7305,
      "step": 16873
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5075892210006714,
      "learning_rate": 9.838034693160401e-06,
      "loss": 0.8645,
      "step": 16874
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6253416538238525,
      "learning_rate": 9.83739577676261e-06,
      "loss": 0.9363,
      "step": 16875
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5730544328689575,
      "learning_rate": 9.836756860364822e-06,
      "loss": 0.8638,
      "step": 16876
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9977089166641235,
      "learning_rate": 9.836117943967032e-06,
      "loss": 0.9784,
      "step": 16877
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7287309169769287,
      "learning_rate": 9.835479027569244e-06,
      "loss": 0.9599,
      "step": 16878
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7543346881866455,
      "learning_rate": 9.834840111171454e-06,
      "loss": 1.0307,
      "step": 16879
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1413960456848145,
      "learning_rate": 9.834201194773665e-06,
      "loss": 1.085,
      "step": 16880
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.173304557800293,
      "learning_rate": 9.833562278375875e-06,
      "loss": 1.0443,
      "step": 16881
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1044750213623047,
      "learning_rate": 9.832923361978087e-06,
      "loss": 0.8923,
      "step": 16882
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5514118671417236,
      "learning_rate": 9.832284445580296e-06,
      "loss": 0.8724,
      "step": 16883
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.0284247398376465,
      "learning_rate": 9.831645529182508e-06,
      "loss": 1.0579,
      "step": 16884
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5018209218978882,
      "learning_rate": 9.831006612784718e-06,
      "loss": 0.3905,
      "step": 16885
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.019681692123413,
      "learning_rate": 9.83036769638693e-06,
      "loss": 0.8935,
      "step": 16886
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.647284746170044,
      "learning_rate": 9.82972877998914e-06,
      "loss": 0.8684,
      "step": 16887
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6691014766693115,
      "learning_rate": 9.82908986359135e-06,
      "loss": 0.9732,
      "step": 16888
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.618299961090088,
      "learning_rate": 9.82845094719356e-06,
      "loss": 1.6104,
      "step": 16889
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7515400648117065,
      "learning_rate": 9.827812030795772e-06,
      "loss": 1.0799,
      "step": 16890
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7993168830871582,
      "learning_rate": 9.827173114397982e-06,
      "loss": 1.1292,
      "step": 16891
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8835508823394775,
      "learning_rate": 9.826534198000192e-06,
      "loss": 0.8489,
      "step": 16892
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7074872255325317,
      "learning_rate": 9.825895281602403e-06,
      "loss": 0.9497,
      "step": 16893
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6322287321090698,
      "learning_rate": 9.825256365204613e-06,
      "loss": 0.9239,
      "step": 16894
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.8653693199157715,
      "learning_rate": 9.824617448806825e-06,
      "loss": 1.0181,
      "step": 16895
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.9176058769226074,
      "learning_rate": 9.823978532409035e-06,
      "loss": 0.8431,
      "step": 16896
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.502366542816162,
      "learning_rate": 9.823339616011246e-06,
      "loss": 0.8629,
      "step": 16897
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7057979106903076,
      "learning_rate": 9.822700699613456e-06,
      "loss": 0.68,
      "step": 16898
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.009364366531372,
      "learning_rate": 9.822061783215668e-06,
      "loss": 1.1784,
      "step": 16899
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7391130924224854,
      "learning_rate": 9.821422866817877e-06,
      "loss": 0.9129,
      "step": 16900
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7390391826629639,
      "learning_rate": 9.820783950420089e-06,
      "loss": 0.798,
      "step": 16901
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.7343368530273438,
      "learning_rate": 9.820145034022299e-06,
      "loss": 0.6538,
      "step": 16902
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9996514320373535,
      "learning_rate": 9.81950611762451e-06,
      "loss": 0.8801,
      "step": 16903
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0894615650177,
      "learning_rate": 9.81886720122672e-06,
      "loss": 0.968,
      "step": 16904
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8220943212509155,
      "learning_rate": 9.818228284828932e-06,
      "loss": 0.9237,
      "step": 16905
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.9107816219329834,
      "learning_rate": 9.817589368431142e-06,
      "loss": 0.7942,
      "step": 16906
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8238732814788818,
      "learning_rate": 9.816950452033353e-06,
      "loss": 0.8218,
      "step": 16907
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7301911115646362,
      "learning_rate": 9.816311535635563e-06,
      "loss": 0.7021,
      "step": 16908
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.008004903793335,
      "learning_rate": 9.815672619237775e-06,
      "loss": 1.1227,
      "step": 16909
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5114402770996094,
      "learning_rate": 9.815033702839984e-06,
      "loss": 0.7712,
      "step": 16910
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.4870786666870117,
      "learning_rate": 9.814394786442194e-06,
      "loss": 0.8971,
      "step": 16911
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.286379337310791,
      "learning_rate": 9.813755870044406e-06,
      "loss": 0.5397,
      "step": 16912
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.514360785484314,
      "learning_rate": 9.813116953646616e-06,
      "loss": 0.7911,
      "step": 16913
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7125189304351807,
      "learning_rate": 9.812478037248827e-06,
      "loss": 0.9816,
      "step": 16914
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.753116250038147,
      "learning_rate": 9.811839120851037e-06,
      "loss": 0.8964,
      "step": 16915
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.086376428604126,
      "learning_rate": 9.811200204453249e-06,
      "loss": 1.0232,
      "step": 16916
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.0838840007781982,
      "learning_rate": 9.810561288055459e-06,
      "loss": 0.7088,
      "step": 16917
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.021749973297119,
      "learning_rate": 9.80992237165767e-06,
      "loss": 0.951,
      "step": 16918
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7631168365478516,
      "learning_rate": 9.80928345525988e-06,
      "loss": 0.9626,
      "step": 16919
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5479989051818848,
      "learning_rate": 9.808644538862091e-06,
      "loss": 0.8725,
      "step": 16920
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.129260778427124,
      "learning_rate": 9.808005622464301e-06,
      "loss": 0.4471,
      "step": 16921
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.3590660095214844,
      "learning_rate": 9.807366706066513e-06,
      "loss": 0.8678,
      "step": 16922
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.563098669052124,
      "learning_rate": 9.806727789668723e-06,
      "loss": 0.8586,
      "step": 16923
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2433323860168457,
      "learning_rate": 9.806088873270934e-06,
      "loss": 0.9486,
      "step": 16924
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8596059083938599,
      "learning_rate": 9.805449956873144e-06,
      "loss": 0.9895,
      "step": 16925
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6381709575653076,
      "learning_rate": 9.804811040475356e-06,
      "loss": 0.4894,
      "step": 16926
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5316851139068604,
      "learning_rate": 9.804172124077566e-06,
      "loss": 0.777,
      "step": 16927
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.71099853515625,
      "learning_rate": 9.803533207679775e-06,
      "loss": 0.8037,
      "step": 16928
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.457245349884033,
      "learning_rate": 9.802894291281987e-06,
      "loss": 1.366,
      "step": 16929
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.329103946685791,
      "learning_rate": 9.802255374884197e-06,
      "loss": 1.1959,
      "step": 16930
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7874548435211182,
      "learning_rate": 9.801616458486408e-06,
      "loss": 0.8248,
      "step": 16931
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.315336227416992,
      "learning_rate": 9.800977542088618e-06,
      "loss": 1.243,
      "step": 16932
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.6441714763641357,
      "learning_rate": 9.80033862569083e-06,
      "loss": 0.9212,
      "step": 16933
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5956913232803345,
      "learning_rate": 9.79969970929304e-06,
      "loss": 0.7401,
      "step": 16934
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9386839866638184,
      "learning_rate": 9.799060792895251e-06,
      "loss": 0.9123,
      "step": 16935
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.117629051208496,
      "learning_rate": 9.798421876497461e-06,
      "loss": 0.9234,
      "step": 16936
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8789620399475098,
      "learning_rate": 9.797782960099673e-06,
      "loss": 0.7998,
      "step": 16937
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.36061692237854,
      "learning_rate": 9.797144043701882e-06,
      "loss": 0.7983,
      "step": 16938
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4506500959396362,
      "learning_rate": 9.796505127304094e-06,
      "loss": 0.8183,
      "step": 16939
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2336783409118652,
      "learning_rate": 9.795866210906304e-06,
      "loss": 1.0218,
      "step": 16940
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.13619327545166,
      "learning_rate": 9.795227294508515e-06,
      "loss": 0.859,
      "step": 16941
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0778563022613525,
      "learning_rate": 9.794588378110725e-06,
      "loss": 0.5943,
      "step": 16942
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.727441787719727,
      "learning_rate": 9.793949461712937e-06,
      "loss": 1.3792,
      "step": 16943
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5947521924972534,
      "learning_rate": 9.793310545315147e-06,
      "loss": 0.8874,
      "step": 16944
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.043515205383301,
      "learning_rate": 9.792671628917358e-06,
      "loss": 1.0613,
      "step": 16945
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7081435918807983,
      "learning_rate": 9.792032712519568e-06,
      "loss": 0.8991,
      "step": 16946
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6079970598220825,
      "learning_rate": 9.791393796121778e-06,
      "loss": 0.8339,
      "step": 16947
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7986071109771729,
      "learning_rate": 9.79075487972399e-06,
      "loss": 0.8527,
      "step": 16948
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.738000750541687,
      "learning_rate": 9.7901159633262e-06,
      "loss": 0.9271,
      "step": 16949
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.756997585296631,
      "learning_rate": 9.78947704692841e-06,
      "loss": 1.1918,
      "step": 16950
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8023375272750854,
      "learning_rate": 9.78883813053062e-06,
      "loss": 0.8641,
      "step": 16951
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3153600692749023,
      "learning_rate": 9.788199214132832e-06,
      "loss": 1.0393,
      "step": 16952
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.0844571590423584,
      "learning_rate": 9.787560297735042e-06,
      "loss": 0.9531,
      "step": 16953
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8671960830688477,
      "learning_rate": 9.786921381337254e-06,
      "loss": 0.8655,
      "step": 16954
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.6225767135620117,
      "learning_rate": 9.786282464939463e-06,
      "loss": 1.039,
      "step": 16955
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8303470611572266,
      "learning_rate": 9.785643548541675e-06,
      "loss": 0.9665,
      "step": 16956
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7438545227050781,
      "learning_rate": 9.785004632143885e-06,
      "loss": 0.8921,
      "step": 16957
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.5689189434051514,
      "learning_rate": 9.784365715746096e-06,
      "loss": 0.6732,
      "step": 16958
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.910776138305664,
      "learning_rate": 9.783726799348306e-06,
      "loss": 0.987,
      "step": 16959
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8118059635162354,
      "learning_rate": 9.783087882950518e-06,
      "loss": 0.8475,
      "step": 16960
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8249555826187134,
      "learning_rate": 9.782448966552728e-06,
      "loss": 0.9047,
      "step": 16961
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.932809591293335,
      "learning_rate": 9.78181005015494e-06,
      "loss": 0.8778,
      "step": 16962
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0554745197296143,
      "learning_rate": 9.781171133757149e-06,
      "loss": 0.8179,
      "step": 16963
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0506787300109863,
      "learning_rate": 9.78053221735936e-06,
      "loss": 0.966,
      "step": 16964
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8451443910598755,
      "learning_rate": 9.77989330096157e-06,
      "loss": 0.9334,
      "step": 16965
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5041831731796265,
      "learning_rate": 9.77925438456378e-06,
      "loss": 0.7369,
      "step": 16966
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.39463210105896,
      "learning_rate": 9.778615468165992e-06,
      "loss": 1.1104,
      "step": 16967
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.2206966876983643,
      "learning_rate": 9.777976551768202e-06,
      "loss": 1.1607,
      "step": 16968
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9282039403915405,
      "learning_rate": 9.777337635370413e-06,
      "loss": 0.949,
      "step": 16969
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.634211540222168,
      "learning_rate": 9.776698718972623e-06,
      "loss": 1.4788,
      "step": 16970
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.745286464691162,
      "learning_rate": 9.776059802574835e-06,
      "loss": 0.4542,
      "step": 16971
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4343174695968628,
      "learning_rate": 9.775420886177044e-06,
      "loss": 0.9118,
      "step": 16972
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6173795461654663,
      "learning_rate": 9.774781969779256e-06,
      "loss": 0.8498,
      "step": 16973
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7086493968963623,
      "learning_rate": 9.774143053381466e-06,
      "loss": 0.9175,
      "step": 16974
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0835981369018555,
      "learning_rate": 9.773504136983677e-06,
      "loss": 0.9748,
      "step": 16975
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1098899841308594,
      "learning_rate": 9.772865220585887e-06,
      "loss": 1.0133,
      "step": 16976
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7628694772720337,
      "learning_rate": 9.772226304188099e-06,
      "loss": 0.8534,
      "step": 16977
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.321213483810425,
      "learning_rate": 9.771587387790309e-06,
      "loss": 0.6698,
      "step": 16978
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.8775696754455566,
      "learning_rate": 9.77094847139252e-06,
      "loss": 0.8567,
      "step": 16979
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5881495475769043,
      "learning_rate": 9.77030955499473e-06,
      "loss": 0.8448,
      "step": 16980
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6457806825637817,
      "learning_rate": 9.769670638596942e-06,
      "loss": 0.9562,
      "step": 16981
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0312280654907227,
      "learning_rate": 9.769031722199151e-06,
      "loss": 0.7586,
      "step": 16982
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3200843334198,
      "learning_rate": 9.768392805801361e-06,
      "loss": 1.0492,
      "step": 16983
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5256495475769043,
      "learning_rate": 9.767753889403573e-06,
      "loss": 0.8831,
      "step": 16984
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.416435480117798,
      "learning_rate": 9.767114973005783e-06,
      "loss": 0.5163,
      "step": 16985
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.701978325843811,
      "learning_rate": 9.766476056607994e-06,
      "loss": 0.9366,
      "step": 16986
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.914888620376587,
      "learning_rate": 9.765837140210204e-06,
      "loss": 0.8241,
      "step": 16987
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1818206310272217,
      "learning_rate": 9.765198223812416e-06,
      "loss": 0.9154,
      "step": 16988
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.089076280593872,
      "learning_rate": 9.764559307414626e-06,
      "loss": 0.9487,
      "step": 16989
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0890204906463623,
      "learning_rate": 9.763920391016837e-06,
      "loss": 0.81,
      "step": 16990
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6131057739257812,
      "learning_rate": 9.763281474619047e-06,
      "loss": 0.9628,
      "step": 16991
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4230222702026367,
      "learning_rate": 9.762642558221258e-06,
      "loss": 0.8709,
      "step": 16992
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.884893536567688,
      "learning_rate": 9.762003641823468e-06,
      "loss": 1.201,
      "step": 16993
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6116812229156494,
      "learning_rate": 9.76136472542568e-06,
      "loss": 0.4794,
      "step": 16994
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8657221794128418,
      "learning_rate": 9.76072580902789e-06,
      "loss": 1.0069,
      "step": 16995
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.104429006576538,
      "learning_rate": 9.760086892630101e-06,
      "loss": 0.7808,
      "step": 16996
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.995674729347229,
      "learning_rate": 9.759447976232311e-06,
      "loss": 0.9391,
      "step": 16997
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6121022701263428,
      "learning_rate": 9.758809059834523e-06,
      "loss": 0.9712,
      "step": 16998
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.516838073730469,
      "learning_rate": 9.758170143436733e-06,
      "loss": 1.1254,
      "step": 16999
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.067435264587402,
      "learning_rate": 9.757531227038944e-06,
      "loss": 1.6296,
      "step": 17000
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.794707179069519,
      "learning_rate": 9.756892310641154e-06,
      "loss": 0.8604,
      "step": 17001
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.474428415298462,
      "learning_rate": 9.756253394243364e-06,
      "loss": 0.9056,
      "step": 17002
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5019266605377197,
      "learning_rate": 9.755614477845575e-06,
      "loss": 1.0165,
      "step": 17003
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2567520141601562,
      "learning_rate": 9.754975561447785e-06,
      "loss": 0.8225,
      "step": 17004
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.3348710536956787,
      "learning_rate": 9.754336645049997e-06,
      "loss": 1.4292,
      "step": 17005
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.463456153869629,
      "learning_rate": 9.753697728652207e-06,
      "loss": 0.8284,
      "step": 17006
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.352062940597534,
      "learning_rate": 9.753058812254416e-06,
      "loss": 1.0918,
      "step": 17007
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.754917025566101,
      "learning_rate": 9.752419895856628e-06,
      "loss": 0.9487,
      "step": 17008
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5072367191314697,
      "learning_rate": 9.751780979458838e-06,
      "loss": 0.9158,
      "step": 17009
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.140256643295288,
      "learning_rate": 9.75114206306105e-06,
      "loss": 0.9083,
      "step": 17010
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.294937252998352,
      "learning_rate": 9.75050314666326e-06,
      "loss": 0.4703,
      "step": 17011
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.659973621368408,
      "learning_rate": 9.74986423026547e-06,
      "loss": 0.9169,
      "step": 17012
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.34820556640625,
      "learning_rate": 9.74922531386768e-06,
      "loss": 0.7677,
      "step": 17013
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.073633670806885,
      "learning_rate": 9.74858639746989e-06,
      "loss": 1.3382,
      "step": 17014
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5243582725524902,
      "learning_rate": 9.747947481072102e-06,
      "loss": 0.8534,
      "step": 17015
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.679640769958496,
      "learning_rate": 9.747308564674312e-06,
      "loss": 0.9135,
      "step": 17016
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4940953254699707,
      "learning_rate": 9.746669648276523e-06,
      "loss": 1.1142,
      "step": 17017
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7792408466339111,
      "learning_rate": 9.746030731878733e-06,
      "loss": 0.9455,
      "step": 17018
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.714323043823242,
      "learning_rate": 9.745391815480945e-06,
      "loss": 1.3283,
      "step": 17019
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2768654823303223,
      "learning_rate": 9.744752899083155e-06,
      "loss": 0.727,
      "step": 17020
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.294001817703247,
      "learning_rate": 9.744113982685366e-06,
      "loss": 0.9327,
      "step": 17021
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4639705419540405,
      "learning_rate": 9.743475066287576e-06,
      "loss": 0.8928,
      "step": 17022
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6734302043914795,
      "learning_rate": 9.742836149889788e-06,
      "loss": 0.7823,
      "step": 17023
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8870223760604858,
      "learning_rate": 9.742197233491997e-06,
      "loss": 0.7865,
      "step": 17024
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.9811007976531982,
      "learning_rate": 9.741558317094209e-06,
      "loss": 1.0588,
      "step": 17025
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.675938606262207,
      "learning_rate": 9.740919400696419e-06,
      "loss": 0.501,
      "step": 17026
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.863685965538025,
      "learning_rate": 9.74028048429863e-06,
      "loss": 0.8523,
      "step": 17027
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7311159372329712,
      "learning_rate": 9.73964156790084e-06,
      "loss": 0.9401,
      "step": 17028
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.2832961082458496,
      "learning_rate": 9.739002651503052e-06,
      "loss": 1.0272,
      "step": 17029
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.80633544921875,
      "learning_rate": 9.738363735105262e-06,
      "loss": 1.0925,
      "step": 17030
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.727203369140625,
      "learning_rate": 9.737724818707473e-06,
      "loss": 0.9824,
      "step": 17031
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.531564950942993,
      "learning_rate": 9.737085902309683e-06,
      "loss": 0.7496,
      "step": 17032
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.854623317718506,
      "learning_rate": 9.736446985911893e-06,
      "loss": 0.9078,
      "step": 17033
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0218703746795654,
      "learning_rate": 9.735808069514104e-06,
      "loss": 0.8138,
      "step": 17034
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.635490655899048,
      "learning_rate": 9.735169153116314e-06,
      "loss": 0.7468,
      "step": 17035
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.8388452529907227,
      "learning_rate": 9.734530236718526e-06,
      "loss": 1.12,
      "step": 17036
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8366252183914185,
      "learning_rate": 9.733891320320736e-06,
      "loss": 0.7913,
      "step": 17037
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8488616943359375,
      "learning_rate": 9.733252403922947e-06,
      "loss": 0.8103,
      "step": 17038
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8870173692703247,
      "learning_rate": 9.732613487525157e-06,
      "loss": 0.785,
      "step": 17039
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4282060861587524,
      "learning_rate": 9.731974571127369e-06,
      "loss": 0.8029,
      "step": 17040
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7518075704574585,
      "learning_rate": 9.731335654729579e-06,
      "loss": 0.9022,
      "step": 17041
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5937180519104004,
      "learning_rate": 9.73069673833179e-06,
      "loss": 0.8633,
      "step": 17042
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.952431559562683,
      "learning_rate": 9.730057821934e-06,
      "loss": 0.941,
      "step": 17043
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.120973587036133,
      "learning_rate": 9.729418905536211e-06,
      "loss": 1.2959,
      "step": 17044
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8006834983825684,
      "learning_rate": 9.728779989138421e-06,
      "loss": 0.8648,
      "step": 17045
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7049469947814941,
      "learning_rate": 9.728141072740633e-06,
      "loss": 0.9087,
      "step": 17046
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2840054035186768,
      "learning_rate": 9.727502156342843e-06,
      "loss": 0.889,
      "step": 17047
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.655942678451538,
      "learning_rate": 9.726863239945054e-06,
      "loss": 0.9686,
      "step": 17048
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0723938941955566,
      "learning_rate": 9.726224323547264e-06,
      "loss": 1.1732,
      "step": 17049
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6503169536590576,
      "learning_rate": 9.725585407149476e-06,
      "loss": 0.8581,
      "step": 17050
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.585379719734192,
      "learning_rate": 9.724946490751686e-06,
      "loss": 0.8707,
      "step": 17051
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5720999240875244,
      "learning_rate": 9.724307574353895e-06,
      "loss": 0.8746,
      "step": 17052
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9498558044433594,
      "learning_rate": 9.723668657956107e-06,
      "loss": 0.8994,
      "step": 17053
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3897478580474854,
      "learning_rate": 9.723029741558317e-06,
      "loss": 0.8292,
      "step": 17054
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6653008460998535,
      "learning_rate": 9.722390825160528e-06,
      "loss": 0.5274,
      "step": 17055
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7487175464630127,
      "learning_rate": 9.721751908762738e-06,
      "loss": 0.907,
      "step": 17056
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7879679203033447,
      "learning_rate": 9.72111299236495e-06,
      "loss": 0.9404,
      "step": 17057
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7658085823059082,
      "learning_rate": 9.72047407596716e-06,
      "loss": 0.921,
      "step": 17058
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6806981563568115,
      "learning_rate": 9.719835159569371e-06,
      "loss": 0.8421,
      "step": 17059
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.990450620651245,
      "learning_rate": 9.719196243171581e-06,
      "loss": 1.0204,
      "step": 17060
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6845844984054565,
      "learning_rate": 9.718557326773793e-06,
      "loss": 0.8692,
      "step": 17061
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.983681321144104,
      "learning_rate": 9.717918410376002e-06,
      "loss": 0.9317,
      "step": 17062
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5822290182113647,
      "learning_rate": 9.717279493978214e-06,
      "loss": 0.4405,
      "step": 17063
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8497618436813354,
      "learning_rate": 9.716640577580424e-06,
      "loss": 0.8382,
      "step": 17064
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2671886682510376,
      "learning_rate": 9.716001661182635e-06,
      "loss": 0.8835,
      "step": 17065
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0573298931121826,
      "learning_rate": 9.715362744784845e-06,
      "loss": 0.8958,
      "step": 17066
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.957629680633545,
      "learning_rate": 9.714723828387057e-06,
      "loss": 0.873,
      "step": 17067
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9726853370666504,
      "learning_rate": 9.714084911989267e-06,
      "loss": 0.934,
      "step": 17068
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.250608444213867,
      "learning_rate": 9.713445995591476e-06,
      "loss": 0.8239,
      "step": 17069
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.924964189529419,
      "learning_rate": 9.712807079193688e-06,
      "loss": 0.9069,
      "step": 17070
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0370328426361084,
      "learning_rate": 9.712168162795898e-06,
      "loss": 0.8712,
      "step": 17071
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5696146488189697,
      "learning_rate": 9.71152924639811e-06,
      "loss": 0.8741,
      "step": 17072
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9197795391082764,
      "learning_rate": 9.71089033000032e-06,
      "loss": 1.2163,
      "step": 17073
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.562825083732605,
      "learning_rate": 9.71025141360253e-06,
      "loss": 0.806,
      "step": 17074
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7733311653137207,
      "learning_rate": 9.70961249720474e-06,
      "loss": 0.9736,
      "step": 17075
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4571425914764404,
      "learning_rate": 9.708973580806952e-06,
      "loss": 0.9376,
      "step": 17076
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4880317449569702,
      "learning_rate": 9.708334664409162e-06,
      "loss": 0.934,
      "step": 17077
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9777801036834717,
      "learning_rate": 9.707695748011374e-06,
      "loss": 0.959,
      "step": 17078
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.8082656860351562,
      "learning_rate": 9.707056831613583e-06,
      "loss": 0.9213,
      "step": 17079
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3535327911376953,
      "learning_rate": 9.706417915215795e-06,
      "loss": 0.7892,
      "step": 17080
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.032528877258301,
      "learning_rate": 9.705778998818005e-06,
      "loss": 0.9996,
      "step": 17081
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.614516019821167,
      "learning_rate": 9.705140082420216e-06,
      "loss": 0.6014,
      "step": 17082
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.442800283432007,
      "learning_rate": 9.704501166022426e-06,
      "loss": 0.7945,
      "step": 17083
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1218574047088623,
      "learning_rate": 9.703862249624638e-06,
      "loss": 1.1168,
      "step": 17084
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0315253734588623,
      "learning_rate": 9.703223333226848e-06,
      "loss": 0.5957,
      "step": 17085
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.878051519393921,
      "learning_rate": 9.70258441682906e-06,
      "loss": 0.9611,
      "step": 17086
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.211935043334961,
      "learning_rate": 9.701945500431269e-06,
      "loss": 0.5517,
      "step": 17087
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.5342164039611816,
      "learning_rate": 9.701306584033479e-06,
      "loss": 0.8624,
      "step": 17088
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8167011737823486,
      "learning_rate": 9.70066766763569e-06,
      "loss": 0.8607,
      "step": 17089
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1780879497528076,
      "learning_rate": 9.7000287512379e-06,
      "loss": 0.5639,
      "step": 17090
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9188106060028076,
      "learning_rate": 9.699389834840112e-06,
      "loss": 0.9945,
      "step": 17091
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.706686496734619,
      "learning_rate": 9.698750918442322e-06,
      "loss": 0.6412,
      "step": 17092
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5255208015441895,
      "learning_rate": 9.698112002044533e-06,
      "loss": 0.8606,
      "step": 17093
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.5181806087493896,
      "learning_rate": 9.697473085646743e-06,
      "loss": 1.1366,
      "step": 17094
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.026186227798462,
      "learning_rate": 9.696834169248955e-06,
      "loss": 0.8214,
      "step": 17095
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6195484399795532,
      "learning_rate": 9.696195252851165e-06,
      "loss": 0.8862,
      "step": 17096
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7008256912231445,
      "learning_rate": 9.695556336453376e-06,
      "loss": 0.7326,
      "step": 17097
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.370206356048584,
      "learning_rate": 9.694917420055586e-06,
      "loss": 0.8724,
      "step": 17098
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7484382390975952,
      "learning_rate": 9.694278503657797e-06,
      "loss": 0.9376,
      "step": 17099
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7498172521591187,
      "learning_rate": 9.693639587260007e-06,
      "loss": 0.8737,
      "step": 17100
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.656609535217285,
      "learning_rate": 9.693000670862219e-06,
      "loss": 1.47,
      "step": 17101
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.412958860397339,
      "learning_rate": 9.692361754464429e-06,
      "loss": 0.7209,
      "step": 17102
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.879665732383728,
      "learning_rate": 9.69172283806664e-06,
      "loss": 0.9887,
      "step": 17103
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9504950046539307,
      "learning_rate": 9.69108392166885e-06,
      "loss": 0.8778,
      "step": 17104
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.628363609313965,
      "learning_rate": 9.69044500527106e-06,
      "loss": 0.9147,
      "step": 17105
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5711736679077148,
      "learning_rate": 9.689806088873272e-06,
      "loss": 0.927,
      "step": 17106
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.5308549404144287,
      "learning_rate": 9.689167172475481e-06,
      "loss": 0.7927,
      "step": 17107
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.653496742248535,
      "learning_rate": 9.688528256077693e-06,
      "loss": 1.155,
      "step": 17108
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.447750210762024,
      "learning_rate": 9.687889339679903e-06,
      "loss": 0.8251,
      "step": 17109
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.1331982612609863,
      "learning_rate": 9.687250423282114e-06,
      "loss": 1.2534,
      "step": 17110
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8519091606140137,
      "learning_rate": 9.686611506884324e-06,
      "loss": 0.8628,
      "step": 17111
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.0947890281677246,
      "learning_rate": 9.685972590486536e-06,
      "loss": 1.1569,
      "step": 17112
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8632004261016846,
      "learning_rate": 9.685333674088746e-06,
      "loss": 0.8426,
      "step": 17113
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7158275842666626,
      "learning_rate": 9.684694757690957e-06,
      "loss": 0.9556,
      "step": 17114
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.721349000930786,
      "learning_rate": 9.684055841293167e-06,
      "loss": 0.8855,
      "step": 17115
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.148716449737549,
      "learning_rate": 9.683416924895379e-06,
      "loss": 1.0708,
      "step": 17116
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5434809923171997,
      "learning_rate": 9.682778008497588e-06,
      "loss": 0.793,
      "step": 17117
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.668673038482666,
      "learning_rate": 9.6821390920998e-06,
      "loss": 1.5305,
      "step": 17118
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6028246879577637,
      "learning_rate": 9.68150017570201e-06,
      "loss": 0.8735,
      "step": 17119
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7236950397491455,
      "learning_rate": 9.680861259304221e-06,
      "loss": 0.4081,
      "step": 17120
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9376838207244873,
      "learning_rate": 9.680222342906431e-06,
      "loss": 0.9355,
      "step": 17121
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2009406089782715,
      "learning_rate": 9.679583426508643e-06,
      "loss": 1.0285,
      "step": 17122
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4681203365325928,
      "learning_rate": 9.678944510110853e-06,
      "loss": 0.8206,
      "step": 17123
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.8915629386901855,
      "learning_rate": 9.678305593713062e-06,
      "loss": 1.2337,
      "step": 17124
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.4495346546173096,
      "learning_rate": 9.677666677315274e-06,
      "loss": 1.2115,
      "step": 17125
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7465993165969849,
      "learning_rate": 9.677027760917484e-06,
      "loss": 0.8792,
      "step": 17126
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9531861543655396,
      "learning_rate": 9.676388844519695e-06,
      "loss": 0.9276,
      "step": 17127
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.833244800567627,
      "learning_rate": 9.675749928121905e-06,
      "loss": 1.0597,
      "step": 17128
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.977156162261963,
      "learning_rate": 9.675111011724117e-06,
      "loss": 1.2604,
      "step": 17129
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4977277517318726,
      "learning_rate": 9.674472095326327e-06,
      "loss": 0.8089,
      "step": 17130
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4978573322296143,
      "learning_rate": 9.673833178928538e-06,
      "loss": 0.8962,
      "step": 17131
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4811153411865234,
      "learning_rate": 9.673194262530748e-06,
      "loss": 1.0638,
      "step": 17132
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.006624221801758,
      "learning_rate": 9.67255534613296e-06,
      "loss": 0.8888,
      "step": 17133
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5992482900619507,
      "learning_rate": 9.67191642973517e-06,
      "loss": 0.8204,
      "step": 17134
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.943244218826294,
      "learning_rate": 9.671277513337381e-06,
      "loss": 0.993,
      "step": 17135
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.725828766822815,
      "learning_rate": 9.67063859693959e-06,
      "loss": 0.9331,
      "step": 17136
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5054658651351929,
      "learning_rate": 9.669999680541802e-06,
      "loss": 0.7367,
      "step": 17137
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.066192626953125,
      "learning_rate": 9.669360764144012e-06,
      "loss": 0.8349,
      "step": 17138
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.990086555480957,
      "learning_rate": 9.668721847746224e-06,
      "loss": 0.9673,
      "step": 17139
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.81716787815094,
      "learning_rate": 9.668082931348434e-06,
      "loss": 0.8267,
      "step": 17140
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4457809925079346,
      "learning_rate": 9.667444014950645e-06,
      "loss": 0.8689,
      "step": 17141
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7689805030822754,
      "learning_rate": 9.666805098552855e-06,
      "loss": 0.7776,
      "step": 17142
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0783145427703857,
      "learning_rate": 9.666166182155065e-06,
      "loss": 0.8881,
      "step": 17143
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5556647777557373,
      "learning_rate": 9.665527265757276e-06,
      "loss": 0.473,
      "step": 17144
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4789650440216064,
      "learning_rate": 9.664888349359486e-06,
      "loss": 0.7784,
      "step": 17145
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5030070543289185,
      "learning_rate": 9.664249432961698e-06,
      "loss": 0.8611,
      "step": 17146
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5428707599639893,
      "learning_rate": 9.663610516563908e-06,
      "loss": 0.864,
      "step": 17147
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.846015453338623,
      "learning_rate": 9.66297160016612e-06,
      "loss": 1.1905,
      "step": 17148
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6313523054122925,
      "learning_rate": 9.662332683768329e-06,
      "loss": 0.842,
      "step": 17149
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4967882633209229,
      "learning_rate": 9.66169376737054e-06,
      "loss": 0.4906,
      "step": 17150
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.364048719406128,
      "learning_rate": 9.66105485097275e-06,
      "loss": 0.9318,
      "step": 17151
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7402970790863037,
      "learning_rate": 9.660415934574962e-06,
      "loss": 0.8644,
      "step": 17152
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5445023775100708,
      "learning_rate": 9.659777018177172e-06,
      "loss": 0.9449,
      "step": 17153
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7272177934646606,
      "learning_rate": 9.659138101779383e-06,
      "loss": 0.9492,
      "step": 17154
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.752013683319092,
      "learning_rate": 9.658499185381593e-06,
      "loss": 0.9482,
      "step": 17155
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8359503746032715,
      "learning_rate": 9.657860268983805e-06,
      "loss": 0.9221,
      "step": 17156
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.824872374534607,
      "learning_rate": 9.657221352586015e-06,
      "loss": 0.8447,
      "step": 17157
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.119993209838867,
      "learning_rate": 9.656582436188226e-06,
      "loss": 0.7876,
      "step": 17158
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.570753574371338,
      "learning_rate": 9.655943519790436e-06,
      "loss": 0.8827,
      "step": 17159
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3322694301605225,
      "learning_rate": 9.655304603392646e-06,
      "loss": 0.7823,
      "step": 17160
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.701779842376709,
      "learning_rate": 9.654665686994857e-06,
      "loss": 0.982,
      "step": 17161
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.201284885406494,
      "learning_rate": 9.654026770597067e-06,
      "loss": 1.0609,
      "step": 17162
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9366626739501953,
      "learning_rate": 9.653387854199279e-06,
      "loss": 0.9845,
      "step": 17163
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.14780330657959,
      "learning_rate": 9.652748937801489e-06,
      "loss": 0.6974,
      "step": 17164
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6744980812072754,
      "learning_rate": 9.6521100214037e-06,
      "loss": 0.9122,
      "step": 17165
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.852291226387024,
      "learning_rate": 9.65147110500591e-06,
      "loss": 0.8972,
      "step": 17166
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.882693290710449,
      "learning_rate": 9.650832188608122e-06,
      "loss": 1.1824,
      "step": 17167
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8132929801940918,
      "learning_rate": 9.650193272210332e-06,
      "loss": 0.9022,
      "step": 17168
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.756724238395691,
      "learning_rate": 9.649554355812543e-06,
      "loss": 0.9358,
      "step": 17169
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.695469856262207,
      "learning_rate": 9.648915439414753e-06,
      "loss": 1.1579,
      "step": 17170
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.0488290786743164,
      "learning_rate": 9.648276523016964e-06,
      "loss": 1.1849,
      "step": 17171
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6229865550994873,
      "learning_rate": 9.647637606619174e-06,
      "loss": 0.8795,
      "step": 17172
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6570281982421875,
      "learning_rate": 9.646998690221386e-06,
      "loss": 1.1818,
      "step": 17173
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8324640989303589,
      "learning_rate": 9.646359773823596e-06,
      "loss": 0.8523,
      "step": 17174
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5288959741592407,
      "learning_rate": 9.645720857425807e-06,
      "loss": 0.4463,
      "step": 17175
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0289740562438965,
      "learning_rate": 9.645081941028017e-06,
      "loss": 0.8679,
      "step": 17176
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8403018712997437,
      "learning_rate": 9.644443024630229e-06,
      "loss": 0.7911,
      "step": 17177
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6087974309921265,
      "learning_rate": 9.643804108232439e-06,
      "loss": 0.7186,
      "step": 17178
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2896945476531982,
      "learning_rate": 9.643165191834648e-06,
      "loss": 0.7901,
      "step": 17179
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.679832100868225,
      "learning_rate": 9.64252627543686e-06,
      "loss": 0.8974,
      "step": 17180
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2063186168670654,
      "learning_rate": 9.64188735903907e-06,
      "loss": 0.6082,
      "step": 17181
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5239391326904297,
      "learning_rate": 9.641248442641281e-06,
      "loss": 0.9327,
      "step": 17182
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.5884127616882324,
      "learning_rate": 9.640609526243491e-06,
      "loss": 0.7927,
      "step": 17183
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6781812906265259,
      "learning_rate": 9.639970609845703e-06,
      "loss": 1.1634,
      "step": 17184
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.133164882659912,
      "learning_rate": 9.639331693447913e-06,
      "loss": 0.9655,
      "step": 17185
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.192112445831299,
      "learning_rate": 9.638692777050124e-06,
      "loss": 0.9256,
      "step": 17186
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.1311123371124268,
      "learning_rate": 9.638053860652334e-06,
      "loss": 0.9057,
      "step": 17187
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8004863262176514,
      "learning_rate": 9.637414944254546e-06,
      "loss": 0.891,
      "step": 17188
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.381397247314453,
      "learning_rate": 9.636776027856755e-06,
      "loss": 1.4006,
      "step": 17189
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7535768747329712,
      "learning_rate": 9.636137111458967e-06,
      "loss": 0.8741,
      "step": 17190
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1063945293426514,
      "learning_rate": 9.635498195061177e-06,
      "loss": 1.0273,
      "step": 17191
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.296389102935791,
      "learning_rate": 9.634859278663388e-06,
      "loss": 0.8358,
      "step": 17192
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5360546112060547,
      "learning_rate": 9.634220362265598e-06,
      "loss": 0.782,
      "step": 17193
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9197908639907837,
      "learning_rate": 9.63358144586781e-06,
      "loss": 0.9514,
      "step": 17194
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.372436761856079,
      "learning_rate": 9.63294252947002e-06,
      "loss": 0.9946,
      "step": 17195
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.283428907394409,
      "learning_rate": 9.632303613072231e-06,
      "loss": 0.62,
      "step": 17196
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.697525143623352,
      "learning_rate": 9.631664696674441e-06,
      "loss": 0.8548,
      "step": 17197
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.106501817703247,
      "learning_rate": 9.63102578027665e-06,
      "loss": 0.9225,
      "step": 17198
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.990818977355957,
      "learning_rate": 9.630386863878862e-06,
      "loss": 1.1243,
      "step": 17199
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.407243251800537,
      "learning_rate": 9.629747947481072e-06,
      "loss": 0.4164,
      "step": 17200
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.426236152648926,
      "learning_rate": 9.629109031083284e-06,
      "loss": 1.6244,
      "step": 17201
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.776456356048584,
      "learning_rate": 9.628470114685494e-06,
      "loss": 0.8012,
      "step": 17202
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.399337887763977,
      "learning_rate": 9.627831198287705e-06,
      "loss": 0.4201,
      "step": 17203
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.242356061935425,
      "learning_rate": 9.627192281889915e-06,
      "loss": 0.7493,
      "step": 17204
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.513731837272644,
      "learning_rate": 9.626553365492127e-06,
      "loss": 0.9151,
      "step": 17205
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3061509132385254,
      "learning_rate": 9.625914449094336e-06,
      "loss": 0.9838,
      "step": 17206
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.024468183517456,
      "learning_rate": 9.625275532696548e-06,
      "loss": 0.8418,
      "step": 17207
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.9390337467193604,
      "learning_rate": 9.624636616298758e-06,
      "loss": 1.1053,
      "step": 17208
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.996769428253174,
      "learning_rate": 9.62399769990097e-06,
      "loss": 0.9115,
      "step": 17209
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7387778759002686,
      "learning_rate": 9.62335878350318e-06,
      "loss": 0.809,
      "step": 17210
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.604684829711914,
      "learning_rate": 9.62271986710539e-06,
      "loss": 0.8402,
      "step": 17211
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.988346576690674,
      "learning_rate": 9.6220809507076e-06,
      "loss": 0.7756,
      "step": 17212
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3281689882278442,
      "learning_rate": 9.621442034309812e-06,
      "loss": 0.8103,
      "step": 17213
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9014478921890259,
      "learning_rate": 9.620803117912022e-06,
      "loss": 0.9368,
      "step": 17214
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7960654497146606,
      "learning_rate": 9.620164201514232e-06,
      "loss": 0.9256,
      "step": 17215
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.522143363952637,
      "learning_rate": 9.619525285116443e-06,
      "loss": 1.5462,
      "step": 17216
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.1611807346343994,
      "learning_rate": 9.618886368718653e-06,
      "loss": 1.2807,
      "step": 17217
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8417935371398926,
      "learning_rate": 9.618247452320865e-06,
      "loss": 0.884,
      "step": 17218
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7412124872207642,
      "learning_rate": 9.617608535923075e-06,
      "loss": 0.7791,
      "step": 17219
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5677796602249146,
      "learning_rate": 9.616969619525286e-06,
      "loss": 0.8381,
      "step": 17220
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2861342430114746,
      "learning_rate": 9.616330703127496e-06,
      "loss": 0.9165,
      "step": 17221
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.457137107849121,
      "learning_rate": 9.615691786729708e-06,
      "loss": 1.5386,
      "step": 17222
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9724669456481934,
      "learning_rate": 9.615052870331917e-06,
      "loss": 0.8907,
      "step": 17223
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5113945007324219,
      "learning_rate": 9.614413953934129e-06,
      "loss": 0.8705,
      "step": 17224
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5394279956817627,
      "learning_rate": 9.613775037536339e-06,
      "loss": 0.9054,
      "step": 17225
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.937894821166992,
      "learning_rate": 9.61313612113855e-06,
      "loss": 1.2021,
      "step": 17226
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6289489269256592,
      "learning_rate": 9.61249720474076e-06,
      "loss": 0.9929,
      "step": 17227
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9593433141708374,
      "learning_rate": 9.611858288342972e-06,
      "loss": 1.2034,
      "step": 17228
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8935105800628662,
      "learning_rate": 9.611219371945182e-06,
      "loss": 0.9223,
      "step": 17229
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1459898948669434,
      "learning_rate": 9.610580455547393e-06,
      "loss": 1.2751,
      "step": 17230
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1700382232666016,
      "learning_rate": 9.609941539149603e-06,
      "loss": 0.8687,
      "step": 17231
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5439789295196533,
      "learning_rate": 9.609302622751815e-06,
      "loss": 0.4146,
      "step": 17232
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6362049579620361,
      "learning_rate": 9.608663706354024e-06,
      "loss": 0.907,
      "step": 17233
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.155334949493408,
      "learning_rate": 9.608024789956234e-06,
      "loss": 0.8427,
      "step": 17234
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.419703722000122,
      "learning_rate": 9.607385873558446e-06,
      "loss": 0.4521,
      "step": 17235
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.3883373737335205,
      "learning_rate": 9.606746957160656e-06,
      "loss": 0.4098,
      "step": 17236
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4560306072235107,
      "learning_rate": 9.606108040762867e-06,
      "loss": 0.8284,
      "step": 17237
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6332296133041382,
      "learning_rate": 9.605469124365077e-06,
      "loss": 0.3955,
      "step": 17238
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7983672618865967,
      "learning_rate": 9.604830207967289e-06,
      "loss": 0.9655,
      "step": 17239
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3970478773117065,
      "learning_rate": 9.604191291569499e-06,
      "loss": 0.8005,
      "step": 17240
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9874985218048096,
      "learning_rate": 9.60355237517171e-06,
      "loss": 0.9067,
      "step": 17241
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6559488773345947,
      "learning_rate": 9.60291345877392e-06,
      "loss": 0.9264,
      "step": 17242
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5329053401947021,
      "learning_rate": 9.602274542376131e-06,
      "loss": 0.9242,
      "step": 17243
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9694430828094482,
      "learning_rate": 9.601635625978341e-06,
      "loss": 0.8917,
      "step": 17244
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.870842456817627,
      "learning_rate": 9.600996709580553e-06,
      "loss": 0.8513,
      "step": 17245
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.071547031402588,
      "learning_rate": 9.600357793182763e-06,
      "loss": 1.17,
      "step": 17246
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.517670154571533,
      "learning_rate": 9.599718876784974e-06,
      "loss": 0.8914,
      "step": 17247
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9588158130645752,
      "learning_rate": 9.599079960387184e-06,
      "loss": 0.9275,
      "step": 17248
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.041968584060669,
      "learning_rate": 9.598441043989396e-06,
      "loss": 0.8503,
      "step": 17249
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3922556638717651,
      "learning_rate": 9.597802127591606e-06,
      "loss": 0.4475,
      "step": 17250
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.082012176513672,
      "learning_rate": 9.597163211193815e-06,
      "loss": 0.9317,
      "step": 17251
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8586695194244385,
      "learning_rate": 9.596524294796027e-06,
      "loss": 0.8662,
      "step": 17252
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.5504767894744873,
      "learning_rate": 9.595885378398237e-06,
      "loss": 0.7961,
      "step": 17253
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9044442176818848,
      "learning_rate": 9.595246462000448e-06,
      "loss": 0.9251,
      "step": 17254
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.6082849502563477,
      "learning_rate": 9.594607545602658e-06,
      "loss": 0.9757,
      "step": 17255
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.694973349571228,
      "learning_rate": 9.59396862920487e-06,
      "loss": 0.8656,
      "step": 17256
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.89150869846344,
      "learning_rate": 9.59332971280708e-06,
      "loss": 0.9547,
      "step": 17257
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.014028310775757,
      "learning_rate": 9.592690796409291e-06,
      "loss": 1.0368,
      "step": 17258
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.3747024536132812,
      "learning_rate": 9.592051880011501e-06,
      "loss": 1.0326,
      "step": 17259
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9479390382766724,
      "learning_rate": 9.591412963613713e-06,
      "loss": 0.8873,
      "step": 17260
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7271497249603271,
      "learning_rate": 9.590774047215922e-06,
      "loss": 0.8517,
      "step": 17261
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.4997453689575195,
      "learning_rate": 9.590135130818134e-06,
      "loss": 0.8331,
      "step": 17262
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4870281219482422,
      "learning_rate": 9.589496214420344e-06,
      "loss": 0.7043,
      "step": 17263
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.035905599594116,
      "learning_rate": 9.588857298022555e-06,
      "loss": 0.9377,
      "step": 17264
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7234827280044556,
      "learning_rate": 9.588218381624765e-06,
      "loss": 0.73,
      "step": 17265
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.041945457458496,
      "learning_rate": 9.587579465226977e-06,
      "loss": 0.931,
      "step": 17266
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8519140481948853,
      "learning_rate": 9.586940548829187e-06,
      "loss": 0.9436,
      "step": 17267
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8847016096115112,
      "learning_rate": 9.586301632431398e-06,
      "loss": 0.8915,
      "step": 17268
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3653056621551514,
      "learning_rate": 9.585662716033608e-06,
      "loss": 0.903,
      "step": 17269
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.892717123031616,
      "learning_rate": 9.585023799635818e-06,
      "loss": 1.7303,
      "step": 17270
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.481956958770752,
      "learning_rate": 9.58438488323803e-06,
      "loss": 1.0571,
      "step": 17271
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1703975200653076,
      "learning_rate": 9.58374596684024e-06,
      "loss": 0.6325,
      "step": 17272
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.6648812294006348,
      "learning_rate": 9.58310705044245e-06,
      "loss": 0.9654,
      "step": 17273
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9481781721115112,
      "learning_rate": 9.58246813404466e-06,
      "loss": 0.8544,
      "step": 17274
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0001401901245117,
      "learning_rate": 9.581829217646872e-06,
      "loss": 0.8108,
      "step": 17275
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.93604576587677,
      "learning_rate": 9.581190301249082e-06,
      "loss": 0.953,
      "step": 17276
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9463714361190796,
      "learning_rate": 9.580551384851294e-06,
      "loss": 0.914,
      "step": 17277
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.162978172302246,
      "learning_rate": 9.579912468453503e-06,
      "loss": 1.2024,
      "step": 17278
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.9707210063934326,
      "learning_rate": 9.579273552055715e-06,
      "loss": 1.1298,
      "step": 17279
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5389301776885986,
      "learning_rate": 9.578634635657925e-06,
      "loss": 1.1758,
      "step": 17280
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.808002233505249,
      "learning_rate": 9.577995719260136e-06,
      "loss": 0.8525,
      "step": 17281
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.915348768234253,
      "learning_rate": 9.577356802862346e-06,
      "loss": 1.0161,
      "step": 17282
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.6378304958343506,
      "learning_rate": 9.576717886464558e-06,
      "loss": 1.0446,
      "step": 17283
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4419548511505127,
      "learning_rate": 9.576078970066768e-06,
      "loss": 0.5,
      "step": 17284
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.06095290184021,
      "learning_rate": 9.57544005366898e-06,
      "loss": 0.9378,
      "step": 17285
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7245842218399048,
      "learning_rate": 9.574801137271189e-06,
      "loss": 0.9739,
      "step": 17286
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.737707495689392,
      "learning_rate": 9.5741622208734e-06,
      "loss": 0.8415,
      "step": 17287
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6847338676452637,
      "learning_rate": 9.57352330447561e-06,
      "loss": 0.8385,
      "step": 17288
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8276675939559937,
      "learning_rate": 9.57288438807782e-06,
      "loss": 0.9846,
      "step": 17289
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5854309797286987,
      "learning_rate": 9.572245471680032e-06,
      "loss": 0.9239,
      "step": 17290
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.516780376434326,
      "learning_rate": 9.571606555282242e-06,
      "loss": 1.0785,
      "step": 17291
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.4570059776306152,
      "learning_rate": 9.570967638884453e-06,
      "loss": 1.1791,
      "step": 17292
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8584725856781006,
      "learning_rate": 9.570328722486663e-06,
      "loss": 0.89,
      "step": 17293
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0664496421813965,
      "learning_rate": 9.569689806088875e-06,
      "loss": 0.9785,
      "step": 17294
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.390408754348755,
      "learning_rate": 9.569050889691084e-06,
      "loss": 0.8731,
      "step": 17295
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1567800045013428,
      "learning_rate": 9.568411973293296e-06,
      "loss": 0.9727,
      "step": 17296
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6922500133514404,
      "learning_rate": 9.567773056895506e-06,
      "loss": 0.8767,
      "step": 17297
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.420023202896118,
      "learning_rate": 9.567134140497717e-06,
      "loss": 0.9959,
      "step": 17298
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.938657283782959,
      "learning_rate": 9.566495224099927e-06,
      "loss": 1.0423,
      "step": 17299
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.59974205493927,
      "learning_rate": 9.565856307702139e-06,
      "loss": 0.9152,
      "step": 17300
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9874169826507568,
      "learning_rate": 9.565217391304349e-06,
      "loss": 0.9473,
      "step": 17301
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8719099760055542,
      "learning_rate": 9.56457847490656e-06,
      "loss": 0.9628,
      "step": 17302
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.005302906036377,
      "learning_rate": 9.56393955850877e-06,
      "loss": 1.101,
      "step": 17303
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.639961004257202,
      "learning_rate": 9.563300642110982e-06,
      "loss": 0.8747,
      "step": 17304
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1134746074676514,
      "learning_rate": 9.562661725713191e-06,
      "loss": 0.6416,
      "step": 17305
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.3952715396881104,
      "learning_rate": 9.562022809315401e-06,
      "loss": 1.2932,
      "step": 17306
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3946348428726196,
      "learning_rate": 9.561383892917613e-06,
      "loss": 0.3865,
      "step": 17307
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4790149927139282,
      "learning_rate": 9.560744976519823e-06,
      "loss": 0.7936,
      "step": 17308
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1832094192504883,
      "learning_rate": 9.560106060122034e-06,
      "loss": 0.9195,
      "step": 17309
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.624375343322754,
      "learning_rate": 9.559467143724244e-06,
      "loss": 1.0921,
      "step": 17310
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1972949504852295,
      "learning_rate": 9.558828227326456e-06,
      "loss": 0.9233,
      "step": 17311
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.252727746963501,
      "learning_rate": 9.558189310928666e-06,
      "loss": 0.9971,
      "step": 17312
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6995259523391724,
      "learning_rate": 9.557550394530877e-06,
      "loss": 0.9027,
      "step": 17313
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0377845764160156,
      "learning_rate": 9.556911478133087e-06,
      "loss": 0.9433,
      "step": 17314
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5684069395065308,
      "learning_rate": 9.556272561735298e-06,
      "loss": 0.7521,
      "step": 17315
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9224034547805786,
      "learning_rate": 9.555633645337508e-06,
      "loss": 0.9055,
      "step": 17316
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.024198293685913,
      "learning_rate": 9.55499472893972e-06,
      "loss": 0.9429,
      "step": 17317
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3663721084594727,
      "learning_rate": 9.55435581254193e-06,
      "loss": 0.9389,
      "step": 17318
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.553200125694275,
      "learning_rate": 9.553716896144141e-06,
      "loss": 0.8459,
      "step": 17319
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7975256443023682,
      "learning_rate": 9.553077979746351e-06,
      "loss": 0.7853,
      "step": 17320
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7787501811981201,
      "learning_rate": 9.552439063348563e-06,
      "loss": 0.835,
      "step": 17321
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8805220127105713,
      "learning_rate": 9.551800146950773e-06,
      "loss": 0.8966,
      "step": 17322
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6236317157745361,
      "learning_rate": 9.551161230552984e-06,
      "loss": 0.8599,
      "step": 17323
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4942492246627808,
      "learning_rate": 9.550522314155194e-06,
      "loss": 0.8118,
      "step": 17324
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6871373653411865,
      "learning_rate": 9.549883397757404e-06,
      "loss": 1.0166,
      "step": 17325
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6378695964813232,
      "learning_rate": 9.549244481359615e-06,
      "loss": 0.8921,
      "step": 17326
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9983781576156616,
      "learning_rate": 9.548605564961825e-06,
      "loss": 0.9389,
      "step": 17327
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8633499145507812,
      "learning_rate": 9.547966648564037e-06,
      "loss": 0.9026,
      "step": 17328
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.747652530670166,
      "learning_rate": 9.547327732166247e-06,
      "loss": 0.8346,
      "step": 17329
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.342129945755005,
      "learning_rate": 9.546688815768458e-06,
      "loss": 0.8754,
      "step": 17330
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.715245246887207,
      "learning_rate": 9.546049899370668e-06,
      "loss": 1.8132,
      "step": 17331
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7521629333496094,
      "learning_rate": 9.54541098297288e-06,
      "loss": 0.8139,
      "step": 17332
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.089507579803467,
      "learning_rate": 9.54477206657509e-06,
      "loss": 0.8974,
      "step": 17333
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9325562715530396,
      "learning_rate": 9.544133150177301e-06,
      "loss": 0.82,
      "step": 17334
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.633715271949768,
      "learning_rate": 9.54349423377951e-06,
      "loss": 0.8175,
      "step": 17335
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4196271896362305,
      "learning_rate": 9.542855317381722e-06,
      "loss": 0.4316,
      "step": 17336
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7579089403152466,
      "learning_rate": 9.542216400983932e-06,
      "loss": 0.8155,
      "step": 17337
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.8728723526000977,
      "learning_rate": 9.541577484586144e-06,
      "loss": 1.2284,
      "step": 17338
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5572348833084106,
      "learning_rate": 9.540938568188354e-06,
      "loss": 0.7249,
      "step": 17339
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.508784532546997,
      "learning_rate": 9.540299651790565e-06,
      "loss": 0.8744,
      "step": 17340
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3355350494384766,
      "learning_rate": 9.539660735392775e-06,
      "loss": 1.0063,
      "step": 17341
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.118898391723633,
      "learning_rate": 9.539021818994987e-06,
      "loss": 0.9121,
      "step": 17342
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5319327116012573,
      "learning_rate": 9.538382902597196e-06,
      "loss": 0.8377,
      "step": 17343
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7270454168319702,
      "learning_rate": 9.537743986199406e-06,
      "loss": 0.4733,
      "step": 17344
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.690895915031433,
      "learning_rate": 9.537105069801618e-06,
      "loss": 0.9008,
      "step": 17345
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.134270668029785,
      "learning_rate": 9.536466153403828e-06,
      "loss": 0.8581,
      "step": 17346
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8873906135559082,
      "learning_rate": 9.53582723700604e-06,
      "loss": 0.7751,
      "step": 17347
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0656051635742188,
      "learning_rate": 9.535188320608249e-06,
      "loss": 1.0578,
      "step": 17348
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.503173589706421,
      "learning_rate": 9.53454940421046e-06,
      "loss": 1.0601,
      "step": 17349
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9368945360183716,
      "learning_rate": 9.53391048781267e-06,
      "loss": 0.9235,
      "step": 17350
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.833298921585083,
      "learning_rate": 9.533271571414882e-06,
      "loss": 0.897,
      "step": 17351
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1072747707366943,
      "learning_rate": 9.532632655017092e-06,
      "loss": 0.8506,
      "step": 17352
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.332068681716919,
      "learning_rate": 9.531993738619303e-06,
      "loss": 0.8198,
      "step": 17353
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.2203118801116943,
      "learning_rate": 9.531354822221513e-06,
      "loss": 0.6226,
      "step": 17354
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.29862117767334,
      "learning_rate": 9.530715905823725e-06,
      "loss": 0.9109,
      "step": 17355
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5813770294189453,
      "learning_rate": 9.530076989425935e-06,
      "loss": 1.2072,
      "step": 17356
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8079560995101929,
      "learning_rate": 9.529438073028146e-06,
      "loss": 0.6822,
      "step": 17357
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.052158832550049,
      "learning_rate": 9.528799156630356e-06,
      "loss": 0.7969,
      "step": 17358
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.729299783706665,
      "learning_rate": 9.528160240232568e-06,
      "loss": 0.7665,
      "step": 17359
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.146228313446045,
      "learning_rate": 9.527521323834777e-06,
      "loss": 1.0492,
      "step": 17360
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.060004472732544,
      "learning_rate": 9.526882407436987e-06,
      "loss": 0.8671,
      "step": 17361
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9771114587783813,
      "learning_rate": 9.526243491039199e-06,
      "loss": 0.9138,
      "step": 17362
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6332348585128784,
      "learning_rate": 9.525604574641409e-06,
      "loss": 0.9185,
      "step": 17363
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7349075078964233,
      "learning_rate": 9.52496565824362e-06,
      "loss": 0.9321,
      "step": 17364
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2145981788635254,
      "learning_rate": 9.52432674184583e-06,
      "loss": 0.7497,
      "step": 17365
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.194794178009033,
      "learning_rate": 9.523687825448042e-06,
      "loss": 1.3775,
      "step": 17366
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6955664157867432,
      "learning_rate": 9.523048909050251e-06,
      "loss": 0.8558,
      "step": 17367
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4903316497802734,
      "learning_rate": 9.522409992652463e-06,
      "loss": 0.8185,
      "step": 17368
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.542060136795044,
      "learning_rate": 9.521771076254673e-06,
      "loss": 0.9492,
      "step": 17369
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9037235975265503,
      "learning_rate": 9.521132159856884e-06,
      "loss": 0.9183,
      "step": 17370
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9304730892181396,
      "learning_rate": 9.520493243459094e-06,
      "loss": 0.898,
      "step": 17371
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9214088916778564,
      "learning_rate": 9.519854327061306e-06,
      "loss": 1.0032,
      "step": 17372
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6458089351654053,
      "learning_rate": 9.519215410663516e-06,
      "loss": 0.7909,
      "step": 17373
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9030264616012573,
      "learning_rate": 9.518576494265727e-06,
      "loss": 0.9325,
      "step": 17374
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.982316493988037,
      "learning_rate": 9.517937577867937e-06,
      "loss": 0.9918,
      "step": 17375
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.0356409549713135,
      "learning_rate": 9.517298661470149e-06,
      "loss": 1.0558,
      "step": 17376
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6292580366134644,
      "learning_rate": 9.516659745072358e-06,
      "loss": 0.8983,
      "step": 17377
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.604604721069336,
      "learning_rate": 9.51602082867457e-06,
      "loss": 0.9729,
      "step": 17378
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7094461917877197,
      "learning_rate": 9.51538191227678e-06,
      "loss": 1.0137,
      "step": 17379
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.0321266651153564,
      "learning_rate": 9.51474299587899e-06,
      "loss": 0.6764,
      "step": 17380
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3635916709899902,
      "learning_rate": 9.514104079481201e-06,
      "loss": 1.0314,
      "step": 17381
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.128818988800049,
      "learning_rate": 9.513465163083411e-06,
      "loss": 1.2386,
      "step": 17382
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.636014938354492,
      "learning_rate": 9.512826246685623e-06,
      "loss": 1.1928,
      "step": 17383
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.850437879562378,
      "learning_rate": 9.512187330287833e-06,
      "loss": 0.8652,
      "step": 17384
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.4101710319519043,
      "learning_rate": 9.511548413890044e-06,
      "loss": 1.0734,
      "step": 17385
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7582345008850098,
      "learning_rate": 9.510909497492254e-06,
      "loss": 0.8741,
      "step": 17386
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5144898891448975,
      "learning_rate": 9.510270581094465e-06,
      "loss": 0.9112,
      "step": 17387
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.599714994430542,
      "learning_rate": 9.509631664696675e-06,
      "loss": 0.6528,
      "step": 17388
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.1762800216674805,
      "learning_rate": 9.508992748298887e-06,
      "loss": 1.4911,
      "step": 17389
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.345682382583618,
      "learning_rate": 9.508353831901097e-06,
      "loss": 0.8621,
      "step": 17390
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9977524280548096,
      "learning_rate": 9.507714915503308e-06,
      "loss": 0.9309,
      "step": 17391
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.34632396697998,
      "learning_rate": 9.507075999105518e-06,
      "loss": 1.5369,
      "step": 17392
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.5539121627807617,
      "learning_rate": 9.50643708270773e-06,
      "loss": 1.2779,
      "step": 17393
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.215529680252075,
      "learning_rate": 9.505798166309938e-06,
      "loss": 0.8982,
      "step": 17394
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0620875358581543,
      "learning_rate": 9.50515924991215e-06,
      "loss": 1.2967,
      "step": 17395
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6168512105941772,
      "learning_rate": 9.50452033351436e-06,
      "loss": 0.76,
      "step": 17396
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.528592109680176,
      "learning_rate": 9.50388141711657e-06,
      "loss": 0.9682,
      "step": 17397
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6837377548217773,
      "learning_rate": 9.50324250071878e-06,
      "loss": 0.8468,
      "step": 17398
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7268688678741455,
      "learning_rate": 9.502603584320992e-06,
      "loss": 0.6641,
      "step": 17399
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2320408821105957,
      "learning_rate": 9.501964667923202e-06,
      "loss": 0.9662,
      "step": 17400
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5045140981674194,
      "learning_rate": 9.501325751525414e-06,
      "loss": 0.4489,
      "step": 17401
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.053273916244507,
      "learning_rate": 9.500686835127623e-06,
      "loss": 1.385,
      "step": 17402
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7160946130752563,
      "learning_rate": 9.500047918729835e-06,
      "loss": 0.8102,
      "step": 17403
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2283904552459717,
      "learning_rate": 9.499409002332045e-06,
      "loss": 0.8534,
      "step": 17404
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.299736499786377,
      "learning_rate": 9.498770085934256e-06,
      "loss": 1.5998,
      "step": 17405
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7483834028244019,
      "learning_rate": 9.498131169536466e-06,
      "loss": 0.4706,
      "step": 17406
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.62740957736969,
      "learning_rate": 9.497492253138678e-06,
      "loss": 0.8164,
      "step": 17407
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9184094667434692,
      "learning_rate": 9.496853336740888e-06,
      "loss": 0.9,
      "step": 17408
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5914933681488037,
      "learning_rate": 9.4962144203431e-06,
      "loss": 1.0475,
      "step": 17409
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0510408878326416,
      "learning_rate": 9.495575503945309e-06,
      "loss": 0.8958,
      "step": 17410
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.85145902633667,
      "learning_rate": 9.494936587547519e-06,
      "loss": 0.8496,
      "step": 17411
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.98384690284729,
      "learning_rate": 9.49429767114973e-06,
      "loss": 0.9485,
      "step": 17412
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7526886463165283,
      "learning_rate": 9.49365875475194e-06,
      "loss": 0.6713,
      "step": 17413
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9029881954193115,
      "learning_rate": 9.493019838354152e-06,
      "loss": 0.9329,
      "step": 17414
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.741620659828186,
      "learning_rate": 9.492380921956362e-06,
      "loss": 0.8983,
      "step": 17415
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.8649210929870605,
      "learning_rate": 9.491742005558573e-06,
      "loss": 1.229,
      "step": 17416
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0241880416870117,
      "learning_rate": 9.491103089160783e-06,
      "loss": 0.9017,
      "step": 17417
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.904942750930786,
      "learning_rate": 9.490464172762995e-06,
      "loss": 1.0344,
      "step": 17418
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.8240809440612793,
      "learning_rate": 9.489825256365205e-06,
      "loss": 0.919,
      "step": 17419
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9286003112792969,
      "learning_rate": 9.489186339967416e-06,
      "loss": 0.874,
      "step": 17420
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8734121322631836,
      "learning_rate": 9.488547423569626e-06,
      "loss": 0.9708,
      "step": 17421
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8324164152145386,
      "learning_rate": 9.487908507171837e-06,
      "loss": 0.9561,
      "step": 17422
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7941325902938843,
      "learning_rate": 9.487269590774047e-06,
      "loss": 0.952,
      "step": 17423
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7489005327224731,
      "learning_rate": 9.486630674376259e-06,
      "loss": 0.9129,
      "step": 17424
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.146660804748535,
      "learning_rate": 9.485991757978469e-06,
      "loss": 0.7965,
      "step": 17425
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1725354194641113,
      "learning_rate": 9.48535284158068e-06,
      "loss": 0.9735,
      "step": 17426
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5402634143829346,
      "learning_rate": 9.48471392518289e-06,
      "loss": 0.914,
      "step": 17427
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8481124639511108,
      "learning_rate": 9.4840750087851e-06,
      "loss": 0.9399,
      "step": 17428
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.87869930267334,
      "learning_rate": 9.483436092387312e-06,
      "loss": 1.0841,
      "step": 17429
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7280880212783813,
      "learning_rate": 9.482797175989521e-06,
      "loss": 0.8542,
      "step": 17430
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.394827365875244,
      "learning_rate": 9.482158259591733e-06,
      "loss": 0.8239,
      "step": 17431
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0494186878204346,
      "learning_rate": 9.481519343193943e-06,
      "loss": 1.1767,
      "step": 17432
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9696725606918335,
      "learning_rate": 9.480880426796154e-06,
      "loss": 0.9765,
      "step": 17433
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.3687102794647217,
      "learning_rate": 9.480241510398364e-06,
      "loss": 0.95,
      "step": 17434
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9848451614379883,
      "learning_rate": 9.479602594000576e-06,
      "loss": 0.9512,
      "step": 17435
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.4681472778320312,
      "learning_rate": 9.478963677602786e-06,
      "loss": 0.993,
      "step": 17436
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.689317226409912,
      "learning_rate": 9.478324761204997e-06,
      "loss": 1.556,
      "step": 17437
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.230368137359619,
      "learning_rate": 9.477685844807207e-06,
      "loss": 0.8906,
      "step": 17438
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5634868144989014,
      "learning_rate": 9.477046928409419e-06,
      "loss": 0.7729,
      "step": 17439
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6710138320922852,
      "learning_rate": 9.476408012011628e-06,
      "loss": 0.4626,
      "step": 17440
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6945061683654785,
      "learning_rate": 9.47576909561384e-06,
      "loss": 0.8407,
      "step": 17441
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.374016046524048,
      "learning_rate": 9.47513017921605e-06,
      "loss": 1.3637,
      "step": 17442
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6673328876495361,
      "learning_rate": 9.474491262818261e-06,
      "loss": 0.9091,
      "step": 17443
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.857609748840332,
      "learning_rate": 9.473852346420471e-06,
      "loss": 0.7874,
      "step": 17444
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7800809144973755,
      "learning_rate": 9.473213430022683e-06,
      "loss": 0.8424,
      "step": 17445
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.555856943130493,
      "learning_rate": 9.472574513624893e-06,
      "loss": 0.8358,
      "step": 17446
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.027292251586914,
      "learning_rate": 9.471935597227102e-06,
      "loss": 0.8135,
      "step": 17447
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.972262978553772,
      "learning_rate": 9.471296680829314e-06,
      "loss": 0.9331,
      "step": 17448
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7940036058425903,
      "learning_rate": 9.470657764431524e-06,
      "loss": 0.8727,
      "step": 17449
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.252639293670654,
      "learning_rate": 9.470018848033735e-06,
      "loss": 1.3825,
      "step": 17450
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7172744274139404,
      "learning_rate": 9.469379931635945e-06,
      "loss": 0.4796,
      "step": 17451
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8534348011016846,
      "learning_rate": 9.468741015238157e-06,
      "loss": 0.9205,
      "step": 17452
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4597238302230835,
      "learning_rate": 9.468102098840367e-06,
      "loss": 0.7473,
      "step": 17453
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0988104343414307,
      "learning_rate": 9.467463182442578e-06,
      "loss": 1.0068,
      "step": 17454
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.436849594116211,
      "learning_rate": 9.466824266044788e-06,
      "loss": 0.9335,
      "step": 17455
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.076796770095825,
      "learning_rate": 9.466185349647e-06,
      "loss": 1.2602,
      "step": 17456
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4778351783752441,
      "learning_rate": 9.46554643324921e-06,
      "loss": 0.771,
      "step": 17457
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6141456365585327,
      "learning_rate": 9.464907516851421e-06,
      "loss": 0.8129,
      "step": 17458
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8062185049057007,
      "learning_rate": 9.46426860045363e-06,
      "loss": 0.9411,
      "step": 17459
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.406679391860962,
      "learning_rate": 9.463629684055842e-06,
      "loss": 0.8361,
      "step": 17460
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9962843656539917,
      "learning_rate": 9.462990767658052e-06,
      "loss": 0.8304,
      "step": 17461
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3257648944854736,
      "learning_rate": 9.462351851260264e-06,
      "loss": 0.9699,
      "step": 17462
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.834869146347046,
      "learning_rate": 9.461712934862474e-06,
      "loss": 0.9579,
      "step": 17463
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.235109806060791,
      "learning_rate": 9.461074018464685e-06,
      "loss": 0.9988,
      "step": 17464
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6346404552459717,
      "learning_rate": 9.460435102066895e-06,
      "loss": 0.8458,
      "step": 17465
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3136487007141113,
      "learning_rate": 9.459796185669105e-06,
      "loss": 0.4606,
      "step": 17466
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7358922958374023,
      "learning_rate": 9.459157269271316e-06,
      "loss": 0.9157,
      "step": 17467
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1158361434936523,
      "learning_rate": 9.458518352873526e-06,
      "loss": 0.8025,
      "step": 17468
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7966983318328857,
      "learning_rate": 9.457879436475738e-06,
      "loss": 0.8256,
      "step": 17469
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.2904930114746094,
      "learning_rate": 9.457240520077948e-06,
      "loss": 1.1002,
      "step": 17470
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.150696039199829,
      "learning_rate": 9.45660160368016e-06,
      "loss": 0.968,
      "step": 17471
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.161262512207031,
      "learning_rate": 9.455962687282369e-06,
      "loss": 1.2334,
      "step": 17472
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9045367240905762,
      "learning_rate": 9.45532377088458e-06,
      "loss": 0.8938,
      "step": 17473
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.769077181816101,
      "learning_rate": 9.45468485448679e-06,
      "loss": 0.9009,
      "step": 17474
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2746469974517822,
      "learning_rate": 9.454045938089002e-06,
      "loss": 0.9899,
      "step": 17475
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.397629737854004,
      "learning_rate": 9.453407021691212e-06,
      "loss": 1.0575,
      "step": 17476
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.81534481048584,
      "learning_rate": 9.452768105293423e-06,
      "loss": 0.5367,
      "step": 17477
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.186354637145996,
      "learning_rate": 9.452129188895633e-06,
      "loss": 0.9962,
      "step": 17478
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.777036428451538,
      "learning_rate": 9.451490272497845e-06,
      "loss": 1.09,
      "step": 17479
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.5042171478271484,
      "learning_rate": 9.450851356100055e-06,
      "loss": 0.9714,
      "step": 17480
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.785819411277771,
      "learning_rate": 9.450212439702266e-06,
      "loss": 0.8893,
      "step": 17481
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6119688749313354,
      "learning_rate": 9.449573523304476e-06,
      "loss": 0.8821,
      "step": 17482
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0566680431365967,
      "learning_rate": 9.448934606906686e-06,
      "loss": 0.8178,
      "step": 17483
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.715343952178955,
      "learning_rate": 9.448295690508897e-06,
      "loss": 1.0142,
      "step": 17484
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.777976989746094,
      "learning_rate": 9.447656774111107e-06,
      "loss": 1.5544,
      "step": 17485
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4851967096328735,
      "learning_rate": 9.447017857713319e-06,
      "loss": 0.8384,
      "step": 17486
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.793475866317749,
      "learning_rate": 9.446378941315529e-06,
      "loss": 0.9245,
      "step": 17487
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1647655963897705,
      "learning_rate": 9.44574002491774e-06,
      "loss": 0.7818,
      "step": 17488
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.613770008087158,
      "learning_rate": 9.44510110851995e-06,
      "loss": 0.7075,
      "step": 17489
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.654496192932129,
      "learning_rate": 9.444462192122162e-06,
      "loss": 1.1519,
      "step": 17490
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.285553455352783,
      "learning_rate": 9.443823275724372e-06,
      "loss": 1.0667,
      "step": 17491
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.8531196117401123,
      "learning_rate": 9.443184359326583e-06,
      "loss": 1.1354,
      "step": 17492
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.838002920150757,
      "learning_rate": 9.442545442928793e-06,
      "loss": 1.0189,
      "step": 17493
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.0705764293670654,
      "learning_rate": 9.441906526531004e-06,
      "loss": 1.0951,
      "step": 17494
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.2969086170196533,
      "learning_rate": 9.441267610133214e-06,
      "loss": 0.9706,
      "step": 17495
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9303748607635498,
      "learning_rate": 9.440628693735426e-06,
      "loss": 0.7965,
      "step": 17496
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5886369943618774,
      "learning_rate": 9.439989777337636e-06,
      "loss": 0.7965,
      "step": 17497
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0798754692077637,
      "learning_rate": 9.439350860939847e-06,
      "loss": 0.9389,
      "step": 17498
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9769935607910156,
      "learning_rate": 9.438711944542057e-06,
      "loss": 0.8268,
      "step": 17499
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7320626974105835,
      "learning_rate": 9.438073028144269e-06,
      "loss": 0.9711,
      "step": 17500
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7548500299453735,
      "learning_rate": 9.437434111746479e-06,
      "loss": 1.0265,
      "step": 17501
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6922699213027954,
      "learning_rate": 9.436795195348688e-06,
      "loss": 0.9123,
      "step": 17502
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.392081618309021,
      "learning_rate": 9.4361562789509e-06,
      "loss": 0.353,
      "step": 17503
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1389408111572266,
      "learning_rate": 9.43551736255311e-06,
      "loss": 0.9828,
      "step": 17504
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.000108003616333,
      "learning_rate": 9.434878446155321e-06,
      "loss": 1.084,
      "step": 17505
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.387083053588867,
      "learning_rate": 9.434239529757531e-06,
      "loss": 1.8936,
      "step": 17506
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0337305068969727,
      "learning_rate": 9.433600613359743e-06,
      "loss": 1.18,
      "step": 17507
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.726872444152832,
      "learning_rate": 9.432961696961953e-06,
      "loss": 1.1357,
      "step": 17508
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.438979625701904,
      "learning_rate": 9.432322780564164e-06,
      "loss": 0.881,
      "step": 17509
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6692988872528076,
      "learning_rate": 9.431683864166374e-06,
      "loss": 0.7823,
      "step": 17510
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6983803510665894,
      "learning_rate": 9.431044947768586e-06,
      "loss": 0.9586,
      "step": 17511
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6430751085281372,
      "learning_rate": 9.430406031370795e-06,
      "loss": 0.937,
      "step": 17512
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.385513424873352,
      "learning_rate": 9.429767114973007e-06,
      "loss": 0.8419,
      "step": 17513
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.626672387123108,
      "learning_rate": 9.429128198575217e-06,
      "loss": 0.9036,
      "step": 17514
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6044275760650635,
      "learning_rate": 9.428489282177428e-06,
      "loss": 0.8351,
      "step": 17515
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8778308629989624,
      "learning_rate": 9.427850365779638e-06,
      "loss": 0.9547,
      "step": 17516
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5403993129730225,
      "learning_rate": 9.42721144938185e-06,
      "loss": 0.9053,
      "step": 17517
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3723777532577515,
      "learning_rate": 9.42657253298406e-06,
      "loss": 0.4371,
      "step": 17518
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.981896162033081,
      "learning_rate": 9.425933616586271e-06,
      "loss": 0.9415,
      "step": 17519
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8472732305526733,
      "learning_rate": 9.425294700188481e-06,
      "loss": 0.9438,
      "step": 17520
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5357695817947388,
      "learning_rate": 9.42465578379069e-06,
      "loss": 0.832,
      "step": 17521
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.627399206161499,
      "learning_rate": 9.424016867392902e-06,
      "loss": 0.9001,
      "step": 17522
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.765781879425049,
      "learning_rate": 9.423377950995112e-06,
      "loss": 0.8495,
      "step": 17523
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6587879657745361,
      "learning_rate": 9.422739034597324e-06,
      "loss": 0.8913,
      "step": 17524
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.448488712310791,
      "learning_rate": 9.422100118199534e-06,
      "loss": 1.4535,
      "step": 17525
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4411894083023071,
      "learning_rate": 9.421461201801745e-06,
      "loss": 0.9288,
      "step": 17526
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9116096496582031,
      "learning_rate": 9.420822285403955e-06,
      "loss": 0.9514,
      "step": 17527
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8388808965682983,
      "learning_rate": 9.420183369006167e-06,
      "loss": 0.9123,
      "step": 17528
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.867354393005371,
      "learning_rate": 9.419544452608376e-06,
      "loss": 1.3872,
      "step": 17529
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.5738022327423096,
      "learning_rate": 9.418905536210588e-06,
      "loss": 0.9639,
      "step": 17530
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0057623386383057,
      "learning_rate": 9.418266619812798e-06,
      "loss": 0.9355,
      "step": 17531
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.187028408050537,
      "learning_rate": 9.41762770341501e-06,
      "loss": 1.1553,
      "step": 17532
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5912331342697144,
      "learning_rate": 9.41698878701722e-06,
      "loss": 0.3981,
      "step": 17533
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9048811197280884,
      "learning_rate": 9.41634987061943e-06,
      "loss": 0.918,
      "step": 17534
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5931044816970825,
      "learning_rate": 9.41571095422164e-06,
      "loss": 0.8261,
      "step": 17535
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4825356006622314,
      "learning_rate": 9.415072037823852e-06,
      "loss": 0.9101,
      "step": 17536
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5203732252120972,
      "learning_rate": 9.414433121426062e-06,
      "loss": 0.541,
      "step": 17537
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.639875650405884,
      "learning_rate": 9.413794205028272e-06,
      "loss": 0.8333,
      "step": 17538
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7675408124923706,
      "learning_rate": 9.413155288630483e-06,
      "loss": 0.8762,
      "step": 17539
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.342806577682495,
      "learning_rate": 9.412516372232693e-06,
      "loss": 0.8618,
      "step": 17540
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.369406223297119,
      "learning_rate": 9.411877455834905e-06,
      "loss": 0.885,
      "step": 17541
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.2026047706604004,
      "learning_rate": 9.411238539437115e-06,
      "loss": 0.9701,
      "step": 17542
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5241481065750122,
      "learning_rate": 9.410599623039326e-06,
      "loss": 0.855,
      "step": 17543
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6702933311462402,
      "learning_rate": 9.409960706641536e-06,
      "loss": 0.9127,
      "step": 17544
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.146562099456787,
      "learning_rate": 9.409321790243748e-06,
      "loss": 1.0485,
      "step": 17545
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.6488349437713623,
      "learning_rate": 9.408682873845957e-06,
      "loss": 0.9709,
      "step": 17546
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5723659992218018,
      "learning_rate": 9.408043957448169e-06,
      "loss": 0.8546,
      "step": 17547
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.274592638015747,
      "learning_rate": 9.407405041050379e-06,
      "loss": 0.9252,
      "step": 17548
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.6985981464385986,
      "learning_rate": 9.40676612465259e-06,
      "loss": 1.0487,
      "step": 17549
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3786081075668335,
      "learning_rate": 9.4061272082548e-06,
      "loss": 0.8883,
      "step": 17550
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6675416231155396,
      "learning_rate": 9.405488291857012e-06,
      "loss": 0.739,
      "step": 17551
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7033429145812988,
      "learning_rate": 9.404849375459222e-06,
      "loss": 0.8662,
      "step": 17552
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0925369262695312,
      "learning_rate": 9.404210459061433e-06,
      "loss": 1.2398,
      "step": 17553
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.771200180053711,
      "learning_rate": 9.403571542663643e-06,
      "loss": 1.5406,
      "step": 17554
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9513895511627197,
      "learning_rate": 9.402932626265855e-06,
      "loss": 0.9598,
      "step": 17555
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0919172763824463,
      "learning_rate": 9.402293709868064e-06,
      "loss": 0.9622,
      "step": 17556
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.508815288543701,
      "learning_rate": 9.401654793470274e-06,
      "loss": 1.9687,
      "step": 17557
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9622544050216675,
      "learning_rate": 9.401015877072486e-06,
      "loss": 0.4963,
      "step": 17558
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9227710962295532,
      "learning_rate": 9.400376960674696e-06,
      "loss": 0.9384,
      "step": 17559
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.747056245803833,
      "learning_rate": 9.399738044276907e-06,
      "loss": 1.3154,
      "step": 17560
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.657164454460144,
      "learning_rate": 9.399099127879117e-06,
      "loss": 0.9495,
      "step": 17561
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.832520604133606,
      "learning_rate": 9.398460211481329e-06,
      "loss": 1.023,
      "step": 17562
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.449278354644775,
      "learning_rate": 9.397821295083539e-06,
      "loss": 1.6456,
      "step": 17563
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.244785785675049,
      "learning_rate": 9.39718237868575e-06,
      "loss": 1.1112,
      "step": 17564
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.033613920211792,
      "learning_rate": 9.39654346228796e-06,
      "loss": 0.7096,
      "step": 17565
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6843187808990479,
      "learning_rate": 9.395904545890171e-06,
      "loss": 0.9208,
      "step": 17566
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.661692500114441,
      "learning_rate": 9.395265629492381e-06,
      "loss": 0.8715,
      "step": 17567
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9991775751113892,
      "learning_rate": 9.394626713094593e-06,
      "loss": 0.9199,
      "step": 17568
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7564828395843506,
      "learning_rate": 9.393987796696803e-06,
      "loss": 1.0827,
      "step": 17569
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9157100915908813,
      "learning_rate": 9.393348880299014e-06,
      "loss": 0.9637,
      "step": 17570
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2859303951263428,
      "learning_rate": 9.392709963901224e-06,
      "loss": 0.9761,
      "step": 17571
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8132041692733765,
      "learning_rate": 9.392071047503436e-06,
      "loss": 0.8117,
      "step": 17572
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.193379878997803,
      "learning_rate": 9.391432131105646e-06,
      "loss": 1.2995,
      "step": 17573
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.079892873764038,
      "learning_rate": 9.390793214707855e-06,
      "loss": 0.9249,
      "step": 17574
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3800735473632812,
      "learning_rate": 9.390154298310067e-06,
      "loss": 0.9757,
      "step": 17575
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7600661516189575,
      "learning_rate": 9.389515381912277e-06,
      "loss": 0.9025,
      "step": 17576
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3058260679244995,
      "learning_rate": 9.388876465514488e-06,
      "loss": 0.8556,
      "step": 17577
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.634767532348633,
      "learning_rate": 9.388237549116698e-06,
      "loss": 0.8591,
      "step": 17578
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3643243312835693,
      "learning_rate": 9.38759863271891e-06,
      "loss": 0.7615,
      "step": 17579
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.032545804977417,
      "learning_rate": 9.38695971632112e-06,
      "loss": 0.9585,
      "step": 17580
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.9763410091400146,
      "learning_rate": 9.386320799923331e-06,
      "loss": 0.8953,
      "step": 17581
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.3616747856140137,
      "learning_rate": 9.385681883525541e-06,
      "loss": 0.7849,
      "step": 17582
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8021681308746338,
      "learning_rate": 9.385042967127753e-06,
      "loss": 0.7846,
      "step": 17583
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.217149019241333,
      "learning_rate": 9.384404050729962e-06,
      "loss": 0.9145,
      "step": 17584
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7396529912948608,
      "learning_rate": 9.383765134332174e-06,
      "loss": 0.7972,
      "step": 17585
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6997629404067993,
      "learning_rate": 9.383126217934384e-06,
      "loss": 0.706,
      "step": 17586
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8377825021743774,
      "learning_rate": 9.382487301536595e-06,
      "loss": 0.9537,
      "step": 17587
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.417308807373047,
      "learning_rate": 9.381848385138805e-06,
      "loss": 0.9871,
      "step": 17588
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9513418674468994,
      "learning_rate": 9.381209468741017e-06,
      "loss": 1.0708,
      "step": 17589
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.156350612640381,
      "learning_rate": 9.380570552343227e-06,
      "loss": 0.9514,
      "step": 17590
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3796623945236206,
      "learning_rate": 9.379931635945438e-06,
      "loss": 0.9094,
      "step": 17591
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3351542949676514,
      "learning_rate": 9.379292719547648e-06,
      "loss": 0.9241,
      "step": 17592
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5176639556884766,
      "learning_rate": 9.378653803149858e-06,
      "loss": 1.2346,
      "step": 17593
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.137573480606079,
      "learning_rate": 9.37801488675207e-06,
      "loss": 1.2233,
      "step": 17594
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7417359352111816,
      "learning_rate": 9.37737597035428e-06,
      "loss": 0.7824,
      "step": 17595
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.453940987586975,
      "learning_rate": 9.37673705395649e-06,
      "loss": 0.8535,
      "step": 17596
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.745678424835205,
      "learning_rate": 9.3760981375587e-06,
      "loss": 0.4367,
      "step": 17597
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6121248006820679,
      "learning_rate": 9.375459221160912e-06,
      "loss": 0.9436,
      "step": 17598
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.359156608581543,
      "learning_rate": 9.374820304763122e-06,
      "loss": 1.4154,
      "step": 17599
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6400580406188965,
      "learning_rate": 9.374181388365334e-06,
      "loss": 0.9559,
      "step": 17600
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.1902594566345215,
      "learning_rate": 9.373542471967543e-06,
      "loss": 1.0784,
      "step": 17601
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5826261043548584,
      "learning_rate": 9.372903555569755e-06,
      "loss": 0.6082,
      "step": 17602
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.416034698486328,
      "learning_rate": 9.372264639171965e-06,
      "loss": 1.4967,
      "step": 17603
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6294485330581665,
      "learning_rate": 9.371625722774176e-06,
      "loss": 0.9815,
      "step": 17604
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4427037239074707,
      "learning_rate": 9.370986806376386e-06,
      "loss": 0.9458,
      "step": 17605
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6808422803878784,
      "learning_rate": 9.370347889978598e-06,
      "loss": 0.8072,
      "step": 17606
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.027254104614258,
      "learning_rate": 9.369708973580808e-06,
      "loss": 1.1803,
      "step": 17607
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7784700393676758,
      "learning_rate": 9.36907005718302e-06,
      "loss": 1.3042,
      "step": 17608
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.700767993927002,
      "learning_rate": 9.368431140785229e-06,
      "loss": 1.9265,
      "step": 17609
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8740400075912476,
      "learning_rate": 9.36779222438744e-06,
      "loss": 0.9673,
      "step": 17610
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.204431533813477,
      "learning_rate": 9.36715330798965e-06,
      "loss": 1.2438,
      "step": 17611
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7132751941680908,
      "learning_rate": 9.36651439159186e-06,
      "loss": 0.9928,
      "step": 17612
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8075803518295288,
      "learning_rate": 9.365875475194072e-06,
      "loss": 0.8616,
      "step": 17613
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5842585563659668,
      "learning_rate": 9.365236558796282e-06,
      "loss": 0.8065,
      "step": 17614
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.425434947013855,
      "learning_rate": 9.364597642398493e-06,
      "loss": 0.3977,
      "step": 17615
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.507565498352051,
      "learning_rate": 9.363958726000703e-06,
      "loss": 1.3721,
      "step": 17616
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.6737141609191895,
      "learning_rate": 9.363319809602915e-06,
      "loss": 0.8015,
      "step": 17617
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.410207986831665,
      "learning_rate": 9.362680893205124e-06,
      "loss": 0.8512,
      "step": 17618
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4784411191940308,
      "learning_rate": 9.362041976807336e-06,
      "loss": 0.4791,
      "step": 17619
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6363801956176758,
      "learning_rate": 9.361403060409546e-06,
      "loss": 0.8555,
      "step": 17620
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7527529001235962,
      "learning_rate": 9.360764144011757e-06,
      "loss": 0.8608,
      "step": 17621
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1685914993286133,
      "learning_rate": 9.360125227613967e-06,
      "loss": 0.9116,
      "step": 17622
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4395473003387451,
      "learning_rate": 9.359486311216179e-06,
      "loss": 0.4544,
      "step": 17623
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.061178684234619,
      "learning_rate": 9.358847394818389e-06,
      "loss": 0.8312,
      "step": 17624
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7753483057022095,
      "learning_rate": 9.3582084784206e-06,
      "loss": 0.8824,
      "step": 17625
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.485467553138733,
      "learning_rate": 9.35756956202281e-06,
      "loss": 0.8274,
      "step": 17626
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6525105237960815,
      "learning_rate": 9.356930645625022e-06,
      "loss": 0.8887,
      "step": 17627
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5634316205978394,
      "learning_rate": 9.356291729227231e-06,
      "loss": 0.9409,
      "step": 17628
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.9435904026031494,
      "learning_rate": 9.355652812829441e-06,
      "loss": 1.1767,
      "step": 17629
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6894941329956055,
      "learning_rate": 9.355013896431653e-06,
      "loss": 0.7882,
      "step": 17630
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5447219610214233,
      "learning_rate": 9.354374980033863e-06,
      "loss": 0.8958,
      "step": 17631
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.659602165222168,
      "learning_rate": 9.353736063636074e-06,
      "loss": 0.8521,
      "step": 17632
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7898005247116089,
      "learning_rate": 9.353097147238284e-06,
      "loss": 0.8798,
      "step": 17633
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.020026683807373,
      "learning_rate": 9.352458230840496e-06,
      "loss": 1.6587,
      "step": 17634
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8884556293487549,
      "learning_rate": 9.351819314442706e-06,
      "loss": 0.9815,
      "step": 17635
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9910467863082886,
      "learning_rate": 9.351180398044917e-06,
      "loss": 1.0101,
      "step": 17636
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.069593667984009,
      "learning_rate": 9.350541481647127e-06,
      "loss": 0.8936,
      "step": 17637
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.951778769493103,
      "learning_rate": 9.349902565249338e-06,
      "loss": 0.8707,
      "step": 17638
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1228232383728027,
      "learning_rate": 9.349263648851548e-06,
      "loss": 0.8559,
      "step": 17639
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5165258646011353,
      "learning_rate": 9.34862473245376e-06,
      "loss": 0.4645,
      "step": 17640
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5822803974151611,
      "learning_rate": 9.34798581605597e-06,
      "loss": 0.8387,
      "step": 17641
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.525753140449524,
      "learning_rate": 9.347346899658181e-06,
      "loss": 0.8122,
      "step": 17642
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.591538906097412,
      "learning_rate": 9.346707983260391e-06,
      "loss": 1.1843,
      "step": 17643
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8045289516448975,
      "learning_rate": 9.346069066862603e-06,
      "loss": 0.8741,
      "step": 17644
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6882894039154053,
      "learning_rate": 9.345430150464813e-06,
      "loss": 1.0013,
      "step": 17645
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.6927871704101562,
      "learning_rate": 9.344791234067024e-06,
      "loss": 1.0494,
      "step": 17646
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.90478777885437,
      "learning_rate": 9.344152317669234e-06,
      "loss": 0.8091,
      "step": 17647
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.427532196044922,
      "learning_rate": 9.343513401271444e-06,
      "loss": 0.9754,
      "step": 17648
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1398873329162598,
      "learning_rate": 9.342874484873655e-06,
      "loss": 0.9045,
      "step": 17649
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.824054479598999,
      "learning_rate": 9.342235568475865e-06,
      "loss": 0.7712,
      "step": 17650
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.414939045906067,
      "learning_rate": 9.341596652078077e-06,
      "loss": 0.5056,
      "step": 17651
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4242515563964844,
      "learning_rate": 9.340957735680287e-06,
      "loss": 0.9302,
      "step": 17652
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9690696001052856,
      "learning_rate": 9.340318819282498e-06,
      "loss": 0.8735,
      "step": 17653
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8627289533615112,
      "learning_rate": 9.339679902884708e-06,
      "loss": 0.9675,
      "step": 17654
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9276491403579712,
      "learning_rate": 9.33904098648692e-06,
      "loss": 1.2503,
      "step": 17655
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.186230421066284,
      "learning_rate": 9.33840207008913e-06,
      "loss": 0.899,
      "step": 17656
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.568572521209717,
      "learning_rate": 9.337763153691341e-06,
      "loss": 1.1113,
      "step": 17657
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3334474563598633,
      "learning_rate": 9.33712423729355e-06,
      "loss": 0.8723,
      "step": 17658
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5504952669143677,
      "learning_rate": 9.336485320895762e-06,
      "loss": 0.8838,
      "step": 17659
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9231574535369873,
      "learning_rate": 9.335846404497972e-06,
      "loss": 0.9422,
      "step": 17660
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.7868213653564453,
      "learning_rate": 9.335207488100184e-06,
      "loss": 1.2142,
      "step": 17661
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.799121379852295,
      "learning_rate": 9.334568571702394e-06,
      "loss": 0.7834,
      "step": 17662
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.439649820327759,
      "learning_rate": 9.333929655304605e-06,
      "loss": 1.1064,
      "step": 17663
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.019280195236206,
      "learning_rate": 9.333290738906815e-06,
      "loss": 0.8606,
      "step": 17664
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.754732370376587,
      "learning_rate": 9.332651822509027e-06,
      "loss": 0.8174,
      "step": 17665
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.9503962993621826,
      "learning_rate": 9.332012906111236e-06,
      "loss": 1.0911,
      "step": 17666
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7692269086837769,
      "learning_rate": 9.331373989713446e-06,
      "loss": 0.7046,
      "step": 17667
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.551548957824707,
      "learning_rate": 9.330735073315658e-06,
      "loss": 0.7858,
      "step": 17668
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.374223232269287,
      "learning_rate": 9.330096156917868e-06,
      "loss": 0.8322,
      "step": 17669
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.589615821838379,
      "learning_rate": 9.32945724052008e-06,
      "loss": 1.0029,
      "step": 17670
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.743083119392395,
      "learning_rate": 9.328818324122289e-06,
      "loss": 0.8719,
      "step": 17671
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6594444513320923,
      "learning_rate": 9.3281794077245e-06,
      "loss": 0.8307,
      "step": 17672
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.6342685222625732,
      "learning_rate": 9.32754049132671e-06,
      "loss": 0.8332,
      "step": 17673
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.620138645172119,
      "learning_rate": 9.326901574928922e-06,
      "loss": 0.7945,
      "step": 17674
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.896026611328125,
      "learning_rate": 9.326262658531132e-06,
      "loss": 0.827,
      "step": 17675
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5292372703552246,
      "learning_rate": 9.325623742133343e-06,
      "loss": 0.7478,
      "step": 17676
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.197389125823975,
      "learning_rate": 9.324984825735553e-06,
      "loss": 1.5206,
      "step": 17677
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7088077068328857,
      "learning_rate": 9.324345909337765e-06,
      "loss": 0.7432,
      "step": 17678
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3808298110961914,
      "learning_rate": 9.323706992939975e-06,
      "loss": 0.8923,
      "step": 17679
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6573091745376587,
      "learning_rate": 9.323068076542186e-06,
      "loss": 0.8028,
      "step": 17680
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.182589292526245,
      "learning_rate": 9.322429160144396e-06,
      "loss": 1.1002,
      "step": 17681
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.9327433109283447,
      "learning_rate": 9.321790243746608e-06,
      "loss": 0.6357,
      "step": 17682
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9965239763259888,
      "learning_rate": 9.321151327348817e-06,
      "loss": 0.8254,
      "step": 17683
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9525846242904663,
      "learning_rate": 9.320512410951027e-06,
      "loss": 0.9049,
      "step": 17684
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.325505495071411,
      "learning_rate": 9.319873494553239e-06,
      "loss": 0.8178,
      "step": 17685
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1346395015716553,
      "learning_rate": 9.319234578155449e-06,
      "loss": 0.9214,
      "step": 17686
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8556941747665405,
      "learning_rate": 9.31859566175766e-06,
      "loss": 0.9474,
      "step": 17687
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7056695222854614,
      "learning_rate": 9.31795674535987e-06,
      "loss": 0.9148,
      "step": 17688
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1497576236724854,
      "learning_rate": 9.317317828962082e-06,
      "loss": 0.9104,
      "step": 17689
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.511099100112915,
      "learning_rate": 9.316678912564291e-06,
      "loss": 0.8492,
      "step": 17690
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6754519939422607,
      "learning_rate": 9.316039996166503e-06,
      "loss": 0.7429,
      "step": 17691
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.730574369430542,
      "learning_rate": 9.315401079768713e-06,
      "loss": 0.8584,
      "step": 17692
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.610837697982788,
      "learning_rate": 9.314762163370924e-06,
      "loss": 1.0623,
      "step": 17693
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1438562870025635,
      "learning_rate": 9.314123246973134e-06,
      "loss": 1.0109,
      "step": 17694
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.762527585029602,
      "learning_rate": 9.313484330575346e-06,
      "loss": 0.8972,
      "step": 17695
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0388176441192627,
      "learning_rate": 9.312845414177556e-06,
      "loss": 0.9358,
      "step": 17696
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.928450345993042,
      "learning_rate": 9.312206497779767e-06,
      "loss": 0.8715,
      "step": 17697
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.591870903968811,
      "learning_rate": 9.311567581381977e-06,
      "loss": 0.8361,
      "step": 17698
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.42760169506073,
      "learning_rate": 9.310928664984189e-06,
      "loss": 0.7038,
      "step": 17699
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.423199415206909,
      "learning_rate": 9.310289748586398e-06,
      "loss": 0.8734,
      "step": 17700
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5124502182006836,
      "learning_rate": 9.30965083218861e-06,
      "loss": 0.8723,
      "step": 17701
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7814332246780396,
      "learning_rate": 9.30901191579082e-06,
      "loss": 0.8425,
      "step": 17702
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.785478353500366,
      "learning_rate": 9.30837299939303e-06,
      "loss": 0.8487,
      "step": 17703
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6933743953704834,
      "learning_rate": 9.307734082995241e-06,
      "loss": 0.7924,
      "step": 17704
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8289624452590942,
      "learning_rate": 9.307095166597451e-06,
      "loss": 0.8327,
      "step": 17705
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9836316108703613,
      "learning_rate": 9.306456250199663e-06,
      "loss": 0.9055,
      "step": 17706
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.769588589668274,
      "learning_rate": 9.305817333801873e-06,
      "loss": 0.8898,
      "step": 17707
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5268633365631104,
      "learning_rate": 9.305178417404084e-06,
      "loss": 0.9167,
      "step": 17708
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7529833316802979,
      "learning_rate": 9.304539501006294e-06,
      "loss": 0.8488,
      "step": 17709
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.253976583480835,
      "learning_rate": 9.303900584608505e-06,
      "loss": 0.9606,
      "step": 17710
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5628020763397217,
      "learning_rate": 9.303261668210715e-06,
      "loss": 0.933,
      "step": 17711
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.030863046646118,
      "learning_rate": 9.302622751812927e-06,
      "loss": 0.9583,
      "step": 17712
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.756973147392273,
      "learning_rate": 9.301983835415137e-06,
      "loss": 0.8654,
      "step": 17713
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.120203733444214,
      "learning_rate": 9.301344919017348e-06,
      "loss": 0.8489,
      "step": 17714
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4816136360168457,
      "learning_rate": 9.300706002619558e-06,
      "loss": 1.1302,
      "step": 17715
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7197582721710205,
      "learning_rate": 9.30006708622177e-06,
      "loss": 0.9228,
      "step": 17716
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.079683303833008,
      "learning_rate": 9.29942816982398e-06,
      "loss": 1.1986,
      "step": 17717
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6179258823394775,
      "learning_rate": 9.298789253426191e-06,
      "loss": 0.7798,
      "step": 17718
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5901261568069458,
      "learning_rate": 9.298150337028401e-06,
      "loss": 0.9072,
      "step": 17719
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8530584573745728,
      "learning_rate": 9.29751142063061e-06,
      "loss": 0.9447,
      "step": 17720
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9289207458496094,
      "learning_rate": 9.296872504232822e-06,
      "loss": 0.8175,
      "step": 17721
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6663557291030884,
      "learning_rate": 9.296233587835032e-06,
      "loss": 0.527,
      "step": 17722
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.064362049102783,
      "learning_rate": 9.295594671437244e-06,
      "loss": 0.7793,
      "step": 17723
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.2272565364837646,
      "learning_rate": 9.294955755039454e-06,
      "loss": 0.7684,
      "step": 17724
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1499807834625244,
      "learning_rate": 9.294316838641665e-06,
      "loss": 0.8305,
      "step": 17725
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.8773064613342285,
      "learning_rate": 9.293677922243875e-06,
      "loss": 0.5197,
      "step": 17726
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.6103923320770264,
      "learning_rate": 9.293039005846087e-06,
      "loss": 0.9568,
      "step": 17727
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.869094729423523,
      "learning_rate": 9.292400089448296e-06,
      "loss": 0.8528,
      "step": 17728
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.2767913341522217,
      "learning_rate": 9.291761173050508e-06,
      "loss": 0.675,
      "step": 17729
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.813025951385498,
      "learning_rate": 9.291122256652718e-06,
      "loss": 1.514,
      "step": 17730
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.875020146369934,
      "learning_rate": 9.29048334025493e-06,
      "loss": 0.9213,
      "step": 17731
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.452878952026367,
      "learning_rate": 9.28984442385714e-06,
      "loss": 1.2481,
      "step": 17732
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.529338002204895,
      "learning_rate": 9.28920550745935e-06,
      "loss": 0.3807,
      "step": 17733
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.2558701038360596,
      "learning_rate": 9.28856659106156e-06,
      "loss": 0.9791,
      "step": 17734
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7534959316253662,
      "learning_rate": 9.287927674663772e-06,
      "loss": 0.6934,
      "step": 17735
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.825244903564453,
      "learning_rate": 9.287288758265982e-06,
      "loss": 1.3457,
      "step": 17736
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.0981903076171875,
      "learning_rate": 9.286649841868194e-06,
      "loss": 1.0413,
      "step": 17737
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.866794109344482,
      "learning_rate": 9.286010925470403e-06,
      "loss": 1.2047,
      "step": 17738
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.251507520675659,
      "learning_rate": 9.285372009072613e-06,
      "loss": 0.7534,
      "step": 17739
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3649542331695557,
      "learning_rate": 9.284733092674825e-06,
      "loss": 0.8378,
      "step": 17740
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.453407049179077,
      "learning_rate": 9.284094176277035e-06,
      "loss": 1.0398,
      "step": 17741
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6072933673858643,
      "learning_rate": 9.283455259879246e-06,
      "loss": 0.8923,
      "step": 17742
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8181918859481812,
      "learning_rate": 9.282816343481456e-06,
      "loss": 0.7608,
      "step": 17743
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.874306678771973,
      "learning_rate": 9.282177427083668e-06,
      "loss": 1.4909,
      "step": 17744
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5804681777954102,
      "learning_rate": 9.281538510685877e-06,
      "loss": 0.8711,
      "step": 17745
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3141212463378906,
      "learning_rate": 9.280899594288089e-06,
      "loss": 1.2278,
      "step": 17746
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0699808597564697,
      "learning_rate": 9.280260677890299e-06,
      "loss": 0.8918,
      "step": 17747
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.180091619491577,
      "learning_rate": 9.27962176149251e-06,
      "loss": 0.8332,
      "step": 17748
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8356385231018066,
      "learning_rate": 9.27898284509472e-06,
      "loss": 0.8758,
      "step": 17749
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.8401806354522705,
      "learning_rate": 9.278343928696932e-06,
      "loss": 0.7703,
      "step": 17750
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.35040807723999,
      "learning_rate": 9.277705012299142e-06,
      "loss": 1.1768,
      "step": 17751
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.709018588066101,
      "learning_rate": 9.277066095901353e-06,
      "loss": 0.8362,
      "step": 17752
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5246636867523193,
      "learning_rate": 9.276427179503563e-06,
      "loss": 0.8971,
      "step": 17753
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.2508699893951416,
      "learning_rate": 9.275788263105775e-06,
      "loss": 1.031,
      "step": 17754
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.968452215194702,
      "learning_rate": 9.275149346707984e-06,
      "loss": 1.481,
      "step": 17755
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9985830783843994,
      "learning_rate": 9.274510430310196e-06,
      "loss": 0.9118,
      "step": 17756
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.2411253452301025,
      "learning_rate": 9.273871513912406e-06,
      "loss": 0.7152,
      "step": 17757
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5578433275222778,
      "learning_rate": 9.273232597514616e-06,
      "loss": 0.5788,
      "step": 17758
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7323203086853027,
      "learning_rate": 9.272593681116827e-06,
      "loss": 0.9362,
      "step": 17759
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5267362594604492,
      "learning_rate": 9.271954764719037e-06,
      "loss": 0.8531,
      "step": 17760
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1794469356536865,
      "learning_rate": 9.271315848321249e-06,
      "loss": 1.037,
      "step": 17761
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6955575942993164,
      "learning_rate": 9.270676931923458e-06,
      "loss": 0.8025,
      "step": 17762
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5630645751953125,
      "learning_rate": 9.27003801552567e-06,
      "loss": 0.8637,
      "step": 17763
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5185048580169678,
      "learning_rate": 9.26939909912788e-06,
      "loss": 0.8279,
      "step": 17764
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7063109874725342,
      "learning_rate": 9.268760182730091e-06,
      "loss": 0.6386,
      "step": 17765
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8185670375823975,
      "learning_rate": 9.268121266332301e-06,
      "loss": 0.8837,
      "step": 17766
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6104730367660522,
      "learning_rate": 9.267482349934513e-06,
      "loss": 0.8254,
      "step": 17767
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5360950231552124,
      "learning_rate": 9.266843433536723e-06,
      "loss": 0.8145,
      "step": 17768
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.165996551513672,
      "learning_rate": 9.266204517138934e-06,
      "loss": 0.9451,
      "step": 17769
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9418535232543945,
      "learning_rate": 9.265565600741144e-06,
      "loss": 0.8806,
      "step": 17770
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3359475135803223,
      "learning_rate": 9.264926684343356e-06,
      "loss": 0.9985,
      "step": 17771
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8510366678237915,
      "learning_rate": 9.264287767945565e-06,
      "loss": 0.9033,
      "step": 17772
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7461776733398438,
      "learning_rate": 9.263648851547777e-06,
      "loss": 0.9005,
      "step": 17773
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3550522327423096,
      "learning_rate": 9.263009935149987e-06,
      "loss": 0.678,
      "step": 17774
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0079233646392822,
      "learning_rate": 9.262371018752197e-06,
      "loss": 0.8494,
      "step": 17775
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.104029417037964,
      "learning_rate": 9.261732102354408e-06,
      "loss": 0.8304,
      "step": 17776
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5740184783935547,
      "learning_rate": 9.261093185956618e-06,
      "loss": 0.8361,
      "step": 17777
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8358572721481323,
      "learning_rate": 9.26045426955883e-06,
      "loss": 0.8799,
      "step": 17778
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.453666687011719,
      "learning_rate": 9.25981535316104e-06,
      "loss": 0.8986,
      "step": 17779
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.001997709274292,
      "learning_rate": 9.25917643676325e-06,
      "loss": 0.9579,
      "step": 17780
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1149673461914062,
      "learning_rate": 9.258537520365461e-06,
      "loss": 0.8652,
      "step": 17781
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.2983715534210205,
      "learning_rate": 9.25789860396767e-06,
      "loss": 0.9179,
      "step": 17782
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9862017631530762,
      "learning_rate": 9.257259687569882e-06,
      "loss": 1.1414,
      "step": 17783
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6623307466506958,
      "learning_rate": 9.256620771172092e-06,
      "loss": 0.794,
      "step": 17784
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9181574583053589,
      "learning_rate": 9.255981854774304e-06,
      "loss": 0.8586,
      "step": 17785
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.322396278381348,
      "learning_rate": 9.255342938376514e-06,
      "loss": 1.0474,
      "step": 17786
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.411839723587036,
      "learning_rate": 9.254704021978725e-06,
      "loss": 1.0099,
      "step": 17787
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.525811195373535,
      "learning_rate": 9.254065105580935e-06,
      "loss": 0.8817,
      "step": 17788
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2526320219039917,
      "learning_rate": 9.253426189183145e-06,
      "loss": 0.3859,
      "step": 17789
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4140390157699585,
      "learning_rate": 9.252787272785356e-06,
      "loss": 0.7717,
      "step": 17790
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9652453660964966,
      "learning_rate": 9.252148356387566e-06,
      "loss": 0.9673,
      "step": 17791
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5754363536834717,
      "learning_rate": 9.251509439989778e-06,
      "loss": 0.7724,
      "step": 17792
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8008241653442383,
      "learning_rate": 9.250870523591988e-06,
      "loss": 0.9192,
      "step": 17793
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.452528476715088,
      "learning_rate": 9.2502316071942e-06,
      "loss": 0.912,
      "step": 17794
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4101529121398926,
      "learning_rate": 9.249592690796409e-06,
      "loss": 0.8579,
      "step": 17795
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.6417348384857178,
      "learning_rate": 9.24895377439862e-06,
      "loss": 1.0355,
      "step": 17796
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.517831802368164,
      "learning_rate": 9.24831485800083e-06,
      "loss": 0.824,
      "step": 17797
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0778074264526367,
      "learning_rate": 9.247675941603042e-06,
      "loss": 0.9655,
      "step": 17798
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.707439422607422,
      "learning_rate": 9.247037025205252e-06,
      "loss": 1.0426,
      "step": 17799
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4295185804367065,
      "learning_rate": 9.246398108807463e-06,
      "loss": 0.8997,
      "step": 17800
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8375307321548462,
      "learning_rate": 9.245759192409673e-06,
      "loss": 0.8559,
      "step": 17801
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7155061960220337,
      "learning_rate": 9.245120276011885e-06,
      "loss": 0.8516,
      "step": 17802
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5767931938171387,
      "learning_rate": 9.244481359614095e-06,
      "loss": 0.8717,
      "step": 17803
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4892582893371582,
      "learning_rate": 9.243842443216306e-06,
      "loss": 1.1361,
      "step": 17804
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9121966361999512,
      "learning_rate": 9.243203526818516e-06,
      "loss": 0.8512,
      "step": 17805
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.118133068084717,
      "learning_rate": 9.242564610420726e-06,
      "loss": 0.7726,
      "step": 17806
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.7176859378814697,
      "learning_rate": 9.241925694022937e-06,
      "loss": 1.0583,
      "step": 17807
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8004125356674194,
      "learning_rate": 9.241286777625147e-06,
      "loss": 0.8885,
      "step": 17808
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0931894779205322,
      "learning_rate": 9.240647861227359e-06,
      "loss": 1.0335,
      "step": 17809
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.976162075996399,
      "learning_rate": 9.240008944829569e-06,
      "loss": 0.8044,
      "step": 17810
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.81761634349823,
      "learning_rate": 9.23937002843178e-06,
      "loss": 0.8381,
      "step": 17811
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6368643045425415,
      "learning_rate": 9.23873111203399e-06,
      "loss": 0.8079,
      "step": 17812
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9177656173706055,
      "learning_rate": 9.238092195636202e-06,
      "loss": 0.8912,
      "step": 17813
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9718858003616333,
      "learning_rate": 9.237453279238412e-06,
      "loss": 0.8701,
      "step": 17814
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.728283166885376,
      "learning_rate": 9.236814362840623e-06,
      "loss": 1.0031,
      "step": 17815
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5584670305252075,
      "learning_rate": 9.236175446442833e-06,
      "loss": 0.8893,
      "step": 17816
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.173346757888794,
      "learning_rate": 9.235536530045044e-06,
      "loss": 0.7692,
      "step": 17817
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.836921215057373,
      "learning_rate": 9.234897613647254e-06,
      "loss": 1.0718,
      "step": 17818
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6893601417541504,
      "learning_rate": 9.234258697249466e-06,
      "loss": 0.895,
      "step": 17819
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9174445867538452,
      "learning_rate": 9.233619780851676e-06,
      "loss": 0.9813,
      "step": 17820
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4783896207809448,
      "learning_rate": 9.232980864453887e-06,
      "loss": 0.8811,
      "step": 17821
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.603245973587036,
      "learning_rate": 9.232341948056097e-06,
      "loss": 0.9459,
      "step": 17822
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8680330514907837,
      "learning_rate": 9.231703031658309e-06,
      "loss": 1.0507,
      "step": 17823
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8811417818069458,
      "learning_rate": 9.231064115260519e-06,
      "loss": 0.9369,
      "step": 17824
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9789303541183472,
      "learning_rate": 9.230425198862728e-06,
      "loss": 1.128,
      "step": 17825
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6827915906906128,
      "learning_rate": 9.22978628246494e-06,
      "loss": 0.8611,
      "step": 17826
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.622774124145508,
      "learning_rate": 9.22914736606715e-06,
      "loss": 0.6923,
      "step": 17827
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.6353299617767334,
      "learning_rate": 9.228508449669361e-06,
      "loss": 1.0142,
      "step": 17828
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.713466167449951,
      "learning_rate": 9.227869533271571e-06,
      "loss": 1.0438,
      "step": 17829
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1978657245635986,
      "learning_rate": 9.227230616873783e-06,
      "loss": 0.9272,
      "step": 17830
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.8225302696228027,
      "learning_rate": 9.226591700475993e-06,
      "loss": 1.0975,
      "step": 17831
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3126089572906494,
      "learning_rate": 9.225952784078204e-06,
      "loss": 0.873,
      "step": 17832
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.270875930786133,
      "learning_rate": 9.225313867680414e-06,
      "loss": 0.8443,
      "step": 17833
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.091809034347534,
      "learning_rate": 9.224674951282626e-06,
      "loss": 0.8275,
      "step": 17834
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.478352665901184,
      "learning_rate": 9.224036034884835e-06,
      "loss": 0.8183,
      "step": 17835
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.813371181488037,
      "learning_rate": 9.223397118487047e-06,
      "loss": 0.8672,
      "step": 17836
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.388883352279663,
      "learning_rate": 9.222758202089257e-06,
      "loss": 1.0737,
      "step": 17837
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7187855243682861,
      "learning_rate": 9.222119285691468e-06,
      "loss": 0.849,
      "step": 17838
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1768498420715332,
      "learning_rate": 9.221480369293678e-06,
      "loss": 0.4144,
      "step": 17839
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.522507905960083,
      "learning_rate": 9.22084145289589e-06,
      "loss": 0.8751,
      "step": 17840
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1917717456817627,
      "learning_rate": 9.2202025364981e-06,
      "loss": 0.8049,
      "step": 17841
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.8604652881622314,
      "learning_rate": 9.21956362010031e-06,
      "loss": 1.2842,
      "step": 17842
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7943763732910156,
      "learning_rate": 9.218924703702521e-06,
      "loss": 0.8793,
      "step": 17843
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5777769088745117,
      "learning_rate": 9.21828578730473e-06,
      "loss": 0.8706,
      "step": 17844
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3899993896484375,
      "learning_rate": 9.217646870906942e-06,
      "loss": 0.9419,
      "step": 17845
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8818614482879639,
      "learning_rate": 9.217007954509152e-06,
      "loss": 0.9713,
      "step": 17846
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6005096435546875,
      "learning_rate": 9.216369038111364e-06,
      "loss": 0.5564,
      "step": 17847
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7430728673934937,
      "learning_rate": 9.215730121713574e-06,
      "loss": 0.904,
      "step": 17848
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.53018319606781,
      "learning_rate": 9.215091205315785e-06,
      "loss": 0.8135,
      "step": 17849
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.608953833580017,
      "learning_rate": 9.214452288917995e-06,
      "loss": 0.8173,
      "step": 17850
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.771066427230835,
      "learning_rate": 9.213813372520207e-06,
      "loss": 0.8634,
      "step": 17851
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7024983167648315,
      "learning_rate": 9.213174456122416e-06,
      "loss": 0.8125,
      "step": 17852
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.020005464553833,
      "learning_rate": 9.212535539724628e-06,
      "loss": 1.029,
      "step": 17853
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9671084880828857,
      "learning_rate": 9.211896623326838e-06,
      "loss": 0.9296,
      "step": 17854
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4262421131134033,
      "learning_rate": 9.21125770692905e-06,
      "loss": 0.5979,
      "step": 17855
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7501988410949707,
      "learning_rate": 9.21061879053126e-06,
      "loss": 0.8701,
      "step": 17856
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4887170791625977,
      "learning_rate": 9.20997987413347e-06,
      "loss": 0.7825,
      "step": 17857
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.647768020629883,
      "learning_rate": 9.20934095773568e-06,
      "loss": 1.0825,
      "step": 17858
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.2481307983398438,
      "learning_rate": 9.208702041337892e-06,
      "loss": 1.0052,
      "step": 17859
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3647598028182983,
      "learning_rate": 9.208063124940102e-06,
      "loss": 0.783,
      "step": 17860
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9605844020843506,
      "learning_rate": 9.207424208542312e-06,
      "loss": 0.9791,
      "step": 17861
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.359625816345215,
      "learning_rate": 9.206785292144523e-06,
      "loss": 0.9917,
      "step": 17862
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.849791407585144,
      "learning_rate": 9.206146375746733e-06,
      "loss": 0.9771,
      "step": 17863
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8732995986938477,
      "learning_rate": 9.205507459348945e-06,
      "loss": 0.9633,
      "step": 17864
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5295945405960083,
      "learning_rate": 9.204868542951155e-06,
      "loss": 0.9473,
      "step": 17865
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.237413167953491,
      "learning_rate": 9.204229626553366e-06,
      "loss": 0.941,
      "step": 17866
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.487194776535034,
      "learning_rate": 9.203590710155576e-06,
      "loss": 0.8395,
      "step": 17867
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.743518590927124,
      "learning_rate": 9.202951793757788e-06,
      "loss": 0.8238,
      "step": 17868
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.9050087928771973,
      "learning_rate": 9.202312877359997e-06,
      "loss": 1.2516,
      "step": 17869
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.02617883682251,
      "learning_rate": 9.201673960962209e-06,
      "loss": 0.8618,
      "step": 17870
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6396559476852417,
      "learning_rate": 9.201035044564419e-06,
      "loss": 0.8614,
      "step": 17871
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.796620488166809,
      "learning_rate": 9.20039612816663e-06,
      "loss": 0.8479,
      "step": 17872
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9456229209899902,
      "learning_rate": 9.19975721176884e-06,
      "loss": 1.1825,
      "step": 17873
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.131955862045288,
      "learning_rate": 9.199118295371052e-06,
      "loss": 0.8549,
      "step": 17874
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6277190446853638,
      "learning_rate": 9.198479378973262e-06,
      "loss": 0.8847,
      "step": 17875
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6652534008026123,
      "learning_rate": 9.197840462575473e-06,
      "loss": 0.8961,
      "step": 17876
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7732516527175903,
      "learning_rate": 9.197201546177683e-06,
      "loss": 0.9691,
      "step": 17877
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.76098370552063,
      "learning_rate": 9.196562629779895e-06,
      "loss": 1.221,
      "step": 17878
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1945371627807617,
      "learning_rate": 9.195923713382104e-06,
      "loss": 0.863,
      "step": 17879
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.706811428070068,
      "learning_rate": 9.195284796984314e-06,
      "loss": 1.4013,
      "step": 17880
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8265960216522217,
      "learning_rate": 9.194645880586526e-06,
      "loss": 0.8314,
      "step": 17881
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.8118834495544434,
      "learning_rate": 9.194006964188736e-06,
      "loss": 0.8368,
      "step": 17882
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7863777875900269,
      "learning_rate": 9.193368047790947e-06,
      "loss": 0.7406,
      "step": 17883
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.6822948455810547,
      "learning_rate": 9.192729131393157e-06,
      "loss": 0.726,
      "step": 17884
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.223067283630371,
      "learning_rate": 9.192090214995369e-06,
      "loss": 0.7004,
      "step": 17885
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.964392900466919,
      "learning_rate": 9.191451298597579e-06,
      "loss": 0.8277,
      "step": 17886
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8408317565917969,
      "learning_rate": 9.19081238219979e-06,
      "loss": 0.8642,
      "step": 17887
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5533223152160645,
      "learning_rate": 9.190173465802e-06,
      "loss": 1.0244,
      "step": 17888
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.862189769744873,
      "learning_rate": 9.189534549404211e-06,
      "loss": 0.8096,
      "step": 17889
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8755674362182617,
      "learning_rate": 9.188895633006421e-06,
      "loss": 0.8098,
      "step": 17890
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6334072351455688,
      "learning_rate": 9.188256716608633e-06,
      "loss": 0.8221,
      "step": 17891
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6810187101364136,
      "learning_rate": 9.187617800210843e-06,
      "loss": 0.8141,
      "step": 17892
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.598028302192688,
      "learning_rate": 9.186978883813054e-06,
      "loss": 0.9088,
      "step": 17893
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.874023914337158,
      "learning_rate": 9.186339967415264e-06,
      "loss": 1.4699,
      "step": 17894
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4901881217956543,
      "learning_rate": 9.185701051017476e-06,
      "loss": 0.9745,
      "step": 17895
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4070963859558105,
      "learning_rate": 9.185062134619686e-06,
      "loss": 0.8534,
      "step": 17896
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9178410768508911,
      "learning_rate": 9.184423218221895e-06,
      "loss": 1.1098,
      "step": 17897
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.837414026260376,
      "learning_rate": 9.183784301824107e-06,
      "loss": 0.8926,
      "step": 17898
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0369255542755127,
      "learning_rate": 9.183145385426317e-06,
      "loss": 0.8696,
      "step": 17899
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5782572031021118,
      "learning_rate": 9.182506469028528e-06,
      "loss": 0.7614,
      "step": 17900
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.013434410095215,
      "learning_rate": 9.181867552630738e-06,
      "loss": 0.8972,
      "step": 17901
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3238601684570312,
      "learning_rate": 9.18122863623295e-06,
      "loss": 0.3737,
      "step": 17902
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0393424034118652,
      "learning_rate": 9.18058971983516e-06,
      "loss": 0.9568,
      "step": 17903
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.853885293006897,
      "learning_rate": 9.179950803437371e-06,
      "loss": 0.7482,
      "step": 17904
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9739161729812622,
      "learning_rate": 9.179311887039581e-06,
      "loss": 0.8473,
      "step": 17905
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.079899787902832,
      "learning_rate": 9.178672970641793e-06,
      "loss": 1.3293,
      "step": 17906
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8448930978775024,
      "learning_rate": 9.178034054244002e-06,
      "loss": 0.8405,
      "step": 17907
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.778706431388855,
      "learning_rate": 9.177395137846214e-06,
      "loss": 0.9644,
      "step": 17908
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.5938916206359863,
      "learning_rate": 9.176756221448424e-06,
      "loss": 0.908,
      "step": 17909
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0137388706207275,
      "learning_rate": 9.176117305050635e-06,
      "loss": 0.7702,
      "step": 17910
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0744388103485107,
      "learning_rate": 9.175478388652845e-06,
      "loss": 1.2674,
      "step": 17911
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8395081758499146,
      "learning_rate": 9.174839472255057e-06,
      "loss": 0.8845,
      "step": 17912
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6217150688171387,
      "learning_rate": 9.174200555857267e-06,
      "loss": 0.4843,
      "step": 17913
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4289515018463135,
      "learning_rate": 9.173561639459478e-06,
      "loss": 1.2551,
      "step": 17914
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7647947072982788,
      "learning_rate": 9.172922723061688e-06,
      "loss": 0.8397,
      "step": 17915
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7656779289245605,
      "learning_rate": 9.172283806663898e-06,
      "loss": 0.3667,
      "step": 17916
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.529402256011963,
      "learning_rate": 9.17164489026611e-06,
      "loss": 1.2561,
      "step": 17917
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7397420406341553,
      "learning_rate": 9.17100597386832e-06,
      "loss": 0.9078,
      "step": 17918
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.720555067062378,
      "learning_rate": 9.17036705747053e-06,
      "loss": 0.8527,
      "step": 17919
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.164034366607666,
      "learning_rate": 9.16972814107274e-06,
      "loss": 1.0795,
      "step": 17920
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.024630546569824,
      "learning_rate": 9.169089224674952e-06,
      "loss": 1.0834,
      "step": 17921
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.170146465301514,
      "learning_rate": 9.168450308277162e-06,
      "loss": 1.0918,
      "step": 17922
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.735437870025635,
      "learning_rate": 9.167811391879374e-06,
      "loss": 1.3835,
      "step": 17923
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.451141595840454,
      "learning_rate": 9.167172475481583e-06,
      "loss": 0.8119,
      "step": 17924
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5345332622528076,
      "learning_rate": 9.166533559083795e-06,
      "loss": 0.9582,
      "step": 17925
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6372514963150024,
      "learning_rate": 9.165894642686005e-06,
      "loss": 0.871,
      "step": 17926
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6977931261062622,
      "learning_rate": 9.165255726288216e-06,
      "loss": 0.7602,
      "step": 17927
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9828667640686035,
      "learning_rate": 9.164616809890426e-06,
      "loss": 0.8484,
      "step": 17928
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7965532541275024,
      "learning_rate": 9.163977893492638e-06,
      "loss": 0.9126,
      "step": 17929
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6758157014846802,
      "learning_rate": 9.163338977094848e-06,
      "loss": 0.4265,
      "step": 17930
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1514980792999268,
      "learning_rate": 9.16270006069706e-06,
      "loss": 0.941,
      "step": 17931
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9703177213668823,
      "learning_rate": 9.162061144299269e-06,
      "loss": 0.8463,
      "step": 17932
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.775586485862732,
      "learning_rate": 9.16142222790148e-06,
      "loss": 0.9286,
      "step": 17933
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.195871591567993,
      "learning_rate": 9.16078331150369e-06,
      "loss": 0.9534,
      "step": 17934
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7455025911331177,
      "learning_rate": 9.1601443951059e-06,
      "loss": 0.9855,
      "step": 17935
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9490715265274048,
      "learning_rate": 9.159505478708112e-06,
      "loss": 0.963,
      "step": 17936
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3939712047576904,
      "learning_rate": 9.158866562310322e-06,
      "loss": 1.1441,
      "step": 17937
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.225858688354492,
      "learning_rate": 9.158227645912533e-06,
      "loss": 0.9892,
      "step": 17938
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.887346863746643,
      "learning_rate": 9.157588729514743e-06,
      "loss": 0.9144,
      "step": 17939
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7464152574539185,
      "learning_rate": 9.156949813116955e-06,
      "loss": 0.7539,
      "step": 17940
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.041880130767822,
      "learning_rate": 9.156310896719164e-06,
      "loss": 0.9282,
      "step": 17941
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9961405992507935,
      "learning_rate": 9.155671980321376e-06,
      "loss": 0.75,
      "step": 17942
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.736493468284607,
      "learning_rate": 9.155033063923586e-06,
      "loss": 0.8868,
      "step": 17943
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7595065832138062,
      "learning_rate": 9.154394147525797e-06,
      "loss": 0.9077,
      "step": 17944
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5606653690338135,
      "learning_rate": 9.153755231128007e-06,
      "loss": 0.8923,
      "step": 17945
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5794905424118042,
      "learning_rate": 9.153116314730219e-06,
      "loss": 0.392,
      "step": 17946
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.6902215480804443,
      "learning_rate": 9.152477398332429e-06,
      "loss": 1.2687,
      "step": 17947
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6889437437057495,
      "learning_rate": 9.15183848193464e-06,
      "loss": 0.8606,
      "step": 17948
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.916893243789673,
      "learning_rate": 9.15119956553685e-06,
      "loss": 1.2198,
      "step": 17949
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5814036130905151,
      "learning_rate": 9.150560649139062e-06,
      "loss": 0.8174,
      "step": 17950
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7351261377334595,
      "learning_rate": 9.149921732741271e-06,
      "loss": 0.9536,
      "step": 17951
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.661491632461548,
      "learning_rate": 9.149282816343481e-06,
      "loss": 0.9061,
      "step": 17952
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2723822593688965,
      "learning_rate": 9.148643899945693e-06,
      "loss": 0.9753,
      "step": 17953
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.9137978553771973,
      "learning_rate": 9.148004983547903e-06,
      "loss": 0.8115,
      "step": 17954
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.260077476501465,
      "learning_rate": 9.147366067150114e-06,
      "loss": 0.8687,
      "step": 17955
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6688032150268555,
      "learning_rate": 9.146727150752324e-06,
      "loss": 0.9194,
      "step": 17956
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.184751987457275,
      "learning_rate": 9.146088234354536e-06,
      "loss": 1.4013,
      "step": 17957
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4994924068450928,
      "learning_rate": 9.145449317956746e-06,
      "loss": 0.9247,
      "step": 17958
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5865482091903687,
      "learning_rate": 9.144810401558957e-06,
      "loss": 0.9574,
      "step": 17959
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6995333433151245,
      "learning_rate": 9.144171485161167e-06,
      "loss": 0.76,
      "step": 17960
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9511853456497192,
      "learning_rate": 9.143532568763378e-06,
      "loss": 1.0412,
      "step": 17961
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6544499397277832,
      "learning_rate": 9.142893652365588e-06,
      "loss": 0.8495,
      "step": 17962
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5131211280822754,
      "learning_rate": 9.1422547359678e-06,
      "loss": 0.5459,
      "step": 17963
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3660471439361572,
      "learning_rate": 9.14161581957001e-06,
      "loss": 0.9323,
      "step": 17964
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.344029188156128,
      "learning_rate": 9.140976903172221e-06,
      "loss": 0.7848,
      "step": 17965
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5845563411712646,
      "learning_rate": 9.140337986774431e-06,
      "loss": 0.869,
      "step": 17966
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7668792009353638,
      "learning_rate": 9.139699070376643e-06,
      "loss": 0.8572,
      "step": 17967
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2877609729766846,
      "learning_rate": 9.139060153978853e-06,
      "loss": 0.9754,
      "step": 17968
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5894170999526978,
      "learning_rate": 9.138421237581064e-06,
      "loss": 0.9196,
      "step": 17969
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0494396686553955,
      "learning_rate": 9.137782321183274e-06,
      "loss": 0.7645,
      "step": 17970
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9492424726486206,
      "learning_rate": 9.137143404785484e-06,
      "loss": 0.8798,
      "step": 17971
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.0811312198638916,
      "learning_rate": 9.136504488387695e-06,
      "loss": 0.8,
      "step": 17972
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.1745705604553223,
      "learning_rate": 9.135865571989905e-06,
      "loss": 0.7428,
      "step": 17973
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0862786769866943,
      "learning_rate": 9.135226655592117e-06,
      "loss": 0.918,
      "step": 17974
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7057634592056274,
      "learning_rate": 9.134587739194327e-06,
      "loss": 0.9007,
      "step": 17975
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1187520027160645,
      "learning_rate": 9.133948822796538e-06,
      "loss": 0.7933,
      "step": 17976
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8891502618789673,
      "learning_rate": 9.133309906398748e-06,
      "loss": 0.8027,
      "step": 17977
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7901808023452759,
      "learning_rate": 9.13267099000096e-06,
      "loss": 0.8806,
      "step": 17978
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5948286056518555,
      "learning_rate": 9.13203207360317e-06,
      "loss": 0.9725,
      "step": 17979
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.72675621509552,
      "learning_rate": 9.131393157205381e-06,
      "loss": 0.8111,
      "step": 17980
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.0375723838806152,
      "learning_rate": 9.13075424080759e-06,
      "loss": 0.8732,
      "step": 17981
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7272485494613647,
      "learning_rate": 9.130115324409802e-06,
      "loss": 0.8791,
      "step": 17982
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0334010124206543,
      "learning_rate": 9.129476408012012e-06,
      "loss": 0.8999,
      "step": 17983
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7486116886138916,
      "learning_rate": 9.128837491614224e-06,
      "loss": 0.6787,
      "step": 17984
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9257750511169434,
      "learning_rate": 9.128198575216434e-06,
      "loss": 1.0533,
      "step": 17985
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.439615249633789,
      "learning_rate": 9.127559658818645e-06,
      "loss": 0.8046,
      "step": 17986
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.190671443939209,
      "learning_rate": 9.126920742420855e-06,
      "loss": 1.0385,
      "step": 17987
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7533689737319946,
      "learning_rate": 9.126281826023065e-06,
      "loss": 0.8259,
      "step": 17988
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.130929470062256,
      "learning_rate": 9.125642909625276e-06,
      "loss": 0.8097,
      "step": 17989
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3440639972686768,
      "learning_rate": 9.125003993227486e-06,
      "loss": 0.9217,
      "step": 17990
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3316240310668945,
      "learning_rate": 9.124365076829698e-06,
      "loss": 1.2711,
      "step": 17991
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8591443300247192,
      "learning_rate": 9.123726160431908e-06,
      "loss": 0.8193,
      "step": 17992
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.874834418296814,
      "learning_rate": 9.12308724403412e-06,
      "loss": 0.8157,
      "step": 17993
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9961966276168823,
      "learning_rate": 9.122448327636329e-06,
      "loss": 0.813,
      "step": 17994
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.7850632667541504,
      "learning_rate": 9.12180941123854e-06,
      "loss": 0.8293,
      "step": 17995
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5358858108520508,
      "learning_rate": 9.12117049484075e-06,
      "loss": 0.442,
      "step": 17996
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.7197625637054443,
      "learning_rate": 9.120531578442962e-06,
      "loss": 0.8372,
      "step": 17997
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.665396213531494,
      "learning_rate": 9.119892662045172e-06,
      "loss": 1.0637,
      "step": 17998
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7324650287628174,
      "learning_rate": 9.119253745647383e-06,
      "loss": 0.939,
      "step": 17999
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0520758628845215,
      "learning_rate": 9.118614829249593e-06,
      "loss": 0.9583,
      "step": 18000
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.151662588119507,
      "learning_rate": 9.117975912851805e-06,
      "loss": 1.0286,
      "step": 18001
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8648661375045776,
      "learning_rate": 9.117336996454015e-06,
      "loss": 1.119,
      "step": 18002
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5396020412445068,
      "learning_rate": 9.116698080056226e-06,
      "loss": 0.8376,
      "step": 18003
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8318623304367065,
      "learning_rate": 9.116059163658436e-06,
      "loss": 0.8444,
      "step": 18004
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3248116970062256,
      "learning_rate": 9.115420247260648e-06,
      "loss": 1.0281,
      "step": 18005
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6546952724456787,
      "learning_rate": 9.114781330862857e-06,
      "loss": 0.9021,
      "step": 18006
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9932528734207153,
      "learning_rate": 9.114142414465067e-06,
      "loss": 0.9577,
      "step": 18007
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8011285066604614,
      "learning_rate": 9.113503498067279e-06,
      "loss": 0.8636,
      "step": 18008
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.232381820678711,
      "learning_rate": 9.112864581669489e-06,
      "loss": 0.9901,
      "step": 18009
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.429683804512024,
      "learning_rate": 9.1122256652717e-06,
      "loss": 0.7448,
      "step": 18010
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7534548044204712,
      "learning_rate": 9.11158674887391e-06,
      "loss": 0.9953,
      "step": 18011
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5036011934280396,
      "learning_rate": 9.110947832476122e-06,
      "loss": 0.9046,
      "step": 18012
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3032963275909424,
      "learning_rate": 9.110308916078331e-06,
      "loss": 0.9842,
      "step": 18013
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9370626211166382,
      "learning_rate": 9.109669999680543e-06,
      "loss": 0.8353,
      "step": 18014
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5498099327087402,
      "learning_rate": 9.109031083282753e-06,
      "loss": 1.0196,
      "step": 18015
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.187714576721191,
      "learning_rate": 9.108392166884964e-06,
      "loss": 1.6646,
      "step": 18016
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.9649856090545654,
      "learning_rate": 9.107753250487174e-06,
      "loss": 1.1958,
      "step": 18017
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3778846263885498,
      "learning_rate": 9.107114334089386e-06,
      "loss": 0.3556,
      "step": 18018
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4723467826843262,
      "learning_rate": 9.106475417691596e-06,
      "loss": 0.8649,
      "step": 18019
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.516811490058899,
      "learning_rate": 9.105836501293807e-06,
      "loss": 0.826,
      "step": 18020
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6568576097488403,
      "learning_rate": 9.105197584896017e-06,
      "loss": 0.7222,
      "step": 18021
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.01230525970459,
      "learning_rate": 9.104558668498229e-06,
      "loss": 1.1023,
      "step": 18022
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8275665044784546,
      "learning_rate": 9.103919752100438e-06,
      "loss": 0.548,
      "step": 18023
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5130138397216797,
      "learning_rate": 9.10328083570265e-06,
      "loss": 0.6419,
      "step": 18024
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9337694644927979,
      "learning_rate": 9.10264191930486e-06,
      "loss": 1.0121,
      "step": 18025
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.565062403678894,
      "learning_rate": 9.10200300290707e-06,
      "loss": 0.8874,
      "step": 18026
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8626736402511597,
      "learning_rate": 9.101364086509281e-06,
      "loss": 0.8346,
      "step": 18027
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6457726955413818,
      "learning_rate": 9.100725170111491e-06,
      "loss": 0.866,
      "step": 18028
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.27683162689209,
      "learning_rate": 9.100086253713703e-06,
      "loss": 1.6381,
      "step": 18029
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5710581541061401,
      "learning_rate": 9.099447337315913e-06,
      "loss": 0.7749,
      "step": 18030
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.405215740203857,
      "learning_rate": 9.098808420918124e-06,
      "loss": 1.0358,
      "step": 18031
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9676927328109741,
      "learning_rate": 9.098169504520334e-06,
      "loss": 0.8907,
      "step": 18032
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.612109422683716,
      "learning_rate": 9.097530588122545e-06,
      "loss": 1.2416,
      "step": 18033
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8387761116027832,
      "learning_rate": 9.096891671724755e-06,
      "loss": 0.881,
      "step": 18034
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.723571538925171,
      "learning_rate": 9.096252755326967e-06,
      "loss": 0.7783,
      "step": 18035
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9198439121246338,
      "learning_rate": 9.095613838929177e-06,
      "loss": 0.9008,
      "step": 18036
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7640907764434814,
      "learning_rate": 9.094974922531388e-06,
      "loss": 0.8341,
      "step": 18037
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.1144938468933105,
      "learning_rate": 9.094336006133598e-06,
      "loss": 1.0154,
      "step": 18038
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1396615505218506,
      "learning_rate": 9.09369708973581e-06,
      "loss": 0.9509,
      "step": 18039
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5765663385391235,
      "learning_rate": 9.09305817333802e-06,
      "loss": 0.8488,
      "step": 18040
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.715924859046936,
      "learning_rate": 9.092419256940231e-06,
      "loss": 0.8049,
      "step": 18041
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.177989959716797,
      "learning_rate": 9.091780340542441e-06,
      "loss": 0.948,
      "step": 18042
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3666892051696777,
      "learning_rate": 9.09114142414465e-06,
      "loss": 0.92,
      "step": 18043
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8222640752792358,
      "learning_rate": 9.090502507746862e-06,
      "loss": 0.9242,
      "step": 18044
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.608627438545227,
      "learning_rate": 9.089863591349072e-06,
      "loss": 0.9033,
      "step": 18045
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6268930435180664,
      "learning_rate": 9.089224674951284e-06,
      "loss": 0.8616,
      "step": 18046
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.039214611053467,
      "learning_rate": 9.088585758553494e-06,
      "loss": 1.7783,
      "step": 18047
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.077459335327148,
      "learning_rate": 9.087946842155705e-06,
      "loss": 1.0705,
      "step": 18048
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4890899658203125,
      "learning_rate": 9.087307925757915e-06,
      "loss": 0.9348,
      "step": 18049
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.14312744140625,
      "learning_rate": 9.086669009360127e-06,
      "loss": 1.2438,
      "step": 18050
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.788315773010254,
      "learning_rate": 9.086030092962336e-06,
      "loss": 1.1784,
      "step": 18051
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.655527114868164,
      "learning_rate": 9.085391176564548e-06,
      "loss": 0.8313,
      "step": 18052
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2039847373962402,
      "learning_rate": 9.084752260166758e-06,
      "loss": 0.942,
      "step": 18053
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8589352369308472,
      "learning_rate": 9.08411334376897e-06,
      "loss": 1.2548,
      "step": 18054
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.236840009689331,
      "learning_rate": 9.08347442737118e-06,
      "loss": 0.7143,
      "step": 18055
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9074437618255615,
      "learning_rate": 9.08283551097339e-06,
      "loss": 0.8656,
      "step": 18056
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9002803564071655,
      "learning_rate": 9.0821965945756e-06,
      "loss": 0.8838,
      "step": 18057
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6291743516921997,
      "learning_rate": 9.081557678177812e-06,
      "loss": 0.8903,
      "step": 18058
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8046162128448486,
      "learning_rate": 9.080918761780022e-06,
      "loss": 0.918,
      "step": 18059
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.652193307876587,
      "learning_rate": 9.080279845382234e-06,
      "loss": 0.9295,
      "step": 18060
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8376450538635254,
      "learning_rate": 9.079640928984443e-06,
      "loss": 0.8691,
      "step": 18061
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.006685733795166,
      "learning_rate": 9.079002012586653e-06,
      "loss": 0.9924,
      "step": 18062
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.233820676803589,
      "learning_rate": 9.078363096188865e-06,
      "loss": 1.1802,
      "step": 18063
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.287891864776611,
      "learning_rate": 9.077724179791075e-06,
      "loss": 1.7324,
      "step": 18064
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8954110145568848,
      "learning_rate": 9.077085263393286e-06,
      "loss": 1.129,
      "step": 18065
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9584904909133911,
      "learning_rate": 9.076446346995496e-06,
      "loss": 0.9159,
      "step": 18066
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2132110595703125,
      "learning_rate": 9.075807430597708e-06,
      "loss": 0.9689,
      "step": 18067
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5421087741851807,
      "learning_rate": 9.075168514199917e-06,
      "loss": 0.8413,
      "step": 18068
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.498252272605896,
      "learning_rate": 9.074529597802129e-06,
      "loss": 0.8005,
      "step": 18069
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.004824161529541,
      "learning_rate": 9.073890681404339e-06,
      "loss": 0.9252,
      "step": 18070
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.798575520515442,
      "learning_rate": 9.07325176500655e-06,
      "loss": 0.8423,
      "step": 18071
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9540950059890747,
      "learning_rate": 9.07261284860876e-06,
      "loss": 0.9695,
      "step": 18072
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1358935832977295,
      "learning_rate": 9.071973932210972e-06,
      "loss": 1.0186,
      "step": 18073
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9409394264221191,
      "learning_rate": 9.071335015813182e-06,
      "loss": 0.8743,
      "step": 18074
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1185977458953857,
      "learning_rate": 9.070696099415393e-06,
      "loss": 1.0209,
      "step": 18075
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.932586669921875,
      "learning_rate": 9.070057183017603e-06,
      "loss": 0.849,
      "step": 18076
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8611770868301392,
      "learning_rate": 9.069418266619815e-06,
      "loss": 0.9553,
      "step": 18077
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.050217151641846,
      "learning_rate": 9.068779350222024e-06,
      "loss": 0.8019,
      "step": 18078
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.403526782989502,
      "learning_rate": 9.068140433824236e-06,
      "loss": 1.105,
      "step": 18079
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2858736515045166,
      "learning_rate": 9.067501517426446e-06,
      "loss": 0.9079,
      "step": 18080
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3723024129867554,
      "learning_rate": 9.066862601028656e-06,
      "loss": 0.7817,
      "step": 18081
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.03194522857666,
      "learning_rate": 9.066223684630867e-06,
      "loss": 0.809,
      "step": 18082
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.59576416015625,
      "learning_rate": 9.065584768233077e-06,
      "loss": 1.4673,
      "step": 18083
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9234509468078613,
      "learning_rate": 9.064945851835289e-06,
      "loss": 0.8967,
      "step": 18084
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.533418655395508,
      "learning_rate": 9.064306935437498e-06,
      "loss": 0.83,
      "step": 18085
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.579856038093567,
      "learning_rate": 9.06366801903971e-06,
      "loss": 0.8312,
      "step": 18086
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0516278743743896,
      "learning_rate": 9.06302910264192e-06,
      "loss": 1.0066,
      "step": 18087
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.329015254974365,
      "learning_rate": 9.062390186244131e-06,
      "loss": 1.4701,
      "step": 18088
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.077094554901123,
      "learning_rate": 9.061751269846341e-06,
      "loss": 0.8946,
      "step": 18089
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.6966373920440674,
      "learning_rate": 9.061112353448553e-06,
      "loss": 1.2647,
      "step": 18090
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0393967628479004,
      "learning_rate": 9.060473437050763e-06,
      "loss": 0.9015,
      "step": 18091
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.73712956905365,
      "learning_rate": 9.059834520652974e-06,
      "loss": 0.7992,
      "step": 18092
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.715195894241333,
      "learning_rate": 9.059195604255184e-06,
      "loss": 1.0631,
      "step": 18093
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.509823799133301,
      "learning_rate": 9.058556687857396e-06,
      "loss": 0.9799,
      "step": 18094
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8205196857452393,
      "learning_rate": 9.057917771459605e-06,
      "loss": 0.7891,
      "step": 18095
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8517377376556396,
      "learning_rate": 9.057278855061817e-06,
      "loss": 0.8243,
      "step": 18096
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1436805725097656,
      "learning_rate": 9.056639938664027e-06,
      "loss": 0.7027,
      "step": 18097
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9085923433303833,
      "learning_rate": 9.056001022266237e-06,
      "loss": 0.8305,
      "step": 18098
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9912625551223755,
      "learning_rate": 9.055362105868448e-06,
      "loss": 0.8943,
      "step": 18099
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7367198467254639,
      "learning_rate": 9.054723189470658e-06,
      "loss": 0.8684,
      "step": 18100
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9856376647949219,
      "learning_rate": 9.05408427307287e-06,
      "loss": 0.9204,
      "step": 18101
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4184411764144897,
      "learning_rate": 9.05344535667508e-06,
      "loss": 0.8786,
      "step": 18102
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.735853672027588,
      "learning_rate": 9.052806440277291e-06,
      "loss": 0.8633,
      "step": 18103
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5325636863708496,
      "learning_rate": 9.052167523879501e-06,
      "loss": 0.5,
      "step": 18104
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.685697078704834,
      "learning_rate": 9.051528607481712e-06,
      "loss": 0.9094,
      "step": 18105
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.777510404586792,
      "learning_rate": 9.050889691083922e-06,
      "loss": 1.0298,
      "step": 18106
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8134089708328247,
      "learning_rate": 9.050250774686134e-06,
      "loss": 0.9551,
      "step": 18107
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.841774582862854,
      "learning_rate": 9.049611858288344e-06,
      "loss": 0.8336,
      "step": 18108
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2427423000335693,
      "learning_rate": 9.048972941890555e-06,
      "loss": 1.2191,
      "step": 18109
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.720101237297058,
      "learning_rate": 9.048334025492765e-06,
      "loss": 0.9333,
      "step": 18110
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5291805267333984,
      "learning_rate": 9.047695109094977e-06,
      "loss": 0.8192,
      "step": 18111
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1062662601470947,
      "learning_rate": 9.047056192697187e-06,
      "loss": 0.8567,
      "step": 18112
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8120192289352417,
      "learning_rate": 9.046417276299398e-06,
      "loss": 0.797,
      "step": 18113
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4144887924194336,
      "learning_rate": 9.045778359901608e-06,
      "loss": 0.9245,
      "step": 18114
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.074185848236084,
      "learning_rate": 9.04513944350382e-06,
      "loss": 0.9575,
      "step": 18115
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.494673728942871,
      "learning_rate": 9.04450052710603e-06,
      "loss": 1.2449,
      "step": 18116
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1353511810302734,
      "learning_rate": 9.04386161070824e-06,
      "loss": 1.1429,
      "step": 18117
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.8567306995391846,
      "learning_rate": 9.04322269431045e-06,
      "loss": 1.1566,
      "step": 18118
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.727536678314209,
      "learning_rate": 9.04258377791266e-06,
      "loss": 1.0873,
      "step": 18119
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.158616065979004,
      "learning_rate": 9.041944861514872e-06,
      "loss": 0.9744,
      "step": 18120
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5456514358520508,
      "learning_rate": 9.041305945117082e-06,
      "loss": 0.4314,
      "step": 18121
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8385838270187378,
      "learning_rate": 9.040667028719294e-06,
      "loss": 0.9461,
      "step": 18122
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6909726858139038,
      "learning_rate": 9.040028112321503e-06,
      "loss": 0.8486,
      "step": 18123
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.2895586490631104,
      "learning_rate": 9.039389195923715e-06,
      "loss": 1.3967,
      "step": 18124
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4501802921295166,
      "learning_rate": 9.038750279525925e-06,
      "loss": 0.9658,
      "step": 18125
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0427803993225098,
      "learning_rate": 9.038111363128136e-06,
      "loss": 0.9593,
      "step": 18126
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.8128433227539062,
      "learning_rate": 9.037472446730346e-06,
      "loss": 0.782,
      "step": 18127
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.333819389343262,
      "learning_rate": 9.036833530332558e-06,
      "loss": 1.9103,
      "step": 18128
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6740803718566895,
      "learning_rate": 9.036194613934768e-06,
      "loss": 0.389,
      "step": 18129
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.565864086151123,
      "learning_rate": 9.035555697536979e-06,
      "loss": 0.8183,
      "step": 18130
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7486652135849,
      "learning_rate": 9.034916781139189e-06,
      "loss": 0.9077,
      "step": 18131
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.99694561958313,
      "learning_rate": 9.0342778647414e-06,
      "loss": 0.6771,
      "step": 18132
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7804917097091675,
      "learning_rate": 9.03363894834361e-06,
      "loss": 0.8015,
      "step": 18133
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6020967960357666,
      "learning_rate": 9.033000031945822e-06,
      "loss": 0.8601,
      "step": 18134
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.573625087738037,
      "learning_rate": 9.032361115548032e-06,
      "loss": 0.8094,
      "step": 18135
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6094945669174194,
      "learning_rate": 9.031722199150242e-06,
      "loss": 0.9164,
      "step": 18136
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.1105635166168213,
      "learning_rate": 9.031083282752453e-06,
      "loss": 0.8149,
      "step": 18137
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.415388822555542,
      "learning_rate": 9.030444366354663e-06,
      "loss": 0.8761,
      "step": 18138
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3190598487854004,
      "learning_rate": 9.029805449956875e-06,
      "loss": 0.9412,
      "step": 18139
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7528679370880127,
      "learning_rate": 9.029166533559084e-06,
      "loss": 0.9011,
      "step": 18140
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7083230018615723,
      "learning_rate": 9.028527617161296e-06,
      "loss": 0.7978,
      "step": 18141
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.563979148864746,
      "learning_rate": 9.027888700763506e-06,
      "loss": 0.7571,
      "step": 18142
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5516066551208496,
      "learning_rate": 9.027249784365717e-06,
      "loss": 0.895,
      "step": 18143
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5796828269958496,
      "learning_rate": 9.026610867967927e-06,
      "loss": 0.8336,
      "step": 18144
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.1534647941589355,
      "learning_rate": 9.025971951570139e-06,
      "loss": 0.868,
      "step": 18145
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.614260196685791,
      "learning_rate": 9.025333035172349e-06,
      "loss": 0.8184,
      "step": 18146
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2589282989501953,
      "learning_rate": 9.02469411877456e-06,
      "loss": 0.5862,
      "step": 18147
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.079108953475952,
      "learning_rate": 9.02405520237677e-06,
      "loss": 1.0067,
      "step": 18148
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5781153440475464,
      "learning_rate": 9.023416285978982e-06,
      "loss": 0.4424,
      "step": 18149
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.955819845199585,
      "learning_rate": 9.022777369581191e-06,
      "loss": 0.8971,
      "step": 18150
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4771902561187744,
      "learning_rate": 9.022138453183403e-06,
      "loss": 1.3152,
      "step": 18151
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.397644281387329,
      "learning_rate": 9.021499536785613e-06,
      "loss": 0.9652,
      "step": 18152
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1280641555786133,
      "learning_rate": 9.020860620387823e-06,
      "loss": 0.926,
      "step": 18153
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8869682550430298,
      "learning_rate": 9.020221703990034e-06,
      "loss": 0.873,
      "step": 18154
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6602742671966553,
      "learning_rate": 9.019582787592244e-06,
      "loss": 0.8781,
      "step": 18155
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.599392890930176,
      "learning_rate": 9.018943871194456e-06,
      "loss": 0.9078,
      "step": 18156
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4620161056518555,
      "learning_rate": 9.018304954796665e-06,
      "loss": 0.8019,
      "step": 18157
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.9232349395751953,
      "learning_rate": 9.017666038398877e-06,
      "loss": 0.8227,
      "step": 18158
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.771726369857788,
      "learning_rate": 9.017027122001087e-06,
      "loss": 0.8837,
      "step": 18159
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7292193174362183,
      "learning_rate": 9.016388205603298e-06,
      "loss": 0.8662,
      "step": 18160
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7913016080856323,
      "learning_rate": 9.015749289205508e-06,
      "loss": 0.8893,
      "step": 18161
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.361145973205566,
      "learning_rate": 9.01511037280772e-06,
      "loss": 1.0512,
      "step": 18162
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.927136182785034,
      "learning_rate": 9.01447145640993e-06,
      "loss": 0.8121,
      "step": 18163
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5527950525283813,
      "learning_rate": 9.013832540012141e-06,
      "loss": 0.9255,
      "step": 18164
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5932631492614746,
      "learning_rate": 9.013193623614351e-06,
      "loss": 1.0206,
      "step": 18165
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.990696668624878,
      "learning_rate": 9.012554707216563e-06,
      "loss": 1.1989,
      "step": 18166
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5867782831192017,
      "learning_rate": 9.01191579081877e-06,
      "loss": 0.8626,
      "step": 18167
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.694073438644409,
      "learning_rate": 9.011276874420982e-06,
      "loss": 0.5598,
      "step": 18168
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0455451011657715,
      "learning_rate": 9.010637958023192e-06,
      "loss": 0.9025,
      "step": 18169
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5406248569488525,
      "learning_rate": 9.009999041625404e-06,
      "loss": 0.7956,
      "step": 18170
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7375236749649048,
      "learning_rate": 9.009360125227614e-06,
      "loss": 0.891,
      "step": 18171
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.465132713317871,
      "learning_rate": 9.008721208829825e-06,
      "loss": 1.4134,
      "step": 18172
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.775547981262207,
      "learning_rate": 9.008082292432035e-06,
      "loss": 0.8505,
      "step": 18173
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4473010301589966,
      "learning_rate": 9.007443376034247e-06,
      "loss": 0.3491,
      "step": 18174
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5592494010925293,
      "learning_rate": 9.006804459636456e-06,
      "loss": 0.9759,
      "step": 18175
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.153724193572998,
      "learning_rate": 9.006165543238668e-06,
      "loss": 0.7936,
      "step": 18176
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.956169843673706,
      "learning_rate": 9.005526626840878e-06,
      "loss": 0.8671,
      "step": 18177
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5428357124328613,
      "learning_rate": 9.00488771044309e-06,
      "loss": 1.0411,
      "step": 18178
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7454179525375366,
      "learning_rate": 9.0042487940453e-06,
      "loss": 0.8655,
      "step": 18179
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.07033371925354,
      "learning_rate": 9.00360987764751e-06,
      "loss": 0.9671,
      "step": 18180
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4937551021575928,
      "learning_rate": 9.00297096124972e-06,
      "loss": 0.8514,
      "step": 18181
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7702792882919312,
      "learning_rate": 9.002332044851932e-06,
      "loss": 0.8917,
      "step": 18182
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4106314182281494,
      "learning_rate": 9.001693128454142e-06,
      "loss": 1.0162,
      "step": 18183
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3492298126220703,
      "learning_rate": 9.001054212056352e-06,
      "loss": 0.9353,
      "step": 18184
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7638990879058838,
      "learning_rate": 9.000415295658563e-06,
      "loss": 1.0586,
      "step": 18185
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0227322578430176,
      "learning_rate": 8.999776379260773e-06,
      "loss": 0.9322,
      "step": 18186
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.980059266090393,
      "learning_rate": 8.999137462862985e-06,
      "loss": 0.5084,
      "step": 18187
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0623726844787598,
      "learning_rate": 8.998498546465195e-06,
      "loss": 0.9464,
      "step": 18188
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7068309783935547,
      "learning_rate": 8.997859630067406e-06,
      "loss": 0.731,
      "step": 18189
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.2352678775787354,
      "learning_rate": 8.997220713669616e-06,
      "loss": 0.966,
      "step": 18190
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7662248611450195,
      "learning_rate": 8.996581797271828e-06,
      "loss": 1.0138,
      "step": 18191
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.5873961448669434,
      "learning_rate": 8.995942880874037e-06,
      "loss": 1.2726,
      "step": 18192
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7443443536758423,
      "learning_rate": 8.995303964476249e-06,
      "loss": 0.889,
      "step": 18193
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.2197329998016357,
      "learning_rate": 8.994665048078459e-06,
      "loss": 0.876,
      "step": 18194
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.959885597229004,
      "learning_rate": 8.99402613168067e-06,
      "loss": 0.6648,
      "step": 18195
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.563320279121399,
      "learning_rate": 8.99338721528288e-06,
      "loss": 0.4037,
      "step": 18196
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.642721652984619,
      "learning_rate": 8.992748298885092e-06,
      "loss": 0.9185,
      "step": 18197
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8792202472686768,
      "learning_rate": 8.992109382487302e-06,
      "loss": 0.9174,
      "step": 18198
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0056049823760986,
      "learning_rate": 8.991470466089513e-06,
      "loss": 0.7422,
      "step": 18199
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.487301826477051,
      "learning_rate": 8.990831549691723e-06,
      "loss": 1.0663,
      "step": 18200
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7064414024353027,
      "learning_rate": 8.990192633293935e-06,
      "loss": 0.8765,
      "step": 18201
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.836260437965393,
      "learning_rate": 8.989553716896144e-06,
      "loss": 0.9556,
      "step": 18202
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7303446531295776,
      "learning_rate": 8.988914800498354e-06,
      "loss": 0.9254,
      "step": 18203
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.215623378753662,
      "learning_rate": 8.988275884100566e-06,
      "loss": 1.1336,
      "step": 18204
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.179650783538818,
      "learning_rate": 8.987636967702776e-06,
      "loss": 1.5659,
      "step": 18205
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7205226421356201,
      "learning_rate": 8.986998051304987e-06,
      "loss": 0.9028,
      "step": 18206
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5962202548980713,
      "learning_rate": 8.986359134907197e-06,
      "loss": 0.9268,
      "step": 18207
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8669672012329102,
      "learning_rate": 8.985720218509409e-06,
      "loss": 0.8703,
      "step": 18208
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8953527212142944,
      "learning_rate": 8.985081302111619e-06,
      "loss": 0.9145,
      "step": 18209
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5551989078521729,
      "learning_rate": 8.98444238571383e-06,
      "loss": 0.8189,
      "step": 18210
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1701200008392334,
      "learning_rate": 8.98380346931604e-06,
      "loss": 0.9252,
      "step": 18211
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9013464450836182,
      "learning_rate": 8.983164552918251e-06,
      "loss": 0.8643,
      "step": 18212
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9101476669311523,
      "learning_rate": 8.982525636520461e-06,
      "loss": 0.7721,
      "step": 18213
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9921674728393555,
      "learning_rate": 8.981886720122673e-06,
      "loss": 0.902,
      "step": 18214
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4463493824005127,
      "learning_rate": 8.981247803724883e-06,
      "loss": 0.7966,
      "step": 18215
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5758917331695557,
      "learning_rate": 8.980608887327094e-06,
      "loss": 0.8741,
      "step": 18216
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.387254238128662,
      "learning_rate": 8.979969970929304e-06,
      "loss": 1.2172,
      "step": 18217
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.9433207511901855,
      "learning_rate": 8.979331054531516e-06,
      "loss": 1.419,
      "step": 18218
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7514069080352783,
      "learning_rate": 8.978692138133726e-06,
      "loss": 0.8427,
      "step": 18219
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2129499912261963,
      "learning_rate": 8.978053221735935e-06,
      "loss": 0.892,
      "step": 18220
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8846471309661865,
      "learning_rate": 8.977414305338147e-06,
      "loss": 1.1218,
      "step": 18221
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.2122628688812256,
      "learning_rate": 8.976775388940357e-06,
      "loss": 1.194,
      "step": 18222
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0140066146850586,
      "learning_rate": 8.976136472542568e-06,
      "loss": 0.8301,
      "step": 18223
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9389256238937378,
      "learning_rate": 8.975497556144778e-06,
      "loss": 0.9399,
      "step": 18224
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.883259654045105,
      "learning_rate": 8.97485863974699e-06,
      "loss": 0.922,
      "step": 18225
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5283918380737305,
      "learning_rate": 8.9742197233492e-06,
      "loss": 0.8893,
      "step": 18226
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.098527193069458,
      "learning_rate": 8.973580806951411e-06,
      "loss": 0.8838,
      "step": 18227
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.823824167251587,
      "learning_rate": 8.972941890553621e-06,
      "loss": 0.8369,
      "step": 18228
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.843259334564209,
      "learning_rate": 8.972302974155833e-06,
      "loss": 0.9108,
      "step": 18229
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9710389375686646,
      "learning_rate": 8.971664057758042e-06,
      "loss": 1.3905,
      "step": 18230
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.550922155380249,
      "learning_rate": 8.971025141360254e-06,
      "loss": 0.8388,
      "step": 18231
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8636963367462158,
      "learning_rate": 8.970386224962464e-06,
      "loss": 0.8814,
      "step": 18232
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1142685413360596,
      "learning_rate": 8.969747308564675e-06,
      "loss": 0.9774,
      "step": 18233
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.8248021602630615,
      "learning_rate": 8.969108392166885e-06,
      "loss": 1.1795,
      "step": 18234
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1990859508514404,
      "learning_rate": 8.968469475769097e-06,
      "loss": 0.8739,
      "step": 18235
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.6938364505767822,
      "learning_rate": 8.967830559371307e-06,
      "loss": 0.9616,
      "step": 18236
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.464532732963562,
      "learning_rate": 8.967191642973518e-06,
      "loss": 0.4666,
      "step": 18237
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6441150903701782,
      "learning_rate": 8.966552726575728e-06,
      "loss": 0.9616,
      "step": 18238
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.155565023422241,
      "learning_rate": 8.965913810177938e-06,
      "loss": 0.9246,
      "step": 18239
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6201810836791992,
      "learning_rate": 8.96527489378015e-06,
      "loss": 0.7973,
      "step": 18240
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8029881715774536,
      "learning_rate": 8.96463597738236e-06,
      "loss": 0.943,
      "step": 18241
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.538689613342285,
      "learning_rate": 8.96399706098457e-06,
      "loss": 1.8149,
      "step": 18242
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.583193302154541,
      "learning_rate": 8.96335814458678e-06,
      "loss": 0.826,
      "step": 18243
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8025789260864258,
      "learning_rate": 8.962719228188992e-06,
      "loss": 0.848,
      "step": 18244
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.948230266571045,
      "learning_rate": 8.962080311791202e-06,
      "loss": 0.8387,
      "step": 18245
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.52470064163208,
      "learning_rate": 8.961441395393414e-06,
      "loss": 0.7879,
      "step": 18246
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.7423439025878906,
      "learning_rate": 8.960802478995623e-06,
      "loss": 0.5809,
      "step": 18247
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.739978313446045,
      "learning_rate": 8.960163562597835e-06,
      "loss": 0.7799,
      "step": 18248
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0616066455841064,
      "learning_rate": 8.959524646200045e-06,
      "loss": 1.0242,
      "step": 18249
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.279247283935547,
      "learning_rate": 8.958885729802256e-06,
      "loss": 1.0195,
      "step": 18250
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2537572383880615,
      "learning_rate": 8.958246813404466e-06,
      "loss": 0.9674,
      "step": 18251
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8716827630996704,
      "learning_rate": 8.957607897006678e-06,
      "loss": 0.919,
      "step": 18252
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0655460357666016,
      "learning_rate": 8.956968980608888e-06,
      "loss": 0.8613,
      "step": 18253
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.012303590774536,
      "learning_rate": 8.9563300642111e-06,
      "loss": 1.0742,
      "step": 18254
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1185669898986816,
      "learning_rate": 8.955691147813309e-06,
      "loss": 0.9008,
      "step": 18255
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.0220043659210205,
      "learning_rate": 8.95505223141552e-06,
      "loss": 1.0547,
      "step": 18256
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.131896734237671,
      "learning_rate": 8.95441331501773e-06,
      "loss": 0.9092,
      "step": 18257
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8388103246688843,
      "learning_rate": 8.95377439861994e-06,
      "loss": 0.9181,
      "step": 18258
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9612174034118652,
      "learning_rate": 8.953135482222152e-06,
      "loss": 0.9509,
      "step": 18259
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.895239233970642,
      "learning_rate": 8.952496565824362e-06,
      "loss": 1.0367,
      "step": 18260
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.8439066410064697,
      "learning_rate": 8.951857649426573e-06,
      "loss": 0.9273,
      "step": 18261
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9570035934448242,
      "learning_rate": 8.951218733028783e-06,
      "loss": 0.8147,
      "step": 18262
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.3524246215820312,
      "learning_rate": 8.950579816630995e-06,
      "loss": 0.5574,
      "step": 18263
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2041683197021484,
      "learning_rate": 8.949940900233204e-06,
      "loss": 1.2781,
      "step": 18264
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7764778137207031,
      "learning_rate": 8.949301983835416e-06,
      "loss": 0.9237,
      "step": 18265
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6119790077209473,
      "learning_rate": 8.948663067437626e-06,
      "loss": 0.7594,
      "step": 18266
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.188973903656006,
      "learning_rate": 8.948024151039837e-06,
      "loss": 1.5703,
      "step": 18267
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8778629302978516,
      "learning_rate": 8.947385234642047e-06,
      "loss": 0.9902,
      "step": 18268
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8973462581634521,
      "learning_rate": 8.946746318244259e-06,
      "loss": 0.8746,
      "step": 18269
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2960398197174072,
      "learning_rate": 8.946107401846469e-06,
      "loss": 1.1048,
      "step": 18270
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.312819242477417,
      "learning_rate": 8.94546848544868e-06,
      "loss": 0.9094,
      "step": 18271
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.018340826034546,
      "learning_rate": 8.94482956905089e-06,
      "loss": 0.9375,
      "step": 18272
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.672646164894104,
      "learning_rate": 8.944190652653102e-06,
      "loss": 1.0079,
      "step": 18273
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.044449806213379,
      "learning_rate": 8.943551736255311e-06,
      "loss": 0.7355,
      "step": 18274
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.04587459564209,
      "learning_rate": 8.942912819857521e-06,
      "loss": 1.5509,
      "step": 18275
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9138256311416626,
      "learning_rate": 8.942273903459733e-06,
      "loss": 0.837,
      "step": 18276
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.944818139076233,
      "learning_rate": 8.941634987061943e-06,
      "loss": 1.233,
      "step": 18277
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9377576112747192,
      "learning_rate": 8.940996070664154e-06,
      "loss": 0.8896,
      "step": 18278
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.134049415588379,
      "learning_rate": 8.940357154266364e-06,
      "loss": 1.1656,
      "step": 18279
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5599985122680664,
      "learning_rate": 8.939718237868576e-06,
      "loss": 0.7762,
      "step": 18280
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7606215476989746,
      "learning_rate": 8.939079321470786e-06,
      "loss": 0.9273,
      "step": 18281
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.95817232131958,
      "learning_rate": 8.938440405072997e-06,
      "loss": 1.1033,
      "step": 18282
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6112014055252075,
      "learning_rate": 8.937801488675207e-06,
      "loss": 0.9244,
      "step": 18283
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.282398223876953,
      "learning_rate": 8.937162572277418e-06,
      "loss": 0.6236,
      "step": 18284
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5726583003997803,
      "learning_rate": 8.936523655879628e-06,
      "loss": 0.8967,
      "step": 18285
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2399513721466064,
      "learning_rate": 8.93588473948184e-06,
      "loss": 0.8791,
      "step": 18286
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.896222472190857,
      "learning_rate": 8.93524582308405e-06,
      "loss": 0.8394,
      "step": 18287
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.619411826133728,
      "learning_rate": 8.934606906686261e-06,
      "loss": 0.8731,
      "step": 18288
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0723862648010254,
      "learning_rate": 8.933967990288471e-06,
      "loss": 0.7812,
      "step": 18289
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9260741472244263,
      "learning_rate": 8.933329073890683e-06,
      "loss": 0.8708,
      "step": 18290
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.3633053302764893,
      "learning_rate": 8.932690157492893e-06,
      "loss": 0.7996,
      "step": 18291
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9578596353530884,
      "learning_rate": 8.932051241095104e-06,
      "loss": 0.5219,
      "step": 18292
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6091936826705933,
      "learning_rate": 8.931412324697314e-06,
      "loss": 0.7614,
      "step": 18293
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.841958999633789,
      "learning_rate": 8.930773408299524e-06,
      "loss": 0.7807,
      "step": 18294
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6995904445648193,
      "learning_rate": 8.930134491901735e-06,
      "loss": 0.9738,
      "step": 18295
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8764514923095703,
      "learning_rate": 8.929495575503945e-06,
      "loss": 1.0758,
      "step": 18296
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0704305171966553,
      "learning_rate": 8.928856659106157e-06,
      "loss": 1.2287,
      "step": 18297
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0012123584747314,
      "learning_rate": 8.928217742708367e-06,
      "loss": 0.9021,
      "step": 18298
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0722556114196777,
      "learning_rate": 8.927578826310578e-06,
      "loss": 0.8972,
      "step": 18299
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8823168277740479,
      "learning_rate": 8.926939909912788e-06,
      "loss": 1.1516,
      "step": 18300
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.757576823234558,
      "learning_rate": 8.926300993515e-06,
      "loss": 0.9185,
      "step": 18301
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7889310121536255,
      "learning_rate": 8.92566207711721e-06,
      "loss": 0.8774,
      "step": 18302
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0377800464630127,
      "learning_rate": 8.925023160719421e-06,
      "loss": 1.1524,
      "step": 18303
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7210676670074463,
      "learning_rate": 8.92438424432163e-06,
      "loss": 0.8717,
      "step": 18304
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.087038278579712,
      "learning_rate": 8.923745327923842e-06,
      "loss": 0.9416,
      "step": 18305
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.9931533336639404,
      "learning_rate": 8.923106411526052e-06,
      "loss": 0.8391,
      "step": 18306
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4980485439300537,
      "learning_rate": 8.922467495128264e-06,
      "loss": 0.8408,
      "step": 18307
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8893194198608398,
      "learning_rate": 8.921828578730474e-06,
      "loss": 0.7763,
      "step": 18308
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5749632120132446,
      "learning_rate": 8.921189662332685e-06,
      "loss": 0.8452,
      "step": 18309
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5568904876708984,
      "learning_rate": 8.920550745934895e-06,
      "loss": 0.7919,
      "step": 18310
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.9043757915496826,
      "learning_rate": 8.919911829537105e-06,
      "loss": 0.8838,
      "step": 18311
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6996111869812012,
      "learning_rate": 8.919272913139316e-06,
      "loss": 0.9093,
      "step": 18312
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8125505447387695,
      "learning_rate": 8.918633996741526e-06,
      "loss": 0.9396,
      "step": 18313
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1104788780212402,
      "learning_rate": 8.917995080343738e-06,
      "loss": 0.8904,
      "step": 18314
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.3376152515411377,
      "learning_rate": 8.917356163945948e-06,
      "loss": 0.8703,
      "step": 18315
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8684624433517456,
      "learning_rate": 8.91671724754816e-06,
      "loss": 0.8992,
      "step": 18316
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0951805114746094,
      "learning_rate": 8.916078331150369e-06,
      "loss": 0.7505,
      "step": 18317
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.134462356567383,
      "learning_rate": 8.91543941475258e-06,
      "loss": 0.9847,
      "step": 18318
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1019420623779297,
      "learning_rate": 8.91480049835479e-06,
      "loss": 0.8673,
      "step": 18319
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.409518241882324,
      "learning_rate": 8.914161581957002e-06,
      "loss": 0.8978,
      "step": 18320
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9603830575942993,
      "learning_rate": 8.913522665559212e-06,
      "loss": 1.061,
      "step": 18321
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5389214754104614,
      "learning_rate": 8.912883749161423e-06,
      "loss": 0.7248,
      "step": 18322
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5127800703048706,
      "learning_rate": 8.912244832763633e-06,
      "loss": 0.7905,
      "step": 18323
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7956293821334839,
      "learning_rate": 8.911605916365845e-06,
      "loss": 0.9471,
      "step": 18324
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.307706356048584,
      "learning_rate": 8.910966999968055e-06,
      "loss": 0.8896,
      "step": 18325
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.3657424449920654,
      "learning_rate": 8.910328083570266e-06,
      "loss": 1.0869,
      "step": 18326
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0878796577453613,
      "learning_rate": 8.909689167172476e-06,
      "loss": 0.7116,
      "step": 18327
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9676066637039185,
      "learning_rate": 8.909050250774688e-06,
      "loss": 0.8452,
      "step": 18328
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6334953308105469,
      "learning_rate": 8.908411334376897e-06,
      "loss": 0.9562,
      "step": 18329
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7991187572479248,
      "learning_rate": 8.907772417979107e-06,
      "loss": 0.9375,
      "step": 18330
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4645206928253174,
      "learning_rate": 8.907133501581319e-06,
      "loss": 0.8384,
      "step": 18331
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.827151894569397,
      "learning_rate": 8.906494585183529e-06,
      "loss": 0.9199,
      "step": 18332
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9700589179992676,
      "learning_rate": 8.90585566878574e-06,
      "loss": 0.4295,
      "step": 18333
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.618504285812378,
      "learning_rate": 8.90521675238795e-06,
      "loss": 0.7849,
      "step": 18334
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6255801916122437,
      "learning_rate": 8.904577835990162e-06,
      "loss": 0.9427,
      "step": 18335
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.447269320487976,
      "learning_rate": 8.903938919592371e-06,
      "loss": 0.8463,
      "step": 18336
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8582175970077515,
      "learning_rate": 8.903300003194583e-06,
      "loss": 0.9182,
      "step": 18337
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9639445543289185,
      "learning_rate": 8.902661086796793e-06,
      "loss": 1.1286,
      "step": 18338
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0352814197540283,
      "learning_rate": 8.902022170399004e-06,
      "loss": 0.9174,
      "step": 18339
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.6465647220611572,
      "learning_rate": 8.901383254001214e-06,
      "loss": 0.7567,
      "step": 18340
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.025841474533081,
      "learning_rate": 8.900744337603426e-06,
      "loss": 0.8399,
      "step": 18341
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.6186842918396,
      "learning_rate": 8.900105421205636e-06,
      "loss": 1.4748,
      "step": 18342
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1732068061828613,
      "learning_rate": 8.899466504807847e-06,
      "loss": 0.935,
      "step": 18343
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9097771644592285,
      "learning_rate": 8.898827588410057e-06,
      "loss": 0.883,
      "step": 18344
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7536729574203491,
      "learning_rate": 8.898188672012269e-06,
      "loss": 0.8094,
      "step": 18345
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9810023307800293,
      "learning_rate": 8.897549755614478e-06,
      "loss": 0.9625,
      "step": 18346
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.190288066864014,
      "learning_rate": 8.89691083921669e-06,
      "loss": 1.0256,
      "step": 18347
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9193757772445679,
      "learning_rate": 8.8962719228189e-06,
      "loss": 0.9228,
      "step": 18348
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2483537197113037,
      "learning_rate": 8.89563300642111e-06,
      "loss": 1.0621,
      "step": 18349
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.489576816558838,
      "learning_rate": 8.894994090023321e-06,
      "loss": 0.8348,
      "step": 18350
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2250473499298096,
      "learning_rate": 8.894355173625531e-06,
      "loss": 0.8133,
      "step": 18351
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6223552227020264,
      "learning_rate": 8.893716257227743e-06,
      "loss": 0.8233,
      "step": 18352
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7863215208053589,
      "learning_rate": 8.893077340829953e-06,
      "loss": 1.0016,
      "step": 18353
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1017708778381348,
      "learning_rate": 8.892438424432164e-06,
      "loss": 0.8338,
      "step": 18354
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.384521484375,
      "learning_rate": 8.891799508034374e-06,
      "loss": 0.4331,
      "step": 18355
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6432629823684692,
      "learning_rate": 8.891160591636585e-06,
      "loss": 0.9087,
      "step": 18356
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.808710098266602,
      "learning_rate": 8.890521675238795e-06,
      "loss": 1.0889,
      "step": 18357
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.083888530731201,
      "learning_rate": 8.889882758841007e-06,
      "loss": 0.7577,
      "step": 18358
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6483498811721802,
      "learning_rate": 8.889243842443217e-06,
      "loss": 0.8447,
      "step": 18359
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5559437274932861,
      "learning_rate": 8.888604926045428e-06,
      "loss": 0.882,
      "step": 18360
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9293173551559448,
      "learning_rate": 8.887966009647638e-06,
      "loss": 0.85,
      "step": 18361
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1323795318603516,
      "learning_rate": 8.88732709324985e-06,
      "loss": 1.1399,
      "step": 18362
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9612153768539429,
      "learning_rate": 8.88668817685206e-06,
      "loss": 0.824,
      "step": 18363
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.7335920333862305,
      "learning_rate": 8.886049260454271e-06,
      "loss": 1.0654,
      "step": 18364
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.596597671508789,
      "learning_rate": 8.885410344056481e-06,
      "loss": 0.8627,
      "step": 18365
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.6049458980560303,
      "learning_rate": 8.88477142765869e-06,
      "loss": 1.0157,
      "step": 18366
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7521318197250366,
      "learning_rate": 8.884132511260902e-06,
      "loss": 0.8525,
      "step": 18367
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7513985633850098,
      "learning_rate": 8.883493594863112e-06,
      "loss": 0.8534,
      "step": 18368
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5576568841934204,
      "learning_rate": 8.882854678465324e-06,
      "loss": 0.7713,
      "step": 18369
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.589125633239746,
      "learning_rate": 8.882215762067534e-06,
      "loss": 1.0033,
      "step": 18370
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6765459775924683,
      "learning_rate": 8.881576845669745e-06,
      "loss": 0.4453,
      "step": 18371
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.3025643825531006,
      "learning_rate": 8.880937929271955e-06,
      "loss": 1.0207,
      "step": 18372
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1325607299804688,
      "learning_rate": 8.880299012874167e-06,
      "loss": 0.9104,
      "step": 18373
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6494818925857544,
      "learning_rate": 8.879660096476376e-06,
      "loss": 0.8466,
      "step": 18374
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4236180782318115,
      "learning_rate": 8.879021180078588e-06,
      "loss": 1.0244,
      "step": 18375
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7354565858840942,
      "learning_rate": 8.878382263680798e-06,
      "loss": 0.8081,
      "step": 18376
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.883766770362854,
      "learning_rate": 8.87774334728301e-06,
      "loss": 1.2565,
      "step": 18377
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6919361352920532,
      "learning_rate": 8.87710443088522e-06,
      "loss": 0.9976,
      "step": 18378
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.541459083557129,
      "learning_rate": 8.87646551448743e-06,
      "loss": 0.8683,
      "step": 18379
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1752192974090576,
      "learning_rate": 8.87582659808964e-06,
      "loss": 0.8965,
      "step": 18380
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5870596170425415,
      "learning_rate": 8.875187681691852e-06,
      "loss": 0.7552,
      "step": 18381
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4722445011138916,
      "learning_rate": 8.874548765294062e-06,
      "loss": 0.8774,
      "step": 18382
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7906908988952637,
      "learning_rate": 8.873909848896274e-06,
      "loss": 0.8149,
      "step": 18383
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9398770332336426,
      "learning_rate": 8.873270932498483e-06,
      "loss": 1.194,
      "step": 18384
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4965956211090088,
      "learning_rate": 8.872632016100693e-06,
      "loss": 0.8568,
      "step": 18385
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0275800228118896,
      "learning_rate": 8.871993099702905e-06,
      "loss": 0.7778,
      "step": 18386
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.677978515625,
      "learning_rate": 8.871354183305115e-06,
      "loss": 0.7937,
      "step": 18387
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5490572452545166,
      "learning_rate": 8.870715266907326e-06,
      "loss": 0.4218,
      "step": 18388
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.715441346168518,
      "learning_rate": 8.870076350509536e-06,
      "loss": 0.8522,
      "step": 18389
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.7666120529174805,
      "learning_rate": 8.869437434111748e-06,
      "loss": 0.8423,
      "step": 18390
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4617443084716797,
      "learning_rate": 8.868798517713957e-06,
      "loss": 0.4843,
      "step": 18391
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6940045356750488,
      "learning_rate": 8.868159601316169e-06,
      "loss": 0.863,
      "step": 18392
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.615392804145813,
      "learning_rate": 8.867520684918379e-06,
      "loss": 0.8773,
      "step": 18393
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.792988657951355,
      "learning_rate": 8.86688176852059e-06,
      "loss": 0.859,
      "step": 18394
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9696550369262695,
      "learning_rate": 8.8662428521228e-06,
      "loss": 0.9763,
      "step": 18395
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5065289735794067,
      "learning_rate": 8.865603935725012e-06,
      "loss": 0.8073,
      "step": 18396
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1030540466308594,
      "learning_rate": 8.864965019327222e-06,
      "loss": 0.9615,
      "step": 18397
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.9129321575164795,
      "learning_rate": 8.864326102929433e-06,
      "loss": 0.9613,
      "step": 18398
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.01456618309021,
      "learning_rate": 8.863687186531643e-06,
      "loss": 1.0105,
      "step": 18399
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9499986171722412,
      "learning_rate": 8.863048270133855e-06,
      "loss": 0.8972,
      "step": 18400
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9312242269515991,
      "learning_rate": 8.862409353736064e-06,
      "loss": 0.9462,
      "step": 18401
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6269961595535278,
      "learning_rate": 8.861770437338276e-06,
      "loss": 0.9349,
      "step": 18402
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.969179153442383,
      "learning_rate": 8.861131520940486e-06,
      "loss": 1.5679,
      "step": 18403
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1137759685516357,
      "learning_rate": 8.860492604542696e-06,
      "loss": 0.9441,
      "step": 18404
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.697655439376831,
      "learning_rate": 8.859853688144907e-06,
      "loss": 0.8511,
      "step": 18405
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7016123533248901,
      "learning_rate": 8.859214771747117e-06,
      "loss": 0.8724,
      "step": 18406
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3279670476913452,
      "learning_rate": 8.858575855349329e-06,
      "loss": 0.4525,
      "step": 18407
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5353881120681763,
      "learning_rate": 8.857936938951538e-06,
      "loss": 0.9096,
      "step": 18408
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.670849084854126,
      "learning_rate": 8.85729802255375e-06,
      "loss": 0.8498,
      "step": 18409
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.7661845684051514,
      "learning_rate": 8.85665910615596e-06,
      "loss": 0.7825,
      "step": 18410
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.193692922592163,
      "learning_rate": 8.856020189758171e-06,
      "loss": 1.2542,
      "step": 18411
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.877498626708984,
      "learning_rate": 8.855381273360381e-06,
      "loss": 1.4807,
      "step": 18412
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.309442520141602,
      "learning_rate": 8.854742356962593e-06,
      "loss": 0.9558,
      "step": 18413
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6669495105743408,
      "learning_rate": 8.854103440564803e-06,
      "loss": 0.9055,
      "step": 18414
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1613669395446777,
      "learning_rate": 8.853464524167014e-06,
      "loss": 1.1103,
      "step": 18415
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5432980060577393,
      "learning_rate": 8.852825607769224e-06,
      "loss": 0.9474,
      "step": 18416
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6141026020050049,
      "learning_rate": 8.852186691371436e-06,
      "loss": 0.8273,
      "step": 18417
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9550303220748901,
      "learning_rate": 8.851547774973645e-06,
      "loss": 0.7354,
      "step": 18418
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.560795783996582,
      "learning_rate": 8.850908858575857e-06,
      "loss": 0.9519,
      "step": 18419
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8327651023864746,
      "learning_rate": 8.850269942178067e-06,
      "loss": 0.8485,
      "step": 18420
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.916715145111084,
      "learning_rate": 8.849631025780277e-06,
      "loss": 0.8442,
      "step": 18421
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.983723521232605,
      "learning_rate": 8.848992109382488e-06,
      "loss": 0.8607,
      "step": 18422
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5735950469970703,
      "learning_rate": 8.848353192984698e-06,
      "loss": 0.8984,
      "step": 18423
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.603430271148682,
      "learning_rate": 8.84771427658691e-06,
      "loss": 1.5217,
      "step": 18424
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.4536936283111572,
      "learning_rate": 8.84707536018912e-06,
      "loss": 1.1962,
      "step": 18425
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1087708473205566,
      "learning_rate": 8.846436443791331e-06,
      "loss": 1.2008,
      "step": 18426
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7217206954956055,
      "learning_rate": 8.845797527393541e-06,
      "loss": 0.8527,
      "step": 18427
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.138808488845825,
      "learning_rate": 8.845158610995752e-06,
      "loss": 1.1807,
      "step": 18428
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.801114797592163,
      "learning_rate": 8.844519694597962e-06,
      "loss": 1.0626,
      "step": 18429
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0541462898254395,
      "learning_rate": 8.843880778200174e-06,
      "loss": 0.9424,
      "step": 18430
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.318005561828613,
      "learning_rate": 8.843241861802384e-06,
      "loss": 1.7887,
      "step": 18431
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.745819330215454,
      "learning_rate": 8.842602945404595e-06,
      "loss": 1.0975,
      "step": 18432
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.832161784172058,
      "learning_rate": 8.841964029006805e-06,
      "loss": 0.9281,
      "step": 18433
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6115641593933105,
      "learning_rate": 8.841325112609017e-06,
      "loss": 0.3607,
      "step": 18434
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.418614387512207,
      "learning_rate": 8.840686196211227e-06,
      "loss": 0.4626,
      "step": 18435
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9003019332885742,
      "learning_rate": 8.840047279813438e-06,
      "loss": 0.9151,
      "step": 18436
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.759428977966309,
      "learning_rate": 8.839408363415648e-06,
      "loss": 1.7036,
      "step": 18437
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.100175619125366,
      "learning_rate": 8.83876944701786e-06,
      "loss": 1.0284,
      "step": 18438
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5529472827911377,
      "learning_rate": 8.83813053062007e-06,
      "loss": 0.65,
      "step": 18439
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.580949306488037,
      "learning_rate": 8.83749161422228e-06,
      "loss": 0.8713,
      "step": 18440
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6359606981277466,
      "learning_rate": 8.83685269782449e-06,
      "loss": 0.8804,
      "step": 18441
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.033827781677246,
      "learning_rate": 8.8362137814267e-06,
      "loss": 1.362,
      "step": 18442
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.9312849044799805,
      "learning_rate": 8.835574865028912e-06,
      "loss": 0.8449,
      "step": 18443
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5183658599853516,
      "learning_rate": 8.834935948631122e-06,
      "loss": 0.9353,
      "step": 18444
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0452663898468018,
      "learning_rate": 8.834297032233334e-06,
      "loss": 0.8754,
      "step": 18445
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.086559772491455,
      "learning_rate": 8.833658115835543e-06,
      "loss": 0.9222,
      "step": 18446
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.346450090408325,
      "learning_rate": 8.833019199437755e-06,
      "loss": 0.9623,
      "step": 18447
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9921987056732178,
      "learning_rate": 8.832380283039965e-06,
      "loss": 1.1411,
      "step": 18448
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.863741397857666,
      "learning_rate": 8.831741366642176e-06,
      "loss": 1.1927,
      "step": 18449
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.737060785293579,
      "learning_rate": 8.831102450244386e-06,
      "loss": 0.8581,
      "step": 18450
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.539189338684082,
      "learning_rate": 8.830463533846598e-06,
      "loss": 0.9847,
      "step": 18451
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6820482015609741,
      "learning_rate": 8.829824617448808e-06,
      "loss": 0.8176,
      "step": 18452
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1831839084625244,
      "learning_rate": 8.829185701051019e-06,
      "loss": 0.9583,
      "step": 18453
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.3361430168151855,
      "learning_rate": 8.828546784653229e-06,
      "loss": 0.9763,
      "step": 18454
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.581125020980835,
      "learning_rate": 8.82790786825544e-06,
      "loss": 0.8877,
      "step": 18455
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2500834465026855,
      "learning_rate": 8.82726895185765e-06,
      "loss": 0.902,
      "step": 18456
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6030583381652832,
      "learning_rate": 8.82663003545986e-06,
      "loss": 0.9418,
      "step": 18457
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.9926598072052,
      "learning_rate": 8.825991119062072e-06,
      "loss": 0.6439,
      "step": 18458
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8900513648986816,
      "learning_rate": 8.825352202664282e-06,
      "loss": 0.8901,
      "step": 18459
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.79580020904541,
      "learning_rate": 8.824713286266493e-06,
      "loss": 0.8295,
      "step": 18460
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.367974281311035,
      "learning_rate": 8.824074369868703e-06,
      "loss": 0.6855,
      "step": 18461
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9931696653366089,
      "learning_rate": 8.823435453470915e-06,
      "loss": 0.4066,
      "step": 18462
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.925633192062378,
      "learning_rate": 8.822796537073124e-06,
      "loss": 0.7214,
      "step": 18463
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9676569700241089,
      "learning_rate": 8.822157620675336e-06,
      "loss": 1.1657,
      "step": 18464
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6476749181747437,
      "learning_rate": 8.821518704277546e-06,
      "loss": 0.8237,
      "step": 18465
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4410245418548584,
      "learning_rate": 8.820879787879757e-06,
      "loss": 0.8591,
      "step": 18466
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.052565097808838,
      "learning_rate": 8.820240871481967e-06,
      "loss": 1.2038,
      "step": 18467
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8307355642318726,
      "learning_rate": 8.819601955084179e-06,
      "loss": 0.5416,
      "step": 18468
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8102399110794067,
      "learning_rate": 8.818963038686389e-06,
      "loss": 0.8904,
      "step": 18469
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9804340600967407,
      "learning_rate": 8.8183241222886e-06,
      "loss": 0.9947,
      "step": 18470
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4323474168777466,
      "learning_rate": 8.81768520589081e-06,
      "loss": 0.8662,
      "step": 18471
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.71914803981781,
      "learning_rate": 8.817046289493022e-06,
      "loss": 0.8454,
      "step": 18472
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2838199138641357,
      "learning_rate": 8.816407373095231e-06,
      "loss": 0.7469,
      "step": 18473
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4651747941970825,
      "learning_rate": 8.815768456697443e-06,
      "loss": 0.4686,
      "step": 18474
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8303683996200562,
      "learning_rate": 8.815129540299653e-06,
      "loss": 0.8037,
      "step": 18475
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.28452730178833,
      "learning_rate": 8.814490623901863e-06,
      "loss": 0.8449,
      "step": 18476
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5107371807098389,
      "learning_rate": 8.813851707504074e-06,
      "loss": 0.3729,
      "step": 18477
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.107856512069702,
      "learning_rate": 8.813212791106284e-06,
      "loss": 0.9556,
      "step": 18478
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4553768634796143,
      "learning_rate": 8.812573874708496e-06,
      "loss": 0.8006,
      "step": 18479
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6586626768112183,
      "learning_rate": 8.811934958310705e-06,
      "loss": 0.9568,
      "step": 18480
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.634096622467041,
      "learning_rate": 8.811296041912917e-06,
      "loss": 0.8179,
      "step": 18481
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5167310237884521,
      "learning_rate": 8.810657125515127e-06,
      "loss": 0.8567,
      "step": 18482
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0236918926239014,
      "learning_rate": 8.810018209117338e-06,
      "loss": 0.8697,
      "step": 18483
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5883597135543823,
      "learning_rate": 8.809379292719548e-06,
      "loss": 0.7799,
      "step": 18484
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.129903793334961,
      "learning_rate": 8.80874037632176e-06,
      "loss": 1.2231,
      "step": 18485
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0437064170837402,
      "learning_rate": 8.80810145992397e-06,
      "loss": 0.8455,
      "step": 18486
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7084038257598877,
      "learning_rate": 8.807462543526181e-06,
      "loss": 0.8595,
      "step": 18487
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.029287099838257,
      "learning_rate": 8.806823627128391e-06,
      "loss": 0.6841,
      "step": 18488
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.8106915950775146,
      "learning_rate": 8.806184710730603e-06,
      "loss": 1.1129,
      "step": 18489
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9657727479934692,
      "learning_rate": 8.805545794332812e-06,
      "loss": 0.8816,
      "step": 18490
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.760270595550537,
      "learning_rate": 8.804906877935024e-06,
      "loss": 0.9547,
      "step": 18491
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3215233087539673,
      "learning_rate": 8.804267961537234e-06,
      "loss": 0.4746,
      "step": 18492
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6715248823165894,
      "learning_rate": 8.803629045139445e-06,
      "loss": 0.9244,
      "step": 18493
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5182392597198486,
      "learning_rate": 8.802990128741655e-06,
      "loss": 0.9798,
      "step": 18494
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.54325795173645,
      "learning_rate": 8.802351212343865e-06,
      "loss": 0.8368,
      "step": 18495
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9142038822174072,
      "learning_rate": 8.801712295946077e-06,
      "loss": 0.8932,
      "step": 18496
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7729154825210571,
      "learning_rate": 8.801073379548287e-06,
      "loss": 0.9481,
      "step": 18497
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0779807567596436,
      "learning_rate": 8.800434463150498e-06,
      "loss": 0.8256,
      "step": 18498
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6552398204803467,
      "learning_rate": 8.799795546752708e-06,
      "loss": 0.9344,
      "step": 18499
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.025665044784546,
      "learning_rate": 8.79915663035492e-06,
      "loss": 0.9945,
      "step": 18500
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.146167278289795,
      "learning_rate": 8.79851771395713e-06,
      "loss": 1.1638,
      "step": 18501
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7672088146209717,
      "learning_rate": 8.797878797559341e-06,
      "loss": 0.879,
      "step": 18502
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6546072959899902,
      "learning_rate": 8.79723988116155e-06,
      "loss": 0.8576,
      "step": 18503
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.384239912033081,
      "learning_rate": 8.796600964763762e-06,
      "loss": 1.0896,
      "step": 18504
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4357335567474365,
      "learning_rate": 8.795962048365972e-06,
      "loss": 0.6684,
      "step": 18505
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4643874168395996,
      "learning_rate": 8.795323131968184e-06,
      "loss": 0.8411,
      "step": 18506
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6822580099105835,
      "learning_rate": 8.794684215570394e-06,
      "loss": 0.8308,
      "step": 18507
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.183115243911743,
      "learning_rate": 8.794045299172605e-06,
      "loss": 1.1767,
      "step": 18508
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2701282501220703,
      "learning_rate": 8.793406382774815e-06,
      "loss": 1.0703,
      "step": 18509
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8764960765838623,
      "learning_rate": 8.792767466377026e-06,
      "loss": 0.951,
      "step": 18510
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.617512822151184,
      "learning_rate": 8.792128549979236e-06,
      "loss": 0.7712,
      "step": 18511
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.139394760131836,
      "learning_rate": 8.791489633581446e-06,
      "loss": 1.6428,
      "step": 18512
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6452198028564453,
      "learning_rate": 8.790850717183658e-06,
      "loss": 0.9408,
      "step": 18513
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.959938883781433,
      "learning_rate": 8.790211800785868e-06,
      "loss": 0.8921,
      "step": 18514
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7039536237716675,
      "learning_rate": 8.789572884388079e-06,
      "loss": 0.9247,
      "step": 18515
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4833933115005493,
      "learning_rate": 8.788933967990289e-06,
      "loss": 0.7792,
      "step": 18516
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.515583515167236,
      "learning_rate": 8.7882950515925e-06,
      "loss": 1.2067,
      "step": 18517
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6306681632995605,
      "learning_rate": 8.78765613519471e-06,
      "loss": 0.5284,
      "step": 18518
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2493510246276855,
      "learning_rate": 8.787017218796922e-06,
      "loss": 0.932,
      "step": 18519
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2445688247680664,
      "learning_rate": 8.786378302399132e-06,
      "loss": 0.8318,
      "step": 18520
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7691254615783691,
      "learning_rate": 8.785739386001343e-06,
      "loss": 0.8679,
      "step": 18521
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.1372475624084473,
      "learning_rate": 8.785100469603553e-06,
      "loss": 0.8641,
      "step": 18522
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2255077362060547,
      "learning_rate": 8.784461553205765e-06,
      "loss": 0.9714,
      "step": 18523
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.6121623516082764,
      "learning_rate": 8.783822636807975e-06,
      "loss": 1.0594,
      "step": 18524
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.084027051925659,
      "learning_rate": 8.783183720410186e-06,
      "loss": 0.8577,
      "step": 18525
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.727764129638672,
      "learning_rate": 8.782544804012396e-06,
      "loss": 0.8616,
      "step": 18526
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5836418867111206,
      "learning_rate": 8.781905887614608e-06,
      "loss": 0.841,
      "step": 18527
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9384559392929077,
      "learning_rate": 8.781266971216817e-06,
      "loss": 0.8731,
      "step": 18528
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9534043073654175,
      "learning_rate": 8.780628054819029e-06,
      "loss": 1.0021,
      "step": 18529
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7220972776412964,
      "learning_rate": 8.779989138421239e-06,
      "loss": 0.8386,
      "step": 18530
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.83871591091156,
      "learning_rate": 8.779350222023449e-06,
      "loss": 0.9052,
      "step": 18531
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.787222146987915,
      "learning_rate": 8.77871130562566e-06,
      "loss": 0.8135,
      "step": 18532
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.983632802963257,
      "learning_rate": 8.77807238922787e-06,
      "loss": 0.668,
      "step": 18533
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7526508569717407,
      "learning_rate": 8.777433472830082e-06,
      "loss": 0.9713,
      "step": 18534
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7995063066482544,
      "learning_rate": 8.776794556432291e-06,
      "loss": 0.9567,
      "step": 18535
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.063565254211426,
      "learning_rate": 8.776155640034503e-06,
      "loss": 1.5198,
      "step": 18536
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.744539260864258,
      "learning_rate": 8.775516723636713e-06,
      "loss": 1.6332,
      "step": 18537
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8352116346359253,
      "learning_rate": 8.774877807238924e-06,
      "loss": 0.8274,
      "step": 18538
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7048442363739014,
      "learning_rate": 8.774238890841134e-06,
      "loss": 0.7742,
      "step": 18539
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.61984920501709,
      "learning_rate": 8.773599974443346e-06,
      "loss": 1.141,
      "step": 18540
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9281103610992432,
      "learning_rate": 8.772961058045556e-06,
      "loss": 0.9005,
      "step": 18541
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9821240901947021,
      "learning_rate": 8.772322141647767e-06,
      "loss": 0.875,
      "step": 18542
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7384212017059326,
      "learning_rate": 8.771683225249977e-06,
      "loss": 0.8956,
      "step": 18543
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7330790758132935,
      "learning_rate": 8.771044308852189e-06,
      "loss": 0.8753,
      "step": 18544
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.393547534942627,
      "learning_rate": 8.770405392454398e-06,
      "loss": 0.7702,
      "step": 18545
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1448936462402344,
      "learning_rate": 8.76976647605661e-06,
      "loss": 1.012,
      "step": 18546
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9130951166152954,
      "learning_rate": 8.76912755965882e-06,
      "loss": 0.8642,
      "step": 18547
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.780905842781067,
      "learning_rate": 8.768488643261031e-06,
      "loss": 0.8829,
      "step": 18548
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5993130207061768,
      "learning_rate": 8.767849726863241e-06,
      "loss": 0.7848,
      "step": 18549
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.01851224899292,
      "learning_rate": 8.767210810465451e-06,
      "loss": 1.111,
      "step": 18550
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8329793214797974,
      "learning_rate": 8.766571894067663e-06,
      "loss": 1.2141,
      "step": 18551
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.4285459518432617,
      "learning_rate": 8.765932977669872e-06,
      "loss": 0.9816,
      "step": 18552
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.476089596748352,
      "learning_rate": 8.765294061272082e-06,
      "loss": 0.4366,
      "step": 18553
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7368552684783936,
      "learning_rate": 8.764655144874294e-06,
      "loss": 0.8066,
      "step": 18554
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2389485836029053,
      "learning_rate": 8.764016228476504e-06,
      "loss": 0.9295,
      "step": 18555
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0159671306610107,
      "learning_rate": 8.763377312078715e-06,
      "loss": 0.9256,
      "step": 18556
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.7116641998291016,
      "learning_rate": 8.762738395680925e-06,
      "loss": 1.1585,
      "step": 18557
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8045451641082764,
      "learning_rate": 8.762099479283137e-06,
      "loss": 0.886,
      "step": 18558
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5092406272888184,
      "learning_rate": 8.761460562885347e-06,
      "loss": 0.9275,
      "step": 18559
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8452458381652832,
      "learning_rate": 8.760821646487558e-06,
      "loss": 0.9135,
      "step": 18560
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3626015186309814,
      "learning_rate": 8.760182730089768e-06,
      "loss": 0.9305,
      "step": 18561
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9652040004730225,
      "learning_rate": 8.759543813691978e-06,
      "loss": 0.8658,
      "step": 18562
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9248069524765015,
      "learning_rate": 8.75890489729419e-06,
      "loss": 0.8948,
      "step": 18563
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9517995119094849,
      "learning_rate": 8.7582659808964e-06,
      "loss": 0.8657,
      "step": 18564
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.773650884628296,
      "learning_rate": 8.75762706449861e-06,
      "loss": 0.8453,
      "step": 18565
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.231191396713257,
      "learning_rate": 8.75698814810082e-06,
      "loss": 0.9289,
      "step": 18566
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9313963651657104,
      "learning_rate": 8.756349231703032e-06,
      "loss": 0.8839,
      "step": 18567
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.141219139099121,
      "learning_rate": 8.755710315305242e-06,
      "loss": 0.6814,
      "step": 18568
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.815943956375122,
      "learning_rate": 8.755071398907454e-06,
      "loss": 0.7458,
      "step": 18569
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1423590183258057,
      "learning_rate": 8.754432482509663e-06,
      "loss": 0.7754,
      "step": 18570
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1938271522521973,
      "learning_rate": 8.753793566111875e-06,
      "loss": 0.9119,
      "step": 18571
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6401678323745728,
      "learning_rate": 8.753154649714085e-06,
      "loss": 0.8146,
      "step": 18572
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7192573547363281,
      "learning_rate": 8.752515733316296e-06,
      "loss": 0.9788,
      "step": 18573
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.7691822052001953,
      "learning_rate": 8.751876816918506e-06,
      "loss": 0.9011,
      "step": 18574
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3330299854278564,
      "learning_rate": 8.751237900520718e-06,
      "loss": 0.8541,
      "step": 18575
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.005476236343384,
      "learning_rate": 8.750598984122928e-06,
      "loss": 0.7789,
      "step": 18576
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.624927043914795,
      "learning_rate": 8.749960067725139e-06,
      "loss": 1.0704,
      "step": 18577
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.107729434967041,
      "learning_rate": 8.749321151327349e-06,
      "loss": 1.211,
      "step": 18578
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.107248544692993,
      "learning_rate": 8.74868223492956e-06,
      "loss": 0.932,
      "step": 18579
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6274397373199463,
      "learning_rate": 8.74804331853177e-06,
      "loss": 0.7514,
      "step": 18580
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.258453607559204,
      "learning_rate": 8.74740440213398e-06,
      "loss": 0.7865,
      "step": 18581
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.003756046295166,
      "learning_rate": 8.746765485736192e-06,
      "loss": 1.2364,
      "step": 18582
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.876487374305725,
      "learning_rate": 8.746126569338402e-06,
      "loss": 0.9046,
      "step": 18583
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6280417442321777,
      "learning_rate": 8.745487652940613e-06,
      "loss": 0.8147,
      "step": 18584
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.4837331771850586,
      "learning_rate": 8.744848736542823e-06,
      "loss": 1.09,
      "step": 18585
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9563331604003906,
      "learning_rate": 8.744209820145035e-06,
      "loss": 0.8621,
      "step": 18586
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.599972248077393,
      "learning_rate": 8.743570903747244e-06,
      "loss": 0.8026,
      "step": 18587
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8461410999298096,
      "learning_rate": 8.742931987349456e-06,
      "loss": 0.9043,
      "step": 18588
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0297605991363525,
      "learning_rate": 8.742293070951666e-06,
      "loss": 0.8235,
      "step": 18589
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.860119342803955,
      "learning_rate": 8.741654154553877e-06,
      "loss": 1.4338,
      "step": 18590
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9264717102050781,
      "learning_rate": 8.741015238156087e-06,
      "loss": 1.354,
      "step": 18591
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9681825637817383,
      "learning_rate": 8.740376321758299e-06,
      "loss": 0.9881,
      "step": 18592
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6780223846435547,
      "learning_rate": 8.739737405360509e-06,
      "loss": 0.9127,
      "step": 18593
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6306074857711792,
      "learning_rate": 8.73909848896272e-06,
      "loss": 0.8497,
      "step": 18594
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.4576170444488525,
      "learning_rate": 8.73845957256493e-06,
      "loss": 0.9941,
      "step": 18595
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0275893211364746,
      "learning_rate": 8.737820656167142e-06,
      "loss": 1.0342,
      "step": 18596
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.333228349685669,
      "learning_rate": 8.737181739769351e-06,
      "loss": 1.0064,
      "step": 18597
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7855602502822876,
      "learning_rate": 8.736542823371561e-06,
      "loss": 0.8026,
      "step": 18598
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.819754958152771,
      "learning_rate": 8.735903906973773e-06,
      "loss": 0.8477,
      "step": 18599
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.016125440597534,
      "learning_rate": 8.735264990575983e-06,
      "loss": 0.788,
      "step": 18600
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.601898193359375,
      "learning_rate": 8.734626074178194e-06,
      "loss": 0.825,
      "step": 18601
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.964028835296631,
      "learning_rate": 8.733987157780404e-06,
      "loss": 0.8726,
      "step": 18602
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.9997270107269287,
      "learning_rate": 8.733348241382616e-06,
      "loss": 0.7669,
      "step": 18603
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.855983257293701,
      "learning_rate": 8.732709324984826e-06,
      "loss": 0.9034,
      "step": 18604
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.724927544593811,
      "learning_rate": 8.732070408587037e-06,
      "loss": 0.8807,
      "step": 18605
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.299349546432495,
      "learning_rate": 8.731431492189247e-06,
      "loss": 1.0341,
      "step": 18606
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9116209745407104,
      "learning_rate": 8.730792575791458e-06,
      "loss": 1.2286,
      "step": 18607
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.716055154800415,
      "learning_rate": 8.730153659393668e-06,
      "loss": 1.0492,
      "step": 18608
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5069472789764404,
      "learning_rate": 8.72951474299588e-06,
      "loss": 0.7514,
      "step": 18609
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4061803817749023,
      "learning_rate": 8.72887582659809e-06,
      "loss": 0.9026,
      "step": 18610
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4856680631637573,
      "learning_rate": 8.728236910200301e-06,
      "loss": 0.8476,
      "step": 18611
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.8451285362243652,
      "learning_rate": 8.727597993802511e-06,
      "loss": 0.9426,
      "step": 18612
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.446878433227539,
      "learning_rate": 8.726959077404723e-06,
      "loss": 0.2927,
      "step": 18613
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8223514556884766,
      "learning_rate": 8.726320161006933e-06,
      "loss": 0.8826,
      "step": 18614
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.529285430908203,
      "learning_rate": 8.725681244609144e-06,
      "loss": 1.3397,
      "step": 18615
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.524773597717285,
      "learning_rate": 8.725042328211354e-06,
      "loss": 2.0203,
      "step": 18616
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6020857095718384,
      "learning_rate": 8.724403411813564e-06,
      "loss": 0.8626,
      "step": 18617
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.084663152694702,
      "learning_rate": 8.723764495415775e-06,
      "loss": 0.7753,
      "step": 18618
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5689412355422974,
      "learning_rate": 8.723125579017985e-06,
      "loss": 0.8451,
      "step": 18619
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.269566297531128,
      "learning_rate": 8.722486662620197e-06,
      "loss": 0.9348,
      "step": 18620
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.699019432067871,
      "learning_rate": 8.721847746222407e-06,
      "loss": 0.8764,
      "step": 18621
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7461258172988892,
      "learning_rate": 8.721208829824618e-06,
      "loss": 0.9014,
      "step": 18622
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9927995204925537,
      "learning_rate": 8.720569913426828e-06,
      "loss": 0.9418,
      "step": 18623
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6930288076400757,
      "learning_rate": 8.71993099702904e-06,
      "loss": 0.957,
      "step": 18624
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.2142488956451416,
      "learning_rate": 8.71929208063125e-06,
      "loss": 0.7266,
      "step": 18625
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9390627145767212,
      "learning_rate": 8.718653164233461e-06,
      "loss": 0.9579,
      "step": 18626
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.036731243133545,
      "learning_rate": 8.71801424783567e-06,
      "loss": 0.8413,
      "step": 18627
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.010708332061768,
      "learning_rate": 8.717375331437882e-06,
      "loss": 0.8849,
      "step": 18628
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.207496404647827,
      "learning_rate": 8.716736415040092e-06,
      "loss": 1.0253,
      "step": 18629
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7570761442184448,
      "learning_rate": 8.716097498642304e-06,
      "loss": 0.8901,
      "step": 18630
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.7585537433624268,
      "learning_rate": 8.715458582244514e-06,
      "loss": 1.0006,
      "step": 18631
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.472551345825195,
      "learning_rate": 8.714819665846725e-06,
      "loss": 1.4876,
      "step": 18632
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.881807565689087,
      "learning_rate": 8.714180749448935e-06,
      "loss": 0.8886,
      "step": 18633
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6069906949996948,
      "learning_rate": 8.713541833051145e-06,
      "loss": 0.8822,
      "step": 18634
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9817893505096436,
      "learning_rate": 8.712902916653356e-06,
      "loss": 0.9535,
      "step": 18635
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7284728288650513,
      "learning_rate": 8.712264000255566e-06,
      "loss": 0.8703,
      "step": 18636
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6354107856750488,
      "learning_rate": 8.711625083857778e-06,
      "loss": 0.7939,
      "step": 18637
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1496498584747314,
      "learning_rate": 8.710986167459988e-06,
      "loss": 0.9522,
      "step": 18638
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6983271837234497,
      "learning_rate": 8.7103472510622e-06,
      "loss": 0.9087,
      "step": 18639
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7998487949371338,
      "learning_rate": 8.709708334664409e-06,
      "loss": 0.9093,
      "step": 18640
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.063026189804077,
      "learning_rate": 8.70906941826662e-06,
      "loss": 0.817,
      "step": 18641
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5154201984405518,
      "learning_rate": 8.70843050186883e-06,
      "loss": 0.4234,
      "step": 18642
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8734767436981201,
      "learning_rate": 8.707791585471042e-06,
      "loss": 0.9085,
      "step": 18643
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9520386457443237,
      "learning_rate": 8.707152669073252e-06,
      "loss": 0.9715,
      "step": 18644
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6597895622253418,
      "learning_rate": 8.706513752675463e-06,
      "loss": 0.9924,
      "step": 18645
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.802720308303833,
      "learning_rate": 8.705874836277673e-06,
      "loss": 0.9652,
      "step": 18646
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.0654256343841553,
      "learning_rate": 8.705235919879885e-06,
      "loss": 0.653,
      "step": 18647
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4895137548446655,
      "learning_rate": 8.704597003482095e-06,
      "loss": 0.8929,
      "step": 18648
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1819305419921875,
      "learning_rate": 8.703958087084306e-06,
      "loss": 1.0207,
      "step": 18649
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1816468238830566,
      "learning_rate": 8.703319170686516e-06,
      "loss": 0.9192,
      "step": 18650
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6235039234161377,
      "learning_rate": 8.702680254288728e-06,
      "loss": 0.8761,
      "step": 18651
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1029160022735596,
      "learning_rate": 8.702041337890937e-06,
      "loss": 0.9439,
      "step": 18652
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.00037145614624,
      "learning_rate": 8.701402421493147e-06,
      "loss": 1.1198,
      "step": 18653
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.654930591583252,
      "learning_rate": 8.700763505095359e-06,
      "loss": 0.8629,
      "step": 18654
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0041024684906006,
      "learning_rate": 8.700124588697569e-06,
      "loss": 0.8742,
      "step": 18655
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9952268600463867,
      "learning_rate": 8.69948567229978e-06,
      "loss": 0.8291,
      "step": 18656
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6999680995941162,
      "learning_rate": 8.69884675590199e-06,
      "loss": 0.8584,
      "step": 18657
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6927441358566284,
      "learning_rate": 8.698207839504202e-06,
      "loss": 0.8525,
      "step": 18658
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.776276707649231,
      "learning_rate": 8.697568923106411e-06,
      "loss": 0.7573,
      "step": 18659
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.756184458732605,
      "learning_rate": 8.696930006708623e-06,
      "loss": 0.8612,
      "step": 18660
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9588911533355713,
      "learning_rate": 8.696291090310833e-06,
      "loss": 0.9609,
      "step": 18661
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3614606857299805,
      "learning_rate": 8.695652173913044e-06,
      "loss": 0.9292,
      "step": 18662
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0026257038116455,
      "learning_rate": 8.695013257515254e-06,
      "loss": 0.896,
      "step": 18663
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.102860927581787,
      "learning_rate": 8.694374341117466e-06,
      "loss": 1.0554,
      "step": 18664
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.912682056427002,
      "learning_rate": 8.693735424719676e-06,
      "loss": 0.8898,
      "step": 18665
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.798513650894165,
      "learning_rate": 8.693096508321887e-06,
      "loss": 0.7892,
      "step": 18666
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.567828893661499,
      "learning_rate": 8.692457591924097e-06,
      "loss": 0.9075,
      "step": 18667
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2854654788970947,
      "learning_rate": 8.691818675526309e-06,
      "loss": 0.9013,
      "step": 18668
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7532669305801392,
      "learning_rate": 8.691179759128518e-06,
      "loss": 0.8264,
      "step": 18669
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1637232303619385,
      "learning_rate": 8.69054084273073e-06,
      "loss": 0.8829,
      "step": 18670
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5926790237426758,
      "learning_rate": 8.68990192633294e-06,
      "loss": 0.8081,
      "step": 18671
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0290772914886475,
      "learning_rate": 8.68926300993515e-06,
      "loss": 0.9192,
      "step": 18672
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0249857902526855,
      "learning_rate": 8.688624093537361e-06,
      "loss": 0.983,
      "step": 18673
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.851194143295288,
      "learning_rate": 8.687985177139571e-06,
      "loss": 0.9931,
      "step": 18674
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6008061170578003,
      "learning_rate": 8.687346260741783e-06,
      "loss": 0.8433,
      "step": 18675
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.3538620471954346,
      "learning_rate": 8.686707344343993e-06,
      "loss": 0.8462,
      "step": 18676
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.018860340118408,
      "learning_rate": 8.686068427946204e-06,
      "loss": 1.2109,
      "step": 18677
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9184621572494507,
      "learning_rate": 8.685429511548414e-06,
      "loss": 0.9146,
      "step": 18678
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.122901439666748,
      "learning_rate": 8.684790595150625e-06,
      "loss": 0.8837,
      "step": 18679
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.6755170822143555,
      "learning_rate": 8.684151678752835e-06,
      "loss": 1.0518,
      "step": 18680
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7448594570159912,
      "learning_rate": 8.683512762355047e-06,
      "loss": 0.9498,
      "step": 18681
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.962120771408081,
      "learning_rate": 8.682873845957257e-06,
      "loss": 1.0049,
      "step": 18682
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7039438486099243,
      "learning_rate": 8.682234929559468e-06,
      "loss": 0.8443,
      "step": 18683
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.112314224243164,
      "learning_rate": 8.681596013161678e-06,
      "loss": 0.7726,
      "step": 18684
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.651625156402588,
      "learning_rate": 8.68095709676389e-06,
      "loss": 0.7349,
      "step": 18685
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0842416286468506,
      "learning_rate": 8.6803181803661e-06,
      "loss": 0.8092,
      "step": 18686
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.032454013824463,
      "learning_rate": 8.679679263968311e-06,
      "loss": 1.2011,
      "step": 18687
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2910566329956055,
      "learning_rate": 8.679040347570521e-06,
      "loss": 0.9278,
      "step": 18688
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6820518970489502,
      "learning_rate": 8.67840143117273e-06,
      "loss": 0.8749,
      "step": 18689
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.237142562866211,
      "learning_rate": 8.677762514774942e-06,
      "loss": 1.1299,
      "step": 18690
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7907298803329468,
      "learning_rate": 8.677123598377152e-06,
      "loss": 0.91,
      "step": 18691
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.699122905731201,
      "learning_rate": 8.676484681979364e-06,
      "loss": 1.2353,
      "step": 18692
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.564439296722412,
      "learning_rate": 8.675845765581574e-06,
      "loss": 0.4165,
      "step": 18693
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0149571895599365,
      "learning_rate": 8.675206849183785e-06,
      "loss": 0.7549,
      "step": 18694
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3486039638519287,
      "learning_rate": 8.674567932785995e-06,
      "loss": 0.751,
      "step": 18695
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.340360164642334,
      "learning_rate": 8.673929016388207e-06,
      "loss": 0.9262,
      "step": 18696
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.715092420578003,
      "learning_rate": 8.673290099990416e-06,
      "loss": 0.8197,
      "step": 18697
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.385736346244812,
      "learning_rate": 8.672651183592628e-06,
      "loss": 0.7636,
      "step": 18698
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6356501579284668,
      "learning_rate": 8.672012267194838e-06,
      "loss": 0.9171,
      "step": 18699
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3497934341430664,
      "learning_rate": 8.67137335079705e-06,
      "loss": 0.6609,
      "step": 18700
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.365375518798828,
      "learning_rate": 8.67073443439926e-06,
      "loss": 0.8035,
      "step": 18701
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7623642683029175,
      "learning_rate": 8.67009551800147e-06,
      "loss": 0.8755,
      "step": 18702
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8846937417984009,
      "learning_rate": 8.66945660160368e-06,
      "loss": 0.9136,
      "step": 18703
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4850302934646606,
      "learning_rate": 8.668817685205892e-06,
      "loss": 0.8565,
      "step": 18704
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1502132415771484,
      "learning_rate": 8.668178768808102e-06,
      "loss": 0.9349,
      "step": 18705
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.138010501861572,
      "learning_rate": 8.667539852410314e-06,
      "loss": 1.194,
      "step": 18706
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5675265789031982,
      "learning_rate": 8.666900936012523e-06,
      "loss": 0.905,
      "step": 18707
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.141569137573242,
      "learning_rate": 8.666262019614733e-06,
      "loss": 0.866,
      "step": 18708
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5403422117233276,
      "learning_rate": 8.665623103216945e-06,
      "loss": 0.863,
      "step": 18709
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.13222336769104,
      "learning_rate": 8.664984186819155e-06,
      "loss": 0.9042,
      "step": 18710
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.811505913734436,
      "learning_rate": 8.664345270421366e-06,
      "loss": 0.9575,
      "step": 18711
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5389513969421387,
      "learning_rate": 8.663706354023576e-06,
      "loss": 0.905,
      "step": 18712
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.74803626537323,
      "learning_rate": 8.663067437625788e-06,
      "loss": 0.8546,
      "step": 18713
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3859152793884277,
      "learning_rate": 8.662428521227997e-06,
      "loss": 1.0392,
      "step": 18714
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1353166103363037,
      "learning_rate": 8.661789604830209e-06,
      "loss": 0.964,
      "step": 18715
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.369649648666382,
      "learning_rate": 8.661150688432419e-06,
      "loss": 0.8387,
      "step": 18716
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1123366355895996,
      "learning_rate": 8.66051177203463e-06,
      "loss": 1.3183,
      "step": 18717
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.132488250732422,
      "learning_rate": 8.65987285563684e-06,
      "loss": 0.8199,
      "step": 18718
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6763099431991577,
      "learning_rate": 8.659233939239052e-06,
      "loss": 0.8982,
      "step": 18719
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7517704963684082,
      "learning_rate": 8.658595022841262e-06,
      "loss": 0.9518,
      "step": 18720
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0813655853271484,
      "learning_rate": 8.657956106443473e-06,
      "loss": 0.8957,
      "step": 18721
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.89693284034729,
      "learning_rate": 8.657317190045683e-06,
      "loss": 0.903,
      "step": 18722
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7560843229293823,
      "learning_rate": 8.656678273647895e-06,
      "loss": 0.9128,
      "step": 18723
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.4829862117767334,
      "learning_rate": 8.656039357250104e-06,
      "loss": 1.0224,
      "step": 18724
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6186565160751343,
      "learning_rate": 8.655400440852316e-06,
      "loss": 0.7585,
      "step": 18725
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0882985591888428,
      "learning_rate": 8.654761524454526e-06,
      "loss": 1.173,
      "step": 18726
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4961647987365723,
      "learning_rate": 8.654122608056736e-06,
      "loss": 0.7663,
      "step": 18727
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5260694026947021,
      "learning_rate": 8.653483691658947e-06,
      "loss": 0.8274,
      "step": 18728
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6782270669937134,
      "learning_rate": 8.652844775261157e-06,
      "loss": 0.9077,
      "step": 18729
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0463476181030273,
      "learning_rate": 8.652205858863369e-06,
      "loss": 0.8895,
      "step": 18730
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.70151686668396,
      "learning_rate": 8.651566942465578e-06,
      "loss": 0.4554,
      "step": 18731
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.615720510482788,
      "learning_rate": 8.65092802606779e-06,
      "loss": 1.1131,
      "step": 18732
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9959388971328735,
      "learning_rate": 8.65028910967e-06,
      "loss": 0.845,
      "step": 18733
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.212484121322632,
      "learning_rate": 8.649650193272211e-06,
      "loss": 0.9576,
      "step": 18734
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9352550506591797,
      "learning_rate": 8.649011276874421e-06,
      "loss": 0.8719,
      "step": 18735
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.817772626876831,
      "learning_rate": 8.648372360476633e-06,
      "loss": 1.1585,
      "step": 18736
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7410924434661865,
      "learning_rate": 8.647733444078843e-06,
      "loss": 0.9786,
      "step": 18737
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0731379985809326,
      "learning_rate": 8.647094527681054e-06,
      "loss": 0.9392,
      "step": 18738
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.391534328460693,
      "learning_rate": 8.646455611283264e-06,
      "loss": 1.6347,
      "step": 18739
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.7693660259246826,
      "learning_rate": 8.645816694885476e-06,
      "loss": 0.8926,
      "step": 18740
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6153088808059692,
      "learning_rate": 8.645177778487685e-06,
      "loss": 0.8406,
      "step": 18741
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9098222255706787,
      "learning_rate": 8.644538862089897e-06,
      "loss": 0.9887,
      "step": 18742
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5630404949188232,
      "learning_rate": 8.643899945692107e-06,
      "loss": 0.9181,
      "step": 18743
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.727937698364258,
      "learning_rate": 8.643261029294317e-06,
      "loss": 1.4088,
      "step": 18744
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2934765815734863,
      "learning_rate": 8.642622112896528e-06,
      "loss": 1.316,
      "step": 18745
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4329205751419067,
      "learning_rate": 8.641983196498738e-06,
      "loss": 0.8135,
      "step": 18746
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8758318424224854,
      "learning_rate": 8.64134428010095e-06,
      "loss": 1.0494,
      "step": 18747
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7627909183502197,
      "learning_rate": 8.64070536370316e-06,
      "loss": 0.83,
      "step": 18748
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7718510627746582,
      "learning_rate": 8.640066447305371e-06,
      "loss": 0.9963,
      "step": 18749
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.9991390705108643,
      "learning_rate": 8.639427530907581e-06,
      "loss": 1.0473,
      "step": 18750
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.047288656234741,
      "learning_rate": 8.638788614509792e-06,
      "loss": 1.001,
      "step": 18751
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.739686012268066,
      "learning_rate": 8.638149698112002e-06,
      "loss": 1.3335,
      "step": 18752
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.61871337890625,
      "learning_rate": 8.637510781714214e-06,
      "loss": 0.8567,
      "step": 18753
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9510037899017334,
      "learning_rate": 8.636871865316424e-06,
      "loss": 0.8837,
      "step": 18754
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.657723307609558,
      "learning_rate": 8.636232948918635e-06,
      "loss": 0.9113,
      "step": 18755
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5550005435943604,
      "learning_rate": 8.635594032520845e-06,
      "loss": 0.8905,
      "step": 18756
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0365216732025146,
      "learning_rate": 8.634955116123057e-06,
      "loss": 0.8405,
      "step": 18757
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5825715065002441,
      "learning_rate": 8.634316199725267e-06,
      "loss": 0.8212,
      "step": 18758
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1697583198547363,
      "learning_rate": 8.633677283327478e-06,
      "loss": 0.7105,
      "step": 18759
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.111069679260254,
      "learning_rate": 8.633038366929688e-06,
      "loss": 0.9671,
      "step": 18760
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.828514575958252,
      "learning_rate": 8.6323994505319e-06,
      "loss": 1.4109,
      "step": 18761
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8518037796020508,
      "learning_rate": 8.63176053413411e-06,
      "loss": 1.0267,
      "step": 18762
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.4895544052124023,
      "learning_rate": 8.63112161773632e-06,
      "loss": 1.2092,
      "step": 18763
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.3249616622924805,
      "learning_rate": 8.63048270133853e-06,
      "loss": 1.0815,
      "step": 18764
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.936242699623108,
      "learning_rate": 8.62984378494074e-06,
      "loss": 0.9025,
      "step": 18765
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.0642950534820557,
      "learning_rate": 8.629204868542952e-06,
      "loss": 0.8616,
      "step": 18766
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.420045375823975,
      "learning_rate": 8.628565952145162e-06,
      "loss": 1.0167,
      "step": 18767
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9304155111312866,
      "learning_rate": 8.627927035747374e-06,
      "loss": 0.8544,
      "step": 18768
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.631280779838562,
      "learning_rate": 8.627288119349583e-06,
      "loss": 0.7935,
      "step": 18769
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8217413425445557,
      "learning_rate": 8.626649202951795e-06,
      "loss": 1.1716,
      "step": 18770
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8081656694412231,
      "learning_rate": 8.626010286554005e-06,
      "loss": 0.952,
      "step": 18771
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.780797004699707,
      "learning_rate": 8.625371370156216e-06,
      "loss": 0.9584,
      "step": 18772
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3705036640167236,
      "learning_rate": 8.624732453758426e-06,
      "loss": 0.6545,
      "step": 18773
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8176954984664917,
      "learning_rate": 8.624093537360638e-06,
      "loss": 0.7431,
      "step": 18774
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.789939522743225,
      "learning_rate": 8.623454620962848e-06,
      "loss": 0.8323,
      "step": 18775
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.610394835472107,
      "learning_rate": 8.622815704565059e-06,
      "loss": 0.8717,
      "step": 18776
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7622346878051758,
      "learning_rate": 8.622176788167269e-06,
      "loss": 0.8966,
      "step": 18777
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.375939130783081,
      "learning_rate": 8.62153787176948e-06,
      "loss": 0.8679,
      "step": 18778
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0534934997558594,
      "learning_rate": 8.62089895537169e-06,
      "loss": 0.9284,
      "step": 18779
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.390418529510498,
      "learning_rate": 8.6202600389739e-06,
      "loss": 0.949,
      "step": 18780
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.7406206130981445,
      "learning_rate": 8.619621122576112e-06,
      "loss": 0.8332,
      "step": 18781
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.255859375,
      "learning_rate": 8.618982206178322e-06,
      "loss": 0.8551,
      "step": 18782
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4250367879867554,
      "learning_rate": 8.618343289780533e-06,
      "loss": 0.9323,
      "step": 18783
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.723863959312439,
      "learning_rate": 8.617704373382743e-06,
      "loss": 0.8301,
      "step": 18784
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9193047285079956,
      "learning_rate": 8.617065456984955e-06,
      "loss": 1.1219,
      "step": 18785
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.4218556880950928,
      "learning_rate": 8.616426540587164e-06,
      "loss": 0.9205,
      "step": 18786
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.119407892227173,
      "learning_rate": 8.615787624189376e-06,
      "loss": 0.9835,
      "step": 18787
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.016003131866455,
      "learning_rate": 8.615148707791586e-06,
      "loss": 1.3494,
      "step": 18788
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9688619375228882,
      "learning_rate": 8.614509791393797e-06,
      "loss": 0.7472,
      "step": 18789
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5339668989181519,
      "learning_rate": 8.613870874996007e-06,
      "loss": 0.8119,
      "step": 18790
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.797393560409546,
      "learning_rate": 8.613231958598219e-06,
      "loss": 0.595,
      "step": 18791
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7277532815933228,
      "learning_rate": 8.612593042200429e-06,
      "loss": 0.9463,
      "step": 18792
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9671199321746826,
      "learning_rate": 8.61195412580264e-06,
      "loss": 0.6597,
      "step": 18793
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6244702339172363,
      "learning_rate": 8.61131520940485e-06,
      "loss": 0.8904,
      "step": 18794
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.028740167617798,
      "learning_rate": 8.610676293007062e-06,
      "loss": 0.8665,
      "step": 18795
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.795646071434021,
      "learning_rate": 8.610037376609271e-06,
      "loss": 0.8549,
      "step": 18796
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0518648624420166,
      "learning_rate": 8.609398460211483e-06,
      "loss": 0.9268,
      "step": 18797
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.488936185836792,
      "learning_rate": 8.608759543813693e-06,
      "loss": 0.4685,
      "step": 18798
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.0836052894592285,
      "learning_rate": 8.608120627415903e-06,
      "loss": 1.1489,
      "step": 18799
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.49116849899292,
      "learning_rate": 8.607481711018114e-06,
      "loss": 0.9199,
      "step": 18800
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5014152526855469,
      "learning_rate": 8.606842794620324e-06,
      "loss": 0.801,
      "step": 18801
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.890167474746704,
      "learning_rate": 8.606203878222536e-06,
      "loss": 0.7442,
      "step": 18802
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7447328567504883,
      "learning_rate": 8.605564961824745e-06,
      "loss": 0.8132,
      "step": 18803
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.973684549331665,
      "learning_rate": 8.604926045426957e-06,
      "loss": 0.7105,
      "step": 18804
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6654834747314453,
      "learning_rate": 8.604287129029167e-06,
      "loss": 0.5115,
      "step": 18805
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.7457525730133057,
      "learning_rate": 8.603648212631378e-06,
      "loss": 0.7471,
      "step": 18806
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.746609091758728,
      "learning_rate": 8.603009296233588e-06,
      "loss": 0.862,
      "step": 18807
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.9441604614257812,
      "learning_rate": 8.6023703798358e-06,
      "loss": 0.8669,
      "step": 18808
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.895009994506836,
      "learning_rate": 8.60173146343801e-06,
      "loss": 0.9675,
      "step": 18809
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4772617816925049,
      "learning_rate": 8.601092547040221e-06,
      "loss": 0.5011,
      "step": 18810
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8276740312576294,
      "learning_rate": 8.600453630642431e-06,
      "loss": 1.1036,
      "step": 18811
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.294084072113037,
      "learning_rate": 8.599814714244643e-06,
      "loss": 0.8013,
      "step": 18812
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.0946478843688965,
      "learning_rate": 8.599175797846852e-06,
      "loss": 1.3303,
      "step": 18813
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.044581890106201,
      "learning_rate": 8.598536881449064e-06,
      "loss": 0.7082,
      "step": 18814
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.558234691619873,
      "learning_rate": 8.597897965051274e-06,
      "loss": 0.8346,
      "step": 18815
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.425905227661133,
      "learning_rate": 8.597259048653485e-06,
      "loss": 0.9144,
      "step": 18816
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1895077228546143,
      "learning_rate": 8.596620132255695e-06,
      "loss": 0.9276,
      "step": 18817
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6973251104354858,
      "learning_rate": 8.595981215857905e-06,
      "loss": 0.7437,
      "step": 18818
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.786177158355713,
      "learning_rate": 8.595342299460117e-06,
      "loss": 0.791,
      "step": 18819
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.838955879211426,
      "learning_rate": 8.594703383062327e-06,
      "loss": 1.175,
      "step": 18820
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.11767578125,
      "learning_rate": 8.594064466664538e-06,
      "loss": 0.8938,
      "step": 18821
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.845677137374878,
      "learning_rate": 8.593425550266748e-06,
      "loss": 0.8921,
      "step": 18822
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.09102463722229,
      "learning_rate": 8.59278663386896e-06,
      "loss": 0.9299,
      "step": 18823
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.352962017059326,
      "learning_rate": 8.59214771747117e-06,
      "loss": 1.1193,
      "step": 18824
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2722408771514893,
      "learning_rate": 8.591508801073381e-06,
      "loss": 0.7616,
      "step": 18825
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.650905728340149,
      "learning_rate": 8.59086988467559e-06,
      "loss": 0.8889,
      "step": 18826
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5596367120742798,
      "learning_rate": 8.590230968277802e-06,
      "loss": 0.9695,
      "step": 18827
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9728527069091797,
      "learning_rate": 8.589592051880012e-06,
      "loss": 0.8531,
      "step": 18828
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.078733444213867,
      "learning_rate": 8.588953135482224e-06,
      "loss": 1.2612,
      "step": 18829
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.084582567214966,
      "learning_rate": 8.588314219084434e-06,
      "loss": 0.922,
      "step": 18830
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6735334396362305,
      "learning_rate": 8.587675302686645e-06,
      "loss": 0.8301,
      "step": 18831
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1479899883270264,
      "learning_rate": 8.587036386288855e-06,
      "loss": 0.9786,
      "step": 18832
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.629764199256897,
      "learning_rate": 8.586397469891066e-06,
      "loss": 0.9325,
      "step": 18833
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6621979475021362,
      "learning_rate": 8.585758553493276e-06,
      "loss": 0.906,
      "step": 18834
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4876072406768799,
      "learning_rate": 8.585119637095486e-06,
      "loss": 0.8603,
      "step": 18835
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.254340648651123,
      "learning_rate": 8.584480720697698e-06,
      "loss": 0.9154,
      "step": 18836
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2191543579101562,
      "learning_rate": 8.583841804299908e-06,
      "loss": 0.8429,
      "step": 18837
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4943691492080688,
      "learning_rate": 8.583202887902119e-06,
      "loss": 0.7506,
      "step": 18838
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.943393349647522,
      "learning_rate": 8.582563971504329e-06,
      "loss": 0.8395,
      "step": 18839
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7589648962020874,
      "learning_rate": 8.58192505510654e-06,
      "loss": 0.9562,
      "step": 18840
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5820614099502563,
      "learning_rate": 8.58128613870875e-06,
      "loss": 0.8569,
      "step": 18841
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5632799863815308,
      "learning_rate": 8.580647222310962e-06,
      "loss": 0.8307,
      "step": 18842
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.467301607131958,
      "learning_rate": 8.580008305913172e-06,
      "loss": 0.4657,
      "step": 18843
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4188637733459473,
      "learning_rate": 8.579369389515383e-06,
      "loss": 0.8738,
      "step": 18844
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.032668113708496,
      "learning_rate": 8.578730473117593e-06,
      "loss": 0.8391,
      "step": 18845
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.155702114105225,
      "learning_rate": 8.578091556719805e-06,
      "loss": 1.2875,
      "step": 18846
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.490312099456787,
      "learning_rate": 8.577452640322015e-06,
      "loss": 1.2609,
      "step": 18847
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6325719356536865,
      "learning_rate": 8.576813723924226e-06,
      "loss": 0.4512,
      "step": 18848
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.528924942016602,
      "learning_rate": 8.576174807526436e-06,
      "loss": 1.5446,
      "step": 18849
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4839272499084473,
      "learning_rate": 8.575535891128648e-06,
      "loss": 0.8294,
      "step": 18850
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7618199586868286,
      "learning_rate": 8.574896974730857e-06,
      "loss": 0.8986,
      "step": 18851
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0007121562957764,
      "learning_rate": 8.574258058333069e-06,
      "loss": 0.9626,
      "step": 18852
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0309197902679443,
      "learning_rate": 8.573619141935279e-06,
      "loss": 0.6824,
      "step": 18853
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1289448738098145,
      "learning_rate": 8.572980225537489e-06,
      "loss": 0.8967,
      "step": 18854
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.712249517440796,
      "learning_rate": 8.5723413091397e-06,
      "loss": 0.9008,
      "step": 18855
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.640092372894287,
      "learning_rate": 8.57170239274191e-06,
      "loss": 1.3491,
      "step": 18856
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5166220664978027,
      "learning_rate": 8.571063476344122e-06,
      "loss": 0.9143,
      "step": 18857
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3124909400939941,
      "learning_rate": 8.570424559946331e-06,
      "loss": 0.3606,
      "step": 18858
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.547492027282715,
      "learning_rate": 8.569785643548543e-06,
      "loss": 0.8619,
      "step": 18859
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2296910285949707,
      "learning_rate": 8.569146727150753e-06,
      "loss": 1.3503,
      "step": 18860
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.514756917953491,
      "learning_rate": 8.568507810752964e-06,
      "loss": 1.1437,
      "step": 18861
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.717392921447754,
      "learning_rate": 8.567868894355174e-06,
      "loss": 0.9172,
      "step": 18862
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.562463641166687,
      "learning_rate": 8.567229977957386e-06,
      "loss": 0.8716,
      "step": 18863
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2108367681503296,
      "learning_rate": 8.566591061559596e-06,
      "loss": 0.7735,
      "step": 18864
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.513638496398926,
      "learning_rate": 8.565952145161807e-06,
      "loss": 1.2441,
      "step": 18865
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.049053907394409,
      "learning_rate": 8.565313228764017e-06,
      "loss": 0.8747,
      "step": 18866
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9305617809295654,
      "learning_rate": 8.564674312366229e-06,
      "loss": 0.838,
      "step": 18867
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7920297384262085,
      "learning_rate": 8.564035395968438e-06,
      "loss": 0.7439,
      "step": 18868
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.136460781097412,
      "learning_rate": 8.56339647957065e-06,
      "loss": 0.8524,
      "step": 18869
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0911898612976074,
      "learning_rate": 8.56275756317286e-06,
      "loss": 1.1927,
      "step": 18870
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.8204147815704346,
      "learning_rate": 8.562118646775071e-06,
      "loss": 1.567,
      "step": 18871
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.635031819343567,
      "learning_rate": 8.561479730377281e-06,
      "loss": 0.9387,
      "step": 18872
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.8405773639678955,
      "learning_rate": 8.560840813979491e-06,
      "loss": 0.683,
      "step": 18873
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.1626031398773193,
      "learning_rate": 8.560201897581703e-06,
      "loss": 1.199,
      "step": 18874
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6937707662582397,
      "learning_rate": 8.559562981183912e-06,
      "loss": 0.8624,
      "step": 18875
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.4495301246643066,
      "learning_rate": 8.558924064786124e-06,
      "loss": 0.6682,
      "step": 18876
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1307461261749268,
      "learning_rate": 8.558285148388334e-06,
      "loss": 1.0544,
      "step": 18877
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.640899896621704,
      "learning_rate": 8.557646231990545e-06,
      "loss": 0.8831,
      "step": 18878
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.3917813301086426,
      "learning_rate": 8.557007315592755e-06,
      "loss": 0.8787,
      "step": 18879
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.745347261428833,
      "learning_rate": 8.556368399194967e-06,
      "loss": 0.8634,
      "step": 18880
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1283915042877197,
      "learning_rate": 8.555729482797177e-06,
      "loss": 0.9546,
      "step": 18881
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8220263719558716,
      "learning_rate": 8.555090566399388e-06,
      "loss": 0.87,
      "step": 18882
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0986318588256836,
      "learning_rate": 8.554451650001598e-06,
      "loss": 1.0278,
      "step": 18883
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3750014305114746,
      "learning_rate": 8.55381273360381e-06,
      "loss": 0.9097,
      "step": 18884
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0478131771087646,
      "learning_rate": 8.55317381720602e-06,
      "loss": 0.9848,
      "step": 18885
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.5291731357574463,
      "learning_rate": 8.552534900808231e-06,
      "loss": 0.8839,
      "step": 18886
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.6514625549316406,
      "learning_rate": 8.551895984410441e-06,
      "loss": 1.117,
      "step": 18887
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9328571557998657,
      "learning_rate": 8.551257068012652e-06,
      "loss": 0.8548,
      "step": 18888
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.86139178276062,
      "learning_rate": 8.550618151614862e-06,
      "loss": 1.2057,
      "step": 18889
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2707117795944214,
      "learning_rate": 8.549979235217072e-06,
      "loss": 0.4588,
      "step": 18890
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.82869553565979,
      "learning_rate": 8.549340318819284e-06,
      "loss": 0.794,
      "step": 18891
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.045438051223755,
      "learning_rate": 8.548701402421494e-06,
      "loss": 0.9674,
      "step": 18892
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4427632093429565,
      "learning_rate": 8.548062486023705e-06,
      "loss": 0.8291,
      "step": 18893
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8366221189498901,
      "learning_rate": 8.547423569625915e-06,
      "loss": 0.8718,
      "step": 18894
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2800824642181396,
      "learning_rate": 8.546784653228126e-06,
      "loss": 0.9306,
      "step": 18895
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.876064658164978,
      "learning_rate": 8.546145736830336e-06,
      "loss": 1.0099,
      "step": 18896
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.171125888824463,
      "learning_rate": 8.545506820432548e-06,
      "loss": 0.9668,
      "step": 18897
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9660478830337524,
      "learning_rate": 8.544867904034758e-06,
      "loss": 0.8227,
      "step": 18898
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.860801100730896,
      "learning_rate": 8.54422898763697e-06,
      "loss": 0.9049,
      "step": 18899
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9102460145950317,
      "learning_rate": 8.543590071239179e-06,
      "loss": 0.8298,
      "step": 18900
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.4196059703826904,
      "learning_rate": 8.54295115484139e-06,
      "loss": 1.0647,
      "step": 18901
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6260085105895996,
      "learning_rate": 8.5423122384436e-06,
      "loss": 0.465,
      "step": 18902
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.519430160522461,
      "learning_rate": 8.541673322045812e-06,
      "loss": 0.898,
      "step": 18903
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9034169912338257,
      "learning_rate": 8.541034405648022e-06,
      "loss": 0.9669,
      "step": 18904
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2527239322662354,
      "learning_rate": 8.540395489250233e-06,
      "loss": 0.881,
      "step": 18905
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6830289363861084,
      "learning_rate": 8.539756572852443e-06,
      "loss": 0.9037,
      "step": 18906
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5357714891433716,
      "learning_rate": 8.539117656454655e-06,
      "loss": 0.8377,
      "step": 18907
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.98655104637146,
      "learning_rate": 8.538478740056865e-06,
      "loss": 1.0116,
      "step": 18908
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7747970819473267,
      "learning_rate": 8.537839823659075e-06,
      "loss": 0.833,
      "step": 18909
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.365612268447876,
      "learning_rate": 8.537200907261286e-06,
      "loss": 0.9394,
      "step": 18910
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5722044706344604,
      "learning_rate": 8.536561990863496e-06,
      "loss": 0.8725,
      "step": 18911
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.6886141300201416,
      "learning_rate": 8.535923074465708e-06,
      "loss": 0.9732,
      "step": 18912
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.848000168800354,
      "learning_rate": 8.535284158067917e-06,
      "loss": 0.8382,
      "step": 18913
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8034745454788208,
      "learning_rate": 8.534645241670129e-06,
      "loss": 0.8097,
      "step": 18914
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.627755880355835,
      "learning_rate": 8.534006325272339e-06,
      "loss": 0.8981,
      "step": 18915
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3443520069122314,
      "learning_rate": 8.53336740887455e-06,
      "loss": 0.7434,
      "step": 18916
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.923073410987854,
      "learning_rate": 8.53272849247676e-06,
      "loss": 0.8419,
      "step": 18917
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8502953052520752,
      "learning_rate": 8.532089576078972e-06,
      "loss": 0.6907,
      "step": 18918
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.766392469406128,
      "learning_rate": 8.531450659681182e-06,
      "loss": 0.6099,
      "step": 18919
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.77120041847229,
      "learning_rate": 8.530811743283393e-06,
      "loss": 0.7779,
      "step": 18920
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2563281059265137,
      "learning_rate": 8.530172826885603e-06,
      "loss": 0.7231,
      "step": 18921
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9088371992111206,
      "learning_rate": 8.529533910487815e-06,
      "loss": 0.891,
      "step": 18922
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2706917524337769,
      "learning_rate": 8.528894994090024e-06,
      "loss": 0.4469,
      "step": 18923
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0737650394439697,
      "learning_rate": 8.528256077692236e-06,
      "loss": 0.8807,
      "step": 18924
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7467539310455322,
      "learning_rate": 8.527617161294446e-06,
      "loss": 0.7672,
      "step": 18925
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.331881523132324,
      "learning_rate": 8.526978244896656e-06,
      "loss": 0.923,
      "step": 18926
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5674679279327393,
      "learning_rate": 8.526339328498867e-06,
      "loss": 0.8425,
      "step": 18927
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.002704620361328,
      "learning_rate": 8.525700412101077e-06,
      "loss": 0.9823,
      "step": 18928
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.514014482498169,
      "learning_rate": 8.525061495703289e-06,
      "loss": 0.3806,
      "step": 18929
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3539628982543945,
      "learning_rate": 8.524422579305498e-06,
      "loss": 1.0955,
      "step": 18930
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.5318193435668945,
      "learning_rate": 8.52378366290771e-06,
      "loss": 1.1948,
      "step": 18931
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9679080247879028,
      "learning_rate": 8.52314474650992e-06,
      "loss": 0.8688,
      "step": 18932
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7805415391921997,
      "learning_rate": 8.522505830112131e-06,
      "loss": 0.887,
      "step": 18933
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9870027303695679,
      "learning_rate": 8.521866913714341e-06,
      "loss": 0.9745,
      "step": 18934
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.048513650894165,
      "learning_rate": 8.521227997316553e-06,
      "loss": 0.848,
      "step": 18935
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.701972246170044,
      "learning_rate": 8.520589080918763e-06,
      "loss": 0.9195,
      "step": 18936
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5895211696624756,
      "learning_rate": 8.519950164520974e-06,
      "loss": 0.5074,
      "step": 18937
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0846035480499268,
      "learning_rate": 8.519311248123184e-06,
      "loss": 0.7966,
      "step": 18938
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8204835653305054,
      "learning_rate": 8.518672331725396e-06,
      "loss": 0.8391,
      "step": 18939
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.4135491847991943,
      "learning_rate": 8.518033415327604e-06,
      "loss": 0.7587,
      "step": 18940
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6573972702026367,
      "learning_rate": 8.517394498929815e-06,
      "loss": 0.8318,
      "step": 18941
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0399866104125977,
      "learning_rate": 8.516755582532025e-06,
      "loss": 0.9359,
      "step": 18942
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1479265689849854,
      "learning_rate": 8.516116666134237e-06,
      "loss": 0.9013,
      "step": 18943
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9384921789169312,
      "learning_rate": 8.515477749736447e-06,
      "loss": 0.7774,
      "step": 18944
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9127118587493896,
      "learning_rate": 8.514838833338658e-06,
      "loss": 1.3156,
      "step": 18945
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5231417417526245,
      "learning_rate": 8.514199916940868e-06,
      "loss": 0.8086,
      "step": 18946
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2400403022766113,
      "learning_rate": 8.51356100054308e-06,
      "loss": 0.4177,
      "step": 18947
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.136021137237549,
      "learning_rate": 8.51292208414529e-06,
      "loss": 1.4186,
      "step": 18948
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.432266116142273,
      "learning_rate": 8.512283167747501e-06,
      "loss": 0.8629,
      "step": 18949
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7946102619171143,
      "learning_rate": 8.51164425134971e-06,
      "loss": 0.9601,
      "step": 18950
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4805686473846436,
      "learning_rate": 8.511005334951922e-06,
      "loss": 0.4599,
      "step": 18951
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7436306476593018,
      "learning_rate": 8.510366418554132e-06,
      "loss": 0.7661,
      "step": 18952
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3314118385314941,
      "learning_rate": 8.509727502156344e-06,
      "loss": 0.5428,
      "step": 18953
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.819216012954712,
      "learning_rate": 8.509088585758554e-06,
      "loss": 0.8557,
      "step": 18954
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7855098247528076,
      "learning_rate": 8.508449669360765e-06,
      "loss": 0.8901,
      "step": 18955
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.5350897312164307,
      "learning_rate": 8.507810752962975e-06,
      "loss": 0.8704,
      "step": 18956
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7226002216339111,
      "learning_rate": 8.507171836565185e-06,
      "loss": 0.792,
      "step": 18957
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.479522466659546,
      "learning_rate": 8.506532920167396e-06,
      "loss": 0.8723,
      "step": 18958
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.757758378982544,
      "learning_rate": 8.505894003769606e-06,
      "loss": 0.8716,
      "step": 18959
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.700745701789856,
      "learning_rate": 8.505255087371818e-06,
      "loss": 0.9007,
      "step": 18960
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.235074996948242,
      "learning_rate": 8.504616170974028e-06,
      "loss": 1.1817,
      "step": 18961
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.833292007446289,
      "learning_rate": 8.503977254576239e-06,
      "loss": 0.853,
      "step": 18962
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9296152591705322,
      "learning_rate": 8.503338338178449e-06,
      "loss": 0.8905,
      "step": 18963
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.674617886543274,
      "learning_rate": 8.50269942178066e-06,
      "loss": 0.8439,
      "step": 18964
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6615041494369507,
      "learning_rate": 8.50206050538287e-06,
      "loss": 0.982,
      "step": 18965
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8575630187988281,
      "learning_rate": 8.501421588985082e-06,
      "loss": 0.9126,
      "step": 18966
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5473045110702515,
      "learning_rate": 8.500782672587292e-06,
      "loss": 0.8176,
      "step": 18967
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.117190361022949,
      "learning_rate": 8.500143756189503e-06,
      "loss": 0.824,
      "step": 18968
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.6529111862182617,
      "learning_rate": 8.499504839791713e-06,
      "loss": 0.904,
      "step": 18969
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5001412630081177,
      "learning_rate": 8.498865923393925e-06,
      "loss": 0.8423,
      "step": 18970
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4306708574295044,
      "learning_rate": 8.498227006996135e-06,
      "loss": 0.645,
      "step": 18971
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.677852153778076,
      "learning_rate": 8.497588090598346e-06,
      "loss": 0.9371,
      "step": 18972
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9602787494659424,
      "learning_rate": 8.496949174200556e-06,
      "loss": 0.9697,
      "step": 18973
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9079846143722534,
      "learning_rate": 8.496310257802768e-06,
      "loss": 0.7945,
      "step": 18974
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7722889184951782,
      "learning_rate": 8.495671341404977e-06,
      "loss": 0.9151,
      "step": 18975
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9010449647903442,
      "learning_rate": 8.495032425007187e-06,
      "loss": 0.8571,
      "step": 18976
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0757896900177,
      "learning_rate": 8.494393508609399e-06,
      "loss": 0.8261,
      "step": 18977
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8042442798614502,
      "learning_rate": 8.493754592211609e-06,
      "loss": 0.8976,
      "step": 18978
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.642902851104736,
      "learning_rate": 8.49311567581382e-06,
      "loss": 1.5366,
      "step": 18979
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0317907333374023,
      "learning_rate": 8.49247675941603e-06,
      "loss": 0.7003,
      "step": 18980
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9683539867401123,
      "learning_rate": 8.491837843018242e-06,
      "loss": 0.9086,
      "step": 18981
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.729654550552368,
      "learning_rate": 8.491198926620451e-06,
      "loss": 1.1473,
      "step": 18982
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.614345073699951,
      "learning_rate": 8.490560010222663e-06,
      "loss": 0.9799,
      "step": 18983
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2836343050003052,
      "learning_rate": 8.489921093824873e-06,
      "loss": 0.4266,
      "step": 18984
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7299810647964478,
      "learning_rate": 8.489282177427084e-06,
      "loss": 0.8896,
      "step": 18985
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0950496196746826,
      "learning_rate": 8.488643261029294e-06,
      "loss": 0.7995,
      "step": 18986
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.8891494274139404,
      "learning_rate": 8.488004344631506e-06,
      "loss": 0.9436,
      "step": 18987
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9755762815475464,
      "learning_rate": 8.487365428233716e-06,
      "loss": 0.9535,
      "step": 18988
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.552418231964111,
      "learning_rate": 8.486726511835927e-06,
      "loss": 0.9569,
      "step": 18989
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8171879053115845,
      "learning_rate": 8.486087595438137e-06,
      "loss": 0.8853,
      "step": 18990
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7934703826904297,
      "learning_rate": 8.485448679040349e-06,
      "loss": 0.8801,
      "step": 18991
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.504823923110962,
      "learning_rate": 8.484809762642558e-06,
      "loss": 0.8383,
      "step": 18992
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0102696418762207,
      "learning_rate": 8.48417084624477e-06,
      "loss": 1.0035,
      "step": 18993
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.404957294464111,
      "learning_rate": 8.48353192984698e-06,
      "loss": 1.2499,
      "step": 18994
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0695559978485107,
      "learning_rate": 8.48289301344919e-06,
      "loss": 0.9111,
      "step": 18995
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8505765199661255,
      "learning_rate": 8.482254097051401e-06,
      "loss": 0.8922,
      "step": 18996
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5170634984970093,
      "learning_rate": 8.481615180653611e-06,
      "loss": 0.7791,
      "step": 18997
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1387760639190674,
      "learning_rate": 8.480976264255823e-06,
      "loss": 0.9324,
      "step": 18998
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3453731536865234,
      "learning_rate": 8.480337347858033e-06,
      "loss": 1.0749,
      "step": 18999
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8020049333572388,
      "learning_rate": 8.479698431460244e-06,
      "loss": 0.8576,
      "step": 19000
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.447689175605774,
      "learning_rate": 8.479059515062454e-06,
      "loss": 0.5058,
      "step": 19001
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7965900897979736,
      "learning_rate": 8.478420598664665e-06,
      "loss": 0.9492,
      "step": 19002
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5697721242904663,
      "learning_rate": 8.477781682266875e-06,
      "loss": 0.7911,
      "step": 19003
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1263785362243652,
      "learning_rate": 8.477142765869087e-06,
      "loss": 0.9237,
      "step": 19004
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7517341375350952,
      "learning_rate": 8.476503849471297e-06,
      "loss": 0.899,
      "step": 19005
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.715060830116272,
      "learning_rate": 8.475864933073508e-06,
      "loss": 0.8633,
      "step": 19006
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.751930594444275,
      "learning_rate": 8.475226016675718e-06,
      "loss": 0.7399,
      "step": 19007
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5023362636566162,
      "learning_rate": 8.47458710027793e-06,
      "loss": 0.8084,
      "step": 19008
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6383458375930786,
      "learning_rate": 8.47394818388014e-06,
      "loss": 0.8725,
      "step": 19009
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6000903844833374,
      "learning_rate": 8.473309267482351e-06,
      "loss": 0.8268,
      "step": 19010
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7285914421081543,
      "learning_rate": 8.472670351084561e-06,
      "loss": 0.8684,
      "step": 19011
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9216970205307007,
      "learning_rate": 8.47203143468677e-06,
      "loss": 1.0064,
      "step": 19012
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8868001699447632,
      "learning_rate": 8.471392518288982e-06,
      "loss": 0.9647,
      "step": 19013
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.628164291381836,
      "learning_rate": 8.470753601891192e-06,
      "loss": 0.8335,
      "step": 19014
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.481949806213379,
      "learning_rate": 8.470114685493404e-06,
      "loss": 1.6491,
      "step": 19015
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.755527973175049,
      "learning_rate": 8.469475769095614e-06,
      "loss": 0.8667,
      "step": 19016
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.852756142616272,
      "learning_rate": 8.468836852697825e-06,
      "loss": 0.8804,
      "step": 19017
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.5326080322265625,
      "learning_rate": 8.468197936300035e-06,
      "loss": 0.8727,
      "step": 19018
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.994241237640381,
      "learning_rate": 8.467559019902247e-06,
      "loss": 0.9391,
      "step": 19019
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9402464628219604,
      "learning_rate": 8.466920103504456e-06,
      "loss": 0.7424,
      "step": 19020
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2822210788726807,
      "learning_rate": 8.466281187106668e-06,
      "loss": 1.1547,
      "step": 19021
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8861894607543945,
      "learning_rate": 8.465642270708878e-06,
      "loss": 0.8629,
      "step": 19022
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6001980304718018,
      "learning_rate": 8.46500335431109e-06,
      "loss": 0.9673,
      "step": 19023
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8280773162841797,
      "learning_rate": 8.4643644379133e-06,
      "loss": 1.2179,
      "step": 19024
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5803078413009644,
      "learning_rate": 8.46372552151551e-06,
      "loss": 0.8042,
      "step": 19025
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.4148545265197754,
      "learning_rate": 8.46308660511772e-06,
      "loss": 1.186,
      "step": 19026
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9695284366607666,
      "learning_rate": 8.462447688719932e-06,
      "loss": 0.747,
      "step": 19027
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3378689289093018,
      "learning_rate": 8.461808772322142e-06,
      "loss": 0.5049,
      "step": 19028
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.50322151184082,
      "learning_rate": 8.461169855924354e-06,
      "loss": 1.2417,
      "step": 19029
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7922230958938599,
      "learning_rate": 8.460530939526563e-06,
      "loss": 1.028,
      "step": 19030
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.183407306671143,
      "learning_rate": 8.459892023128773e-06,
      "loss": 1.415,
      "step": 19031
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0539002418518066,
      "learning_rate": 8.459253106730985e-06,
      "loss": 0.8558,
      "step": 19032
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.6126534938812256,
      "learning_rate": 8.458614190333195e-06,
      "loss": 0.9428,
      "step": 19033
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.7200632095336914,
      "learning_rate": 8.457975273935406e-06,
      "loss": 1.079,
      "step": 19034
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.1280808448791504,
      "learning_rate": 8.457336357537616e-06,
      "loss": 1.0957,
      "step": 19035
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8983813524246216,
      "learning_rate": 8.456697441139828e-06,
      "loss": 0.8607,
      "step": 19036
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.761468768119812,
      "learning_rate": 8.456058524742037e-06,
      "loss": 0.7992,
      "step": 19037
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9388591051101685,
      "learning_rate": 8.455419608344249e-06,
      "loss": 0.9349,
      "step": 19038
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4913315773010254,
      "learning_rate": 8.454780691946459e-06,
      "loss": 0.853,
      "step": 19039
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6293939352035522,
      "learning_rate": 8.45414177554867e-06,
      "loss": 0.9575,
      "step": 19040
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.563651442527771,
      "learning_rate": 8.45350285915088e-06,
      "loss": 0.9261,
      "step": 19041
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.8502182960510254,
      "learning_rate": 8.452863942753092e-06,
      "loss": 1.2673,
      "step": 19042
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8507840633392334,
      "learning_rate": 8.452225026355302e-06,
      "loss": 0.9202,
      "step": 19043
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.248828411102295,
      "learning_rate": 8.451586109957513e-06,
      "loss": 1.2215,
      "step": 19044
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.803041934967041,
      "learning_rate": 8.450947193559723e-06,
      "loss": 0.806,
      "step": 19045
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.163980722427368,
      "learning_rate": 8.450308277161935e-06,
      "loss": 0.6388,
      "step": 19046
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3050334453582764,
      "learning_rate": 8.449669360764144e-06,
      "loss": 1.1597,
      "step": 19047
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.719672679901123,
      "learning_rate": 8.449030444366354e-06,
      "loss": 0.8304,
      "step": 19048
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7125493288040161,
      "learning_rate": 8.448391527968566e-06,
      "loss": 0.7647,
      "step": 19049
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.127976894378662,
      "learning_rate": 8.447752611570776e-06,
      "loss": 1.0705,
      "step": 19050
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6492154598236084,
      "learning_rate": 8.447113695172987e-06,
      "loss": 0.8528,
      "step": 19051
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9446524381637573,
      "learning_rate": 8.446474778775197e-06,
      "loss": 1.0214,
      "step": 19052
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.956006646156311,
      "learning_rate": 8.445835862377409e-06,
      "loss": 0.9992,
      "step": 19053
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.4123008251190186,
      "learning_rate": 8.445196945979618e-06,
      "loss": 1.0436,
      "step": 19054
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0608272552490234,
      "learning_rate": 8.44455802958183e-06,
      "loss": 1.0174,
      "step": 19055
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9182121753692627,
      "learning_rate": 8.44391911318404e-06,
      "loss": 0.6295,
      "step": 19056
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.784716248512268,
      "learning_rate": 8.443280196786251e-06,
      "loss": 0.9093,
      "step": 19057
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2063088417053223,
      "learning_rate": 8.442641280388461e-06,
      "loss": 0.938,
      "step": 19058
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9633432626724243,
      "learning_rate": 8.442002363990673e-06,
      "loss": 0.9973,
      "step": 19059
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.190748929977417,
      "learning_rate": 8.441363447592883e-06,
      "loss": 0.7986,
      "step": 19060
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.053356409072876,
      "learning_rate": 8.440724531195094e-06,
      "loss": 0.9641,
      "step": 19061
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6973986625671387,
      "learning_rate": 8.440085614797304e-06,
      "loss": 0.8161,
      "step": 19062
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.3262834548950195,
      "learning_rate": 8.439446698399516e-06,
      "loss": 1.0612,
      "step": 19063
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6911972761154175,
      "learning_rate": 8.438807782001725e-06,
      "loss": 0.8706,
      "step": 19064
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.9019505977630615,
      "learning_rate": 8.438168865603937e-06,
      "loss": 1.3607,
      "step": 19065
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.837875247001648,
      "learning_rate": 8.437529949206147e-06,
      "loss": 1.1002,
      "step": 19066
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.5312650203704834,
      "learning_rate": 8.436891032808357e-06,
      "loss": 1.0677,
      "step": 19067
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0105373859405518,
      "learning_rate": 8.436252116410568e-06,
      "loss": 1.0043,
      "step": 19068
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8274840116500854,
      "learning_rate": 8.435613200012778e-06,
      "loss": 0.8659,
      "step": 19069
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.786895990371704,
      "learning_rate": 8.43497428361499e-06,
      "loss": 0.8127,
      "step": 19070
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9257850646972656,
      "learning_rate": 8.4343353672172e-06,
      "loss": 0.9275,
      "step": 19071
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5686830282211304,
      "learning_rate": 8.433696450819411e-06,
      "loss": 0.9104,
      "step": 19072
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.28533935546875,
      "learning_rate": 8.433057534421621e-06,
      "loss": 0.9351,
      "step": 19073
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7013723850250244,
      "learning_rate": 8.432418618023832e-06,
      "loss": 0.9105,
      "step": 19074
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6092376708984375,
      "learning_rate": 8.431779701626042e-06,
      "loss": 0.8025,
      "step": 19075
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0268750190734863,
      "learning_rate": 8.431140785228254e-06,
      "loss": 0.8989,
      "step": 19076
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.024974822998047,
      "learning_rate": 8.430501868830464e-06,
      "loss": 0.9673,
      "step": 19077
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.6005804538726807,
      "learning_rate": 8.429862952432675e-06,
      "loss": 0.4801,
      "step": 19078
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.841664433479309,
      "learning_rate": 8.429224036034885e-06,
      "loss": 0.7715,
      "step": 19079
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8713288307189941,
      "learning_rate": 8.428585119637097e-06,
      "loss": 0.8418,
      "step": 19080
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.992125153541565,
      "learning_rate": 8.427946203239307e-06,
      "loss": 0.8681,
      "step": 19081
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.31508207321167,
      "learning_rate": 8.427307286841518e-06,
      "loss": 0.8536,
      "step": 19082
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4647881984710693,
      "learning_rate": 8.426668370443728e-06,
      "loss": 0.8571,
      "step": 19083
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1138522624969482,
      "learning_rate": 8.42602945404594e-06,
      "loss": 0.9695,
      "step": 19084
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.078730344772339,
      "learning_rate": 8.42539053764815e-06,
      "loss": 0.8862,
      "step": 19085
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7291512489318848,
      "learning_rate": 8.42475162125036e-06,
      "loss": 0.867,
      "step": 19086
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.244138479232788,
      "learning_rate": 8.42411270485257e-06,
      "loss": 0.7711,
      "step": 19087
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9413989782333374,
      "learning_rate": 8.42347378845478e-06,
      "loss": 0.9147,
      "step": 19088
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9280452728271484,
      "learning_rate": 8.422834872056992e-06,
      "loss": 0.8726,
      "step": 19089
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2758731842041016,
      "learning_rate": 8.422195955659202e-06,
      "loss": 1.0813,
      "step": 19090
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4889494180679321,
      "learning_rate": 8.421557039261414e-06,
      "loss": 0.8472,
      "step": 19091
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1849756240844727,
      "learning_rate": 8.420918122863623e-06,
      "loss": 0.8044,
      "step": 19092
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.213791847229004,
      "learning_rate": 8.420279206465835e-06,
      "loss": 0.9583,
      "step": 19093
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.741566777229309,
      "learning_rate": 8.419640290068045e-06,
      "loss": 0.9355,
      "step": 19094
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.616845965385437,
      "learning_rate": 8.419001373670256e-06,
      "loss": 0.8463,
      "step": 19095
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7461374998092651,
      "learning_rate": 8.418362457272466e-06,
      "loss": 0.8543,
      "step": 19096
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6100236177444458,
      "learning_rate": 8.417723540874678e-06,
      "loss": 0.3835,
      "step": 19097
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6802756786346436,
      "learning_rate": 8.417084624476888e-06,
      "loss": 0.815,
      "step": 19098
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9206739664077759,
      "learning_rate": 8.416445708079099e-06,
      "loss": 0.807,
      "step": 19099
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8679077625274658,
      "learning_rate": 8.415806791681309e-06,
      "loss": 0.9111,
      "step": 19100
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3932428359985352,
      "learning_rate": 8.41516787528352e-06,
      "loss": 0.4999,
      "step": 19101
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2030200958251953,
      "learning_rate": 8.41452895888573e-06,
      "loss": 1.0412,
      "step": 19102
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.450568675994873,
      "learning_rate": 8.41389004248794e-06,
      "loss": 0.6151,
      "step": 19103
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.8990485668182373,
      "learning_rate": 8.413251126090152e-06,
      "loss": 0.748,
      "step": 19104
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.196011781692505,
      "learning_rate": 8.412612209692362e-06,
      "loss": 0.8352,
      "step": 19105
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.931362271308899,
      "learning_rate": 8.411973293294573e-06,
      "loss": 0.8785,
      "step": 19106
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3232996463775635,
      "learning_rate": 8.411334376896783e-06,
      "loss": 0.9019,
      "step": 19107
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.843259572982788,
      "learning_rate": 8.410695460498995e-06,
      "loss": 0.8383,
      "step": 19108
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.419704794883728,
      "learning_rate": 8.410056544101204e-06,
      "loss": 0.4936,
      "step": 19109
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6252601146697998,
      "learning_rate": 8.409417627703416e-06,
      "loss": 0.9388,
      "step": 19110
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.913872241973877,
      "learning_rate": 8.408778711305626e-06,
      "loss": 0.8067,
      "step": 19111
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4644675254821777,
      "learning_rate": 8.408139794907837e-06,
      "loss": 0.7986,
      "step": 19112
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0338637828826904,
      "learning_rate": 8.407500878510047e-06,
      "loss": 0.6545,
      "step": 19113
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.7471365928649902,
      "learning_rate": 8.406861962112259e-06,
      "loss": 1.0949,
      "step": 19114
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9657623767852783,
      "learning_rate": 8.406223045714469e-06,
      "loss": 1.0903,
      "step": 19115
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7510147094726562,
      "learning_rate": 8.40558412931668e-06,
      "loss": 0.7937,
      "step": 19116
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.846670389175415,
      "learning_rate": 8.40494521291889e-06,
      "loss": 0.5345,
      "step": 19117
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5780541896820068,
      "learning_rate": 8.404306296521102e-06,
      "loss": 0.8689,
      "step": 19118
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.819315791130066,
      "learning_rate": 8.403667380123311e-06,
      "loss": 0.8785,
      "step": 19119
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7267554998397827,
      "learning_rate": 8.403028463725523e-06,
      "loss": 0.8701,
      "step": 19120
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.487185001373291,
      "learning_rate": 8.402389547327733e-06,
      "loss": 0.7841,
      "step": 19121
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7120037078857422,
      "learning_rate": 8.401750630929943e-06,
      "loss": 0.8213,
      "step": 19122
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8246102333068848,
      "learning_rate": 8.401111714532154e-06,
      "loss": 0.7468,
      "step": 19123
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5073027610778809,
      "learning_rate": 8.400472798134364e-06,
      "loss": 0.3699,
      "step": 19124
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.5577433109283447,
      "learning_rate": 8.399833881736576e-06,
      "loss": 1.0343,
      "step": 19125
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.453784227371216,
      "learning_rate": 8.399194965338785e-06,
      "loss": 0.9693,
      "step": 19126
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.057767391204834,
      "learning_rate": 8.398556048940997e-06,
      "loss": 1.0984,
      "step": 19127
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.372905731201172,
      "learning_rate": 8.397917132543207e-06,
      "loss": 0.894,
      "step": 19128
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8965134620666504,
      "learning_rate": 8.397278216145418e-06,
      "loss": 0.8138,
      "step": 19129
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7359118461608887,
      "learning_rate": 8.396639299747628e-06,
      "loss": 1.0098,
      "step": 19130
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.955888271331787,
      "learning_rate": 8.39600038334984e-06,
      "loss": 0.6902,
      "step": 19131
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8051937818527222,
      "learning_rate": 8.39536146695205e-06,
      "loss": 0.5146,
      "step": 19132
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.157452344894409,
      "learning_rate": 8.394722550554261e-06,
      "loss": 0.8086,
      "step": 19133
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0203800201416016,
      "learning_rate": 8.394083634156471e-06,
      "loss": 0.9973,
      "step": 19134
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.249889850616455,
      "learning_rate": 8.393444717758683e-06,
      "loss": 0.6134,
      "step": 19135
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5176540613174438,
      "learning_rate": 8.392805801360892e-06,
      "loss": 0.8237,
      "step": 19136
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7513549327850342,
      "learning_rate": 8.392166884963104e-06,
      "loss": 0.9188,
      "step": 19137
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0723257064819336,
      "learning_rate": 8.391527968565314e-06,
      "loss": 0.9311,
      "step": 19138
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.869553089141846,
      "learning_rate": 8.390889052167525e-06,
      "loss": 1.6616,
      "step": 19139
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8701050281524658,
      "learning_rate": 8.390250135769735e-06,
      "loss": 0.8491,
      "step": 19140
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.412055015563965,
      "learning_rate": 8.389611219371945e-06,
      "loss": 0.8947,
      "step": 19141
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.830012559890747,
      "learning_rate": 8.388972302974157e-06,
      "loss": 0.7733,
      "step": 19142
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7713068723678589,
      "learning_rate": 8.388333386576367e-06,
      "loss": 0.9959,
      "step": 19143
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0533335208892822,
      "learning_rate": 8.387694470178578e-06,
      "loss": 1.3551,
      "step": 19144
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8428410291671753,
      "learning_rate": 8.387055553780788e-06,
      "loss": 0.8389,
      "step": 19145
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5909085273742676,
      "learning_rate": 8.386416637383e-06,
      "loss": 0.8916,
      "step": 19146
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8524785041809082,
      "learning_rate": 8.38577772098521e-06,
      "loss": 1.2412,
      "step": 19147
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5794134140014648,
      "learning_rate": 8.385138804587421e-06,
      "loss": 0.7873,
      "step": 19148
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9047129154205322,
      "learning_rate": 8.38449988818963e-06,
      "loss": 0.9738,
      "step": 19149
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8102569580078125,
      "learning_rate": 8.383860971791842e-06,
      "loss": 0.966,
      "step": 19150
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3062652349472046,
      "learning_rate": 8.383222055394052e-06,
      "loss": 0.7381,
      "step": 19151
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.548029899597168,
      "learning_rate": 8.382583138996264e-06,
      "loss": 0.5503,
      "step": 19152
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7465356588363647,
      "learning_rate": 8.381944222598474e-06,
      "loss": 0.8545,
      "step": 19153
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.847346067428589,
      "learning_rate": 8.381305306200685e-06,
      "loss": 1.1519,
      "step": 19154
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4850239753723145,
      "learning_rate": 8.380666389802895e-06,
      "loss": 0.9625,
      "step": 19155
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.7812249660491943,
      "learning_rate": 8.380027473405106e-06,
      "loss": 1.602,
      "step": 19156
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6138731241226196,
      "learning_rate": 8.379388557007316e-06,
      "loss": 0.8994,
      "step": 19157
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7381054162979126,
      "learning_rate": 8.378749640609526e-06,
      "loss": 0.8582,
      "step": 19158
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.263606071472168,
      "learning_rate": 8.378110724211738e-06,
      "loss": 1.0574,
      "step": 19159
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9997683763504028,
      "learning_rate": 8.377471807813948e-06,
      "loss": 0.9158,
      "step": 19160
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.08774995803833,
      "learning_rate": 8.376832891416159e-06,
      "loss": 0.8815,
      "step": 19161
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.843796968460083,
      "learning_rate": 8.376193975018369e-06,
      "loss": 0.8385,
      "step": 19162
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9939837455749512,
      "learning_rate": 8.37555505862058e-06,
      "loss": 0.738,
      "step": 19163
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2211408615112305,
      "learning_rate": 8.37491614222279e-06,
      "loss": 0.7037,
      "step": 19164
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7897990942001343,
      "learning_rate": 8.374277225825002e-06,
      "loss": 1.1748,
      "step": 19165
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8796725273132324,
      "learning_rate": 8.373638309427212e-06,
      "loss": 0.9733,
      "step": 19166
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6199203729629517,
      "learning_rate": 8.372999393029423e-06,
      "loss": 0.9292,
      "step": 19167
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.1399736404418945,
      "learning_rate": 8.372360476631633e-06,
      "loss": 1.0982,
      "step": 19168
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7971436977386475,
      "learning_rate": 8.371721560233845e-06,
      "loss": 0.8771,
      "step": 19169
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8956154584884644,
      "learning_rate": 8.371082643836055e-06,
      "loss": 0.7703,
      "step": 19170
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6241626739501953,
      "learning_rate": 8.370443727438266e-06,
      "loss": 0.8619,
      "step": 19171
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.794724941253662,
      "learning_rate": 8.369804811040476e-06,
      "loss": 0.5516,
      "step": 19172
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6419755220413208,
      "learning_rate": 8.369165894642688e-06,
      "loss": 0.922,
      "step": 19173
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.970400333404541,
      "learning_rate": 8.368526978244897e-06,
      "loss": 0.9626,
      "step": 19174
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1526947021484375,
      "learning_rate": 8.367888061847109e-06,
      "loss": 0.5125,
      "step": 19175
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9218882322311401,
      "learning_rate": 8.367249145449319e-06,
      "loss": 1.2431,
      "step": 19176
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8486541509628296,
      "learning_rate": 8.366610229051529e-06,
      "loss": 0.861,
      "step": 19177
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.978127360343933,
      "learning_rate": 8.36597131265374e-06,
      "loss": 0.9448,
      "step": 19178
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.21958589553833,
      "learning_rate": 8.36533239625595e-06,
      "loss": 0.8415,
      "step": 19179
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9348098039627075,
      "learning_rate": 8.364693479858162e-06,
      "loss": 1.292,
      "step": 19180
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.180185317993164,
      "learning_rate": 8.364054563460371e-06,
      "loss": 0.9235,
      "step": 19181
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5336569547653198,
      "learning_rate": 8.363415647062583e-06,
      "loss": 0.9346,
      "step": 19182
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6706295013427734,
      "learning_rate": 8.362776730664793e-06,
      "loss": 0.776,
      "step": 19183
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6452957391738892,
      "learning_rate": 8.362137814267004e-06,
      "loss": 0.8513,
      "step": 19184
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3958752155303955,
      "learning_rate": 8.361498897869214e-06,
      "loss": 0.3871,
      "step": 19185
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0675835609436035,
      "learning_rate": 8.360859981471426e-06,
      "loss": 1.0969,
      "step": 19186
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9646347761154175,
      "learning_rate": 8.360221065073636e-06,
      "loss": 0.8321,
      "step": 19187
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5282189846038818,
      "learning_rate": 8.359582148675847e-06,
      "loss": 0.7595,
      "step": 19188
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8522510528564453,
      "learning_rate": 8.358943232278057e-06,
      "loss": 0.8409,
      "step": 19189
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5923423767089844,
      "learning_rate": 8.358304315880269e-06,
      "loss": 0.8897,
      "step": 19190
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7504167556762695,
      "learning_rate": 8.357665399482478e-06,
      "loss": 0.8575,
      "step": 19191
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6036059856414795,
      "learning_rate": 8.35702648308469e-06,
      "loss": 0.8605,
      "step": 19192
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.324727773666382,
      "learning_rate": 8.3563875666869e-06,
      "loss": 0.6508,
      "step": 19193
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7802623510360718,
      "learning_rate": 8.35574865028911e-06,
      "loss": 0.8812,
      "step": 19194
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3107143640518188,
      "learning_rate": 8.355109733891321e-06,
      "loss": 0.4021,
      "step": 19195
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.217186212539673,
      "learning_rate": 8.354470817493531e-06,
      "loss": 0.5631,
      "step": 19196
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.3470940589904785,
      "learning_rate": 8.353831901095743e-06,
      "loss": 1.5994,
      "step": 19197
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8443710803985596,
      "learning_rate": 8.353192984697952e-06,
      "loss": 0.9765,
      "step": 19198
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7350612878799438,
      "learning_rate": 8.352554068300164e-06,
      "loss": 0.7976,
      "step": 19199
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9896855354309082,
      "learning_rate": 8.351915151902374e-06,
      "loss": 0.8437,
      "step": 19200
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.680820465087891,
      "learning_rate": 8.351276235504585e-06,
      "loss": 1.1037,
      "step": 19201
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5411094427108765,
      "learning_rate": 8.350637319106795e-06,
      "loss": 0.7917,
      "step": 19202
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4867140054702759,
      "learning_rate": 8.349998402709007e-06,
      "loss": 0.8258,
      "step": 19203
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7853089570999146,
      "learning_rate": 8.349359486311217e-06,
      "loss": 0.8837,
      "step": 19204
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7681154012680054,
      "learning_rate": 8.348720569913428e-06,
      "loss": 0.9183,
      "step": 19205
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.206851005554199,
      "learning_rate": 8.348081653515638e-06,
      "loss": 1.0671,
      "step": 19206
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.484488844871521,
      "learning_rate": 8.34744273711785e-06,
      "loss": 0.7954,
      "step": 19207
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.827078104019165,
      "learning_rate": 8.34680382072006e-06,
      "loss": 0.8325,
      "step": 19208
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.363739252090454,
      "learning_rate": 8.346164904322271e-06,
      "loss": 1.0641,
      "step": 19209
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2423367500305176,
      "learning_rate": 8.345525987924481e-06,
      "loss": 0.9128,
      "step": 19210
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.774363398551941,
      "learning_rate": 8.344887071526692e-06,
      "loss": 0.8252,
      "step": 19211
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4595214128494263,
      "learning_rate": 8.344248155128902e-06,
      "loss": 0.8497,
      "step": 19212
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.298179864883423,
      "learning_rate": 8.343609238731112e-06,
      "loss": 1.026,
      "step": 19213
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4575366973876953,
      "learning_rate": 8.342970322333324e-06,
      "loss": 0.8653,
      "step": 19214
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9060304164886475,
      "learning_rate": 8.342331405935534e-06,
      "loss": 0.6843,
      "step": 19215
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.932483434677124,
      "learning_rate": 8.341692489537745e-06,
      "loss": 0.9867,
      "step": 19216
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7623982429504395,
      "learning_rate": 8.341053573139955e-06,
      "loss": 0.8787,
      "step": 19217
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7826745510101318,
      "learning_rate": 8.340414656742166e-06,
      "loss": 0.8941,
      "step": 19218
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.71675705909729,
      "learning_rate": 8.339775740344376e-06,
      "loss": 0.997,
      "step": 19219
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4441251754760742,
      "learning_rate": 8.339136823946588e-06,
      "loss": 0.3891,
      "step": 19220
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.5857138633728027,
      "learning_rate": 8.338497907548798e-06,
      "loss": 0.5359,
      "step": 19221
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0942063331604004,
      "learning_rate": 8.33785899115101e-06,
      "loss": 1.0321,
      "step": 19222
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.679234266281128,
      "learning_rate": 8.337220074753219e-06,
      "loss": 0.884,
      "step": 19223
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.612922191619873,
      "learning_rate": 8.33658115835543e-06,
      "loss": 0.8211,
      "step": 19224
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1514017581939697,
      "learning_rate": 8.33594224195764e-06,
      "loss": 1.0103,
      "step": 19225
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.99799382686615,
      "learning_rate": 8.335303325559852e-06,
      "loss": 0.9785,
      "step": 19226
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9235813617706299,
      "learning_rate": 8.334664409162062e-06,
      "loss": 0.9805,
      "step": 19227
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2690539360046387,
      "learning_rate": 8.334025492764273e-06,
      "loss": 0.9546,
      "step": 19228
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7273436784744263,
      "learning_rate": 8.333386576366483e-06,
      "loss": 0.8988,
      "step": 19229
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.184447288513184,
      "learning_rate": 8.332747659968695e-06,
      "loss": 1.2342,
      "step": 19230
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.908942699432373,
      "learning_rate": 8.332108743570905e-06,
      "loss": 1.185,
      "step": 19231
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9874898195266724,
      "learning_rate": 8.331469827173115e-06,
      "loss": 0.9689,
      "step": 19232
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3208136558532715,
      "learning_rate": 8.330830910775326e-06,
      "loss": 0.9948,
      "step": 19233
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.2335875034332275,
      "learning_rate": 8.330191994377536e-06,
      "loss": 0.9273,
      "step": 19234
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.6638870239257812,
      "learning_rate": 8.329553077979748e-06,
      "loss": 1.3391,
      "step": 19235
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.027777671813965,
      "learning_rate": 8.328914161581957e-06,
      "loss": 0.6692,
      "step": 19236
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8137472867965698,
      "learning_rate": 8.328275245184169e-06,
      "loss": 0.8994,
      "step": 19237
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.934950351715088,
      "learning_rate": 8.327636328786379e-06,
      "loss": 0.787,
      "step": 19238
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8220769166946411,
      "learning_rate": 8.32699741238859e-06,
      "loss": 1.1188,
      "step": 19239
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9235621690750122,
      "learning_rate": 8.3263584959908e-06,
      "loss": 0.8868,
      "step": 19240
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8797374963760376,
      "learning_rate": 8.325719579593012e-06,
      "loss": 0.8801,
      "step": 19241
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.988051652908325,
      "learning_rate": 8.325080663195222e-06,
      "loss": 1.6114,
      "step": 19242
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6305861473083496,
      "learning_rate": 8.324441746797433e-06,
      "loss": 0.7562,
      "step": 19243
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.674317717552185,
      "learning_rate": 8.323802830399643e-06,
      "loss": 0.8531,
      "step": 19244
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4930682182312012,
      "learning_rate": 8.323163914001855e-06,
      "loss": 0.4555,
      "step": 19245
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9999016523361206,
      "learning_rate": 8.322524997604064e-06,
      "loss": 0.9112,
      "step": 19246
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.714032769203186,
      "learning_rate": 8.321886081206276e-06,
      "loss": 0.8069,
      "step": 19247
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.159304618835449,
      "learning_rate": 8.321247164808486e-06,
      "loss": 0.8507,
      "step": 19248
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.555650234222412,
      "learning_rate": 8.320608248410696e-06,
      "loss": 0.829,
      "step": 19249
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7475764751434326,
      "learning_rate": 8.319969332012907e-06,
      "loss": 0.8161,
      "step": 19250
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0462005138397217,
      "learning_rate": 8.319330415615117e-06,
      "loss": 0.8628,
      "step": 19251
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0788235664367676,
      "learning_rate": 8.318691499217329e-06,
      "loss": 0.9035,
      "step": 19252
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8373311758041382,
      "learning_rate": 8.318052582819538e-06,
      "loss": 0.8763,
      "step": 19253
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4776545763015747,
      "learning_rate": 8.31741366642175e-06,
      "loss": 0.8677,
      "step": 19254
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.808525085449219,
      "learning_rate": 8.31677475002396e-06,
      "loss": 1.3948,
      "step": 19255
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.5848264694213867,
      "learning_rate": 8.316135833626171e-06,
      "loss": 0.8797,
      "step": 19256
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.945308804512024,
      "learning_rate": 8.315496917228381e-06,
      "loss": 0.8041,
      "step": 19257
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.518843650817871,
      "learning_rate": 8.314858000830593e-06,
      "loss": 0.8257,
      "step": 19258
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.01926326751709,
      "learning_rate": 8.314219084432803e-06,
      "loss": 0.8693,
      "step": 19259
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4705573320388794,
      "learning_rate": 8.313580168035014e-06,
      "loss": 0.8048,
      "step": 19260
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.580589771270752,
      "learning_rate": 8.312941251637224e-06,
      "loss": 0.759,
      "step": 19261
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1058449745178223,
      "learning_rate": 8.312302335239436e-06,
      "loss": 0.8615,
      "step": 19262
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2788143157958984,
      "learning_rate": 8.311663418841645e-06,
      "loss": 0.5248,
      "step": 19263
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7625796794891357,
      "learning_rate": 8.311024502443857e-06,
      "loss": 0.8426,
      "step": 19264
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9769881963729858,
      "learning_rate": 8.310385586046067e-06,
      "loss": 0.9462,
      "step": 19265
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7297337055206299,
      "learning_rate": 8.309746669648278e-06,
      "loss": 0.8034,
      "step": 19266
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0704195499420166,
      "learning_rate": 8.309107753250488e-06,
      "loss": 0.8106,
      "step": 19267
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4098291397094727,
      "learning_rate": 8.308468836852698e-06,
      "loss": 1.0813,
      "step": 19268
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.887185573577881,
      "learning_rate": 8.30782992045491e-06,
      "loss": 0.8077,
      "step": 19269
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6910960674285889,
      "learning_rate": 8.30719100405712e-06,
      "loss": 0.8241,
      "step": 19270
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.841261625289917,
      "learning_rate": 8.306552087659331e-06,
      "loss": 0.9901,
      "step": 19271
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6443148851394653,
      "learning_rate": 8.305913171261541e-06,
      "loss": 0.8413,
      "step": 19272
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7862021923065186,
      "learning_rate": 8.305274254863752e-06,
      "loss": 0.8785,
      "step": 19273
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.959977149963379,
      "learning_rate": 8.304635338465962e-06,
      "loss": 1.5033,
      "step": 19274
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9552221298217773,
      "learning_rate": 8.303996422068174e-06,
      "loss": 0.8345,
      "step": 19275
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2638638019561768,
      "learning_rate": 8.303357505670384e-06,
      "loss": 1.0899,
      "step": 19276
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7124608755111694,
      "learning_rate": 8.302718589272595e-06,
      "loss": 0.8797,
      "step": 19277
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.919446587562561,
      "learning_rate": 8.302079672874805e-06,
      "loss": 0.9479,
      "step": 19278
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9825544357299805,
      "learning_rate": 8.301440756477017e-06,
      "loss": 0.7276,
      "step": 19279
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.0614755153656006,
      "learning_rate": 8.300801840079226e-06,
      "loss": 0.7458,
      "step": 19280
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9167530536651611,
      "learning_rate": 8.300162923681438e-06,
      "loss": 0.8579,
      "step": 19281
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.964754104614258,
      "learning_rate": 8.299524007283648e-06,
      "loss": 0.6874,
      "step": 19282
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.996367454528809,
      "learning_rate": 8.29888509088586e-06,
      "loss": 1.575,
      "step": 19283
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5730202198028564,
      "learning_rate": 8.29824617448807e-06,
      "loss": 0.8956,
      "step": 19284
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6122487783432007,
      "learning_rate": 8.29760725809028e-06,
      "loss": 0.5043,
      "step": 19285
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8468636274337769,
      "learning_rate": 8.29696834169249e-06,
      "loss": 0.9361,
      "step": 19286
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.9919466972351074,
      "learning_rate": 8.2963294252947e-06,
      "loss": 0.9287,
      "step": 19287
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.445870280265808,
      "learning_rate": 8.295690508896912e-06,
      "loss": 0.776,
      "step": 19288
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1200859546661377,
      "learning_rate": 8.295051592499122e-06,
      "loss": 0.6558,
      "step": 19289
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.741819977760315,
      "learning_rate": 8.294412676101333e-06,
      "loss": 0.8371,
      "step": 19290
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.1865179538726807,
      "learning_rate": 8.293773759703543e-06,
      "loss": 0.9179,
      "step": 19291
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4744420051574707,
      "learning_rate": 8.293134843305755e-06,
      "loss": 0.8659,
      "step": 19292
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8105510473251343,
      "learning_rate": 8.292495926907965e-06,
      "loss": 0.8436,
      "step": 19293
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2120471000671387,
      "learning_rate": 8.291857010510176e-06,
      "loss": 0.5554,
      "step": 19294
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2211415767669678,
      "learning_rate": 8.291218094112386e-06,
      "loss": 0.9351,
      "step": 19295
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9064263105392456,
      "learning_rate": 8.290579177714598e-06,
      "loss": 0.995,
      "step": 19296
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.212656021118164,
      "learning_rate": 8.289940261316808e-06,
      "loss": 0.971,
      "step": 19297
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9483976364135742,
      "learning_rate": 8.289301344919019e-06,
      "loss": 0.8877,
      "step": 19298
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7974873781204224,
      "learning_rate": 8.288662428521229e-06,
      "loss": 0.7711,
      "step": 19299
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5297898054122925,
      "learning_rate": 8.28802351212344e-06,
      "loss": 0.8736,
      "step": 19300
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.088787317276001,
      "learning_rate": 8.28738459572565e-06,
      "loss": 0.9671,
      "step": 19301
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.6702966690063477,
      "learning_rate": 8.286745679327862e-06,
      "loss": 0.8515,
      "step": 19302
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.58424711227417,
      "learning_rate": 8.286106762930072e-06,
      "loss": 1.1015,
      "step": 19303
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.140665054321289,
      "learning_rate": 8.285467846532282e-06,
      "loss": 1.0369,
      "step": 19304
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0302212238311768,
      "learning_rate": 8.284828930134493e-06,
      "loss": 0.8647,
      "step": 19305
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9557480812072754,
      "learning_rate": 8.284190013736703e-06,
      "loss": 0.8385,
      "step": 19306
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1595771312713623,
      "learning_rate": 8.283551097338915e-06,
      "loss": 0.7966,
      "step": 19307
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.945255994796753,
      "learning_rate": 8.282912180941124e-06,
      "loss": 0.919,
      "step": 19308
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7672679424285889,
      "learning_rate": 8.282273264543336e-06,
      "loss": 0.879,
      "step": 19309
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2613308429718018,
      "learning_rate": 8.281634348145546e-06,
      "loss": 0.9003,
      "step": 19310
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8708782196044922,
      "learning_rate": 8.280995431747757e-06,
      "loss": 0.7089,
      "step": 19311
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.916182279586792,
      "learning_rate": 8.280356515349967e-06,
      "loss": 0.9486,
      "step": 19312
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.690347909927368,
      "learning_rate": 8.279717598952179e-06,
      "loss": 0.9091,
      "step": 19313
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0472190380096436,
      "learning_rate": 8.279078682554389e-06,
      "loss": 1.1673,
      "step": 19314
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4894330501556396,
      "learning_rate": 8.2784397661566e-06,
      "loss": 0.7885,
      "step": 19315
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8385313749313354,
      "learning_rate": 8.27780084975881e-06,
      "loss": 0.8401,
      "step": 19316
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7006651163101196,
      "learning_rate": 8.277161933361022e-06,
      "loss": 0.8308,
      "step": 19317
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.470956563949585,
      "learning_rate": 8.276523016963231e-06,
      "loss": 0.8079,
      "step": 19318
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3952560424804688,
      "learning_rate": 8.275884100565443e-06,
      "loss": 0.7568,
      "step": 19319
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7340428829193115,
      "learning_rate": 8.275245184167653e-06,
      "loss": 0.8093,
      "step": 19320
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0111336708068848,
      "learning_rate": 8.274606267769864e-06,
      "loss": 0.9,
      "step": 19321
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9074429273605347,
      "learning_rate": 8.273967351372074e-06,
      "loss": 0.816,
      "step": 19322
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8264894485473633,
      "learning_rate": 8.273328434974284e-06,
      "loss": 1.0602,
      "step": 19323
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5921320915222168,
      "learning_rate": 8.272689518576496e-06,
      "loss": 0.8581,
      "step": 19324
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.483176350593567,
      "learning_rate": 8.272050602178705e-06,
      "loss": 0.7848,
      "step": 19325
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.480466604232788,
      "learning_rate": 8.271411685780915e-06,
      "loss": 0.7834,
      "step": 19326
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7809990644454956,
      "learning_rate": 8.270772769383127e-06,
      "loss": 0.9244,
      "step": 19327
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.871285319328308,
      "learning_rate": 8.270133852985337e-06,
      "loss": 1.1055,
      "step": 19328
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1656460762023926,
      "learning_rate": 8.269494936587548e-06,
      "loss": 0.9752,
      "step": 19329
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2973599433898926,
      "learning_rate": 8.268856020189758e-06,
      "loss": 0.9263,
      "step": 19330
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6442316770553589,
      "learning_rate": 8.26821710379197e-06,
      "loss": 0.8483,
      "step": 19331
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.0508668422698975,
      "learning_rate": 8.26757818739418e-06,
      "loss": 0.8235,
      "step": 19332
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.503129482269287,
      "learning_rate": 8.266939270996391e-06,
      "loss": 0.8938,
      "step": 19333
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0241076946258545,
      "learning_rate": 8.266300354598601e-06,
      "loss": 0.8843,
      "step": 19334
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6926831007003784,
      "learning_rate": 8.26566143820081e-06,
      "loss": 0.9482,
      "step": 19335
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5753496885299683,
      "learning_rate": 8.265022521803022e-06,
      "loss": 0.9407,
      "step": 19336
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.546579122543335,
      "learning_rate": 8.264383605405232e-06,
      "loss": 0.7053,
      "step": 19337
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7659591436386108,
      "learning_rate": 8.263744689007444e-06,
      "loss": 0.9405,
      "step": 19338
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9676679372787476,
      "learning_rate": 8.263105772609654e-06,
      "loss": 0.8904,
      "step": 19339
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7034457921981812,
      "learning_rate": 8.262466856211865e-06,
      "loss": 1.1415,
      "step": 19340
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.2595882415771484,
      "learning_rate": 8.261827939814075e-06,
      "loss": 0.6469,
      "step": 19341
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8388692140579224,
      "learning_rate": 8.261189023416287e-06,
      "loss": 0.9591,
      "step": 19342
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9402101039886475,
      "learning_rate": 8.260550107018496e-06,
      "loss": 0.9936,
      "step": 19343
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.554291844367981,
      "learning_rate": 8.259911190620708e-06,
      "loss": 0.7582,
      "step": 19344
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.494517207145691,
      "learning_rate": 8.259272274222918e-06,
      "loss": 0.4252,
      "step": 19345
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5746465921401978,
      "learning_rate": 8.25863335782513e-06,
      "loss": 0.9384,
      "step": 19346
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2807185649871826,
      "learning_rate": 8.25799444142734e-06,
      "loss": 0.8959,
      "step": 19347
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1930127143859863,
      "learning_rate": 8.25735552502955e-06,
      "loss": 0.7904,
      "step": 19348
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8184165954589844,
      "learning_rate": 8.25671660863176e-06,
      "loss": 0.8858,
      "step": 19349
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9791476726531982,
      "learning_rate": 8.256077692233972e-06,
      "loss": 0.9933,
      "step": 19350
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.0518717765808105,
      "learning_rate": 8.255438775836182e-06,
      "loss": 0.5912,
      "step": 19351
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.019780158996582,
      "learning_rate": 8.254799859438394e-06,
      "loss": 0.9384,
      "step": 19352
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.5820353031158447,
      "learning_rate": 8.254160943040603e-06,
      "loss": 0.634,
      "step": 19353
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1749181747436523,
      "learning_rate": 8.253522026642813e-06,
      "loss": 0.8063,
      "step": 19354
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7831017971038818,
      "learning_rate": 8.252883110245025e-06,
      "loss": 0.8869,
      "step": 19355
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.260347604751587,
      "learning_rate": 8.252244193847235e-06,
      "loss": 0.896,
      "step": 19356
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.6241681575775146,
      "learning_rate": 8.251605277449446e-06,
      "loss": 0.6346,
      "step": 19357
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.677024483680725,
      "learning_rate": 8.250966361051656e-06,
      "loss": 0.891,
      "step": 19358
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9752647876739502,
      "learning_rate": 8.250327444653868e-06,
      "loss": 0.9125,
      "step": 19359
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7612569332122803,
      "learning_rate": 8.249688528256077e-06,
      "loss": 0.8492,
      "step": 19360
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1068105697631836,
      "learning_rate": 8.249049611858289e-06,
      "loss": 0.899,
      "step": 19361
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.930976152420044,
      "learning_rate": 8.248410695460499e-06,
      "loss": 0.7629,
      "step": 19362
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5368632078170776,
      "learning_rate": 8.24777177906271e-06,
      "loss": 0.8691,
      "step": 19363
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.032817840576172,
      "learning_rate": 8.24713286266492e-06,
      "loss": 0.7682,
      "step": 19364
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6657930612564087,
      "learning_rate": 8.246493946267132e-06,
      "loss": 0.8796,
      "step": 19365
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.729065418243408,
      "learning_rate": 8.245855029869342e-06,
      "loss": 0.9148,
      "step": 19366
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6123429536819458,
      "learning_rate": 8.245216113471553e-06,
      "loss": 0.9186,
      "step": 19367
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7327942848205566,
      "learning_rate": 8.244577197073763e-06,
      "loss": 1.3787,
      "step": 19368
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3502731323242188,
      "learning_rate": 8.243938280675975e-06,
      "loss": 0.3283,
      "step": 19369
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.2983238697052,
      "learning_rate": 8.243299364278184e-06,
      "loss": 1.0936,
      "step": 19370
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7766942977905273,
      "learning_rate": 8.242660447880394e-06,
      "loss": 0.898,
      "step": 19371
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5995351076126099,
      "learning_rate": 8.242021531482606e-06,
      "loss": 0.9283,
      "step": 19372
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.7740888595581055,
      "learning_rate": 8.241382615084816e-06,
      "loss": 1.2328,
      "step": 19373
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4351983070373535,
      "learning_rate": 8.240743698687027e-06,
      "loss": 0.5679,
      "step": 19374
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5351251363754272,
      "learning_rate": 8.240104782289237e-06,
      "loss": 0.8749,
      "step": 19375
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.023533344268799,
      "learning_rate": 8.239465865891449e-06,
      "loss": 0.8328,
      "step": 19376
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4234402179718018,
      "learning_rate": 8.238826949493658e-06,
      "loss": 0.8249,
      "step": 19377
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1687891483306885,
      "learning_rate": 8.23818803309587e-06,
      "loss": 0.7101,
      "step": 19378
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2040469646453857,
      "learning_rate": 8.23754911669808e-06,
      "loss": 0.9399,
      "step": 19379
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.504953384399414,
      "learning_rate": 8.236910200300291e-06,
      "loss": 0.772,
      "step": 19380
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7523105144500732,
      "learning_rate": 8.236271283902501e-06,
      "loss": 0.6269,
      "step": 19381
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5583027601242065,
      "learning_rate": 8.235632367504713e-06,
      "loss": 0.8463,
      "step": 19382
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4947270154953003,
      "learning_rate": 8.234993451106923e-06,
      "loss": 0.9183,
      "step": 19383
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8336234092712402,
      "learning_rate": 8.234354534709134e-06,
      "loss": 0.8554,
      "step": 19384
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.927598237991333,
      "learning_rate": 8.233715618311344e-06,
      "loss": 1.1057,
      "step": 19385
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3565844297409058,
      "learning_rate": 8.233076701913556e-06,
      "loss": 0.8139,
      "step": 19386
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0456173419952393,
      "learning_rate": 8.232437785515765e-06,
      "loss": 1.0477,
      "step": 19387
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.113687753677368,
      "learning_rate": 8.231798869117977e-06,
      "loss": 0.7155,
      "step": 19388
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.49406099319458,
      "learning_rate": 8.231159952720187e-06,
      "loss": 1.0419,
      "step": 19389
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.534018874168396,
      "learning_rate": 8.230521036322397e-06,
      "loss": 0.3951,
      "step": 19390
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.681094765663147,
      "learning_rate": 8.229882119924608e-06,
      "loss": 1.1985,
      "step": 19391
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9713761806488037,
      "learning_rate": 8.229243203526818e-06,
      "loss": 0.847,
      "step": 19392
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.852638840675354,
      "learning_rate": 8.22860428712903e-06,
      "loss": 0.7925,
      "step": 19393
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5259895324707031,
      "learning_rate": 8.22796537073124e-06,
      "loss": 0.7938,
      "step": 19394
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.487835645675659,
      "learning_rate": 8.227326454333451e-06,
      "loss": 1.0448,
      "step": 19395
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.733371376991272,
      "learning_rate": 8.226687537935661e-06,
      "loss": 0.8135,
      "step": 19396
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5671757459640503,
      "learning_rate": 8.226048621537872e-06,
      "loss": 0.7841,
      "step": 19397
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5316815376281738,
      "learning_rate": 8.225409705140082e-06,
      "loss": 0.8557,
      "step": 19398
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7272870540618896,
      "learning_rate": 8.224770788742294e-06,
      "loss": 0.9076,
      "step": 19399
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.584506869316101,
      "learning_rate": 8.224131872344504e-06,
      "loss": 0.8741,
      "step": 19400
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.570015788078308,
      "learning_rate": 8.223492955946715e-06,
      "loss": 0.9278,
      "step": 19401
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.785269856452942,
      "learning_rate": 8.222854039548925e-06,
      "loss": 0.9105,
      "step": 19402
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8525909185409546,
      "learning_rate": 8.222215123151137e-06,
      "loss": 0.8287,
      "step": 19403
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5922795534133911,
      "learning_rate": 8.221576206753347e-06,
      "loss": 0.8603,
      "step": 19404
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5688972473144531,
      "learning_rate": 8.220937290355558e-06,
      "loss": 0.8103,
      "step": 19405
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.48117196559906,
      "learning_rate": 8.220298373957768e-06,
      "loss": 0.4796,
      "step": 19406
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5507792234420776,
      "learning_rate": 8.21965945755998e-06,
      "loss": 0.8364,
      "step": 19407
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7243597507476807,
      "learning_rate": 8.21902054116219e-06,
      "loss": 0.8138,
      "step": 19408
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5037989616394043,
      "learning_rate": 8.2183816247644e-06,
      "loss": 0.8428,
      "step": 19409
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8297146558761597,
      "learning_rate": 8.21774270836661e-06,
      "loss": 0.8714,
      "step": 19410
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6792608499526978,
      "learning_rate": 8.21710379196882e-06,
      "loss": 0.9758,
      "step": 19411
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.8108856678009033,
      "learning_rate": 8.216464875571032e-06,
      "loss": 1.3879,
      "step": 19412
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4398269653320312,
      "learning_rate": 8.215825959173242e-06,
      "loss": 0.9691,
      "step": 19413
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0306472778320312,
      "learning_rate": 8.215187042775454e-06,
      "loss": 0.9163,
      "step": 19414
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.721620082855225,
      "learning_rate": 8.214548126377663e-06,
      "loss": 1.4492,
      "step": 19415
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.528868556022644,
      "learning_rate": 8.213909209979875e-06,
      "loss": 0.8196,
      "step": 19416
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6937906742095947,
      "learning_rate": 8.213270293582085e-06,
      "loss": 0.8975,
      "step": 19417
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7424060106277466,
      "learning_rate": 8.212631377184296e-06,
      "loss": 0.8899,
      "step": 19418
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1951019763946533,
      "learning_rate": 8.211992460786506e-06,
      "loss": 0.8503,
      "step": 19419
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6895811557769775,
      "learning_rate": 8.211353544388718e-06,
      "loss": 0.8462,
      "step": 19420
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.356835126876831,
      "learning_rate": 8.210714627990928e-06,
      "loss": 0.9729,
      "step": 19421
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7556438446044922,
      "learning_rate": 8.210075711593139e-06,
      "loss": 0.9063,
      "step": 19422
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.906606435775757,
      "learning_rate": 8.209436795195349e-06,
      "loss": 0.8309,
      "step": 19423
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.656805396080017,
      "learning_rate": 8.20879787879756e-06,
      "loss": 0.884,
      "step": 19424
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.498173236846924,
      "learning_rate": 8.20815896239977e-06,
      "loss": 0.751,
      "step": 19425
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1780598163604736,
      "learning_rate": 8.20752004600198e-06,
      "loss": 0.8669,
      "step": 19426
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0042941570281982,
      "learning_rate": 8.206881129604192e-06,
      "loss": 0.8341,
      "step": 19427
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0019967555999756,
      "learning_rate": 8.206242213206402e-06,
      "loss": 0.8593,
      "step": 19428
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8489937782287598,
      "learning_rate": 8.205603296808613e-06,
      "loss": 1.2112,
      "step": 19429
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.013784646987915,
      "learning_rate": 8.204964380410823e-06,
      "loss": 0.7598,
      "step": 19430
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.014440536499023,
      "learning_rate": 8.204325464013035e-06,
      "loss": 1.5193,
      "step": 19431
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.031243324279785,
      "learning_rate": 8.203686547615244e-06,
      "loss": 0.9026,
      "step": 19432
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9312818050384521,
      "learning_rate": 8.203047631217456e-06,
      "loss": 0.8726,
      "step": 19433
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6820244789123535,
      "learning_rate": 8.202408714819666e-06,
      "loss": 0.8509,
      "step": 19434
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.790305256843567,
      "learning_rate": 8.201769798421877e-06,
      "loss": 0.9463,
      "step": 19435
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6889681816101074,
      "learning_rate": 8.201130882024087e-06,
      "loss": 0.8927,
      "step": 19436
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7469903230667114,
      "learning_rate": 8.200491965626299e-06,
      "loss": 0.8493,
      "step": 19437
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.388650417327881,
      "learning_rate": 8.199853049228509e-06,
      "loss": 1.0862,
      "step": 19438
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2547624111175537,
      "learning_rate": 8.19921413283072e-06,
      "loss": 0.8661,
      "step": 19439
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.3353750705718994,
      "learning_rate": 8.19857521643293e-06,
      "loss": 0.8387,
      "step": 19440
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3565967082977295,
      "learning_rate": 8.197936300035142e-06,
      "loss": 0.7454,
      "step": 19441
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8480031490325928,
      "learning_rate": 8.197297383637351e-06,
      "loss": 1.011,
      "step": 19442
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3051449060440063,
      "learning_rate": 8.196658467239563e-06,
      "loss": 0.7958,
      "step": 19443
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6582720279693604,
      "learning_rate": 8.196019550841773e-06,
      "loss": 0.8246,
      "step": 19444
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.683605432510376,
      "learning_rate": 8.195380634443983e-06,
      "loss": 0.9344,
      "step": 19445
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.928091049194336,
      "learning_rate": 8.194741718046194e-06,
      "loss": 1.5313,
      "step": 19446
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.047630786895752,
      "learning_rate": 8.194102801648404e-06,
      "loss": 0.8416,
      "step": 19447
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.689759612083435,
      "learning_rate": 8.193463885250616e-06,
      "loss": 0.4056,
      "step": 19448
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.16935396194458,
      "learning_rate": 8.192824968852825e-06,
      "loss": 1.5746,
      "step": 19449
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.037182331085205,
      "learning_rate": 8.192186052455037e-06,
      "loss": 0.7183,
      "step": 19450
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.176316261291504,
      "learning_rate": 8.191547136057247e-06,
      "loss": 0.8512,
      "step": 19451
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0499839782714844,
      "learning_rate": 8.190908219659458e-06,
      "loss": 0.9198,
      "step": 19452
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6224470138549805,
      "learning_rate": 8.190269303261668e-06,
      "loss": 0.8071,
      "step": 19453
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.785583972930908,
      "learning_rate": 8.18963038686388e-06,
      "loss": 1.2294,
      "step": 19454
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7056561708450317,
      "learning_rate": 8.18899147046609e-06,
      "loss": 0.9175,
      "step": 19455
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4966920614242554,
      "learning_rate": 8.188352554068301e-06,
      "loss": 0.379,
      "step": 19456
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6339465379714966,
      "learning_rate": 8.187713637670511e-06,
      "loss": 0.8967,
      "step": 19457
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6909942626953125,
      "learning_rate": 8.187074721272723e-06,
      "loss": 0.8531,
      "step": 19458
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.934483528137207,
      "learning_rate": 8.186435804874932e-06,
      "loss": 0.9281,
      "step": 19459
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.655167579650879,
      "learning_rate": 8.185796888477144e-06,
      "loss": 1.005,
      "step": 19460
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8052836656570435,
      "learning_rate": 8.185157972079354e-06,
      "loss": 0.8689,
      "step": 19461
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4793930053710938,
      "learning_rate": 8.184519055681565e-06,
      "loss": 0.8531,
      "step": 19462
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7589021921157837,
      "learning_rate": 8.183880139283775e-06,
      "loss": 0.7771,
      "step": 19463
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.3348493576049805,
      "learning_rate": 8.183241222885985e-06,
      "loss": 0.9781,
      "step": 19464
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7621774673461914,
      "learning_rate": 8.182602306488197e-06,
      "loss": 0.8934,
      "step": 19465
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7393325567245483,
      "learning_rate": 8.181963390090407e-06,
      "loss": 0.8997,
      "step": 19466
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.7805917263031006,
      "learning_rate": 8.181324473692618e-06,
      "loss": 0.626,
      "step": 19467
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.4147775173187256,
      "learning_rate": 8.180685557294828e-06,
      "loss": 0.9338,
      "step": 19468
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0423033237457275,
      "learning_rate": 8.18004664089704e-06,
      "loss": 0.8215,
      "step": 19469
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.016568183898926,
      "learning_rate": 8.17940772449925e-06,
      "loss": 0.8963,
      "step": 19470
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7221531867980957,
      "learning_rate": 8.178768808101461e-06,
      "loss": 0.89,
      "step": 19471
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8191125392913818,
      "learning_rate": 8.17812989170367e-06,
      "loss": 0.8578,
      "step": 19472
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7900099754333496,
      "learning_rate": 8.177490975305882e-06,
      "loss": 0.6703,
      "step": 19473
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4581609964370728,
      "learning_rate": 8.176852058908092e-06,
      "loss": 0.8809,
      "step": 19474
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7564904689788818,
      "learning_rate": 8.176213142510304e-06,
      "loss": 0.8451,
      "step": 19475
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3699138164520264,
      "learning_rate": 8.175574226112514e-06,
      "loss": 0.8272,
      "step": 19476
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.401998996734619,
      "learning_rate": 8.174935309714725e-06,
      "loss": 1.0187,
      "step": 19477
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.5385208129882812,
      "learning_rate": 8.174296393316935e-06,
      "loss": 0.9894,
      "step": 19478
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1444129943847656,
      "learning_rate": 8.173657476919146e-06,
      "loss": 0.9056,
      "step": 19479
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8165984153747559,
      "learning_rate": 8.173018560521356e-06,
      "loss": 0.945,
      "step": 19480
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8207653760910034,
      "learning_rate": 8.172379644123566e-06,
      "loss": 0.921,
      "step": 19481
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4185587167739868,
      "learning_rate": 8.171740727725778e-06,
      "loss": 0.8375,
      "step": 19482
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7769367694854736,
      "learning_rate": 8.171101811327988e-06,
      "loss": 0.827,
      "step": 19483
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.868821382522583,
      "learning_rate": 8.170462894930199e-06,
      "loss": 0.9213,
      "step": 19484
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0262222290039062,
      "learning_rate": 8.169823978532409e-06,
      "loss": 1.1557,
      "step": 19485
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.898768424987793,
      "learning_rate": 8.16918506213462e-06,
      "loss": 0.9069,
      "step": 19486
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5450462102890015,
      "learning_rate": 8.16854614573683e-06,
      "loss": 0.8231,
      "step": 19487
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9280260801315308,
      "learning_rate": 8.167907229339042e-06,
      "loss": 0.7277,
      "step": 19488
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.263153314590454,
      "learning_rate": 8.167268312941252e-06,
      "loss": 0.9217,
      "step": 19489
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.161947727203369,
      "learning_rate": 8.166629396543463e-06,
      "loss": 1.2129,
      "step": 19490
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6698720455169678,
      "learning_rate": 8.165990480145673e-06,
      "loss": 0.7842,
      "step": 19491
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.026625871658325,
      "learning_rate": 8.165351563747885e-06,
      "loss": 0.9705,
      "step": 19492
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3538591861724854,
      "learning_rate": 8.164712647350095e-06,
      "loss": 0.7407,
      "step": 19493
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7372467517852783,
      "learning_rate": 8.164073730952306e-06,
      "loss": 0.8515,
      "step": 19494
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9073748588562012,
      "learning_rate": 8.163434814554516e-06,
      "loss": 0.9275,
      "step": 19495
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.448722004890442,
      "learning_rate": 8.162795898156728e-06,
      "loss": 0.476,
      "step": 19496
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.782535433769226,
      "learning_rate": 8.162156981758937e-06,
      "loss": 0.4845,
      "step": 19497
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7761856317520142,
      "learning_rate": 8.161518065361149e-06,
      "loss": 0.7838,
      "step": 19498
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.836761713027954,
      "learning_rate": 8.160879148963359e-06,
      "loss": 0.904,
      "step": 19499
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4750546216964722,
      "learning_rate": 8.160240232565569e-06,
      "loss": 0.8401,
      "step": 19500
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.268173933029175,
      "learning_rate": 8.15960131616778e-06,
      "loss": 1.1614,
      "step": 19501
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.610865831375122,
      "learning_rate": 8.15896239976999e-06,
      "loss": 0.8299,
      "step": 19502
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.936820387840271,
      "learning_rate": 8.158323483372202e-06,
      "loss": 0.9389,
      "step": 19503
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5805718898773193,
      "learning_rate": 8.157684566974411e-06,
      "loss": 0.8069,
      "step": 19504
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.53161883354187,
      "learning_rate": 8.157045650576623e-06,
      "loss": 0.7365,
      "step": 19505
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.014145851135254,
      "learning_rate": 8.156406734178833e-06,
      "loss": 1.383,
      "step": 19506
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7717547416687012,
      "learning_rate": 8.155767817781044e-06,
      "loss": 0.8911,
      "step": 19507
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8454817533493042,
      "learning_rate": 8.155128901383254e-06,
      "loss": 0.8508,
      "step": 19508
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9000678062438965,
      "learning_rate": 8.154489984985466e-06,
      "loss": 0.9677,
      "step": 19509
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.8160808086395264,
      "learning_rate": 8.153851068587676e-06,
      "loss": 1.1857,
      "step": 19510
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6474673748016357,
      "learning_rate": 8.153212152189887e-06,
      "loss": 0.9276,
      "step": 19511
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.6903815269470215,
      "learning_rate": 8.152573235792097e-06,
      "loss": 1.2353,
      "step": 19512
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6603602170944214,
      "learning_rate": 8.151934319394309e-06,
      "loss": 0.8897,
      "step": 19513
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5214308500289917,
      "learning_rate": 8.151295402996518e-06,
      "loss": 0.9432,
      "step": 19514
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.020823001861572,
      "learning_rate": 8.15065648659873e-06,
      "loss": 1.3852,
      "step": 19515
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.957688570022583,
      "learning_rate": 8.15001757020094e-06,
      "loss": 0.9381,
      "step": 19516
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.76385498046875,
      "learning_rate": 8.14937865380315e-06,
      "loss": 0.8252,
      "step": 19517
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.388556480407715,
      "learning_rate": 8.148739737405361e-06,
      "loss": 1.5325,
      "step": 19518
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.214478015899658,
      "learning_rate": 8.148100821007571e-06,
      "loss": 1.1678,
      "step": 19519
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.600549578666687,
      "learning_rate": 8.147461904609783e-06,
      "loss": 0.856,
      "step": 19520
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1086156368255615,
      "learning_rate": 8.146822988211992e-06,
      "loss": 0.9784,
      "step": 19521
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4172407388687134,
      "learning_rate": 8.146184071814204e-06,
      "loss": 0.9376,
      "step": 19522
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8193774223327637,
      "learning_rate": 8.145545155416414e-06,
      "loss": 0.726,
      "step": 19523
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3936309814453125,
      "learning_rate": 8.144906239018625e-06,
      "loss": 0.3836,
      "step": 19524
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.7417778968811035,
      "learning_rate": 8.144267322620835e-06,
      "loss": 1.4807,
      "step": 19525
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4150655269622803,
      "learning_rate": 8.143628406223047e-06,
      "loss": 0.6331,
      "step": 19526
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.492549180984497,
      "learning_rate": 8.142989489825257e-06,
      "loss": 0.4778,
      "step": 19527
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.030115842819214,
      "learning_rate": 8.142350573427468e-06,
      "loss": 0.8871,
      "step": 19528
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8493634462356567,
      "learning_rate": 8.141711657029678e-06,
      "loss": 0.8186,
      "step": 19529
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.124903202056885,
      "learning_rate": 8.14107274063189e-06,
      "loss": 1.4909,
      "step": 19530
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4672454595565796,
      "learning_rate": 8.1404338242341e-06,
      "loss": 0.8064,
      "step": 19531
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.265943765640259,
      "learning_rate": 8.139794907836311e-06,
      "loss": 0.7421,
      "step": 19532
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.085475206375122,
      "learning_rate": 8.139155991438521e-06,
      "loss": 0.7502,
      "step": 19533
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.71679162979126,
      "learning_rate": 8.138517075040732e-06,
      "loss": 1.3208,
      "step": 19534
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.917037010192871,
      "learning_rate": 8.137878158642942e-06,
      "loss": 1.5767,
      "step": 19535
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7602298259735107,
      "learning_rate": 8.137239242245152e-06,
      "loss": 0.8821,
      "step": 19536
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8761793375015259,
      "learning_rate": 8.136600325847364e-06,
      "loss": 0.7361,
      "step": 19537
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.619389533996582,
      "learning_rate": 8.135961409449574e-06,
      "loss": 0.542,
      "step": 19538
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7213187217712402,
      "learning_rate": 8.135322493051785e-06,
      "loss": 0.8201,
      "step": 19539
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8657745122909546,
      "learning_rate": 8.134683576653995e-06,
      "loss": 0.7935,
      "step": 19540
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7176896333694458,
      "learning_rate": 8.134044660256206e-06,
      "loss": 0.8584,
      "step": 19541
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.077631711959839,
      "learning_rate": 8.133405743858416e-06,
      "loss": 0.8688,
      "step": 19542
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9032998085021973,
      "learning_rate": 8.132766827460628e-06,
      "loss": 0.9596,
      "step": 19543
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.403552532196045,
      "learning_rate": 8.132127911062838e-06,
      "loss": 1.1471,
      "step": 19544
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.637242555618286,
      "learning_rate": 8.13148899466505e-06,
      "loss": 0.907,
      "step": 19545
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6486587524414062,
      "learning_rate": 8.130850078267259e-06,
      "loss": 0.9727,
      "step": 19546
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.6192212104797363,
      "learning_rate": 8.13021116186947e-06,
      "loss": 0.937,
      "step": 19547
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.104590892791748,
      "learning_rate": 8.12957224547168e-06,
      "loss": 1.0923,
      "step": 19548
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1795318126678467,
      "learning_rate": 8.128933329073892e-06,
      "loss": 0.9715,
      "step": 19549
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.580600738525391,
      "learning_rate": 8.128294412676102e-06,
      "loss": 1.4349,
      "step": 19550
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0704898834228516,
      "learning_rate": 8.127655496278313e-06,
      "loss": 0.6979,
      "step": 19551
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7681429386138916,
      "learning_rate": 8.127016579880523e-06,
      "loss": 0.8437,
      "step": 19552
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.4480390548706055,
      "learning_rate": 8.126377663482735e-06,
      "loss": 1.2527,
      "step": 19553
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7650890350341797,
      "learning_rate": 8.125738747084945e-06,
      "loss": 0.9093,
      "step": 19554
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.615972638130188,
      "learning_rate": 8.125099830687155e-06,
      "loss": 0.7862,
      "step": 19555
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.639440536499023,
      "learning_rate": 8.124460914289366e-06,
      "loss": 1.0454,
      "step": 19556
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4427082538604736,
      "learning_rate": 8.123821997891576e-06,
      "loss": 0.9747,
      "step": 19557
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8780162334442139,
      "learning_rate": 8.123183081493788e-06,
      "loss": 1.0981,
      "step": 19558
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8389592170715332,
      "learning_rate": 8.122544165095997e-06,
      "loss": 1.2032,
      "step": 19559
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.416847229003906,
      "learning_rate": 8.121905248698209e-06,
      "loss": 1.1631,
      "step": 19560
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8473179340362549,
      "learning_rate": 8.121266332300419e-06,
      "loss": 0.9048,
      "step": 19561
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3979278802871704,
      "learning_rate": 8.12062741590263e-06,
      "loss": 0.5694,
      "step": 19562
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.6257636547088623,
      "learning_rate": 8.11998849950484e-06,
      "loss": 0.7365,
      "step": 19563
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.964735269546509,
      "learning_rate": 8.119349583107052e-06,
      "loss": 1.3491,
      "step": 19564
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9455775022506714,
      "learning_rate": 8.118710666709262e-06,
      "loss": 0.8643,
      "step": 19565
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9534757137298584,
      "learning_rate": 8.118071750311473e-06,
      "loss": 1.2173,
      "step": 19566
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6663103103637695,
      "learning_rate": 8.117432833913683e-06,
      "loss": 0.8071,
      "step": 19567
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2831332683563232,
      "learning_rate": 8.116793917515895e-06,
      "loss": 0.8744,
      "step": 19568
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6751998662948608,
      "learning_rate": 8.116155001118104e-06,
      "loss": 0.3171,
      "step": 19569
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.2507972717285156,
      "learning_rate": 8.115516084720316e-06,
      "loss": 0.6043,
      "step": 19570
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5459885597229004,
      "learning_rate": 8.114877168322526e-06,
      "loss": 0.4822,
      "step": 19571
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7969701290130615,
      "learning_rate": 8.114238251924736e-06,
      "loss": 0.8827,
      "step": 19572
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2824113368988037,
      "learning_rate": 8.113599335526947e-06,
      "loss": 0.9949,
      "step": 19573
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5533004999160767,
      "learning_rate": 8.112960419129157e-06,
      "loss": 0.783,
      "step": 19574
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.977799892425537,
      "learning_rate": 8.112321502731369e-06,
      "loss": 0.7274,
      "step": 19575
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.729240894317627,
      "learning_rate": 8.111682586333578e-06,
      "loss": 0.6436,
      "step": 19576
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9491323232650757,
      "learning_rate": 8.11104366993579e-06,
      "loss": 0.8882,
      "step": 19577
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.3309743404388428,
      "learning_rate": 8.110404753538e-06,
      "loss": 1.0123,
      "step": 19578
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5268185138702393,
      "learning_rate": 8.109765837140211e-06,
      "loss": 0.8359,
      "step": 19579
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.039213180541992,
      "learning_rate": 8.109126920742421e-06,
      "loss": 0.8875,
      "step": 19580
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.237553596496582,
      "learning_rate": 8.108488004344633e-06,
      "loss": 0.9433,
      "step": 19581
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8098279237747192,
      "learning_rate": 8.107849087946843e-06,
      "loss": 1.1226,
      "step": 19582
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.863701581954956,
      "learning_rate": 8.107210171549054e-06,
      "loss": 0.8099,
      "step": 19583
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8424654006958008,
      "learning_rate": 8.106571255151264e-06,
      "loss": 0.8448,
      "step": 19584
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0423474311828613,
      "learning_rate": 8.105932338753476e-06,
      "loss": 0.8073,
      "step": 19585
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7278600931167603,
      "learning_rate": 8.105293422355685e-06,
      "loss": 0.7216,
      "step": 19586
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9095739126205444,
      "learning_rate": 8.104654505957897e-06,
      "loss": 0.8146,
      "step": 19587
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.593759536743164,
      "learning_rate": 8.104015589560107e-06,
      "loss": 0.8037,
      "step": 19588
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.653531789779663,
      "learning_rate": 8.103376673162318e-06,
      "loss": 0.7969,
      "step": 19589
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3032803535461426,
      "learning_rate": 8.102737756764528e-06,
      "loss": 0.9554,
      "step": 19590
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6778860092163086,
      "learning_rate": 8.102098840366738e-06,
      "loss": 0.8193,
      "step": 19591
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8216923475265503,
      "learning_rate": 8.10145992396895e-06,
      "loss": 0.8487,
      "step": 19592
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.551423192024231,
      "learning_rate": 8.10082100757116e-06,
      "loss": 0.8856,
      "step": 19593
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0928800106048584,
      "learning_rate": 8.100182091173371e-06,
      "loss": 1.0194,
      "step": 19594
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4184730052948,
      "learning_rate": 8.099543174775581e-06,
      "loss": 0.6602,
      "step": 19595
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.347867012023926,
      "learning_rate": 8.098904258377792e-06,
      "loss": 1.4737,
      "step": 19596
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5233880281448364,
      "learning_rate": 8.098265341980002e-06,
      "loss": 0.726,
      "step": 19597
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4958763122558594,
      "learning_rate": 8.097626425582214e-06,
      "loss": 0.824,
      "step": 19598
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.934603214263916,
      "learning_rate": 8.096987509184424e-06,
      "loss": 0.7036,
      "step": 19599
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.226369380950928,
      "learning_rate": 8.096348592786635e-06,
      "loss": 1.3607,
      "step": 19600
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7695608139038086,
      "learning_rate": 8.095709676388845e-06,
      "loss": 0.836,
      "step": 19601
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7767823934555054,
      "learning_rate": 8.095070759991057e-06,
      "loss": 0.9073,
      "step": 19602
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6096829175949097,
      "learning_rate": 8.094431843593266e-06,
      "loss": 0.4432,
      "step": 19603
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.021368980407715,
      "learning_rate": 8.093792927195478e-06,
      "loss": 0.9392,
      "step": 19604
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0495433807373047,
      "learning_rate": 8.093154010797688e-06,
      "loss": 1.1124,
      "step": 19605
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.138805866241455,
      "learning_rate": 8.0925150943999e-06,
      "loss": 0.8771,
      "step": 19606
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.391594409942627,
      "learning_rate": 8.09187617800211e-06,
      "loss": 1.209,
      "step": 19607
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9990592002868652,
      "learning_rate": 8.09123726160432e-06,
      "loss": 1.0838,
      "step": 19608
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.013186454772949,
      "learning_rate": 8.09059834520653e-06,
      "loss": 0.9009,
      "step": 19609
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8189659118652344,
      "learning_rate": 8.08995942880874e-06,
      "loss": 0.8274,
      "step": 19610
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.107992649078369,
      "learning_rate": 8.089320512410952e-06,
      "loss": 1.2726,
      "step": 19611
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.5686757564544678,
      "learning_rate": 8.088681596013162e-06,
      "loss": 1.1643,
      "step": 19612
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8335314989089966,
      "learning_rate": 8.088042679615373e-06,
      "loss": 0.9012,
      "step": 19613
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7650604248046875,
      "learning_rate": 8.087403763217583e-06,
      "loss": 0.8818,
      "step": 19614
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.3866500854492188,
      "learning_rate": 8.086764846819795e-06,
      "loss": 0.876,
      "step": 19615
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8735086917877197,
      "learning_rate": 8.086125930422005e-06,
      "loss": 0.9891,
      "step": 19616
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7446106672286987,
      "learning_rate": 8.085487014024216e-06,
      "loss": 0.7978,
      "step": 19617
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.118924140930176,
      "learning_rate": 8.084848097626426e-06,
      "loss": 1.1319,
      "step": 19618
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0327744483947754,
      "learning_rate": 8.084209181228638e-06,
      "loss": 1.2112,
      "step": 19619
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2880403995513916,
      "learning_rate": 8.083570264830848e-06,
      "loss": 0.7431,
      "step": 19620
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6133415699005127,
      "learning_rate": 8.082931348433059e-06,
      "loss": 0.8113,
      "step": 19621
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6060138940811157,
      "learning_rate": 8.082292432035269e-06,
      "loss": 0.9214,
      "step": 19622
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.815971851348877,
      "learning_rate": 8.08165351563748e-06,
      "loss": 0.8887,
      "step": 19623
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.469619631767273,
      "learning_rate": 8.08101459923969e-06,
      "loss": 0.8158,
      "step": 19624
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.044551372528076,
      "learning_rate": 8.080375682841902e-06,
      "loss": 1.5792,
      "step": 19625
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7112600803375244,
      "learning_rate": 8.079736766444112e-06,
      "loss": 0.8299,
      "step": 19626
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8614591360092163,
      "learning_rate": 8.079097850046322e-06,
      "loss": 0.8424,
      "step": 19627
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.5325300693511963,
      "learning_rate": 8.078458933648533e-06,
      "loss": 0.8421,
      "step": 19628
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9824639558792114,
      "learning_rate": 8.077820017250743e-06,
      "loss": 1.0662,
      "step": 19629
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8744629621505737,
      "learning_rate": 8.077181100852955e-06,
      "loss": 0.8668,
      "step": 19630
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9973816871643066,
      "learning_rate": 8.076542184455164e-06,
      "loss": 0.8174,
      "step": 19631
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.7723031044006348,
      "learning_rate": 8.075903268057376e-06,
      "loss": 0.9453,
      "step": 19632
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1233344078063965,
      "learning_rate": 8.075264351659586e-06,
      "loss": 1.2598,
      "step": 19633
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9949795007705688,
      "learning_rate": 8.074625435261797e-06,
      "loss": 0.8502,
      "step": 19634
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7094221115112305,
      "learning_rate": 8.073986518864007e-06,
      "loss": 0.7673,
      "step": 19635
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2809858322143555,
      "learning_rate": 8.073347602466219e-06,
      "loss": 0.9911,
      "step": 19636
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.2554407119750977,
      "learning_rate": 8.072708686068429e-06,
      "loss": 0.7585,
      "step": 19637
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8677738904953003,
      "learning_rate": 8.07206976967064e-06,
      "loss": 0.937,
      "step": 19638
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6079696416854858,
      "learning_rate": 8.07143085327285e-06,
      "loss": 0.9681,
      "step": 19639
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0488674640655518,
      "learning_rate": 8.070791936875062e-06,
      "loss": 0.9939,
      "step": 19640
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6831974983215332,
      "learning_rate": 8.070153020477271e-06,
      "loss": 0.8644,
      "step": 19641
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.721121072769165,
      "learning_rate": 8.069514104079483e-06,
      "loss": 0.8123,
      "step": 19642
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.491716384887695,
      "learning_rate": 8.068875187681693e-06,
      "loss": 1.1758,
      "step": 19643
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1320559978485107,
      "learning_rate": 8.068236271283904e-06,
      "loss": 1.1959,
      "step": 19644
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6621469259262085,
      "learning_rate": 8.067597354886114e-06,
      "loss": 0.8873,
      "step": 19645
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.913862466812134,
      "learning_rate": 8.066958438488324e-06,
      "loss": 0.9658,
      "step": 19646
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2612175941467285,
      "learning_rate": 8.066319522090536e-06,
      "loss": 0.8747,
      "step": 19647
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.914111852645874,
      "learning_rate": 8.065680605692745e-06,
      "loss": 0.8367,
      "step": 19648
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.643860340118408,
      "learning_rate": 8.065041689294957e-06,
      "loss": 1.4914,
      "step": 19649
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9524314403533936,
      "learning_rate": 8.064402772897167e-06,
      "loss": 0.8122,
      "step": 19650
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0865938663482666,
      "learning_rate": 8.063763856499378e-06,
      "loss": 1.3052,
      "step": 19651
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.710955262184143,
      "learning_rate": 8.063124940101588e-06,
      "loss": 0.8252,
      "step": 19652
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.711340069770813,
      "learning_rate": 8.0624860237038e-06,
      "loss": 0.892,
      "step": 19653
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.4775807857513428,
      "learning_rate": 8.06184710730601e-06,
      "loss": 0.9976,
      "step": 19654
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.169861316680908,
      "learning_rate": 8.061208190908221e-06,
      "loss": 0.855,
      "step": 19655
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6896955966949463,
      "learning_rate": 8.060569274510431e-06,
      "loss": 0.9183,
      "step": 19656
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0427517890930176,
      "learning_rate": 8.059930358112643e-06,
      "loss": 0.953,
      "step": 19657
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7371163368225098,
      "learning_rate": 8.059291441714852e-06,
      "loss": 0.8311,
      "step": 19658
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.761702299118042,
      "learning_rate": 8.058652525317064e-06,
      "loss": 0.872,
      "step": 19659
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.729910135269165,
      "learning_rate": 8.058013608919274e-06,
      "loss": 1.0518,
      "step": 19660
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.033212900161743,
      "learning_rate": 8.057374692521485e-06,
      "loss": 0.883,
      "step": 19661
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.6388797760009766,
      "learning_rate": 8.056735776123695e-06,
      "loss": 0.7821,
      "step": 19662
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.649843692779541,
      "learning_rate": 8.056096859725905e-06,
      "loss": 1.316,
      "step": 19663
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.588781476020813,
      "learning_rate": 8.055457943328117e-06,
      "loss": 0.8855,
      "step": 19664
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1859843730926514,
      "learning_rate": 8.054819026930326e-06,
      "loss": 0.7888,
      "step": 19665
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.837933897972107,
      "learning_rate": 8.054180110532538e-06,
      "loss": 0.7621,
      "step": 19666
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0456910133361816,
      "learning_rate": 8.053541194134748e-06,
      "loss": 0.9561,
      "step": 19667
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9245967864990234,
      "learning_rate": 8.05290227773696e-06,
      "loss": 0.902,
      "step": 19668
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.689335823059082,
      "learning_rate": 8.05226336133917e-06,
      "loss": 0.9442,
      "step": 19669
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9873394966125488,
      "learning_rate": 8.05162444494138e-06,
      "loss": 0.7896,
      "step": 19670
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.170757293701172,
      "learning_rate": 8.05098552854359e-06,
      "loss": 0.7387,
      "step": 19671
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8594257831573486,
      "learning_rate": 8.050346612145802e-06,
      "loss": 0.8476,
      "step": 19672
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.040252447128296,
      "learning_rate": 8.049707695748012e-06,
      "loss": 0.8789,
      "step": 19673
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6054372787475586,
      "learning_rate": 8.049068779350224e-06,
      "loss": 0.9522,
      "step": 19674
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.154500722885132,
      "learning_rate": 8.048429862952433e-06,
      "loss": 0.8888,
      "step": 19675
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.1029012203216553,
      "learning_rate": 8.047790946554645e-06,
      "loss": 0.5962,
      "step": 19676
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.508315086364746,
      "learning_rate": 8.047152030156855e-06,
      "loss": 1.0691,
      "step": 19677
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6445112228393555,
      "learning_rate": 8.046513113759066e-06,
      "loss": 0.9437,
      "step": 19678
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.799908995628357,
      "learning_rate": 8.045874197361276e-06,
      "loss": 0.8129,
      "step": 19679
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6414172649383545,
      "learning_rate": 8.045235280963488e-06,
      "loss": 0.7271,
      "step": 19680
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.7207748889923096,
      "learning_rate": 8.044596364565698e-06,
      "loss": 0.851,
      "step": 19681
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3975754976272583,
      "learning_rate": 8.043957448167908e-06,
      "loss": 0.4782,
      "step": 19682
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7824913263320923,
      "learning_rate": 8.043318531770119e-06,
      "loss": 0.7548,
      "step": 19683
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.6116690635681152,
      "learning_rate": 8.042679615372329e-06,
      "loss": 1.406,
      "step": 19684
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.704831600189209,
      "learning_rate": 8.04204069897454e-06,
      "loss": 0.9929,
      "step": 19685
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7816523313522339,
      "learning_rate": 8.04140178257675e-06,
      "loss": 0.8264,
      "step": 19686
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.003843069076538,
      "learning_rate": 8.040762866178962e-06,
      "loss": 0.9893,
      "step": 19687
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.92811918258667,
      "learning_rate": 8.040123949781172e-06,
      "loss": 0.8807,
      "step": 19688
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.2417054176330566,
      "learning_rate": 8.039485033383383e-06,
      "loss": 1.0587,
      "step": 19689
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6444107294082642,
      "learning_rate": 8.038846116985593e-06,
      "loss": 0.9571,
      "step": 19690
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6082019805908203,
      "learning_rate": 8.038207200587805e-06,
      "loss": 0.8771,
      "step": 19691
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3752570152282715,
      "learning_rate": 8.037568284190015e-06,
      "loss": 0.8846,
      "step": 19692
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.552540898323059,
      "learning_rate": 8.036929367792226e-06,
      "loss": 0.8376,
      "step": 19693
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.257094144821167,
      "learning_rate": 8.036290451394436e-06,
      "loss": 0.9207,
      "step": 19694
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7998039722442627,
      "learning_rate": 8.035651534996647e-06,
      "loss": 0.8233,
      "step": 19695
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.597099781036377,
      "learning_rate": 8.035012618598857e-06,
      "loss": 1.1977,
      "step": 19696
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.828513741493225,
      "learning_rate": 8.034373702201069e-06,
      "loss": 0.81,
      "step": 19697
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7754651308059692,
      "learning_rate": 8.033734785803279e-06,
      "loss": 0.8857,
      "step": 19698
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3508036136627197,
      "learning_rate": 8.03309586940549e-06,
      "loss": 0.4292,
      "step": 19699
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4638969898223877,
      "learning_rate": 8.0324569530077e-06,
      "loss": 0.8381,
      "step": 19700
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8454718589782715,
      "learning_rate": 8.03181803660991e-06,
      "loss": 0.8884,
      "step": 19701
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6234972476959229,
      "learning_rate": 8.031179120212122e-06,
      "loss": 0.8106,
      "step": 19702
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.60606050491333,
      "learning_rate": 8.030540203814331e-06,
      "loss": 0.8737,
      "step": 19703
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1247620582580566,
      "learning_rate": 8.029901287416543e-06,
      "loss": 0.9001,
      "step": 19704
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1595094203948975,
      "learning_rate": 8.029262371018753e-06,
      "loss": 1.2919,
      "step": 19705
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2880159616470337,
      "learning_rate": 8.028623454620964e-06,
      "loss": 0.4805,
      "step": 19706
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.235205173492432,
      "learning_rate": 8.027984538223174e-06,
      "loss": 1.2966,
      "step": 19707
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2534406185150146,
      "learning_rate": 8.027345621825386e-06,
      "loss": 0.872,
      "step": 19708
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.561328411102295,
      "learning_rate": 8.026706705427596e-06,
      "loss": 0.6527,
      "step": 19709
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9403276443481445,
      "learning_rate": 8.026067789029807e-06,
      "loss": 0.8535,
      "step": 19710
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6602450609207153,
      "learning_rate": 8.025428872632017e-06,
      "loss": 0.8718,
      "step": 19711
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.3627278804779053,
      "learning_rate": 8.024789956234229e-06,
      "loss": 0.8613,
      "step": 19712
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.811938285827637,
      "learning_rate": 8.024151039836437e-06,
      "loss": 1.6325,
      "step": 19713
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5357822179794312,
      "learning_rate": 8.023512123438648e-06,
      "loss": 0.7444,
      "step": 19714
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.514460802078247,
      "learning_rate": 8.022873207040858e-06,
      "loss": 0.8779,
      "step": 19715
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6548925638198853,
      "learning_rate": 8.02223429064307e-06,
      "loss": 0.77,
      "step": 19716
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.02007794380188,
      "learning_rate": 8.02159537424528e-06,
      "loss": 0.876,
      "step": 19717
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.34786319732666,
      "learning_rate": 8.020956457847491e-06,
      "loss": 0.975,
      "step": 19718
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.327968597412109,
      "learning_rate": 8.020317541449701e-06,
      "loss": 1.0307,
      "step": 19719
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.565493106842041,
      "learning_rate": 8.019678625051912e-06,
      "loss": 0.6888,
      "step": 19720
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6186429262161255,
      "learning_rate": 8.019039708654122e-06,
      "loss": 1.2267,
      "step": 19721
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.53534996509552,
      "learning_rate": 8.018400792256334e-06,
      "loss": 0.3888,
      "step": 19722
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.960430383682251,
      "learning_rate": 8.017761875858544e-06,
      "loss": 0.7777,
      "step": 19723
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4803160429000854,
      "learning_rate": 8.017122959460755e-06,
      "loss": 0.7765,
      "step": 19724
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1062252521514893,
      "learning_rate": 8.016484043062965e-06,
      "loss": 0.9027,
      "step": 19725
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6913238763809204,
      "learning_rate": 8.015845126665177e-06,
      "loss": 0.9155,
      "step": 19726
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2532427310943604,
      "learning_rate": 8.015206210267387e-06,
      "loss": 0.9776,
      "step": 19727
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7557786703109741,
      "learning_rate": 8.014567293869598e-06,
      "loss": 0.8169,
      "step": 19728
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9151719808578491,
      "learning_rate": 8.013928377471808e-06,
      "loss": 0.8784,
      "step": 19729
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.720149040222168,
      "learning_rate": 8.01328946107402e-06,
      "loss": 0.9083,
      "step": 19730
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.443650245666504,
      "learning_rate": 8.01265054467623e-06,
      "loss": 0.7854,
      "step": 19731
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0961246490478516,
      "learning_rate": 8.01201162827844e-06,
      "loss": 1.0981,
      "step": 19732
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7326931953430176,
      "learning_rate": 8.01137271188065e-06,
      "loss": 0.8851,
      "step": 19733
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4891380071640015,
      "learning_rate": 8.01073379548286e-06,
      "loss": 0.4623,
      "step": 19734
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5973371267318726,
      "learning_rate": 8.010094879085072e-06,
      "loss": 0.7615,
      "step": 19735
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0732719898223877,
      "learning_rate": 8.009455962687282e-06,
      "loss": 0.859,
      "step": 19736
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.695436477661133,
      "learning_rate": 8.008817046289494e-06,
      "loss": 0.7059,
      "step": 19737
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7106636762619019,
      "learning_rate": 8.008178129891703e-06,
      "loss": 0.7912,
      "step": 19738
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9821252822875977,
      "learning_rate": 8.007539213493915e-06,
      "loss": 0.8188,
      "step": 19739
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8548145294189453,
      "learning_rate": 8.006900297096125e-06,
      "loss": 0.874,
      "step": 19740
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2148208618164062,
      "learning_rate": 8.006261380698336e-06,
      "loss": 0.9743,
      "step": 19741
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6159697771072388,
      "learning_rate": 8.005622464300546e-06,
      "loss": 0.828,
      "step": 19742
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5508842468261719,
      "learning_rate": 8.004983547902758e-06,
      "loss": 0.8015,
      "step": 19743
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2064497470855713,
      "learning_rate": 8.004344631504968e-06,
      "loss": 0.77,
      "step": 19744
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.911400318145752,
      "learning_rate": 8.003705715107179e-06,
      "loss": 0.9034,
      "step": 19745
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8825534582138062,
      "learning_rate": 8.003066798709389e-06,
      "loss": 0.9734,
      "step": 19746
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.212661027908325,
      "learning_rate": 8.0024278823116e-06,
      "loss": 1.0167,
      "step": 19747
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1787681579589844,
      "learning_rate": 8.00178896591381e-06,
      "loss": 0.9392,
      "step": 19748
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0532984733581543,
      "learning_rate": 8.00115004951602e-06,
      "loss": 0.6033,
      "step": 19749
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6902930736541748,
      "learning_rate": 8.000511133118232e-06,
      "loss": 0.9024,
      "step": 19750
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5624161958694458,
      "learning_rate": 7.999872216720442e-06,
      "loss": 0.8038,
      "step": 19751
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.905945062637329,
      "learning_rate": 7.999233300322653e-06,
      "loss": 0.7694,
      "step": 19752
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8451604843139648,
      "learning_rate": 7.998594383924863e-06,
      "loss": 1.173,
      "step": 19753
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.5068247318267822,
      "learning_rate": 7.997955467527075e-06,
      "loss": 0.7843,
      "step": 19754
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5598410367965698,
      "learning_rate": 7.997316551129284e-06,
      "loss": 0.3679,
      "step": 19755
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.390175461769104,
      "learning_rate": 7.996677634731496e-06,
      "loss": 0.4483,
      "step": 19756
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0311288833618164,
      "learning_rate": 7.996038718333706e-06,
      "loss": 0.9143,
      "step": 19757
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2474887371063232,
      "learning_rate": 7.995399801935917e-06,
      "loss": 0.9831,
      "step": 19758
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.288710832595825,
      "learning_rate": 7.994760885538127e-06,
      "loss": 0.9627,
      "step": 19759
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5726550817489624,
      "learning_rate": 7.994121969140339e-06,
      "loss": 0.8543,
      "step": 19760
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4133354425430298,
      "learning_rate": 7.993483052742549e-06,
      "loss": 0.836,
      "step": 19761
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9264174699783325,
      "learning_rate": 7.99284413634476e-06,
      "loss": 0.8462,
      "step": 19762
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4025543928146362,
      "learning_rate": 7.99220521994697e-06,
      "loss": 0.7664,
      "step": 19763
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.98207426071167,
      "learning_rate": 7.991566303549182e-06,
      "loss": 1.1621,
      "step": 19764
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8325932025909424,
      "learning_rate": 7.990927387151391e-06,
      "loss": 0.8678,
      "step": 19765
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0850577354431152,
      "learning_rate": 7.990288470753603e-06,
      "loss": 0.9848,
      "step": 19766
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.555713176727295,
      "learning_rate": 7.989649554355813e-06,
      "loss": 1.24,
      "step": 19767
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.749936819076538,
      "learning_rate": 7.989010637958023e-06,
      "loss": 0.7768,
      "step": 19768
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.309954881668091,
      "learning_rate": 7.988371721560234e-06,
      "loss": 0.8497,
      "step": 19769
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.941906452178955,
      "learning_rate": 7.987732805162444e-06,
      "loss": 0.8207,
      "step": 19770
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6300686597824097,
      "learning_rate": 7.987093888764656e-06,
      "loss": 0.866,
      "step": 19771
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.708736777305603,
      "learning_rate": 7.986454972366865e-06,
      "loss": 0.8898,
      "step": 19772
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2199153900146484,
      "learning_rate": 7.985816055969077e-06,
      "loss": 1.0783,
      "step": 19773
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7211886644363403,
      "learning_rate": 7.985177139571287e-06,
      "loss": 0.8862,
      "step": 19774
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.015141248703003,
      "learning_rate": 7.984538223173498e-06,
      "loss": 1.1885,
      "step": 19775
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7448862791061401,
      "learning_rate": 7.983899306775708e-06,
      "loss": 0.8901,
      "step": 19776
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.707826018333435,
      "learning_rate": 7.98326039037792e-06,
      "loss": 0.8163,
      "step": 19777
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7630265951156616,
      "learning_rate": 7.98262147398013e-06,
      "loss": 0.7966,
      "step": 19778
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.303811550140381,
      "learning_rate": 7.981982557582341e-06,
      "loss": 0.9664,
      "step": 19779
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0460598468780518,
      "learning_rate": 7.981343641184551e-06,
      "loss": 1.0568,
      "step": 19780
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9780410528182983,
      "learning_rate": 7.980704724786763e-06,
      "loss": 0.8439,
      "step": 19781
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.120758533477783,
      "learning_rate": 7.980065808388972e-06,
      "loss": 0.7754,
      "step": 19782
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.690476655960083,
      "learning_rate": 7.979426891991184e-06,
      "loss": 0.5174,
      "step": 19783
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.4679908752441406,
      "learning_rate": 7.978787975593394e-06,
      "loss": 1.4223,
      "step": 19784
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.811339378356934,
      "learning_rate": 7.978149059195605e-06,
      "loss": 1.1167,
      "step": 19785
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.835601806640625,
      "learning_rate": 7.977510142797815e-06,
      "loss": 0.9253,
      "step": 19786
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.0466132164001465,
      "learning_rate": 7.976871226400025e-06,
      "loss": 1.0173,
      "step": 19787
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.028801918029785,
      "learning_rate": 7.976232310002237e-06,
      "loss": 0.9358,
      "step": 19788
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.1970858573913574,
      "learning_rate": 7.975593393604447e-06,
      "loss": 0.5611,
      "step": 19789
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.424289226531982,
      "learning_rate": 7.974954477206658e-06,
      "loss": 1.3273,
      "step": 19790
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5833640098571777,
      "learning_rate": 7.974315560808868e-06,
      "loss": 0.9025,
      "step": 19791
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6293628215789795,
      "learning_rate": 7.97367664441108e-06,
      "loss": 0.7846,
      "step": 19792
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.851443886756897,
      "learning_rate": 7.97303772801329e-06,
      "loss": 1.2858,
      "step": 19793
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2178640365600586,
      "learning_rate": 7.972398811615501e-06,
      "loss": 0.9295,
      "step": 19794
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.286448955535889,
      "learning_rate": 7.97175989521771e-06,
      "loss": 1.2434,
      "step": 19795
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.494284749031067,
      "learning_rate": 7.971120978819922e-06,
      "loss": 0.7588,
      "step": 19796
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.6263277530670166,
      "learning_rate": 7.970482062422132e-06,
      "loss": 1.0604,
      "step": 19797
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.241246223449707,
      "learning_rate": 7.969843146024344e-06,
      "loss": 1.232,
      "step": 19798
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9168040752410889,
      "learning_rate": 7.969204229626554e-06,
      "loss": 0.8702,
      "step": 19799
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7623445987701416,
      "learning_rate": 7.968565313228765e-06,
      "loss": 0.7901,
      "step": 19800
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5826282501220703,
      "learning_rate": 7.967926396830975e-06,
      "loss": 0.7772,
      "step": 19801
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8311831951141357,
      "learning_rate": 7.967287480433186e-06,
      "loss": 1.3363,
      "step": 19802
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5811731815338135,
      "learning_rate": 7.966648564035396e-06,
      "loss": 0.9085,
      "step": 19803
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.099320888519287,
      "learning_rate": 7.966009647637606e-06,
      "loss": 1.1236,
      "step": 19804
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5166759490966797,
      "learning_rate": 7.965370731239818e-06,
      "loss": 0.7864,
      "step": 19805
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.767536163330078,
      "learning_rate": 7.964731814842028e-06,
      "loss": 1.1135,
      "step": 19806
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.126708507537842,
      "learning_rate": 7.964092898444239e-06,
      "loss": 1.4638,
      "step": 19807
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0646398067474365,
      "learning_rate": 7.963453982046449e-06,
      "loss": 1.077,
      "step": 19808
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4905821084976196,
      "learning_rate": 7.96281506564866e-06,
      "loss": 0.4437,
      "step": 19809
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.031606674194336,
      "learning_rate": 7.96217614925087e-06,
      "loss": 1.0704,
      "step": 19810
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8550434112548828,
      "learning_rate": 7.961537232853082e-06,
      "loss": 0.8456,
      "step": 19811
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9341975450515747,
      "learning_rate": 7.960898316455292e-06,
      "loss": 0.7715,
      "step": 19812
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.585348129272461,
      "learning_rate": 7.960259400057503e-06,
      "loss": 0.9206,
      "step": 19813
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8595267534255981,
      "learning_rate": 7.959620483659713e-06,
      "loss": 0.8601,
      "step": 19814
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7969698905944824,
      "learning_rate": 7.958981567261925e-06,
      "loss": 0.8743,
      "step": 19815
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6820393800735474,
      "learning_rate": 7.958342650864135e-06,
      "loss": 0.9635,
      "step": 19816
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4028992652893066,
      "learning_rate": 7.957703734466346e-06,
      "loss": 0.8946,
      "step": 19817
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9932340383529663,
      "learning_rate": 7.957064818068556e-06,
      "loss": 0.9044,
      "step": 19818
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.295347690582275,
      "learning_rate": 7.956425901670768e-06,
      "loss": 1.0378,
      "step": 19819
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7745076417922974,
      "learning_rate": 7.955786985272977e-06,
      "loss": 0.8295,
      "step": 19820
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4851043224334717,
      "learning_rate": 7.955148068875189e-06,
      "loss": 0.961,
      "step": 19821
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.3876140117645264,
      "learning_rate": 7.954509152477399e-06,
      "loss": 0.8822,
      "step": 19822
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9349350929260254,
      "learning_rate": 7.953870236079609e-06,
      "loss": 0.922,
      "step": 19823
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6102626323699951,
      "learning_rate": 7.95323131968182e-06,
      "loss": 0.7896,
      "step": 19824
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1931753158569336,
      "learning_rate": 7.95259240328403e-06,
      "loss": 1.0319,
      "step": 19825
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2501583099365234,
      "learning_rate": 7.951953486886242e-06,
      "loss": 0.9885,
      "step": 19826
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.45835542678833,
      "learning_rate": 7.951314570488451e-06,
      "loss": 1.3071,
      "step": 19827
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.192209243774414,
      "learning_rate": 7.950675654090663e-06,
      "loss": 1.0561,
      "step": 19828
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.047950267791748,
      "learning_rate": 7.950036737692873e-06,
      "loss": 0.8208,
      "step": 19829
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1143908500671387,
      "learning_rate": 7.949397821295084e-06,
      "loss": 0.991,
      "step": 19830
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.651334524154663,
      "learning_rate": 7.948758904897294e-06,
      "loss": 0.8784,
      "step": 19831
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2533767223358154,
      "learning_rate": 7.948119988499506e-06,
      "loss": 0.5112,
      "step": 19832
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7555041313171387,
      "learning_rate": 7.947481072101716e-06,
      "loss": 0.8416,
      "step": 19833
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5716131925582886,
      "learning_rate": 7.946842155703927e-06,
      "loss": 0.8346,
      "step": 19834
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6617039442062378,
      "learning_rate": 7.946203239306137e-06,
      "loss": 0.8951,
      "step": 19835
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.2127528190612793,
      "learning_rate": 7.945564322908349e-06,
      "loss": 0.8263,
      "step": 19836
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.696295976638794,
      "learning_rate": 7.944925406510558e-06,
      "loss": 0.8107,
      "step": 19837
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8700618743896484,
      "learning_rate": 7.94428649011277e-06,
      "loss": 0.907,
      "step": 19838
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.696613073348999,
      "learning_rate": 7.94364757371498e-06,
      "loss": 1.006,
      "step": 19839
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.473394513130188,
      "learning_rate": 7.94300865731719e-06,
      "loss": 0.8477,
      "step": 19840
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9167832136154175,
      "learning_rate": 7.942369740919401e-06,
      "loss": 0.9751,
      "step": 19841
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4369664192199707,
      "learning_rate": 7.941730824521611e-06,
      "loss": 0.3906,
      "step": 19842
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7857873439788818,
      "learning_rate": 7.941091908123823e-06,
      "loss": 0.8681,
      "step": 19843
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5592451095581055,
      "learning_rate": 7.940452991726032e-06,
      "loss": 0.8794,
      "step": 19844
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8267134428024292,
      "learning_rate": 7.939814075328244e-06,
      "loss": 0.6249,
      "step": 19845
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.613102674484253,
      "learning_rate": 7.939175158930454e-06,
      "loss": 0.8996,
      "step": 19846
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7852436304092407,
      "learning_rate": 7.938536242532665e-06,
      "loss": 0.8904,
      "step": 19847
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.1816344261169434,
      "learning_rate": 7.937897326134875e-06,
      "loss": 1.2183,
      "step": 19848
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9557605981826782,
      "learning_rate": 7.937258409737087e-06,
      "loss": 0.9114,
      "step": 19849
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.985466718673706,
      "learning_rate": 7.936619493339297e-06,
      "loss": 1.0049,
      "step": 19850
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.167369842529297,
      "learning_rate": 7.935980576941508e-06,
      "loss": 0.9974,
      "step": 19851
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.939365029335022,
      "learning_rate": 7.935341660543718e-06,
      "loss": 0.7941,
      "step": 19852
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2735912799835205,
      "learning_rate": 7.93470274414593e-06,
      "loss": 0.773,
      "step": 19853
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2244091033935547,
      "learning_rate": 7.93406382774814e-06,
      "loss": 0.9762,
      "step": 19854
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.682185411453247,
      "learning_rate": 7.933424911350351e-06,
      "loss": 0.8661,
      "step": 19855
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0882627964019775,
      "learning_rate": 7.932785994952561e-06,
      "loss": 0.9658,
      "step": 19856
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.532437562942505,
      "learning_rate": 7.932147078554772e-06,
      "loss": 1.01,
      "step": 19857
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6802375316619873,
      "learning_rate": 7.931508162156982e-06,
      "loss": 0.873,
      "step": 19858
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.753690719604492,
      "learning_rate": 7.930869245759192e-06,
      "loss": 0.9284,
      "step": 19859
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7801376581192017,
      "learning_rate": 7.930230329361404e-06,
      "loss": 0.9383,
      "step": 19860
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7335128784179688,
      "learning_rate": 7.929591412963614e-06,
      "loss": 0.9816,
      "step": 19861
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7077980041503906,
      "learning_rate": 7.928952496565825e-06,
      "loss": 0.85,
      "step": 19862
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4007534980773926,
      "learning_rate": 7.928313580168035e-06,
      "loss": 0.8566,
      "step": 19863
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8926646709442139,
      "learning_rate": 7.927674663770246e-06,
      "loss": 1.0898,
      "step": 19864
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2625553607940674,
      "learning_rate": 7.927035747372456e-06,
      "loss": 1.281,
      "step": 19865
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7567896842956543,
      "learning_rate": 7.926396830974668e-06,
      "loss": 0.9749,
      "step": 19866
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.708206057548523,
      "learning_rate": 7.925757914576878e-06,
      "loss": 0.8999,
      "step": 19867
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6696065664291382,
      "learning_rate": 7.92511899817909e-06,
      "loss": 0.9597,
      "step": 19868
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.543704509735107,
      "learning_rate": 7.924480081781299e-06,
      "loss": 1.4972,
      "step": 19869
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6976864337921143,
      "learning_rate": 7.92384116538351e-06,
      "loss": 0.7853,
      "step": 19870
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7373697757720947,
      "learning_rate": 7.92320224898572e-06,
      "loss": 0.8481,
      "step": 19871
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.5314626693725586,
      "learning_rate": 7.922563332587932e-06,
      "loss": 0.6919,
      "step": 19872
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.343921184539795,
      "learning_rate": 7.921924416190142e-06,
      "loss": 0.7277,
      "step": 19873
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8690900802612305,
      "learning_rate": 7.921285499792353e-06,
      "loss": 0.8232,
      "step": 19874
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.001129388809204,
      "learning_rate": 7.920646583394563e-06,
      "loss": 0.8583,
      "step": 19875
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5878528356552124,
      "learning_rate": 7.920007666996775e-06,
      "loss": 0.9913,
      "step": 19876
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8337100744247437,
      "learning_rate": 7.919368750598985e-06,
      "loss": 0.8239,
      "step": 19877
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.895097255706787,
      "learning_rate": 7.918729834201195e-06,
      "loss": 1.4041,
      "step": 19878
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.881442666053772,
      "learning_rate": 7.918090917803406e-06,
      "loss": 0.9495,
      "step": 19879
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.419798493385315,
      "learning_rate": 7.917452001405616e-06,
      "loss": 0.8935,
      "step": 19880
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.732964038848877,
      "learning_rate": 7.916813085007828e-06,
      "loss": 0.6966,
      "step": 19881
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3424959182739258,
      "learning_rate": 7.916174168610037e-06,
      "loss": 0.3904,
      "step": 19882
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1205711364746094,
      "learning_rate": 7.915535252212249e-06,
      "loss": 0.9114,
      "step": 19883
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.328955054283142,
      "learning_rate": 7.914896335814459e-06,
      "loss": 0.4254,
      "step": 19884
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9339267015457153,
      "learning_rate": 7.91425741941667e-06,
      "loss": 0.8176,
      "step": 19885
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9595296382904053,
      "learning_rate": 7.91361850301888e-06,
      "loss": 0.8744,
      "step": 19886
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.179030179977417,
      "learning_rate": 7.912979586621092e-06,
      "loss": 0.9374,
      "step": 19887
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.1330435276031494,
      "learning_rate": 7.912340670223302e-06,
      "loss": 0.7752,
      "step": 19888
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.814422607421875,
      "learning_rate": 7.911701753825513e-06,
      "loss": 0.8492,
      "step": 19889
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7986191511154175,
      "learning_rate": 7.911062837427723e-06,
      "loss": 1.1479,
      "step": 19890
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.877038598060608,
      "learning_rate": 7.910423921029935e-06,
      "loss": 0.9026,
      "step": 19891
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5689365863800049,
      "learning_rate": 7.909785004632144e-06,
      "loss": 0.8039,
      "step": 19892
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9677507877349854,
      "learning_rate": 7.909146088234356e-06,
      "loss": 0.899,
      "step": 19893
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8842904567718506,
      "learning_rate": 7.908507171836566e-06,
      "loss": 1.0829,
      "step": 19894
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.015949249267578,
      "learning_rate": 7.907868255438776e-06,
      "loss": 1.4566,
      "step": 19895
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9434192180633545,
      "learning_rate": 7.907229339040987e-06,
      "loss": 0.9411,
      "step": 19896
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6489064693450928,
      "learning_rate": 7.906590422643197e-06,
      "loss": 0.7803,
      "step": 19897
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.7804315090179443,
      "learning_rate": 7.905951506245409e-06,
      "loss": 0.8979,
      "step": 19898
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6096574068069458,
      "learning_rate": 7.905312589847618e-06,
      "loss": 0.918,
      "step": 19899
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9676424264907837,
      "learning_rate": 7.90467367344983e-06,
      "loss": 1.0545,
      "step": 19900
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1852657794952393,
      "learning_rate": 7.90403475705204e-06,
      "loss": 0.8949,
      "step": 19901
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5518946647644043,
      "learning_rate": 7.903395840654251e-06,
      "loss": 0.8485,
      "step": 19902
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2247352600097656,
      "learning_rate": 7.902756924256461e-06,
      "loss": 0.6104,
      "step": 19903
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7859718799591064,
      "learning_rate": 7.902118007858673e-06,
      "loss": 0.9262,
      "step": 19904
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9884761571884155,
      "learning_rate": 7.901479091460883e-06,
      "loss": 0.7848,
      "step": 19905
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1319315433502197,
      "learning_rate": 7.900840175063094e-06,
      "loss": 1.1472,
      "step": 19906
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.776669979095459,
      "learning_rate": 7.900201258665304e-06,
      "loss": 1.1401,
      "step": 19907
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8213353157043457,
      "learning_rate": 7.899562342267516e-06,
      "loss": 1.0154,
      "step": 19908
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.981050491333008,
      "learning_rate": 7.898923425869725e-06,
      "loss": 0.9198,
      "step": 19909
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.064335823059082,
      "learning_rate": 7.898284509471937e-06,
      "loss": 1.5203,
      "step": 19910
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.749359130859375,
      "learning_rate": 7.897645593074147e-06,
      "loss": 0.9472,
      "step": 19911
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.525609254837036,
      "learning_rate": 7.897006676676358e-06,
      "loss": 1.0358,
      "step": 19912
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5016483068466187,
      "learning_rate": 7.896367760278568e-06,
      "loss": 0.7103,
      "step": 19913
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6117550134658813,
      "learning_rate": 7.895728843880778e-06,
      "loss": 0.9552,
      "step": 19914
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.364089250564575,
      "learning_rate": 7.89508992748299e-06,
      "loss": 0.5453,
      "step": 19915
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4280750751495361,
      "learning_rate": 7.8944510110852e-06,
      "loss": 0.7772,
      "step": 19916
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6829791069030762,
      "learning_rate": 7.893812094687411e-06,
      "loss": 0.9258,
      "step": 19917
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1803431510925293,
      "learning_rate": 7.893173178289621e-06,
      "loss": 0.9896,
      "step": 19918
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8868210315704346,
      "learning_rate": 7.892534261891832e-06,
      "loss": 0.7735,
      "step": 19919
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9687018394470215,
      "learning_rate": 7.891895345494042e-06,
      "loss": 1.3993,
      "step": 19920
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6905393600463867,
      "learning_rate": 7.891256429096254e-06,
      "loss": 0.9705,
      "step": 19921
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6434636116027832,
      "learning_rate": 7.890617512698464e-06,
      "loss": 0.8104,
      "step": 19922
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.992599606513977,
      "learning_rate": 7.889978596300675e-06,
      "loss": 0.8576,
      "step": 19923
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9888246059417725,
      "learning_rate": 7.889339679902885e-06,
      "loss": 0.9021,
      "step": 19924
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.689028024673462,
      "learning_rate": 7.888700763505097e-06,
      "loss": 0.7362,
      "step": 19925
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9102123975753784,
      "learning_rate": 7.888061847107306e-06,
      "loss": 0.7878,
      "step": 19926
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8598062992095947,
      "learning_rate": 7.887422930709518e-06,
      "loss": 0.8553,
      "step": 19927
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.919060230255127,
      "learning_rate": 7.886784014311728e-06,
      "loss": 0.7937,
      "step": 19928
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.343437433242798,
      "learning_rate": 7.88614509791394e-06,
      "loss": 1.0016,
      "step": 19929
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0313327312469482,
      "learning_rate": 7.88550618151615e-06,
      "loss": 0.8893,
      "step": 19930
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.784474849700928,
      "learning_rate": 7.88486726511836e-06,
      "loss": 1.462,
      "step": 19931
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.3864777088165283,
      "learning_rate": 7.88422834872057e-06,
      "loss": 0.7346,
      "step": 19932
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8168576955795288,
      "learning_rate": 7.88358943232278e-06,
      "loss": 0.9277,
      "step": 19933
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2266626358032227,
      "learning_rate": 7.882950515924992e-06,
      "loss": 0.9445,
      "step": 19934
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9199087619781494,
      "learning_rate": 7.882311599527202e-06,
      "loss": 0.8868,
      "step": 19935
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.144686698913574,
      "learning_rate": 7.881672683129413e-06,
      "loss": 0.893,
      "step": 19936
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5615513324737549,
      "learning_rate": 7.881033766731623e-06,
      "loss": 0.8384,
      "step": 19937
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5265108346939087,
      "learning_rate": 7.880394850333835e-06,
      "loss": 0.8535,
      "step": 19938
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.372326850891113,
      "learning_rate": 7.879755933936045e-06,
      "loss": 0.9087,
      "step": 19939
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6489813327789307,
      "learning_rate": 7.879117017538256e-06,
      "loss": 0.8389,
      "step": 19940
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.016575813293457,
      "learning_rate": 7.878478101140466e-06,
      "loss": 0.9018,
      "step": 19941
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8194997310638428,
      "learning_rate": 7.877839184742678e-06,
      "loss": 0.9293,
      "step": 19942
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8545366525650024,
      "learning_rate": 7.877200268344888e-06,
      "loss": 0.8931,
      "step": 19943
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9082043170928955,
      "learning_rate": 7.876561351947099e-06,
      "loss": 0.9652,
      "step": 19944
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.043796062469482,
      "learning_rate": 7.875922435549309e-06,
      "loss": 0.9304,
      "step": 19945
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9535634517669678,
      "learning_rate": 7.87528351915152e-06,
      "loss": 1.1692,
      "step": 19946
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4548051357269287,
      "learning_rate": 7.87464460275373e-06,
      "loss": 0.9638,
      "step": 19947
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.298478126525879,
      "learning_rate": 7.874005686355942e-06,
      "loss": 0.6639,
      "step": 19948
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9978309869766235,
      "learning_rate": 7.873366769958152e-06,
      "loss": 0.7355,
      "step": 19949
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0061824321746826,
      "learning_rate": 7.872727853560362e-06,
      "loss": 0.8346,
      "step": 19950
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8883402347564697,
      "learning_rate": 7.872088937162573e-06,
      "loss": 0.9114,
      "step": 19951
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.5978610515594482,
      "learning_rate": 7.871450020764783e-06,
      "loss": 0.8722,
      "step": 19952
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.753704309463501,
      "learning_rate": 7.870811104366995e-06,
      "loss": 0.5124,
      "step": 19953
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6351138353347778,
      "learning_rate": 7.870172187969204e-06,
      "loss": 0.7998,
      "step": 19954
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.106078624725342,
      "learning_rate": 7.869533271571416e-06,
      "loss": 1.3392,
      "step": 19955
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8894330263137817,
      "learning_rate": 7.868894355173626e-06,
      "loss": 0.8972,
      "step": 19956
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4045501947402954,
      "learning_rate": 7.868255438775837e-06,
      "loss": 0.8123,
      "step": 19957
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.71830415725708,
      "learning_rate": 7.867616522378047e-06,
      "loss": 0.8152,
      "step": 19958
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8192861080169678,
      "learning_rate": 7.866977605980259e-06,
      "loss": 0.8767,
      "step": 19959
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8976737260818481,
      "learning_rate": 7.866338689582469e-06,
      "loss": 0.8543,
      "step": 19960
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.584325909614563,
      "learning_rate": 7.86569977318468e-06,
      "loss": 0.835,
      "step": 19961
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9666635990142822,
      "learning_rate": 7.86506085678689e-06,
      "loss": 0.9647,
      "step": 19962
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6170836687088013,
      "learning_rate": 7.864421940389102e-06,
      "loss": 0.8213,
      "step": 19963
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9771381616592407,
      "learning_rate": 7.863783023991311e-06,
      "loss": 0.995,
      "step": 19964
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8430181741714478,
      "learning_rate": 7.863144107593523e-06,
      "loss": 0.8356,
      "step": 19965
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.836996078491211,
      "learning_rate": 7.862505191195733e-06,
      "loss": 0.8999,
      "step": 19966
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7034597396850586,
      "learning_rate": 7.861866274797944e-06,
      "loss": 0.8458,
      "step": 19967
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.1728410720825195,
      "learning_rate": 7.861227358400154e-06,
      "loss": 0.938,
      "step": 19968
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0910186767578125,
      "learning_rate": 7.860588442002364e-06,
      "loss": 0.7221,
      "step": 19969
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8966872692108154,
      "learning_rate": 7.859949525604576e-06,
      "loss": 0.936,
      "step": 19970
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.246504545211792,
      "learning_rate": 7.859310609206785e-06,
      "loss": 0.8246,
      "step": 19971
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4456334114074707,
      "learning_rate": 7.858671692808997e-06,
      "loss": 1.3003,
      "step": 19972
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7268880605697632,
      "learning_rate": 7.858032776411207e-06,
      "loss": 0.8462,
      "step": 19973
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9296321868896484,
      "learning_rate": 7.857393860013418e-06,
      "loss": 0.9064,
      "step": 19974
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4884121417999268,
      "learning_rate": 7.856754943615628e-06,
      "loss": 1.0821,
      "step": 19975
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1016690731048584,
      "learning_rate": 7.85611602721784e-06,
      "loss": 0.7873,
      "step": 19976
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.6776926517486572,
      "learning_rate": 7.85547711082005e-06,
      "loss": 1.0615,
      "step": 19977
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7539112567901611,
      "learning_rate": 7.854838194422261e-06,
      "loss": 0.8266,
      "step": 19978
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.135160207748413,
      "learning_rate": 7.854199278024471e-06,
      "loss": 1.0437,
      "step": 19979
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9335236549377441,
      "learning_rate": 7.853560361626683e-06,
      "loss": 0.8495,
      "step": 19980
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.385117769241333,
      "learning_rate": 7.852921445228892e-06,
      "loss": 0.8249,
      "step": 19981
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8271607160568237,
      "learning_rate": 7.852282528831104e-06,
      "loss": 0.9234,
      "step": 19982
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5774322748184204,
      "learning_rate": 7.851643612433314e-06,
      "loss": 0.841,
      "step": 19983
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9077296257019043,
      "learning_rate": 7.851004696035525e-06,
      "loss": 0.8762,
      "step": 19984
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.048008441925049,
      "learning_rate": 7.850365779637735e-06,
      "loss": 1.072,
      "step": 19985
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0728676319122314,
      "learning_rate": 7.849726863239945e-06,
      "loss": 1.126,
      "step": 19986
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8300668001174927,
      "learning_rate": 7.849087946842157e-06,
      "loss": 0.8077,
      "step": 19987
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6351311206817627,
      "learning_rate": 7.848449030444366e-06,
      "loss": 1.0185,
      "step": 19988
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6862924098968506,
      "learning_rate": 7.847810114046578e-06,
      "loss": 0.6455,
      "step": 19989
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.625027179718018,
      "learning_rate": 7.847171197648788e-06,
      "loss": 1.2104,
      "step": 19990
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.780600666999817,
      "learning_rate": 7.846532281251e-06,
      "loss": 1.048,
      "step": 19991
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7119265794754028,
      "learning_rate": 7.84589336485321e-06,
      "loss": 0.7807,
      "step": 19992
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.385928153991699,
      "learning_rate": 7.84525444845542e-06,
      "loss": 1.1772,
      "step": 19993
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.686076045036316,
      "learning_rate": 7.84461553205763e-06,
      "loss": 0.8795,
      "step": 19994
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6504907608032227,
      "learning_rate": 7.843976615659842e-06,
      "loss": 0.9763,
      "step": 19995
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.855108380317688,
      "learning_rate": 7.843337699262052e-06,
      "loss": 0.9236,
      "step": 19996
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8443405628204346,
      "learning_rate": 7.842698782864264e-06,
      "loss": 1.0297,
      "step": 19997
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.053511381149292,
      "learning_rate": 7.842059866466473e-06,
      "loss": 0.9105,
      "step": 19998
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8116648197174072,
      "learning_rate": 7.841420950068685e-06,
      "loss": 0.9028,
      "step": 19999
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.060546875,
      "learning_rate": 7.840782033670895e-06,
      "loss": 0.5441,
      "step": 20000
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.129398822784424,
      "learning_rate": 7.840143117273106e-06,
      "loss": 1.5628,
      "step": 20001
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.842947006225586,
      "learning_rate": 7.839504200875316e-06,
      "loss": 0.977,
      "step": 20002
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9058618545532227,
      "learning_rate": 7.838865284477528e-06,
      "loss": 0.6839,
      "step": 20003
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.062997817993164,
      "learning_rate": 7.838226368079738e-06,
      "loss": 0.9027,
      "step": 20004
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7784218788146973,
      "learning_rate": 7.837587451681948e-06,
      "loss": 0.8181,
      "step": 20005
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.733503818511963,
      "learning_rate": 7.836948535284159e-06,
      "loss": 0.9136,
      "step": 20006
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.162299156188965,
      "learning_rate": 7.836309618886369e-06,
      "loss": 0.8401,
      "step": 20007
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9002243280410767,
      "learning_rate": 7.83567070248858e-06,
      "loss": 0.877,
      "step": 20008
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8358440399169922,
      "learning_rate": 7.83503178609079e-06,
      "loss": 0.8279,
      "step": 20009
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.634825348854065,
      "learning_rate": 7.834392869693002e-06,
      "loss": 0.7812,
      "step": 20010
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9555718898773193,
      "learning_rate": 7.833753953295212e-06,
      "loss": 0.8595,
      "step": 20011
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5526161193847656,
      "learning_rate": 7.833115036897423e-06,
      "loss": 0.7625,
      "step": 20012
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4340531826019287,
      "learning_rate": 7.832476120499633e-06,
      "loss": 0.8798,
      "step": 20013
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.493602991104126,
      "learning_rate": 7.831837204101845e-06,
      "loss": 0.8196,
      "step": 20014
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.7758944034576416,
      "learning_rate": 7.831198287704055e-06,
      "loss": 0.8182,
      "step": 20015
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9908188581466675,
      "learning_rate": 7.830559371306266e-06,
      "loss": 1.2431,
      "step": 20016
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8019483089447021,
      "learning_rate": 7.829920454908476e-06,
      "loss": 1.0402,
      "step": 20017
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0700438022613525,
      "learning_rate": 7.829281538510687e-06,
      "loss": 0.9311,
      "step": 20018
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.102270126342773,
      "learning_rate": 7.828642622112897e-06,
      "loss": 1.5414,
      "step": 20019
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9011563062667847,
      "learning_rate": 7.828003705715109e-06,
      "loss": 0.9407,
      "step": 20020
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9507941007614136,
      "learning_rate": 7.827364789317319e-06,
      "loss": 0.9531,
      "step": 20021
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2965714931488037,
      "learning_rate": 7.82672587291953e-06,
      "loss": 1.1959,
      "step": 20022
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.86579430103302,
      "learning_rate": 7.82608695652174e-06,
      "loss": 0.8789,
      "step": 20023
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6911965608596802,
      "learning_rate": 7.82544804012395e-06,
      "loss": 0.8175,
      "step": 20024
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6477898359298706,
      "learning_rate": 7.824809123726162e-06,
      "loss": 0.8354,
      "step": 20025
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0150742530822754,
      "learning_rate": 7.824170207328371e-06,
      "loss": 1.0836,
      "step": 20026
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5359227657318115,
      "learning_rate": 7.823531290930583e-06,
      "loss": 0.872,
      "step": 20027
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.724889874458313,
      "learning_rate": 7.822892374532793e-06,
      "loss": 0.9167,
      "step": 20028
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3799018859863281,
      "learning_rate": 7.822253458135004e-06,
      "loss": 0.3948,
      "step": 20029
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.299346685409546,
      "learning_rate": 7.821614541737214e-06,
      "loss": 0.7941,
      "step": 20030
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9204552173614502,
      "learning_rate": 7.820975625339426e-06,
      "loss": 0.9448,
      "step": 20031
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.892748832702637,
      "learning_rate": 7.820336708941636e-06,
      "loss": 1.4681,
      "step": 20032
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.195754289627075,
      "learning_rate": 7.819697792543847e-06,
      "loss": 1.1122,
      "step": 20033
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0052967071533203,
      "learning_rate": 7.819058876146057e-06,
      "loss": 0.9051,
      "step": 20034
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.585312604904175,
      "learning_rate": 7.818419959748269e-06,
      "loss": 0.7574,
      "step": 20035
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8294174671173096,
      "learning_rate": 7.817781043350478e-06,
      "loss": 0.8146,
      "step": 20036
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5558832883834839,
      "learning_rate": 7.81714212695269e-06,
      "loss": 0.8165,
      "step": 20037
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3120546340942383,
      "learning_rate": 7.8165032105549e-06,
      "loss": 0.3854,
      "step": 20038
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7338416576385498,
      "learning_rate": 7.815864294157111e-06,
      "loss": 0.9533,
      "step": 20039
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6937083005905151,
      "learning_rate": 7.815225377759321e-06,
      "loss": 0.809,
      "step": 20040
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3886311054229736,
      "learning_rate": 7.814586461361531e-06,
      "loss": 0.3662,
      "step": 20041
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9922144412994385,
      "learning_rate": 7.813947544963743e-06,
      "loss": 0.898,
      "step": 20042
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7827683687210083,
      "learning_rate": 7.813308628565952e-06,
      "loss": 0.8159,
      "step": 20043
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8393630981445312,
      "learning_rate": 7.812669712168164e-06,
      "loss": 0.9664,
      "step": 20044
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6557749509811401,
      "learning_rate": 7.812030795770374e-06,
      "loss": 0.8812,
      "step": 20045
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6470911502838135,
      "learning_rate": 7.811391879372585e-06,
      "loss": 0.91,
      "step": 20046
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8745286464691162,
      "learning_rate": 7.810752962974795e-06,
      "loss": 0.8214,
      "step": 20047
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1348111629486084,
      "learning_rate": 7.810114046577007e-06,
      "loss": 0.7947,
      "step": 20048
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5664759874343872,
      "learning_rate": 7.809475130179217e-06,
      "loss": 0.8872,
      "step": 20049
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.651344656944275,
      "learning_rate": 7.808836213781428e-06,
      "loss": 0.8314,
      "step": 20050
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9108883142471313,
      "learning_rate": 7.808197297383638e-06,
      "loss": 0.8755,
      "step": 20051
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6419882774353027,
      "learning_rate": 7.80755838098585e-06,
      "loss": 0.8664,
      "step": 20052
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.704008936882019,
      "learning_rate": 7.80691946458806e-06,
      "loss": 0.9181,
      "step": 20053
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0650227069854736,
      "learning_rate": 7.806280548190271e-06,
      "loss": 0.7912,
      "step": 20054
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9349071979522705,
      "learning_rate": 7.80564163179248e-06,
      "loss": 0.9201,
      "step": 20055
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.593429088592529,
      "learning_rate": 7.805002715394692e-06,
      "loss": 1.2286,
      "step": 20056
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.797708511352539,
      "learning_rate": 7.804363798996902e-06,
      "loss": 1.5366,
      "step": 20057
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4517916440963745,
      "learning_rate": 7.803724882599114e-06,
      "loss": 0.7902,
      "step": 20058
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7883858680725098,
      "learning_rate": 7.803085966201324e-06,
      "loss": 0.9231,
      "step": 20059
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8745063543319702,
      "learning_rate": 7.802447049803533e-06,
      "loss": 0.9122,
      "step": 20060
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6260567903518677,
      "learning_rate": 7.801808133405745e-06,
      "loss": 0.7646,
      "step": 20061
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6133053302764893,
      "learning_rate": 7.801169217007955e-06,
      "loss": 0.7855,
      "step": 20062
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8859572410583496,
      "learning_rate": 7.800530300610166e-06,
      "loss": 0.9987,
      "step": 20063
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.692225456237793,
      "learning_rate": 7.799891384212376e-06,
      "loss": 0.8211,
      "step": 20064
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.790339708328247,
      "learning_rate": 7.799252467814588e-06,
      "loss": 1.0555,
      "step": 20065
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5828101634979248,
      "learning_rate": 7.798613551416798e-06,
      "loss": 0.7796,
      "step": 20066
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6416882276535034,
      "learning_rate": 7.79797463501901e-06,
      "loss": 0.8555,
      "step": 20067
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.738670825958252,
      "learning_rate": 7.797335718621219e-06,
      "loss": 0.9056,
      "step": 20068
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5083125829696655,
      "learning_rate": 7.79669680222343e-06,
      "loss": 0.8616,
      "step": 20069
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.527141809463501,
      "learning_rate": 7.79605788582564e-06,
      "loss": 0.7627,
      "step": 20070
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.8738808631896973,
      "learning_rate": 7.795418969427852e-06,
      "loss": 1.0944,
      "step": 20071
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0151522159576416,
      "learning_rate": 7.794780053030062e-06,
      "loss": 0.8986,
      "step": 20072
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.594139337539673,
      "learning_rate": 7.794141136632273e-06,
      "loss": 0.9432,
      "step": 20073
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9343899488449097,
      "learning_rate": 7.793502220234483e-06,
      "loss": 0.5643,
      "step": 20074
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8270639181137085,
      "learning_rate": 7.792863303836695e-06,
      "loss": 1.1042,
      "step": 20075
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6958796977996826,
      "learning_rate": 7.792224387438905e-06,
      "loss": 0.7717,
      "step": 20076
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1868629455566406,
      "learning_rate": 7.791585471041116e-06,
      "loss": 0.893,
      "step": 20077
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.821945309638977,
      "learning_rate": 7.790946554643326e-06,
      "loss": 0.9844,
      "step": 20078
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8803484439849854,
      "learning_rate": 7.790307638245536e-06,
      "loss": 0.9861,
      "step": 20079
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7538480758666992,
      "learning_rate": 7.789668721847747e-06,
      "loss": 0.896,
      "step": 20080
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7496742010116577,
      "learning_rate": 7.789029805449957e-06,
      "loss": 1.1483,
      "step": 20081
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.496720314025879,
      "learning_rate": 7.788390889052169e-06,
      "loss": 1.4267,
      "step": 20082
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4192018508911133,
      "learning_rate": 7.787751972654379e-06,
      "loss": 0.9442,
      "step": 20083
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.0996010303497314,
      "learning_rate": 7.78711305625659e-06,
      "loss": 0.9177,
      "step": 20084
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.197268962860107,
      "learning_rate": 7.7864741398588e-06,
      "loss": 0.7495,
      "step": 20085
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.348001480102539,
      "learning_rate": 7.785835223461012e-06,
      "loss": 1.2224,
      "step": 20086
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6338835954666138,
      "learning_rate": 7.785196307063222e-06,
      "loss": 0.4531,
      "step": 20087
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1623010635375977,
      "learning_rate": 7.784557390665433e-06,
      "loss": 1.1137,
      "step": 20088
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.190354824066162,
      "learning_rate": 7.783918474267643e-06,
      "loss": 0.8805,
      "step": 20089
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0826728343963623,
      "learning_rate": 7.783279557869854e-06,
      "loss": 0.7303,
      "step": 20090
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.949388861656189,
      "learning_rate": 7.782640641472064e-06,
      "loss": 0.9883,
      "step": 20091
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.763710379600525,
      "learning_rate": 7.782001725074276e-06,
      "loss": 0.9683,
      "step": 20092
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2050669193267822,
      "learning_rate": 7.781362808676486e-06,
      "loss": 0.9369,
      "step": 20093
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6332167387008667,
      "learning_rate": 7.780723892278697e-06,
      "loss": 0.8739,
      "step": 20094
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8718805313110352,
      "learning_rate": 7.780084975880907e-06,
      "loss": 0.9076,
      "step": 20095
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.153470039367676,
      "learning_rate": 7.779446059483117e-06,
      "loss": 1.0596,
      "step": 20096
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8982512950897217,
      "learning_rate": 7.778807143085329e-06,
      "loss": 0.9278,
      "step": 20097
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6420520544052124,
      "learning_rate": 7.778168226687538e-06,
      "loss": 0.876,
      "step": 20098
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5983901023864746,
      "learning_rate": 7.777529310289748e-06,
      "loss": 0.8184,
      "step": 20099
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.05735445022583,
      "learning_rate": 7.77689039389196e-06,
      "loss": 0.7498,
      "step": 20100
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7313971519470215,
      "learning_rate": 7.77625147749417e-06,
      "loss": 0.889,
      "step": 20101
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.699774146080017,
      "learning_rate": 7.775612561096381e-06,
      "loss": 0.9455,
      "step": 20102
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.94782018661499,
      "learning_rate": 7.774973644698591e-06,
      "loss": 1.5189,
      "step": 20103
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3771291971206665,
      "learning_rate": 7.774334728300803e-06,
      "loss": 0.7584,
      "step": 20104
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9607008695602417,
      "learning_rate": 7.773695811903012e-06,
      "loss": 0.9418,
      "step": 20105
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5771305561065674,
      "learning_rate": 7.773056895505224e-06,
      "loss": 0.84,
      "step": 20106
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8979108333587646,
      "learning_rate": 7.772417979107434e-06,
      "loss": 0.5937,
      "step": 20107
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6869430541992188,
      "learning_rate": 7.771779062709644e-06,
      "loss": 0.8819,
      "step": 20108
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4408626556396484,
      "learning_rate": 7.771140146311855e-06,
      "loss": 0.4261,
      "step": 20109
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8310397863388062,
      "learning_rate": 7.770501229914065e-06,
      "loss": 0.7699,
      "step": 20110
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.388292074203491,
      "learning_rate": 7.769862313516277e-06,
      "loss": 0.9334,
      "step": 20111
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6677175760269165,
      "learning_rate": 7.769223397118487e-06,
      "loss": 0.8785,
      "step": 20112
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.131375789642334,
      "learning_rate": 7.768584480720698e-06,
      "loss": 1.07,
      "step": 20113
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.3286995887756348,
      "learning_rate": 7.767945564322908e-06,
      "loss": 0.8995,
      "step": 20114
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.1939191818237305,
      "learning_rate": 7.76730664792512e-06,
      "loss": 0.9404,
      "step": 20115
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0964560508728027,
      "learning_rate": 7.76666773152733e-06,
      "loss": 0.8188,
      "step": 20116
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.825495481491089,
      "learning_rate": 7.766028815129541e-06,
      "loss": 0.7828,
      "step": 20117
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.0487091541290283,
      "learning_rate": 7.76538989873175e-06,
      "loss": 1.036,
      "step": 20118
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.3301661014556885,
      "learning_rate": 7.764750982333962e-06,
      "loss": 1.2327,
      "step": 20119
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.505674123764038,
      "learning_rate": 7.764112065936172e-06,
      "loss": 1.0878,
      "step": 20120
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7432897090911865,
      "learning_rate": 7.763473149538384e-06,
      "loss": 0.9561,
      "step": 20121
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8165463209152222,
      "learning_rate": 7.762834233140594e-06,
      "loss": 0.8994,
      "step": 20122
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7351658344268799,
      "learning_rate": 7.762195316742805e-06,
      "loss": 0.8155,
      "step": 20123
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1740424633026123,
      "learning_rate": 7.761556400345015e-06,
      "loss": 0.8925,
      "step": 20124
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7689259052276611,
      "learning_rate": 7.760917483947226e-06,
      "loss": 0.402,
      "step": 20125
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6424269676208496,
      "learning_rate": 7.760278567549436e-06,
      "loss": 0.807,
      "step": 20126
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0749762058258057,
      "learning_rate": 7.759639651151646e-06,
      "loss": 0.8334,
      "step": 20127
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3228546380996704,
      "learning_rate": 7.759000734753858e-06,
      "loss": 0.4115,
      "step": 20128
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7844380140304565,
      "learning_rate": 7.758361818356068e-06,
      "loss": 0.8064,
      "step": 20129
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8123208284378052,
      "learning_rate": 7.757722901958279e-06,
      "loss": 0.8244,
      "step": 20130
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6057324409484863,
      "learning_rate": 7.757083985560489e-06,
      "loss": 0.7453,
      "step": 20131
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.878151297569275,
      "learning_rate": 7.7564450691627e-06,
      "loss": 1.0201,
      "step": 20132
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.477813482284546,
      "learning_rate": 7.75580615276491e-06,
      "loss": 0.4132,
      "step": 20133
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6772425174713135,
      "learning_rate": 7.755167236367122e-06,
      "loss": 0.5797,
      "step": 20134
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.112421989440918,
      "learning_rate": 7.754528319969332e-06,
      "loss": 0.6024,
      "step": 20135
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6111259460449219,
      "learning_rate": 7.753889403571543e-06,
      "loss": 0.8298,
      "step": 20136
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9377398490905762,
      "learning_rate": 7.753250487173753e-06,
      "loss": 0.8337,
      "step": 20137
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5304911136627197,
      "learning_rate": 7.752611570775965e-06,
      "loss": 0.6595,
      "step": 20138
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8422009944915771,
      "learning_rate": 7.751972654378175e-06,
      "loss": 0.7647,
      "step": 20139
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.9571447372436523,
      "learning_rate": 7.751333737980386e-06,
      "loss": 1.0934,
      "step": 20140
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.028820753097534,
      "learning_rate": 7.750694821582596e-06,
      "loss": 1.1834,
      "step": 20141
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9570908546447754,
      "learning_rate": 7.750055905184808e-06,
      "loss": 0.9227,
      "step": 20142
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.088183879852295,
      "learning_rate": 7.749416988787017e-06,
      "loss": 0.8076,
      "step": 20143
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6722288131713867,
      "learning_rate": 7.748778072389229e-06,
      "loss": 0.836,
      "step": 20144
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7221970558166504,
      "learning_rate": 7.748139155991439e-06,
      "loss": 0.8879,
      "step": 20145
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.397619366645813,
      "learning_rate": 7.747500239593649e-06,
      "loss": 0.4274,
      "step": 20146
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.113699197769165,
      "learning_rate": 7.74686132319586e-06,
      "loss": 0.9745,
      "step": 20147
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9513397216796875,
      "learning_rate": 7.74622240679807e-06,
      "loss": 1.2002,
      "step": 20148
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8434478044509888,
      "learning_rate": 7.745583490400282e-06,
      "loss": 0.8037,
      "step": 20149
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6407561302185059,
      "learning_rate": 7.744944574002491e-06,
      "loss": 0.4971,
      "step": 20150
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7978650331497192,
      "learning_rate": 7.744305657604703e-06,
      "loss": 0.8636,
      "step": 20151
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.195878028869629,
      "learning_rate": 7.743666741206913e-06,
      "loss": 0.907,
      "step": 20152
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6498498916625977,
      "learning_rate": 7.743027824809124e-06,
      "loss": 0.7906,
      "step": 20153
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.449657917022705,
      "learning_rate": 7.742388908411334e-06,
      "loss": 1.0251,
      "step": 20154
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.976378321647644,
      "learning_rate": 7.741749992013546e-06,
      "loss": 1.0706,
      "step": 20155
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1016597747802734,
      "learning_rate": 7.741111075615756e-06,
      "loss": 0.8198,
      "step": 20156
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3935590982437134,
      "learning_rate": 7.740472159217967e-06,
      "loss": 0.4153,
      "step": 20157
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1440625190734863,
      "learning_rate": 7.739833242820177e-06,
      "loss": 0.7488,
      "step": 20158
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.6309356689453125,
      "learning_rate": 7.739194326422389e-06,
      "loss": 0.6918,
      "step": 20159
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.1156795024871826,
      "learning_rate": 7.738555410024598e-06,
      "loss": 0.7842,
      "step": 20160
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.730026125907898,
      "learning_rate": 7.73791649362681e-06,
      "loss": 0.8617,
      "step": 20161
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.830942153930664,
      "learning_rate": 7.73727757722902e-06,
      "loss": 1.4908,
      "step": 20162
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8517652750015259,
      "learning_rate": 7.73663866083123e-06,
      "loss": 0.9144,
      "step": 20163
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.039170026779175,
      "learning_rate": 7.735999744433441e-06,
      "loss": 0.8405,
      "step": 20164
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8597393035888672,
      "learning_rate": 7.735360828035651e-06,
      "loss": 0.8046,
      "step": 20165
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.426869511604309,
      "learning_rate": 7.734721911637863e-06,
      "loss": 0.7371,
      "step": 20166
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4691855907440186,
      "learning_rate": 7.734082995240072e-06,
      "loss": 0.4542,
      "step": 20167
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8978203535079956,
      "learning_rate": 7.733444078842284e-06,
      "loss": 0.995,
      "step": 20168
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2572755813598633,
      "learning_rate": 7.732805162444494e-06,
      "loss": 0.9061,
      "step": 20169
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.468240261077881,
      "learning_rate": 7.732166246046705e-06,
      "loss": 0.7693,
      "step": 20170
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.975019097328186,
      "learning_rate": 7.731527329648915e-06,
      "loss": 1.2326,
      "step": 20171
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9451607465744019,
      "learning_rate": 7.730888413251127e-06,
      "loss": 0.9323,
      "step": 20172
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4101173877716064,
      "learning_rate": 7.730249496853337e-06,
      "loss": 0.4684,
      "step": 20173
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1053879261016846,
      "learning_rate": 7.729610580455548e-06,
      "loss": 0.8841,
      "step": 20174
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1807198524475098,
      "learning_rate": 7.728971664057758e-06,
      "loss": 0.9833,
      "step": 20175
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7527790069580078,
      "learning_rate": 7.72833274765997e-06,
      "loss": 0.8601,
      "step": 20176
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7326244115829468,
      "learning_rate": 7.72769383126218e-06,
      "loss": 0.9212,
      "step": 20177
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4487287998199463,
      "learning_rate": 7.727054914864391e-06,
      "loss": 0.763,
      "step": 20178
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7994754314422607,
      "learning_rate": 7.726415998466601e-06,
      "loss": 0.8365,
      "step": 20179
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.1900441646575928,
      "learning_rate": 7.725777082068812e-06,
      "loss": 0.8216,
      "step": 20180
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1344363689422607,
      "learning_rate": 7.725138165671022e-06,
      "loss": 1.2389,
      "step": 20181
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8297905921936035,
      "learning_rate": 7.724499249273232e-06,
      "loss": 0.7062,
      "step": 20182
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8985627889633179,
      "learning_rate": 7.723860332875444e-06,
      "loss": 0.8794,
      "step": 20183
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.275003433227539,
      "learning_rate": 7.723221416477654e-06,
      "loss": 0.8258,
      "step": 20184
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0869393348693848,
      "learning_rate": 7.722582500079865e-06,
      "loss": 1.1856,
      "step": 20185
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.630671739578247,
      "learning_rate": 7.721943583682075e-06,
      "loss": 0.9503,
      "step": 20186
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8314789533615112,
      "learning_rate": 7.721304667284286e-06,
      "loss": 0.9939,
      "step": 20187
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.048600673675537,
      "learning_rate": 7.720665750886496e-06,
      "loss": 0.7975,
      "step": 20188
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5806827545166016,
      "learning_rate": 7.720026834488708e-06,
      "loss": 0.4465,
      "step": 20189
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.655950665473938,
      "learning_rate": 7.719387918090918e-06,
      "loss": 0.8864,
      "step": 20190
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4888837337493896,
      "learning_rate": 7.71874900169313e-06,
      "loss": 0.9536,
      "step": 20191
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.361251711845398,
      "learning_rate": 7.718110085295339e-06,
      "loss": 0.4111,
      "step": 20192
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.54045033454895,
      "learning_rate": 7.71747116889755e-06,
      "loss": 0.9332,
      "step": 20193
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.521447777748108,
      "learning_rate": 7.71683225249976e-06,
      "loss": 0.8799,
      "step": 20194
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.418696403503418,
      "learning_rate": 7.716193336101972e-06,
      "loss": 1.4555,
      "step": 20195
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6726350784301758,
      "learning_rate": 7.715554419704182e-06,
      "loss": 0.829,
      "step": 20196
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8530710935592651,
      "learning_rate": 7.714915503306393e-06,
      "loss": 0.8282,
      "step": 20197
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.2642672061920166,
      "learning_rate": 7.714276586908603e-06,
      "loss": 0.7981,
      "step": 20198
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6517857313156128,
      "learning_rate": 7.713637670510815e-06,
      "loss": 0.9127,
      "step": 20199
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3614660501480103,
      "learning_rate": 7.712998754113025e-06,
      "loss": 0.4324,
      "step": 20200
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5436159372329712,
      "learning_rate": 7.712359837715235e-06,
      "loss": 0.8422,
      "step": 20201
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1021316051483154,
      "learning_rate": 7.711720921317446e-06,
      "loss": 1.0439,
      "step": 20202
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.9714882373809814,
      "learning_rate": 7.711082004919656e-06,
      "loss": 1.2656,
      "step": 20203
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.963744878768921,
      "learning_rate": 7.710443088521868e-06,
      "loss": 0.9215,
      "step": 20204
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8467628955841064,
      "learning_rate": 7.709804172124077e-06,
      "loss": 0.7558,
      "step": 20205
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4161289930343628,
      "learning_rate": 7.709165255726289e-06,
      "loss": 0.3635,
      "step": 20206
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.962764859199524,
      "learning_rate": 7.708526339328499e-06,
      "loss": 0.9034,
      "step": 20207
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.64652156829834,
      "learning_rate": 7.70788742293071e-06,
      "loss": 1.2934,
      "step": 20208
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8265511989593506,
      "learning_rate": 7.70724850653292e-06,
      "loss": 0.8282,
      "step": 20209
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.990358829498291,
      "learning_rate": 7.706609590135132e-06,
      "loss": 0.946,
      "step": 20210
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8791499137878418,
      "learning_rate": 7.705970673737342e-06,
      "loss": 0.9389,
      "step": 20211
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7463001012802124,
      "learning_rate": 7.705331757339553e-06,
      "loss": 0.8426,
      "step": 20212
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.570943832397461,
      "learning_rate": 7.704692840941763e-06,
      "loss": 1.2969,
      "step": 20213
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7260829210281372,
      "learning_rate": 7.704053924543975e-06,
      "loss": 0.9093,
      "step": 20214
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5368201732635498,
      "learning_rate": 7.703415008146184e-06,
      "loss": 0.4023,
      "step": 20215
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.2386727333068848,
      "learning_rate": 7.702776091748396e-06,
      "loss": 0.991,
      "step": 20216
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.313756227493286,
      "learning_rate": 7.702137175350606e-06,
      "loss": 1.297,
      "step": 20217
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.2618303298950195,
      "learning_rate": 7.701498258952816e-06,
      "loss": 1.1886,
      "step": 20218
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7469819784164429,
      "learning_rate": 7.700859342555027e-06,
      "loss": 1.1398,
      "step": 20219
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4874433279037476,
      "learning_rate": 7.700220426157237e-06,
      "loss": 0.8507,
      "step": 20220
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0381462574005127,
      "learning_rate": 7.699581509759449e-06,
      "loss": 0.962,
      "step": 20221
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.645462155342102,
      "learning_rate": 7.698942593361658e-06,
      "loss": 0.8517,
      "step": 20222
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.715649127960205,
      "learning_rate": 7.69830367696387e-06,
      "loss": 1.4849,
      "step": 20223
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3453400135040283,
      "learning_rate": 7.69766476056608e-06,
      "loss": 0.8294,
      "step": 20224
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.074270248413086,
      "learning_rate": 7.697025844168291e-06,
      "loss": 0.9566,
      "step": 20225
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0468313694000244,
      "learning_rate": 7.696386927770501e-06,
      "loss": 0.8102,
      "step": 20226
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.227098226547241,
      "learning_rate": 7.695748011372713e-06,
      "loss": 0.8962,
      "step": 20227
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1806845664978027,
      "learning_rate": 7.695109094974923e-06,
      "loss": 0.9934,
      "step": 20228
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8358426094055176,
      "learning_rate": 7.694470178577134e-06,
      "loss": 0.8412,
      "step": 20229
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8765977621078491,
      "learning_rate": 7.693831262179344e-06,
      "loss": 0.9185,
      "step": 20230
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7990294694900513,
      "learning_rate": 7.693192345781556e-06,
      "loss": 1.1741,
      "step": 20231
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.8545567989349365,
      "learning_rate": 7.692553429383765e-06,
      "loss": 1.4411,
      "step": 20232
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7132693529129028,
      "learning_rate": 7.691914512985977e-06,
      "loss": 0.7303,
      "step": 20233
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5253173112869263,
      "learning_rate": 7.691275596588187e-06,
      "loss": 0.9761,
      "step": 20234
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3810875415802,
      "learning_rate": 7.690636680190398e-06,
      "loss": 0.7592,
      "step": 20235
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.451727271080017,
      "learning_rate": 7.689997763792608e-06,
      "loss": 0.7799,
      "step": 20236
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8331122398376465,
      "learning_rate": 7.689358847394818e-06,
      "loss": 0.8378,
      "step": 20237
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.863603591918945,
      "learning_rate": 7.68871993099703e-06,
      "loss": 0.9434,
      "step": 20238
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8594626188278198,
      "learning_rate": 7.68808101459924e-06,
      "loss": 0.9524,
      "step": 20239
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0000665187835693,
      "learning_rate": 7.687442098201451e-06,
      "loss": 0.8061,
      "step": 20240
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.769715785980225,
      "learning_rate": 7.686803181803661e-06,
      "loss": 1.1814,
      "step": 20241
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8483421802520752,
      "learning_rate": 7.686164265405872e-06,
      "loss": 0.9179,
      "step": 20242
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.987496256828308,
      "learning_rate": 7.685525349008082e-06,
      "loss": 0.8392,
      "step": 20243
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5326910018920898,
      "learning_rate": 7.684886432610294e-06,
      "loss": 0.8255,
      "step": 20244
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.662592649459839,
      "learning_rate": 7.684247516212504e-06,
      "loss": 0.6283,
      "step": 20245
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9535174369812012,
      "learning_rate": 7.683608599814715e-06,
      "loss": 0.8842,
      "step": 20246
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.750994086265564,
      "learning_rate": 7.682969683416925e-06,
      "loss": 0.9148,
      "step": 20247
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.9198431968688965,
      "learning_rate": 7.682330767019137e-06,
      "loss": 1.2026,
      "step": 20248
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1654858589172363,
      "learning_rate": 7.681691850621346e-06,
      "loss": 0.8336,
      "step": 20249
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.119424343109131,
      "learning_rate": 7.681052934223558e-06,
      "loss": 1.0218,
      "step": 20250
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7241398096084595,
      "learning_rate": 7.680414017825768e-06,
      "loss": 0.8534,
      "step": 20251
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6082876920700073,
      "learning_rate": 7.67977510142798e-06,
      "loss": 0.8997,
      "step": 20252
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.705814003944397,
      "learning_rate": 7.67913618503019e-06,
      "loss": 0.8454,
      "step": 20253
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.237579822540283,
      "learning_rate": 7.678497268632399e-06,
      "loss": 0.6503,
      "step": 20254
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6685833930969238,
      "learning_rate": 7.67785835223461e-06,
      "loss": 0.91,
      "step": 20255
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.961065649986267,
      "learning_rate": 7.67721943583682e-06,
      "loss": 0.9955,
      "step": 20256
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6324177980422974,
      "learning_rate": 7.676580519439032e-06,
      "loss": 0.8384,
      "step": 20257
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.6602470874786377,
      "learning_rate": 7.675941603041242e-06,
      "loss": 0.7192,
      "step": 20258
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8137226104736328,
      "learning_rate": 7.675302686643453e-06,
      "loss": 0.8938,
      "step": 20259
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.803930401802063,
      "learning_rate": 7.674663770245663e-06,
      "loss": 1.1302,
      "step": 20260
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4832444190979004,
      "learning_rate": 7.674024853847875e-06,
      "loss": 0.8681,
      "step": 20261
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.01898455619812,
      "learning_rate": 7.673385937450085e-06,
      "loss": 1.1068,
      "step": 20262
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9652007818222046,
      "learning_rate": 7.672747021052296e-06,
      "loss": 0.9073,
      "step": 20263
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3416281938552856,
      "learning_rate": 7.672108104654506e-06,
      "loss": 0.4428,
      "step": 20264
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8605443239212036,
      "learning_rate": 7.671469188256718e-06,
      "loss": 0.9813,
      "step": 20265
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.8326632976531982,
      "learning_rate": 7.670830271858928e-06,
      "loss": 1.3481,
      "step": 20266
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7340234518051147,
      "learning_rate": 7.670191355461139e-06,
      "loss": 0.8683,
      "step": 20267
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3725438117980957,
      "learning_rate": 7.669552439063349e-06,
      "loss": 0.7177,
      "step": 20268
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.033381700515747,
      "learning_rate": 7.66891352266556e-06,
      "loss": 1.3033,
      "step": 20269
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.9187591075897217,
      "learning_rate": 7.66827460626777e-06,
      "loss": 0.9115,
      "step": 20270
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.924262285232544,
      "learning_rate": 7.667635689869982e-06,
      "loss": 0.8406,
      "step": 20271
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.093015432357788,
      "learning_rate": 7.666996773472192e-06,
      "loss": 1.1583,
      "step": 20272
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.167788505554199,
      "learning_rate": 7.666357857074402e-06,
      "loss": 1.3956,
      "step": 20273
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8477528095245361,
      "learning_rate": 7.665718940676613e-06,
      "loss": 1.0061,
      "step": 20274
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7429312467575073,
      "learning_rate": 7.665080024278823e-06,
      "loss": 0.6449,
      "step": 20275
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7393016815185547,
      "learning_rate": 7.664441107881035e-06,
      "loss": 0.5229,
      "step": 20276
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5159711837768555,
      "learning_rate": 7.663802191483244e-06,
      "loss": 0.5176,
      "step": 20277
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.9298481941223145,
      "learning_rate": 7.663163275085456e-06,
      "loss": 1.0718,
      "step": 20278
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.999699592590332,
      "learning_rate": 7.662524358687666e-06,
      "loss": 0.9234,
      "step": 20279
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.482304573059082,
      "learning_rate": 7.661885442289877e-06,
      "loss": 1.3525,
      "step": 20280
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5197935104370117,
      "learning_rate": 7.661246525892087e-06,
      "loss": 0.9098,
      "step": 20281
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8040634393692017,
      "learning_rate": 7.660607609494299e-06,
      "loss": 0.7687,
      "step": 20282
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6056290864944458,
      "learning_rate": 7.659968693096509e-06,
      "loss": 1.1825,
      "step": 20283
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.629787802696228,
      "learning_rate": 7.65932977669872e-06,
      "loss": 0.8333,
      "step": 20284
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8493316173553467,
      "learning_rate": 7.65869086030093e-06,
      "loss": 0.8853,
      "step": 20285
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8713797330856323,
      "learning_rate": 7.658051943903142e-06,
      "loss": 0.9613,
      "step": 20286
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8764296770095825,
      "learning_rate": 7.657413027505351e-06,
      "loss": 1.1069,
      "step": 20287
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.264213800430298,
      "learning_rate": 7.656774111107563e-06,
      "loss": 1.0289,
      "step": 20288
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8340566158294678,
      "learning_rate": 7.656135194709773e-06,
      "loss": 0.8703,
      "step": 20289
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.467236042022705,
      "learning_rate": 7.655496278311984e-06,
      "loss": 1.4478,
      "step": 20290
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1834347248077393,
      "learning_rate": 7.654857361914194e-06,
      "loss": 0.9818,
      "step": 20291
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7375823259353638,
      "learning_rate": 7.654218445516404e-06,
      "loss": 0.756,
      "step": 20292
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7294799089431763,
      "learning_rate": 7.653579529118616e-06,
      "loss": 0.8826,
      "step": 20293
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5009968280792236,
      "learning_rate": 7.652940612720825e-06,
      "loss": 0.9957,
      "step": 20294
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1654818058013916,
      "learning_rate": 7.652301696323037e-06,
      "loss": 0.79,
      "step": 20295
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7030001878738403,
      "learning_rate": 7.651662779925247e-06,
      "loss": 0.993,
      "step": 20296
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8758105039596558,
      "learning_rate": 7.651023863527458e-06,
      "loss": 0.7721,
      "step": 20297
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.8552610874176025,
      "learning_rate": 7.650384947129668e-06,
      "loss": 0.7846,
      "step": 20298
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3179101943969727,
      "learning_rate": 7.64974603073188e-06,
      "loss": 0.9143,
      "step": 20299
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.027355909347534,
      "learning_rate": 7.64910711433409e-06,
      "loss": 0.9237,
      "step": 20300
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.346644401550293,
      "learning_rate": 7.648468197936301e-06,
      "loss": 0.7719,
      "step": 20301
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.347649574279785,
      "learning_rate": 7.647829281538511e-06,
      "loss": 0.5458,
      "step": 20302
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2013437747955322,
      "learning_rate": 7.647190365140723e-06,
      "loss": 0.9256,
      "step": 20303
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4834873676300049,
      "learning_rate": 7.646551448742932e-06,
      "loss": 0.8352,
      "step": 20304
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7846509218215942,
      "learning_rate": 7.645912532345144e-06,
      "loss": 0.8654,
      "step": 20305
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1115987300872803,
      "learning_rate": 7.645273615947354e-06,
      "loss": 0.9231,
      "step": 20306
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.697007656097412,
      "learning_rate": 7.644634699549565e-06,
      "loss": 0.8132,
      "step": 20307
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6119669675827026,
      "learning_rate": 7.643995783151775e-06,
      "loss": 0.832,
      "step": 20308
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4929354190826416,
      "learning_rate": 7.643356866753985e-06,
      "loss": 0.9434,
      "step": 20309
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3815557956695557,
      "learning_rate": 7.642717950356197e-06,
      "loss": 1.0216,
      "step": 20310
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7462856769561768,
      "learning_rate": 7.642079033958406e-06,
      "loss": 0.8469,
      "step": 20311
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5617482662200928,
      "learning_rate": 7.641440117560618e-06,
      "loss": 0.8103,
      "step": 20312
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7253509759902954,
      "learning_rate": 7.640801201162828e-06,
      "loss": 0.8111,
      "step": 20313
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5666683912277222,
      "learning_rate": 7.64016228476504e-06,
      "loss": 0.8348,
      "step": 20314
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8358523845672607,
      "learning_rate": 7.63952336836725e-06,
      "loss": 0.7878,
      "step": 20315
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0993571281433105,
      "learning_rate": 7.63888445196946e-06,
      "loss": 0.669,
      "step": 20316
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.8897664546966553,
      "learning_rate": 7.63824553557167e-06,
      "loss": 1.2425,
      "step": 20317
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.440526008605957,
      "learning_rate": 7.637606619173882e-06,
      "loss": 0.8057,
      "step": 20318
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6407244205474854,
      "learning_rate": 7.636967702776092e-06,
      "loss": 0.8311,
      "step": 20319
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3910741806030273,
      "learning_rate": 7.636328786378304e-06,
      "loss": 0.7482,
      "step": 20320
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.181175470352173,
      "learning_rate": 7.635689869980513e-06,
      "loss": 0.8901,
      "step": 20321
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1729445457458496,
      "learning_rate": 7.635050953582725e-06,
      "loss": 0.9145,
      "step": 20322
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2884106636047363,
      "learning_rate": 7.634412037184935e-06,
      "loss": 0.9161,
      "step": 20323
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.837939977645874,
      "learning_rate": 7.633773120787146e-06,
      "loss": 1.0944,
      "step": 20324
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7000306844711304,
      "learning_rate": 7.633134204389356e-06,
      "loss": 0.9094,
      "step": 20325
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8064080476760864,
      "learning_rate": 7.632495287991568e-06,
      "loss": 0.9241,
      "step": 20326
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5995079278945923,
      "learning_rate": 7.631856371593778e-06,
      "loss": 0.7724,
      "step": 20327
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.599175214767456,
      "learning_rate": 7.631217455195988e-06,
      "loss": 1.0857,
      "step": 20328
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.866617441177368,
      "learning_rate": 7.630578538798199e-06,
      "loss": 0.6119,
      "step": 20329
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.775815486907959,
      "learning_rate": 7.629939622400409e-06,
      "loss": 0.9089,
      "step": 20330
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.235888481140137,
      "learning_rate": 7.6293007060026205e-06,
      "loss": 1.4564,
      "step": 20331
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4572912454605103,
      "learning_rate": 7.628661789604831e-06,
      "loss": 0.8624,
      "step": 20332
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.738979339599609,
      "learning_rate": 7.628022873207042e-06,
      "loss": 1.3593,
      "step": 20333
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.406280755996704,
      "learning_rate": 7.627383956809253e-06,
      "loss": 0.4326,
      "step": 20334
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.709544062614441,
      "learning_rate": 7.626745040411463e-06,
      "loss": 0.8692,
      "step": 20335
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.125120162963867,
      "learning_rate": 7.626106124013674e-06,
      "loss": 1.1646,
      "step": 20336
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.506412982940674,
      "learning_rate": 7.625467207615884e-06,
      "loss": 1.5411,
      "step": 20337
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.683516502380371,
      "learning_rate": 7.6248282912180945e-06,
      "loss": 0.8268,
      "step": 20338
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6422522068023682,
      "learning_rate": 7.624189374820305e-06,
      "loss": 0.9365,
      "step": 20339
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5540862083435059,
      "learning_rate": 7.623550458422516e-06,
      "loss": 0.8339,
      "step": 20340
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4271514415740967,
      "learning_rate": 7.622911542024727e-06,
      "loss": 0.7925,
      "step": 20341
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0718750953674316,
      "learning_rate": 7.622272625626937e-06,
      "loss": 0.9973,
      "step": 20342
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3964636325836182,
      "learning_rate": 7.621633709229148e-06,
      "loss": 0.8708,
      "step": 20343
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.410902261734009,
      "learning_rate": 7.620994792831359e-06,
      "loss": 0.7379,
      "step": 20344
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0183563232421875,
      "learning_rate": 7.6203558764335694e-06,
      "loss": 0.8339,
      "step": 20345
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8001154661178589,
      "learning_rate": 7.61971696003578e-06,
      "loss": 0.9045,
      "step": 20346
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.5152430534362793,
      "learning_rate": 7.619078043637991e-06,
      "loss": 1.1908,
      "step": 20347
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.29793119430542,
      "learning_rate": 7.6184391272402015e-06,
      "loss": 0.9989,
      "step": 20348
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.316452980041504,
      "learning_rate": 7.617800210842412e-06,
      "loss": 1.0096,
      "step": 20349
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9004204273223877,
      "learning_rate": 7.617161294444623e-06,
      "loss": 0.8906,
      "step": 20350
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6556774377822876,
      "learning_rate": 7.616522378046834e-06,
      "loss": 0.9798,
      "step": 20351
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1454620361328125,
      "learning_rate": 7.615883461649044e-06,
      "loss": 0.9148,
      "step": 20352
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3910279273986816,
      "learning_rate": 7.615244545251255e-06,
      "loss": 1.1192,
      "step": 20353
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8771030902862549,
      "learning_rate": 7.614605628853466e-06,
      "loss": 1.1854,
      "step": 20354
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6921330690383911,
      "learning_rate": 7.613966712455676e-06,
      "loss": 0.9091,
      "step": 20355
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4442743062973022,
      "learning_rate": 7.613327796057886e-06,
      "loss": 0.7924,
      "step": 20356
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.276768445968628,
      "learning_rate": 7.612688879660097e-06,
      "loss": 1.2472,
      "step": 20357
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.963728666305542,
      "learning_rate": 7.612049963262308e-06,
      "loss": 0.9104,
      "step": 20358
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7606502771377563,
      "learning_rate": 7.611411046864518e-06,
      "loss": 0.7642,
      "step": 20359
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5661084651947021,
      "learning_rate": 7.610772130466729e-06,
      "loss": 0.3732,
      "step": 20360
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.441385269165039,
      "learning_rate": 7.61013321406894e-06,
      "loss": 0.8961,
      "step": 20361
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5421593189239502,
      "learning_rate": 7.6094942976711505e-06,
      "loss": 0.8135,
      "step": 20362
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.83630108833313,
      "learning_rate": 7.608855381273361e-06,
      "loss": 0.8848,
      "step": 20363
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6492853164672852,
      "learning_rate": 7.608216464875572e-06,
      "loss": 0.7613,
      "step": 20364
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7131810188293457,
      "learning_rate": 7.607577548477783e-06,
      "loss": 0.82,
      "step": 20365
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6798619031906128,
      "learning_rate": 7.606938632079993e-06,
      "loss": 0.87,
      "step": 20366
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.696918249130249,
      "learning_rate": 7.606299715682204e-06,
      "loss": 0.697,
      "step": 20367
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6315855979919434,
      "learning_rate": 7.605660799284415e-06,
      "loss": 0.6378,
      "step": 20368
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.359375476837158,
      "learning_rate": 7.605021882886625e-06,
      "loss": 0.8009,
      "step": 20369
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.370119094848633,
      "learning_rate": 7.604382966488836e-06,
      "loss": 0.8361,
      "step": 20370
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7637163400650024,
      "learning_rate": 7.603744050091047e-06,
      "loss": 0.8669,
      "step": 20371
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9858115911483765,
      "learning_rate": 7.6031051336932575e-06,
      "loss": 0.7387,
      "step": 20372
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.753194570541382,
      "learning_rate": 7.602466217295467e-06,
      "loss": 0.8575,
      "step": 20373
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9043647050857544,
      "learning_rate": 7.601827300897678e-06,
      "loss": 0.9749,
      "step": 20374
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.801598072052002,
      "learning_rate": 7.601188384499889e-06,
      "loss": 0.75,
      "step": 20375
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5452113151550293,
      "learning_rate": 7.6005494681020994e-06,
      "loss": 0.8486,
      "step": 20376
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6739616394042969,
      "learning_rate": 7.59991055170431e-06,
      "loss": 0.8381,
      "step": 20377
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.266336679458618,
      "learning_rate": 7.599271635306521e-06,
      "loss": 0.9723,
      "step": 20378
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.540807247161865,
      "learning_rate": 7.5986327189087315e-06,
      "loss": 0.986,
      "step": 20379
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.652642011642456,
      "learning_rate": 7.597993802510942e-06,
      "loss": 0.9379,
      "step": 20380
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.080704689025879,
      "learning_rate": 7.597354886113153e-06,
      "loss": 0.8293,
      "step": 20381
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.492564082145691,
      "learning_rate": 7.596715969715364e-06,
      "loss": 0.8353,
      "step": 20382
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.024311065673828,
      "learning_rate": 7.596077053317574e-06,
      "loss": 0.9242,
      "step": 20383
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.562725305557251,
      "learning_rate": 7.595438136919785e-06,
      "loss": 1.013,
      "step": 20384
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.949846267700195,
      "learning_rate": 7.594799220521996e-06,
      "loss": 1.453,
      "step": 20385
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.895062804222107,
      "learning_rate": 7.5941603041242064e-06,
      "loss": 0.8341,
      "step": 20386
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4736052751541138,
      "learning_rate": 7.593521387726417e-06,
      "loss": 0.8379,
      "step": 20387
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.6607913970947266,
      "learning_rate": 7.592882471328628e-06,
      "loss": 1.04,
      "step": 20388
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.5168421268463135,
      "learning_rate": 7.5922435549308385e-06,
      "loss": 1.3284,
      "step": 20389
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1265459060668945,
      "learning_rate": 7.591604638533049e-06,
      "loss": 0.562,
      "step": 20390
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.356367588043213,
      "learning_rate": 7.590965722135259e-06,
      "loss": 0.8373,
      "step": 20391
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.616729497909546,
      "learning_rate": 7.59032680573747e-06,
      "loss": 0.7751,
      "step": 20392
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.802938461303711,
      "learning_rate": 7.5896878893396805e-06,
      "loss": 0.8354,
      "step": 20393
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7232627868652344,
      "learning_rate": 7.589048972941891e-06,
      "loss": 0.9003,
      "step": 20394
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7882133722305298,
      "learning_rate": 7.588410056544102e-06,
      "loss": 0.8574,
      "step": 20395
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.882331371307373,
      "learning_rate": 7.587771140146313e-06,
      "loss": 0.8506,
      "step": 20396
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8758530616760254,
      "learning_rate": 7.587132223748523e-06,
      "loss": 0.8421,
      "step": 20397
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4976673126220703,
      "learning_rate": 7.586493307350734e-06,
      "loss": 0.8699,
      "step": 20398
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6099138259887695,
      "learning_rate": 7.585854390952945e-06,
      "loss": 0.851,
      "step": 20399
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4504534006118774,
      "learning_rate": 7.585215474555155e-06,
      "loss": 0.8254,
      "step": 20400
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9836009740829468,
      "learning_rate": 7.584576558157366e-06,
      "loss": 0.8928,
      "step": 20401
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2879016399383545,
      "learning_rate": 7.583937641759577e-06,
      "loss": 1.0507,
      "step": 20402
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.904561161994934,
      "learning_rate": 7.5832987253617875e-06,
      "loss": 0.8834,
      "step": 20403
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.628336429595947,
      "learning_rate": 7.582659808963998e-06,
      "loss": 1.1285,
      "step": 20404
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3146893978118896,
      "learning_rate": 7.582020892566209e-06,
      "loss": 0.9694,
      "step": 20405
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8548743724822998,
      "learning_rate": 7.58138197616842e-06,
      "loss": 0.8984,
      "step": 20406
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0155117511749268,
      "learning_rate": 7.58074305977063e-06,
      "loss": 0.8791,
      "step": 20407
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.006452798843384,
      "learning_rate": 7.580104143372841e-06,
      "loss": 1.0089,
      "step": 20408
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.4105091094970703,
      "learning_rate": 7.579465226975052e-06,
      "loss": 1.0572,
      "step": 20409
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8044992685317993,
      "learning_rate": 7.5788263105772615e-06,
      "loss": 1.1419,
      "step": 20410
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7337520122528076,
      "learning_rate": 7.578187394179472e-06,
      "loss": 1.1463,
      "step": 20411
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.3762896060943604,
      "learning_rate": 7.577548477781683e-06,
      "loss": 0.7827,
      "step": 20412
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7413820028305054,
      "learning_rate": 7.576909561383894e-06,
      "loss": 0.8495,
      "step": 20413
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7024357318878174,
      "learning_rate": 7.576270644986104e-06,
      "loss": 0.8504,
      "step": 20414
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7212775945663452,
      "learning_rate": 7.575631728588315e-06,
      "loss": 0.9103,
      "step": 20415
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5804017782211304,
      "learning_rate": 7.574992812190526e-06,
      "loss": 0.9183,
      "step": 20416
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6443125009536743,
      "learning_rate": 7.5743538957927364e-06,
      "loss": 0.7897,
      "step": 20417
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6342315673828125,
      "learning_rate": 7.573714979394947e-06,
      "loss": 0.8254,
      "step": 20418
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4785919189453125,
      "learning_rate": 7.573076062997158e-06,
      "loss": 0.8884,
      "step": 20419
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.256409168243408,
      "learning_rate": 7.5724371465993685e-06,
      "loss": 1.0799,
      "step": 20420
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.563332200050354,
      "learning_rate": 7.571798230201579e-06,
      "loss": 0.8455,
      "step": 20421
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2027533054351807,
      "learning_rate": 7.57115931380379e-06,
      "loss": 0.8778,
      "step": 20422
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.5942673683166504,
      "learning_rate": 7.570520397406001e-06,
      "loss": 0.9995,
      "step": 20423
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.85134220123291,
      "learning_rate": 7.569881481008211e-06,
      "loss": 0.9217,
      "step": 20424
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9666526317596436,
      "learning_rate": 7.569242564610422e-06,
      "loss": 0.8694,
      "step": 20425
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.182094097137451,
      "learning_rate": 7.568603648212633e-06,
      "loss": 0.9207,
      "step": 20426
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.617919683456421,
      "learning_rate": 7.5679647318148434e-06,
      "loss": 1.0396,
      "step": 20427
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4485591650009155,
      "learning_rate": 7.567325815417053e-06,
      "loss": 0.7592,
      "step": 20428
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.758800745010376,
      "learning_rate": 7.566686899019264e-06,
      "loss": 0.955,
      "step": 20429
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5969852209091187,
      "learning_rate": 7.566047982621475e-06,
      "loss": 0.8183,
      "step": 20430
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.898337364196777,
      "learning_rate": 7.565409066223685e-06,
      "loss": 1.6676,
      "step": 20431
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.647736072540283,
      "learning_rate": 7.564770149825896e-06,
      "loss": 1.1747,
      "step": 20432
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.375718593597412,
      "learning_rate": 7.564131233428107e-06,
      "loss": 0.9184,
      "step": 20433
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5409575700759888,
      "learning_rate": 7.5634923170303175e-06,
      "loss": 0.9448,
      "step": 20434
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7305998802185059,
      "learning_rate": 7.562853400632528e-06,
      "loss": 0.8404,
      "step": 20435
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.474213123321533,
      "learning_rate": 7.562214484234739e-06,
      "loss": 1.5876,
      "step": 20436
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8604968786239624,
      "learning_rate": 7.56157556783695e-06,
      "loss": 0.87,
      "step": 20437
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6856987476348877,
      "learning_rate": 7.56093665143916e-06,
      "loss": 0.7955,
      "step": 20438
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.712203860282898,
      "learning_rate": 7.560297735041371e-06,
      "loss": 0.7555,
      "step": 20439
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9256912469863892,
      "learning_rate": 7.559658818643582e-06,
      "loss": 0.9856,
      "step": 20440
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.140205144882202,
      "learning_rate": 7.559019902245792e-06,
      "loss": 0.8765,
      "step": 20441
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.704349160194397,
      "learning_rate": 7.558380985848003e-06,
      "loss": 0.9727,
      "step": 20442
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4663125276565552,
      "learning_rate": 7.557742069450214e-06,
      "loss": 0.8536,
      "step": 20443
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5938087701797485,
      "learning_rate": 7.5571031530524245e-06,
      "loss": 0.9251,
      "step": 20444
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.979084849357605,
      "learning_rate": 7.556464236654635e-06,
      "loss": 0.8327,
      "step": 20445
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8764511346817017,
      "learning_rate": 7.555825320256845e-06,
      "loss": 0.8172,
      "step": 20446
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4840538501739502,
      "learning_rate": 7.555186403859056e-06,
      "loss": 0.8574,
      "step": 20447
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3050626516342163,
      "learning_rate": 7.5545474874612664e-06,
      "loss": 0.4664,
      "step": 20448
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.5428385734558105,
      "learning_rate": 7.553908571063477e-06,
      "loss": 0.8156,
      "step": 20449
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5523388385772705,
      "learning_rate": 7.553269654665688e-06,
      "loss": 0.8337,
      "step": 20450
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9293524026870728,
      "learning_rate": 7.5526307382678985e-06,
      "loss": 0.9536,
      "step": 20451
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.7472591400146484,
      "learning_rate": 7.551991821870109e-06,
      "loss": 0.8467,
      "step": 20452
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6345698833465576,
      "learning_rate": 7.55135290547232e-06,
      "loss": 0.5557,
      "step": 20453
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7277872562408447,
      "learning_rate": 7.550713989074531e-06,
      "loss": 1.0016,
      "step": 20454
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8813199996948242,
      "learning_rate": 7.550075072676741e-06,
      "loss": 0.7457,
      "step": 20455
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7354512214660645,
      "learning_rate": 7.549436156278952e-06,
      "loss": 0.7015,
      "step": 20456
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7820851802825928,
      "learning_rate": 7.548797239881163e-06,
      "loss": 0.9232,
      "step": 20457
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4144911766052246,
      "learning_rate": 7.5481583234833734e-06,
      "loss": 1.0154,
      "step": 20458
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7865521907806396,
      "learning_rate": 7.547519407085584e-06,
      "loss": 0.7663,
      "step": 20459
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6640641689300537,
      "learning_rate": 7.546880490687795e-06,
      "loss": 0.8235,
      "step": 20460
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6345081329345703,
      "learning_rate": 7.5462415742900055e-06,
      "loss": 0.9469,
      "step": 20461
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.9566352367401123,
      "learning_rate": 7.545602657892216e-06,
      "loss": 1.344,
      "step": 20462
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8917962312698364,
      "learning_rate": 7.544963741494427e-06,
      "loss": 0.9309,
      "step": 20463
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.145984649658203,
      "learning_rate": 7.544324825096637e-06,
      "loss": 1.0306,
      "step": 20464
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6920192241668701,
      "learning_rate": 7.5436859086988475e-06,
      "loss": 0.7794,
      "step": 20465
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1762337684631348,
      "learning_rate": 7.543046992301058e-06,
      "loss": 0.8344,
      "step": 20466
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.173087120056152,
      "learning_rate": 7.542408075903269e-06,
      "loss": 1.0907,
      "step": 20467
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2274184226989746,
      "learning_rate": 7.54176915950548e-06,
      "loss": 0.9293,
      "step": 20468
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.177073955535889,
      "learning_rate": 7.54113024310769e-06,
      "loss": 1.3423,
      "step": 20469
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.110722780227661,
      "learning_rate": 7.540491326709901e-06,
      "loss": 0.9208,
      "step": 20470
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4827572107315063,
      "learning_rate": 7.539852410312112e-06,
      "loss": 0.8064,
      "step": 20471
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.7492120265960693,
      "learning_rate": 7.539213493914322e-06,
      "loss": 1.293,
      "step": 20472
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4703285694122314,
      "learning_rate": 7.538574577516533e-06,
      "loss": 0.7473,
      "step": 20473
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9963897466659546,
      "learning_rate": 7.537935661118744e-06,
      "loss": 1.1592,
      "step": 20474
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.825793981552124,
      "learning_rate": 7.5372967447209545e-06,
      "loss": 0.8478,
      "step": 20475
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8842577934265137,
      "learning_rate": 7.536657828323165e-06,
      "loss": 1.2157,
      "step": 20476
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8772008419036865,
      "learning_rate": 7.536018911925376e-06,
      "loss": 0.9713,
      "step": 20477
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7219905853271484,
      "learning_rate": 7.535379995527587e-06,
      "loss": 0.7613,
      "step": 20478
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8389676809310913,
      "learning_rate": 7.534741079129797e-06,
      "loss": 0.9344,
      "step": 20479
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6833763122558594,
      "learning_rate": 7.534102162732008e-06,
      "loss": 1.249,
      "step": 20480
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8458441495895386,
      "learning_rate": 7.533463246334219e-06,
      "loss": 0.931,
      "step": 20481
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7133054733276367,
      "learning_rate": 7.532824329936429e-06,
      "loss": 0.7714,
      "step": 20482
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8331832885742188,
      "learning_rate": 7.532185413538639e-06,
      "loss": 0.8963,
      "step": 20483
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9267323017120361,
      "learning_rate": 7.53154649714085e-06,
      "loss": 0.9537,
      "step": 20484
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.669844388961792,
      "learning_rate": 7.530907580743061e-06,
      "loss": 0.9019,
      "step": 20485
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.598612904548645,
      "learning_rate": 7.5302686643452705e-06,
      "loss": 0.817,
      "step": 20486
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.9017109870910645,
      "learning_rate": 7.529629747947481e-06,
      "loss": 1.3279,
      "step": 20487
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.285520553588867,
      "learning_rate": 7.528990831549692e-06,
      "loss": 0.8314,
      "step": 20488
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7723197937011719,
      "learning_rate": 7.528351915151903e-06,
      "loss": 0.8372,
      "step": 20489
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4746719598770142,
      "learning_rate": 7.527712998754113e-06,
      "loss": 0.7287,
      "step": 20490
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8468784093856812,
      "learning_rate": 7.527074082356324e-06,
      "loss": 0.8129,
      "step": 20491
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5459758043289185,
      "learning_rate": 7.526435165958535e-06,
      "loss": 0.9491,
      "step": 20492
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6458669900894165,
      "learning_rate": 7.525796249560745e-06,
      "loss": 0.909,
      "step": 20493
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0875182151794434,
      "learning_rate": 7.525157333162956e-06,
      "loss": 0.8812,
      "step": 20494
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.201697587966919,
      "learning_rate": 7.524518416765166e-06,
      "loss": 0.7042,
      "step": 20495
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9946027994155884,
      "learning_rate": 7.523879500367377e-06,
      "loss": 1.1896,
      "step": 20496
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.15492057800293,
      "learning_rate": 7.523240583969587e-06,
      "loss": 1.6471,
      "step": 20497
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.876235008239746,
      "learning_rate": 7.522601667571798e-06,
      "loss": 0.9369,
      "step": 20498
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4800679683685303,
      "learning_rate": 7.521962751174009e-06,
      "loss": 0.9112,
      "step": 20499
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.015401363372803,
      "learning_rate": 7.5213238347762195e-06,
      "loss": 1.3099,
      "step": 20500
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6438796520233154,
      "learning_rate": 7.52068491837843e-06,
      "loss": 0.8851,
      "step": 20501
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.814070224761963,
      "learning_rate": 7.520046001980641e-06,
      "loss": 0.9869,
      "step": 20502
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.671822190284729,
      "learning_rate": 7.5194070855828516e-06,
      "loss": 0.8366,
      "step": 20503
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.553867816925049,
      "learning_rate": 7.518768169185062e-06,
      "loss": 0.8773,
      "step": 20504
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5122846364974976,
      "learning_rate": 7.518129252787273e-06,
      "loss": 0.8449,
      "step": 20505
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0059306621551514,
      "learning_rate": 7.517490336389484e-06,
      "loss": 0.8538,
      "step": 20506
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.234692573547363,
      "learning_rate": 7.516851419991694e-06,
      "loss": 1.2481,
      "step": 20507
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3845480680465698,
      "learning_rate": 7.516212503593905e-06,
      "loss": 0.7928,
      "step": 20508
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8494181632995605,
      "learning_rate": 7.515573587196116e-06,
      "loss": 0.8856,
      "step": 20509
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.141406297683716,
      "learning_rate": 7.5149346707983265e-06,
      "loss": 0.8979,
      "step": 20510
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7897413969039917,
      "learning_rate": 7.514295754400537e-06,
      "loss": 0.8649,
      "step": 20511
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6082563400268555,
      "learning_rate": 7.513656838002748e-06,
      "loss": 0.8646,
      "step": 20512
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.223003387451172,
      "learning_rate": 7.513017921604958e-06,
      "loss": 1.2558,
      "step": 20513
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7241028547286987,
      "learning_rate": 7.512379005207168e-06,
      "loss": 0.8378,
      "step": 20514
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0392329692840576,
      "learning_rate": 7.511740088809379e-06,
      "loss": 0.7276,
      "step": 20515
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4367958307266235,
      "learning_rate": 7.51110117241159e-06,
      "loss": 0.4531,
      "step": 20516
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.938053607940674,
      "learning_rate": 7.5104622560138005e-06,
      "loss": 0.9538,
      "step": 20517
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0982728004455566,
      "learning_rate": 7.509823339616011e-06,
      "loss": 0.8289,
      "step": 20518
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0861990451812744,
      "learning_rate": 7.509184423218222e-06,
      "loss": 0.9864,
      "step": 20519
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.572772741317749,
      "learning_rate": 7.508545506820433e-06,
      "loss": 0.9459,
      "step": 20520
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0932111740112305,
      "learning_rate": 7.507906590422643e-06,
      "loss": 0.8249,
      "step": 20521
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7979785203933716,
      "learning_rate": 7.507267674024854e-06,
      "loss": 0.9446,
      "step": 20522
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7353789806365967,
      "learning_rate": 7.506628757627065e-06,
      "loss": 0.9944,
      "step": 20523
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9235222339630127,
      "learning_rate": 7.505989841229275e-06,
      "loss": 0.7396,
      "step": 20524
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4366097450256348,
      "learning_rate": 7.505350924831486e-06,
      "loss": 0.8326,
      "step": 20525
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.8418705463409424,
      "learning_rate": 7.504712008433697e-06,
      "loss": 0.9896,
      "step": 20526
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4728223085403442,
      "learning_rate": 7.5040730920359075e-06,
      "loss": 0.4275,
      "step": 20527
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1933982372283936,
      "learning_rate": 7.503434175638118e-06,
      "loss": 0.9045,
      "step": 20528
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.761795163154602,
      "learning_rate": 7.502795259240329e-06,
      "loss": 0.9926,
      "step": 20529
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4999245405197144,
      "learning_rate": 7.50215634284254e-06,
      "loss": 0.7519,
      "step": 20530
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6841871738433838,
      "learning_rate": 7.50151742644475e-06,
      "loss": 0.8397,
      "step": 20531
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8999125957489014,
      "learning_rate": 7.50087851004696e-06,
      "loss": 0.7933,
      "step": 20532
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.796522855758667,
      "learning_rate": 7.500239593649171e-06,
      "loss": 0.7502,
      "step": 20533
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9445788860321045,
      "learning_rate": 7.4996006772513816e-06,
      "loss": 0.7875,
      "step": 20534
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5798097848892212,
      "learning_rate": 7.498961760853592e-06,
      "loss": 0.8597,
      "step": 20535
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6702525615692139,
      "learning_rate": 7.498322844455803e-06,
      "loss": 1.0387,
      "step": 20536
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.847947359085083,
      "learning_rate": 7.497683928058014e-06,
      "loss": 1.1333,
      "step": 20537
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8293702602386475,
      "learning_rate": 7.497045011660224e-06,
      "loss": 0.9078,
      "step": 20538
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.6629092693328857,
      "learning_rate": 7.496406095262435e-06,
      "loss": 0.9723,
      "step": 20539
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7852431535720825,
      "learning_rate": 7.495767178864646e-06,
      "loss": 0.3888,
      "step": 20540
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0732085704803467,
      "learning_rate": 7.4951282624668565e-06,
      "loss": 0.8659,
      "step": 20541
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.259157657623291,
      "learning_rate": 7.494489346069067e-06,
      "loss": 0.757,
      "step": 20542
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.944779872894287,
      "learning_rate": 7.493850429671278e-06,
      "loss": 1.0107,
      "step": 20543
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.714035987854004,
      "learning_rate": 7.4932115132734886e-06,
      "loss": 1.0381,
      "step": 20544
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9637131690979004,
      "learning_rate": 7.492572596875699e-06,
      "loss": 0.947,
      "step": 20545
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.9348747730255127,
      "learning_rate": 7.49193368047791e-06,
      "loss": 1.4232,
      "step": 20546
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5649635791778564,
      "learning_rate": 7.491294764080121e-06,
      "loss": 1.0072,
      "step": 20547
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5951555967330933,
      "learning_rate": 7.490655847682331e-06,
      "loss": 0.8018,
      "step": 20548
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6835095882415771,
      "learning_rate": 7.490016931284542e-06,
      "loss": 0.7942,
      "step": 20549
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.373795747756958,
      "learning_rate": 7.489378014886752e-06,
      "loss": 1.0744,
      "step": 20550
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.076755523681641,
      "learning_rate": 7.488739098488963e-06,
      "loss": 1.5291,
      "step": 20551
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1130688190460205,
      "learning_rate": 7.488100182091173e-06,
      "loss": 0.9084,
      "step": 20552
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6927649974822998,
      "learning_rate": 7.487461265693384e-06,
      "loss": 0.8453,
      "step": 20553
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.531374216079712,
      "learning_rate": 7.486822349295595e-06,
      "loss": 0.8876,
      "step": 20554
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.877020835876465,
      "learning_rate": 7.486183432897805e-06,
      "loss": 1.5429,
      "step": 20555
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.29472017288208,
      "learning_rate": 7.485544516500016e-06,
      "loss": 0.9179,
      "step": 20556
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.079944610595703,
      "learning_rate": 7.484905600102227e-06,
      "loss": 1.3497,
      "step": 20557
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.610841989517212,
      "learning_rate": 7.4842666837044375e-06,
      "loss": 0.9659,
      "step": 20558
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0680150985717773,
      "learning_rate": 7.483627767306648e-06,
      "loss": 0.7832,
      "step": 20559
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8639793395996094,
      "learning_rate": 7.482988850908859e-06,
      "loss": 0.8379,
      "step": 20560
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9579988718032837,
      "learning_rate": 7.48234993451107e-06,
      "loss": 1.0407,
      "step": 20561
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.1143691539764404,
      "learning_rate": 7.48171101811328e-06,
      "loss": 0.8755,
      "step": 20562
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9368317127227783,
      "learning_rate": 7.481072101715491e-06,
      "loss": 0.6949,
      "step": 20563
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.882009506225586,
      "learning_rate": 7.480433185317702e-06,
      "loss": 0.8831,
      "step": 20564
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.033001184463501,
      "learning_rate": 7.479794268919912e-06,
      "loss": 0.9142,
      "step": 20565
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6365190744400024,
      "learning_rate": 7.479155352522123e-06,
      "loss": 0.9129,
      "step": 20566
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6713614463806152,
      "learning_rate": 7.478516436124334e-06,
      "loss": 0.7892,
      "step": 20567
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.0329806804656982,
      "learning_rate": 7.477877519726544e-06,
      "loss": 1.0624,
      "step": 20568
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2512099742889404,
      "learning_rate": 7.477238603328754e-06,
      "loss": 0.9408,
      "step": 20569
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.1631505489349365,
      "learning_rate": 7.476599686930965e-06,
      "loss": 0.8214,
      "step": 20570
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.081686019897461,
      "learning_rate": 7.475960770533176e-06,
      "loss": 0.9256,
      "step": 20571
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7478258609771729,
      "learning_rate": 7.4753218541353865e-06,
      "loss": 0.8674,
      "step": 20572
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0382556915283203,
      "learning_rate": 7.474682937737597e-06,
      "loss": 0.9229,
      "step": 20573
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3885350227355957,
      "learning_rate": 7.474044021339808e-06,
      "loss": 0.6844,
      "step": 20574
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9781773090362549,
      "learning_rate": 7.4734051049420186e-06,
      "loss": 0.8532,
      "step": 20575
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8176771402359009,
      "learning_rate": 7.472766188544229e-06,
      "loss": 0.8205,
      "step": 20576
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2631757259368896,
      "learning_rate": 7.47212727214644e-06,
      "loss": 0.798,
      "step": 20577
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9443557262420654,
      "learning_rate": 7.471488355748651e-06,
      "loss": 0.928,
      "step": 20578
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9647259712219238,
      "learning_rate": 7.470849439350861e-06,
      "loss": 0.8829,
      "step": 20579
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1996285915374756,
      "learning_rate": 7.470210522953072e-06,
      "loss": 0.9674,
      "step": 20580
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4731013774871826,
      "learning_rate": 7.469571606555283e-06,
      "loss": 0.8487,
      "step": 20581
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0829145908355713,
      "learning_rate": 7.4689326901574935e-06,
      "loss": 1.1103,
      "step": 20582
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1117799282073975,
      "learning_rate": 7.468293773759704e-06,
      "loss": 1.2266,
      "step": 20583
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6673587560653687,
      "learning_rate": 7.467654857361915e-06,
      "loss": 0.5173,
      "step": 20584
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8752412796020508,
      "learning_rate": 7.4670159409641256e-06,
      "loss": 0.7968,
      "step": 20585
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7638185024261475,
      "learning_rate": 7.466377024566336e-06,
      "loss": 0.9155,
      "step": 20586
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7442330121994019,
      "learning_rate": 7.465738108168546e-06,
      "loss": 0.8325,
      "step": 20587
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9323439598083496,
      "learning_rate": 7.465099191770757e-06,
      "loss": 0.7667,
      "step": 20588
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8633259534835815,
      "learning_rate": 7.4644602753729675e-06,
      "loss": 1.0202,
      "step": 20589
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0841925144195557,
      "learning_rate": 7.463821358975178e-06,
      "loss": 1.0336,
      "step": 20590
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4517312049865723,
      "learning_rate": 7.463182442577389e-06,
      "loss": 0.9363,
      "step": 20591
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8788726329803467,
      "learning_rate": 7.4625435261796e-06,
      "loss": 1.1518,
      "step": 20592
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9206541776657104,
      "learning_rate": 7.46190460978181e-06,
      "loss": 1.0188,
      "step": 20593
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3247063159942627,
      "learning_rate": 7.461265693384021e-06,
      "loss": 0.8617,
      "step": 20594
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.238311290740967,
      "learning_rate": 7.460626776986232e-06,
      "loss": 0.7678,
      "step": 20595
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4835575819015503,
      "learning_rate": 7.459987860588442e-06,
      "loss": 0.9017,
      "step": 20596
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8849847316741943,
      "learning_rate": 7.459348944190653e-06,
      "loss": 0.5105,
      "step": 20597
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5520663261413574,
      "learning_rate": 7.458710027792864e-06,
      "loss": 1.0676,
      "step": 20598
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.24478006362915,
      "learning_rate": 7.4580711113950745e-06,
      "loss": 0.8687,
      "step": 20599
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7065809965133667,
      "learning_rate": 7.457432194997285e-06,
      "loss": 0.9393,
      "step": 20600
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6896271705627441,
      "learning_rate": 7.456793278599496e-06,
      "loss": 0.9088,
      "step": 20601
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.935626029968262,
      "learning_rate": 7.456154362201707e-06,
      "loss": 0.9967,
      "step": 20602
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.449519157409668,
      "learning_rate": 7.455515445803917e-06,
      "loss": 0.6132,
      "step": 20603
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2413601875305176,
      "learning_rate": 7.454876529406128e-06,
      "loss": 0.6351,
      "step": 20604
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5822815895080566,
      "learning_rate": 7.454237613008338e-06,
      "loss": 0.9138,
      "step": 20605
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.745190978050232,
      "learning_rate": 7.4535986966105486e-06,
      "loss": 0.8163,
      "step": 20606
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8024832010269165,
      "learning_rate": 7.452959780212759e-06,
      "loss": 0.8542,
      "step": 20607
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.299942970275879,
      "learning_rate": 7.45232086381497e-06,
      "loss": 0.9271,
      "step": 20608
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.23551344871521,
      "learning_rate": 7.451681947417181e-06,
      "loss": 1.0334,
      "step": 20609
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9264647960662842,
      "learning_rate": 7.451043031019391e-06,
      "loss": 0.7769,
      "step": 20610
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9075199365615845,
      "learning_rate": 7.450404114621602e-06,
      "loss": 0.9422,
      "step": 20611
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.972301959991455,
      "learning_rate": 7.449765198223813e-06,
      "loss": 1.3431,
      "step": 20612
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.535617709159851,
      "learning_rate": 7.4491262818260235e-06,
      "loss": 0.8444,
      "step": 20613
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.45835542678833,
      "learning_rate": 7.448487365428234e-06,
      "loss": 1.0612,
      "step": 20614
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.64035964012146,
      "learning_rate": 7.447848449030445e-06,
      "loss": 0.8247,
      "step": 20615
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6967281103134155,
      "learning_rate": 7.4472095326326556e-06,
      "loss": 0.8771,
      "step": 20616
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4832788705825806,
      "learning_rate": 7.446570616234866e-06,
      "loss": 0.8427,
      "step": 20617
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9910945892333984,
      "learning_rate": 7.445931699837077e-06,
      "loss": 1.1855,
      "step": 20618
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5159757137298584,
      "learning_rate": 7.445292783439288e-06,
      "loss": 0.8619,
      "step": 20619
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2958779335021973,
      "learning_rate": 7.444653867041498e-06,
      "loss": 1.2261,
      "step": 20620
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6775922775268555,
      "learning_rate": 7.444014950643709e-06,
      "loss": 0.8447,
      "step": 20621
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.529820203781128,
      "learning_rate": 7.44337603424592e-06,
      "loss": 0.9185,
      "step": 20622
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.506961345672607,
      "learning_rate": 7.44273711784813e-06,
      "loss": 1.1522,
      "step": 20623
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6138575077056885,
      "learning_rate": 7.44209820145034e-06,
      "loss": 0.8733,
      "step": 20624
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.065985202789307,
      "learning_rate": 7.441459285052551e-06,
      "loss": 1.2841,
      "step": 20625
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6550544500350952,
      "learning_rate": 7.440820368654762e-06,
      "loss": 0.8113,
      "step": 20626
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.300924062728882,
      "learning_rate": 7.440181452256972e-06,
      "loss": 0.5336,
      "step": 20627
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.916546583175659,
      "learning_rate": 7.439542535859183e-06,
      "loss": 0.7728,
      "step": 20628
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.04069185256958,
      "learning_rate": 7.438903619461394e-06,
      "loss": 1.1347,
      "step": 20629
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.708805561065674,
      "learning_rate": 7.4382647030636045e-06,
      "loss": 1.2746,
      "step": 20630
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9900031089782715,
      "learning_rate": 7.437625786665815e-06,
      "loss": 0.8679,
      "step": 20631
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4038015604019165,
      "learning_rate": 7.436986870268026e-06,
      "loss": 0.7904,
      "step": 20632
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.6356680393218994,
      "learning_rate": 7.436347953870237e-06,
      "loss": 1.0416,
      "step": 20633
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.69083833694458,
      "learning_rate": 7.435709037472447e-06,
      "loss": 0.9673,
      "step": 20634
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.9026455879211426,
      "learning_rate": 7.435070121074658e-06,
      "loss": 1.2518,
      "step": 20635
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8382328748703003,
      "learning_rate": 7.434431204676869e-06,
      "loss": 0.8366,
      "step": 20636
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7836560010910034,
      "learning_rate": 7.433792288279079e-06,
      "loss": 0.8853,
      "step": 20637
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.0665388107299805,
      "learning_rate": 7.43315337188129e-06,
      "loss": 0.6669,
      "step": 20638
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.323637962341309,
      "learning_rate": 7.432514455483501e-06,
      "loss": 1.4621,
      "step": 20639
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9541831016540527,
      "learning_rate": 7.4318755390857115e-06,
      "loss": 0.8278,
      "step": 20640
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9745402336120605,
      "learning_rate": 7.431236622687921e-06,
      "loss": 0.9685,
      "step": 20641
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8024672269821167,
      "learning_rate": 7.430597706290132e-06,
      "loss": 0.8745,
      "step": 20642
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.497349739074707,
      "learning_rate": 7.429958789892343e-06,
      "loss": 0.9817,
      "step": 20643
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.853020668029785,
      "learning_rate": 7.4293198734945535e-06,
      "loss": 1.0207,
      "step": 20644
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5315279960632324,
      "learning_rate": 7.428680957096764e-06,
      "loss": 0.7637,
      "step": 20645
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8766385316848755,
      "learning_rate": 7.428042040698975e-06,
      "loss": 0.9233,
      "step": 20646
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.971875786781311,
      "learning_rate": 7.4274031243011856e-06,
      "loss": 0.9092,
      "step": 20647
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8752840757369995,
      "learning_rate": 7.426764207903396e-06,
      "loss": 0.8404,
      "step": 20648
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.468212604522705,
      "learning_rate": 7.426125291505607e-06,
      "loss": 0.7628,
      "step": 20649
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.251467227935791,
      "learning_rate": 7.425486375107818e-06,
      "loss": 0.6799,
      "step": 20650
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.798872947692871,
      "learning_rate": 7.424847458710028e-06,
      "loss": 0.9037,
      "step": 20651
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1901092529296875,
      "learning_rate": 7.424208542312239e-06,
      "loss": 0.9401,
      "step": 20652
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7945114374160767,
      "learning_rate": 7.42356962591445e-06,
      "loss": 0.9156,
      "step": 20653
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2469518184661865,
      "learning_rate": 7.4229307095166605e-06,
      "loss": 0.9541,
      "step": 20654
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.79433274269104,
      "learning_rate": 7.422291793118871e-06,
      "loss": 1.0103,
      "step": 20655
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.089993476867676,
      "learning_rate": 7.421652876721082e-06,
      "loss": 1.0992,
      "step": 20656
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4286936521530151,
      "learning_rate": 7.4210139603232926e-06,
      "loss": 0.8947,
      "step": 20657
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.407865524291992,
      "learning_rate": 7.420375043925503e-06,
      "loss": 1.047,
      "step": 20658
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9527451992034912,
      "learning_rate": 7.419736127527714e-06,
      "loss": 0.9229,
      "step": 20659
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.283372402191162,
      "learning_rate": 7.419097211129924e-06,
      "loss": 1.2262,
      "step": 20660
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6768299341201782,
      "learning_rate": 7.4184582947321345e-06,
      "loss": 0.7645,
      "step": 20661
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.234318971633911,
      "learning_rate": 7.417819378334345e-06,
      "loss": 0.9367,
      "step": 20662
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4860045909881592,
      "learning_rate": 7.417180461936556e-06,
      "loss": 0.7981,
      "step": 20663
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.150219678878784,
      "learning_rate": 7.416541545538767e-06,
      "loss": 0.6833,
      "step": 20664
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7925503253936768,
      "learning_rate": 7.415902629140977e-06,
      "loss": 0.932,
      "step": 20665
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7381423711776733,
      "learning_rate": 7.415263712743188e-06,
      "loss": 0.8447,
      "step": 20666
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.697182297706604,
      "learning_rate": 7.414624796345399e-06,
      "loss": 0.9146,
      "step": 20667
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8732167482376099,
      "learning_rate": 7.4139858799476094e-06,
      "loss": 0.844,
      "step": 20668
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.035922527313232,
      "learning_rate": 7.41334696354982e-06,
      "loss": 1.1377,
      "step": 20669
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2731883525848389,
      "learning_rate": 7.412708047152031e-06,
      "loss": 0.651,
      "step": 20670
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.132781028747559,
      "learning_rate": 7.4120691307542415e-06,
      "loss": 0.9601,
      "step": 20671
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2428793907165527,
      "learning_rate": 7.411430214356452e-06,
      "loss": 0.7882,
      "step": 20672
    },
    {
      "epoch": 0.64,
      "grad_norm": 14.279778480529785,
      "learning_rate": 7.410791297958663e-06,
      "loss": 1.4851,
      "step": 20673
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8931103944778442,
      "learning_rate": 7.410152381560874e-06,
      "loss": 1.0194,
      "step": 20674
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.641318678855896,
      "learning_rate": 7.409513465163084e-06,
      "loss": 0.8916,
      "step": 20675
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0521905422210693,
      "learning_rate": 7.408874548765295e-06,
      "loss": 0.8936,
      "step": 20676
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.435199499130249,
      "learning_rate": 7.408235632367506e-06,
      "loss": 0.3545,
      "step": 20677
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0268726348876953,
      "learning_rate": 7.407596715969716e-06,
      "loss": 0.9335,
      "step": 20678
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8553001880645752,
      "learning_rate": 7.406957799571926e-06,
      "loss": 0.8221,
      "step": 20679
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.436142086982727,
      "learning_rate": 7.406318883174137e-06,
      "loss": 0.8161,
      "step": 20680
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5358548164367676,
      "learning_rate": 7.405679966776348e-06,
      "loss": 0.8185,
      "step": 20681
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.943167805671692,
      "learning_rate": 7.405041050378558e-06,
      "loss": 0.9063,
      "step": 20682
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.534724235534668,
      "learning_rate": 7.404402133980769e-06,
      "loss": 0.8001,
      "step": 20683
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.773953914642334,
      "learning_rate": 7.40376321758298e-06,
      "loss": 0.9102,
      "step": 20684
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6469476222991943,
      "learning_rate": 7.4031243011851905e-06,
      "loss": 0.8384,
      "step": 20685
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6163469552993774,
      "learning_rate": 7.402485384787401e-06,
      "loss": 0.3587,
      "step": 20686
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.939571738243103,
      "learning_rate": 7.401846468389612e-06,
      "loss": 0.8671,
      "step": 20687
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.323933124542236,
      "learning_rate": 7.401207551991823e-06,
      "loss": 0.9496,
      "step": 20688
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5883166790008545,
      "learning_rate": 7.400568635594033e-06,
      "loss": 0.4429,
      "step": 20689
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5544695854187012,
      "learning_rate": 7.399929719196244e-06,
      "loss": 0.8289,
      "step": 20690
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8028465509414673,
      "learning_rate": 7.399290802798455e-06,
      "loss": 0.9503,
      "step": 20691
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.550333023071289,
      "learning_rate": 7.398651886400665e-06,
      "loss": 0.8491,
      "step": 20692
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8256590366363525,
      "learning_rate": 7.398012970002876e-06,
      "loss": 0.8504,
      "step": 20693
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0002360343933105,
      "learning_rate": 7.397374053605087e-06,
      "loss": 0.8567,
      "step": 20694
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6505202054977417,
      "learning_rate": 7.3967351372072975e-06,
      "loss": 0.6642,
      "step": 20695
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1742970943450928,
      "learning_rate": 7.396096220809507e-06,
      "loss": 1.0692,
      "step": 20696
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2105965614318848,
      "learning_rate": 7.395457304411718e-06,
      "loss": 0.7159,
      "step": 20697
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.752981424331665,
      "learning_rate": 7.394818388013929e-06,
      "loss": 0.9278,
      "step": 20698
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.641801357269287,
      "learning_rate": 7.3941794716161394e-06,
      "loss": 0.3943,
      "step": 20699
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5194692611694336,
      "learning_rate": 7.39354055521835e-06,
      "loss": 0.7641,
      "step": 20700
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.079373359680176,
      "learning_rate": 7.392901638820561e-06,
      "loss": 0.9517,
      "step": 20701
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.350126028060913,
      "learning_rate": 7.3922627224227715e-06,
      "loss": 0.7781,
      "step": 20702
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7061816453933716,
      "learning_rate": 7.391623806024982e-06,
      "loss": 0.9802,
      "step": 20703
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8295978307724,
      "learning_rate": 7.390984889627193e-06,
      "loss": 0.872,
      "step": 20704
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9137356281280518,
      "learning_rate": 7.390345973229404e-06,
      "loss": 0.8117,
      "step": 20705
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0866053104400635,
      "learning_rate": 7.389707056831614e-06,
      "loss": 0.9364,
      "step": 20706
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.612892508506775,
      "learning_rate": 7.389068140433825e-06,
      "loss": 0.8331,
      "step": 20707
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.424477577209473,
      "learning_rate": 7.388429224036036e-06,
      "loss": 1.1123,
      "step": 20708
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0184972286224365,
      "learning_rate": 7.3877903076382464e-06,
      "loss": 1.0565,
      "step": 20709
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6404438018798828,
      "learning_rate": 7.387151391240457e-06,
      "loss": 0.8711,
      "step": 20710
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5596556663513184,
      "learning_rate": 7.386512474842668e-06,
      "loss": 0.8455,
      "step": 20711
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2975621223449707,
      "learning_rate": 7.3858735584448785e-06,
      "loss": 0.9314,
      "step": 20712
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9227089881896973,
      "learning_rate": 7.385234642047089e-06,
      "loss": 0.8516,
      "step": 20713
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.22453236579895,
      "learning_rate": 7.384595725649299e-06,
      "loss": 0.7182,
      "step": 20714
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.348391532897949,
      "learning_rate": 7.38395680925151e-06,
      "loss": 0.737,
      "step": 20715
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.655001640319824,
      "learning_rate": 7.3833178928537205e-06,
      "loss": 1.1743,
      "step": 20716
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.00248384475708,
      "learning_rate": 7.382678976455931e-06,
      "loss": 1.015,
      "step": 20717
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9289486408233643,
      "learning_rate": 7.382040060058142e-06,
      "loss": 0.8715,
      "step": 20718
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6399574279785156,
      "learning_rate": 7.381401143660353e-06,
      "loss": 0.8741,
      "step": 20719
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.717043876647949,
      "learning_rate": 7.380762227262563e-06,
      "loss": 1.3867,
      "step": 20720
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8431047201156616,
      "learning_rate": 7.380123310864774e-06,
      "loss": 1.0283,
      "step": 20721
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.3258564472198486,
      "learning_rate": 7.379484394466985e-06,
      "loss": 0.8288,
      "step": 20722
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.480180501937866,
      "learning_rate": 7.378845478069195e-06,
      "loss": 0.8404,
      "step": 20723
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8496003150939941,
      "learning_rate": 7.378206561671406e-06,
      "loss": 0.7513,
      "step": 20724
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5010942220687866,
      "learning_rate": 7.377567645273617e-06,
      "loss": 0.8551,
      "step": 20725
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.778289794921875,
      "learning_rate": 7.3769287288758275e-06,
      "loss": 0.803,
      "step": 20726
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.434738636016846,
      "learning_rate": 7.376289812478038e-06,
      "loss": 1.2505,
      "step": 20727
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.432727813720703,
      "learning_rate": 7.375650896080249e-06,
      "loss": 1.7256,
      "step": 20728
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7285555601119995,
      "learning_rate": 7.37501197968246e-06,
      "loss": 0.9242,
      "step": 20729
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.888714075088501,
      "learning_rate": 7.37437306328467e-06,
      "loss": 0.9086,
      "step": 20730
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.054534435272217,
      "learning_rate": 7.373734146886881e-06,
      "loss": 0.7452,
      "step": 20731
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.924115777015686,
      "learning_rate": 7.373095230489092e-06,
      "loss": 0.8639,
      "step": 20732
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2253057956695557,
      "learning_rate": 7.3724563140913015e-06,
      "loss": 0.7369,
      "step": 20733
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5092707872390747,
      "learning_rate": 7.371817397693512e-06,
      "loss": 0.4015,
      "step": 20734
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.4516987800598145,
      "learning_rate": 7.371178481295723e-06,
      "loss": 1.1141,
      "step": 20735
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2960243225097656,
      "learning_rate": 7.370539564897934e-06,
      "loss": 0.9639,
      "step": 20736
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.140256643295288,
      "learning_rate": 7.369900648500144e-06,
      "loss": 0.9364,
      "step": 20737
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.205491542816162,
      "learning_rate": 7.369261732102355e-06,
      "loss": 0.794,
      "step": 20738
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.345459222793579,
      "learning_rate": 7.368622815704566e-06,
      "loss": 0.6644,
      "step": 20739
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4838505983352661,
      "learning_rate": 7.3679838993067764e-06,
      "loss": 0.8736,
      "step": 20740
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7717612981796265,
      "learning_rate": 7.367344982908987e-06,
      "loss": 1.0483,
      "step": 20741
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.321840763092041,
      "learning_rate": 7.366706066511198e-06,
      "loss": 1.221,
      "step": 20742
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.429909706115723,
      "learning_rate": 7.3660671501134085e-06,
      "loss": 0.6764,
      "step": 20743
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0466854572296143,
      "learning_rate": 7.365428233715619e-06,
      "loss": 0.9034,
      "step": 20744
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4550631046295166,
      "learning_rate": 7.36478931731783e-06,
      "loss": 0.9086,
      "step": 20745
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.6272199153900146,
      "learning_rate": 7.364150400920041e-06,
      "loss": 1.0139,
      "step": 20746
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.20936918258667,
      "learning_rate": 7.363511484522251e-06,
      "loss": 1.2636,
      "step": 20747
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4948668479919434,
      "learning_rate": 7.362872568124462e-06,
      "loss": 0.7657,
      "step": 20748
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5385252237319946,
      "learning_rate": 7.362233651726673e-06,
      "loss": 0.8886,
      "step": 20749
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5730012655258179,
      "learning_rate": 7.3615947353288834e-06,
      "loss": 0.9188,
      "step": 20750
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7156944274902344,
      "learning_rate": 7.360955818931093e-06,
      "loss": 0.8402,
      "step": 20751
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.213378667831421,
      "learning_rate": 7.360316902533304e-06,
      "loss": 1.2123,
      "step": 20752
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0428924560546875,
      "learning_rate": 7.359677986135515e-06,
      "loss": 0.8517,
      "step": 20753
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5983402729034424,
      "learning_rate": 7.359039069737725e-06,
      "loss": 1.2301,
      "step": 20754
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.926071286201477,
      "learning_rate": 7.358400153339936e-06,
      "loss": 0.9454,
      "step": 20755
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.999868392944336,
      "learning_rate": 7.357761236942147e-06,
      "loss": 0.928,
      "step": 20756
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7830719947814941,
      "learning_rate": 7.3571223205443575e-06,
      "loss": 0.8694,
      "step": 20757
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.042633056640625,
      "learning_rate": 7.356483404146568e-06,
      "loss": 0.7712,
      "step": 20758
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.270099639892578,
      "learning_rate": 7.355844487748779e-06,
      "loss": 0.9645,
      "step": 20759
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.393657684326172,
      "learning_rate": 7.35520557135099e-06,
      "loss": 0.8833,
      "step": 20760
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.693718671798706,
      "learning_rate": 7.3545666549532e-06,
      "loss": 0.7614,
      "step": 20761
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9285985231399536,
      "learning_rate": 7.353927738555411e-06,
      "loss": 0.8502,
      "step": 20762
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.093469142913818,
      "learning_rate": 7.353288822157622e-06,
      "loss": 1.3958,
      "step": 20763
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7971203327178955,
      "learning_rate": 7.352649905759832e-06,
      "loss": 0.9669,
      "step": 20764
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.120834827423096,
      "learning_rate": 7.352010989362043e-06,
      "loss": 1.1655,
      "step": 20765
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.715736150741577,
      "learning_rate": 7.351372072964254e-06,
      "loss": 0.9859,
      "step": 20766
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7846797704696655,
      "learning_rate": 7.3507331565664645e-06,
      "loss": 0.888,
      "step": 20767
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6944999694824219,
      "learning_rate": 7.350094240168675e-06,
      "loss": 0.8956,
      "step": 20768
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8059560060501099,
      "learning_rate": 7.349455323770885e-06,
      "loss": 0.8637,
      "step": 20769
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4098470211029053,
      "learning_rate": 7.348816407373096e-06,
      "loss": 0.9657,
      "step": 20770
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.810877799987793,
      "learning_rate": 7.3481774909753064e-06,
      "loss": 0.9746,
      "step": 20771
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7889320850372314,
      "learning_rate": 7.347538574577517e-06,
      "loss": 0.9812,
      "step": 20772
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5796096324920654,
      "learning_rate": 7.346899658179728e-06,
      "loss": 0.9065,
      "step": 20773
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.067701816558838,
      "learning_rate": 7.3462607417819385e-06,
      "loss": 1.2885,
      "step": 20774
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.407242774963379,
      "learning_rate": 7.345621825384149e-06,
      "loss": 0.6563,
      "step": 20775
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7103478908538818,
      "learning_rate": 7.34498290898636e-06,
      "loss": 0.8916,
      "step": 20776
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8863539695739746,
      "learning_rate": 7.344343992588571e-06,
      "loss": 0.9079,
      "step": 20777
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.705064296722412,
      "learning_rate": 7.343705076190781e-06,
      "loss": 1.3991,
      "step": 20778
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6855144500732422,
      "learning_rate": 7.343066159792992e-06,
      "loss": 0.9622,
      "step": 20779
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7390356063842773,
      "learning_rate": 7.342427243395203e-06,
      "loss": 0.7657,
      "step": 20780
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2993738651275635,
      "learning_rate": 7.3417883269974134e-06,
      "loss": 0.3973,
      "step": 20781
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1712710857391357,
      "learning_rate": 7.341149410599624e-06,
      "loss": 0.9198,
      "step": 20782
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.863272190093994,
      "learning_rate": 7.340510494201835e-06,
      "loss": 0.6191,
      "step": 20783
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6342923641204834,
      "learning_rate": 7.3398715778040455e-06,
      "loss": 0.8949,
      "step": 20784
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4708616733551025,
      "learning_rate": 7.339232661406256e-06,
      "loss": 1.0656,
      "step": 20785
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5555572509765625,
      "learning_rate": 7.338593745008467e-06,
      "loss": 0.7804,
      "step": 20786
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.554874420166016,
      "learning_rate": 7.337954828610677e-06,
      "loss": 1.2008,
      "step": 20787
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9605761766433716,
      "learning_rate": 7.3373159122128875e-06,
      "loss": 0.7011,
      "step": 20788
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.111351490020752,
      "learning_rate": 7.336676995815098e-06,
      "loss": 0.9616,
      "step": 20789
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7151951789855957,
      "learning_rate": 7.336038079417309e-06,
      "loss": 0.9167,
      "step": 20790
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7802894115447998,
      "learning_rate": 7.33539916301952e-06,
      "loss": 0.8504,
      "step": 20791
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.947798252105713,
      "learning_rate": 7.33476024662173e-06,
      "loss": 1.2852,
      "step": 20792
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6489086151123047,
      "learning_rate": 7.334121330223941e-06,
      "loss": 0.8172,
      "step": 20793
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.742532730102539,
      "learning_rate": 7.333482413826152e-06,
      "loss": 0.8963,
      "step": 20794
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9683507680892944,
      "learning_rate": 7.332843497428362e-06,
      "loss": 0.9123,
      "step": 20795
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.267101287841797,
      "learning_rate": 7.332204581030573e-06,
      "loss": 0.879,
      "step": 20796
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2372889518737793,
      "learning_rate": 7.331565664632784e-06,
      "loss": 1.2547,
      "step": 20797
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7247989177703857,
      "learning_rate": 7.3309267482349945e-06,
      "loss": 0.9088,
      "step": 20798
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2045395374298096,
      "learning_rate": 7.330287831837205e-06,
      "loss": 0.8521,
      "step": 20799
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.4052207469940186,
      "learning_rate": 7.329648915439416e-06,
      "loss": 0.5054,
      "step": 20800
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7190419435501099,
      "learning_rate": 7.329009999041627e-06,
      "loss": 0.9069,
      "step": 20801
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.6985232830047607,
      "learning_rate": 7.328371082643837e-06,
      "loss": 1.0586,
      "step": 20802
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0686848163604736,
      "learning_rate": 7.327732166246048e-06,
      "loss": 0.8559,
      "step": 20803
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5871609449386597,
      "learning_rate": 7.327093249848259e-06,
      "loss": 1.1091,
      "step": 20804
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.595780849456787,
      "learning_rate": 7.326454333450469e-06,
      "loss": 0.8934,
      "step": 20805
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9962453842163086,
      "learning_rate": 7.325815417052679e-06,
      "loss": 0.8347,
      "step": 20806
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9709068536758423,
      "learning_rate": 7.32517650065489e-06,
      "loss": 0.9381,
      "step": 20807
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3159682750701904,
      "learning_rate": 7.324537584257101e-06,
      "loss": 1.098,
      "step": 20808
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.07371187210083,
      "learning_rate": 7.323898667859311e-06,
      "loss": 0.9001,
      "step": 20809
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7935913801193237,
      "learning_rate": 7.323259751461522e-06,
      "loss": 0.8882,
      "step": 20810
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8409883975982666,
      "learning_rate": 7.322620835063733e-06,
      "loss": 0.881,
      "step": 20811
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0354015827178955,
      "learning_rate": 7.3219819186659434e-06,
      "loss": 0.9482,
      "step": 20812
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6544867753982544,
      "learning_rate": 7.321343002268154e-06,
      "loss": 0.8358,
      "step": 20813
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1036791801452637,
      "learning_rate": 7.320704085870365e-06,
      "loss": 0.9743,
      "step": 20814
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.1050705909729,
      "learning_rate": 7.3200651694725755e-06,
      "loss": 0.902,
      "step": 20815
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7976951599121094,
      "learning_rate": 7.319426253074786e-06,
      "loss": 0.8478,
      "step": 20816
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.551718711853027,
      "learning_rate": 7.318787336676997e-06,
      "loss": 1.6069,
      "step": 20817
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.085324764251709,
      "learning_rate": 7.318148420279208e-06,
      "loss": 1.0191,
      "step": 20818
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.110154390335083,
      "learning_rate": 7.317509503881418e-06,
      "loss": 1.1337,
      "step": 20819
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.0648932456970215,
      "learning_rate": 7.316870587483629e-06,
      "loss": 0.8953,
      "step": 20820
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.907550096511841,
      "learning_rate": 7.31623167108584e-06,
      "loss": 0.7933,
      "step": 20821
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.6841304302215576,
      "learning_rate": 7.3155927546880504e-06,
      "loss": 0.7654,
      "step": 20822
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8022327423095703,
      "learning_rate": 7.314953838290261e-06,
      "loss": 0.3696,
      "step": 20823
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0520706176757812,
      "learning_rate": 7.314314921892471e-06,
      "loss": 0.9921,
      "step": 20824
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6142878532409668,
      "learning_rate": 7.313676005494682e-06,
      "loss": 0.8743,
      "step": 20825
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6191810369491577,
      "learning_rate": 7.313037089096892e-06,
      "loss": 0.9052,
      "step": 20826
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1294193267822266,
      "learning_rate": 7.312398172699103e-06,
      "loss": 0.8052,
      "step": 20827
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5110477209091187,
      "learning_rate": 7.311759256301314e-06,
      "loss": 0.7881,
      "step": 20828
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6381880044937134,
      "learning_rate": 7.3111203399035245e-06,
      "loss": 0.8649,
      "step": 20829
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.655327320098877,
      "learning_rate": 7.310481423505735e-06,
      "loss": 0.8555,
      "step": 20830
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2976770401000977,
      "learning_rate": 7.309842507107946e-06,
      "loss": 0.8051,
      "step": 20831
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.789757490158081,
      "learning_rate": 7.309203590710157e-06,
      "loss": 1.2048,
      "step": 20832
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.119607925415039,
      "learning_rate": 7.308564674312367e-06,
      "loss": 0.7962,
      "step": 20833
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8059049844741821,
      "learning_rate": 7.307925757914578e-06,
      "loss": 0.8132,
      "step": 20834
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.198992967605591,
      "learning_rate": 7.307286841516789e-06,
      "loss": 1.0745,
      "step": 20835
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.2562601566314697,
      "learning_rate": 7.306647925118999e-06,
      "loss": 1.0212,
      "step": 20836
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.186435222625732,
      "learning_rate": 7.30600900872121e-06,
      "loss": 1.1647,
      "step": 20837
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6448513269424438,
      "learning_rate": 7.305370092323421e-06,
      "loss": 0.8046,
      "step": 20838
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.56553316116333,
      "learning_rate": 7.3047311759256315e-06,
      "loss": 0.8305,
      "step": 20839
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5183525085449219,
      "learning_rate": 7.304092259527842e-06,
      "loss": 0.8799,
      "step": 20840
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4421944618225098,
      "learning_rate": 7.303453343130053e-06,
      "loss": 0.7291,
      "step": 20841
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6826705932617188,
      "learning_rate": 7.302814426732263e-06,
      "loss": 0.8216,
      "step": 20842
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.278724193572998,
      "learning_rate": 7.3021755103344734e-06,
      "loss": 0.7755,
      "step": 20843
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9820698499679565,
      "learning_rate": 7.301536593936684e-06,
      "loss": 0.819,
      "step": 20844
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1525826454162598,
      "learning_rate": 7.300897677538895e-06,
      "loss": 0.997,
      "step": 20845
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5866498947143555,
      "learning_rate": 7.3002587611411055e-06,
      "loss": 0.8642,
      "step": 20846
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.6940033435821533,
      "learning_rate": 7.299619844743316e-06,
      "loss": 1.1247,
      "step": 20847
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1633553504943848,
      "learning_rate": 7.298980928345527e-06,
      "loss": 0.8934,
      "step": 20848
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6190552711486816,
      "learning_rate": 7.298342011947738e-06,
      "loss": 0.8571,
      "step": 20849
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5504151582717896,
      "learning_rate": 7.297703095549948e-06,
      "loss": 0.7679,
      "step": 20850
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.2083065509796143,
      "learning_rate": 7.297064179152159e-06,
      "loss": 0.7889,
      "step": 20851
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4882068634033203,
      "learning_rate": 7.29642526275437e-06,
      "loss": 0.9497,
      "step": 20852
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9523510932922363,
      "learning_rate": 7.2957863463565804e-06,
      "loss": 0.7295,
      "step": 20853
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.381450653076172,
      "learning_rate": 7.295147429958791e-06,
      "loss": 1.0896,
      "step": 20854
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1628310680389404,
      "learning_rate": 7.294508513561002e-06,
      "loss": 0.9625,
      "step": 20855
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9614639282226562,
      "learning_rate": 7.2938695971632125e-06,
      "loss": 0.7048,
      "step": 20856
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.106748342514038,
      "learning_rate": 7.293230680765423e-06,
      "loss": 0.8551,
      "step": 20857
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0646350383758545,
      "learning_rate": 7.292591764367634e-06,
      "loss": 0.8947,
      "step": 20858
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.110930919647217,
      "learning_rate": 7.291952847969845e-06,
      "loss": 0.6669,
      "step": 20859
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.134766101837158,
      "learning_rate": 7.2913139315720545e-06,
      "loss": 1.0749,
      "step": 20860
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.704061269760132,
      "learning_rate": 7.290675015174265e-06,
      "loss": 0.5702,
      "step": 20861
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3789161443710327,
      "learning_rate": 7.290036098776476e-06,
      "loss": 0.7794,
      "step": 20862
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1368167400360107,
      "learning_rate": 7.289397182378687e-06,
      "loss": 0.8227,
      "step": 20863
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7828530073165894,
      "learning_rate": 7.288758265980897e-06,
      "loss": 0.879,
      "step": 20864
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7537152767181396,
      "learning_rate": 7.288119349583108e-06,
      "loss": 0.8666,
      "step": 20865
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.7077858448028564,
      "learning_rate": 7.287480433185319e-06,
      "loss": 0.8002,
      "step": 20866
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.92414391040802,
      "learning_rate": 7.286841516787529e-06,
      "loss": 0.7878,
      "step": 20867
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.4450793266296387,
      "learning_rate": 7.28620260038974e-06,
      "loss": 1.0962,
      "step": 20868
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7551978826522827,
      "learning_rate": 7.285563683991951e-06,
      "loss": 0.9169,
      "step": 20869
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3852951526641846,
      "learning_rate": 7.2849247675941615e-06,
      "loss": 1.0781,
      "step": 20870
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3266196250915527,
      "learning_rate": 7.284285851196372e-06,
      "loss": 0.906,
      "step": 20871
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.852372169494629,
      "learning_rate": 7.283646934798582e-06,
      "loss": 1.0387,
      "step": 20872
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0958805084228516,
      "learning_rate": 7.283008018400792e-06,
      "loss": 1.004,
      "step": 20873
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.0426747798919678,
      "learning_rate": 7.282369102003003e-06,
      "loss": 0.5788,
      "step": 20874
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.3065104484558105,
      "learning_rate": 7.281730185605213e-06,
      "loss": 1.0611,
      "step": 20875
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5400148630142212,
      "learning_rate": 7.281091269207424e-06,
      "loss": 0.8636,
      "step": 20876
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.0340213775634766,
      "learning_rate": 7.280452352809635e-06,
      "loss": 1.0063,
      "step": 20877
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.589708685874939,
      "learning_rate": 7.279813436411845e-06,
      "loss": 0.8848,
      "step": 20878
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7173759937286377,
      "learning_rate": 7.279174520014056e-06,
      "loss": 0.8798,
      "step": 20879
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8286725282669067,
      "learning_rate": 7.278535603616267e-06,
      "loss": 0.9246,
      "step": 20880
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9253547191619873,
      "learning_rate": 7.2778966872184775e-06,
      "loss": 0.8047,
      "step": 20881
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9354394674301147,
      "learning_rate": 7.277257770820688e-06,
      "loss": 0.8914,
      "step": 20882
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2841622829437256,
      "learning_rate": 7.276618854422899e-06,
      "loss": 1.0034,
      "step": 20883
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7991986274719238,
      "learning_rate": 7.27597993802511e-06,
      "loss": 0.815,
      "step": 20884
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6689401865005493,
      "learning_rate": 7.27534102162732e-06,
      "loss": 0.7559,
      "step": 20885
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.638986587524414,
      "learning_rate": 7.274702105229531e-06,
      "loss": 0.8598,
      "step": 20886
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.244641065597534,
      "learning_rate": 7.274063188831742e-06,
      "loss": 0.7687,
      "step": 20887
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1024529933929443,
      "learning_rate": 7.273424272433952e-06,
      "loss": 0.8551,
      "step": 20888
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0277249813079834,
      "learning_rate": 7.272785356036163e-06,
      "loss": 0.8933,
      "step": 20889
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5883843898773193,
      "learning_rate": 7.272146439638374e-06,
      "loss": 0.907,
      "step": 20890
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4354287385940552,
      "learning_rate": 7.271507523240584e-06,
      "loss": 0.8459,
      "step": 20891
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4207608699798584,
      "learning_rate": 7.270868606842794e-06,
      "loss": 0.8779,
      "step": 20892
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5659537315368652,
      "learning_rate": 7.270229690445005e-06,
      "loss": 0.8153,
      "step": 20893
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.09617018699646,
      "learning_rate": 7.269590774047216e-06,
      "loss": 0.9765,
      "step": 20894
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.118854522705078,
      "learning_rate": 7.2689518576494265e-06,
      "loss": 1.318,
      "step": 20895
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1562552452087402,
      "learning_rate": 7.268312941251637e-06,
      "loss": 1.0011,
      "step": 20896
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.688690423965454,
      "learning_rate": 7.267674024853848e-06,
      "loss": 0.9214,
      "step": 20897
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9437901973724365,
      "learning_rate": 7.2670351084560586e-06,
      "loss": 0.8596,
      "step": 20898
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3854987621307373,
      "learning_rate": 7.266396192058269e-06,
      "loss": 0.7446,
      "step": 20899
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9430428743362427,
      "learning_rate": 7.26575727566048e-06,
      "loss": 0.94,
      "step": 20900
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.6634182929992676,
      "learning_rate": 7.265118359262691e-06,
      "loss": 1.0667,
      "step": 20901
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6752570867538452,
      "learning_rate": 7.264479442864901e-06,
      "loss": 0.9275,
      "step": 20902
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.36220383644104,
      "learning_rate": 7.263840526467112e-06,
      "loss": 0.7602,
      "step": 20903
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9542067050933838,
      "learning_rate": 7.263201610069323e-06,
      "loss": 1.2604,
      "step": 20904
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9471052885055542,
      "learning_rate": 7.2625626936715335e-06,
      "loss": 0.8538,
      "step": 20905
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.922425627708435,
      "learning_rate": 7.261923777273744e-06,
      "loss": 0.9777,
      "step": 20906
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4357359409332275,
      "learning_rate": 7.261284860875955e-06,
      "loss": 0.8514,
      "step": 20907
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.989288091659546,
      "learning_rate": 7.2606459444781656e-06,
      "loss": 1.2452,
      "step": 20908
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6915780305862427,
      "learning_rate": 7.260007028080375e-06,
      "loss": 0.8296,
      "step": 20909
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8775795698165894,
      "learning_rate": 7.259368111682586e-06,
      "loss": 0.9307,
      "step": 20910
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.580162286758423,
      "learning_rate": 7.258729195284797e-06,
      "loss": 0.7566,
      "step": 20911
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5125596523284912,
      "learning_rate": 7.2580902788870075e-06,
      "loss": 0.7884,
      "step": 20912
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.668586015701294,
      "learning_rate": 7.257451362489218e-06,
      "loss": 0.774,
      "step": 20913
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8331669569015503,
      "learning_rate": 7.256812446091429e-06,
      "loss": 0.7899,
      "step": 20914
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.686794400215149,
      "learning_rate": 7.25617352969364e-06,
      "loss": 0.7939,
      "step": 20915
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4891750812530518,
      "learning_rate": 7.25553461329585e-06,
      "loss": 1.0577,
      "step": 20916
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.327239990234375,
      "learning_rate": 7.254895696898061e-06,
      "loss": 0.8721,
      "step": 20917
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7822465896606445,
      "learning_rate": 7.254256780500272e-06,
      "loss": 0.9135,
      "step": 20918
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.677922010421753,
      "learning_rate": 7.253617864102482e-06,
      "loss": 1.0598,
      "step": 20919
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.78098464012146,
      "learning_rate": 7.252978947704693e-06,
      "loss": 0.8452,
      "step": 20920
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8740854263305664,
      "learning_rate": 7.252340031306904e-06,
      "loss": 0.9394,
      "step": 20921
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5522887706756592,
      "learning_rate": 7.2517011149091145e-06,
      "loss": 0.4069,
      "step": 20922
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5652787685394287,
      "learning_rate": 7.251062198511325e-06,
      "loss": 0.8955,
      "step": 20923
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7102264165878296,
      "learning_rate": 7.250423282113536e-06,
      "loss": 0.8717,
      "step": 20924
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.874439001083374,
      "learning_rate": 7.249784365715747e-06,
      "loss": 0.8326,
      "step": 20925
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6397804021835327,
      "learning_rate": 7.249145449317957e-06,
      "loss": 0.8551,
      "step": 20926
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1130588054656982,
      "learning_rate": 7.248506532920168e-06,
      "loss": 0.736,
      "step": 20927
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.284822463989258,
      "learning_rate": 7.247867616522378e-06,
      "loss": 0.8947,
      "step": 20928
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.687735676765442,
      "learning_rate": 7.2472287001245886e-06,
      "loss": 0.8503,
      "step": 20929
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.832861423492432,
      "learning_rate": 7.246589783726799e-06,
      "loss": 1.1455,
      "step": 20930
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0799317359924316,
      "learning_rate": 7.24595086732901e-06,
      "loss": 0.9178,
      "step": 20931
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7578368186950684,
      "learning_rate": 7.245311950931221e-06,
      "loss": 0.7486,
      "step": 20932
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.912849187850952,
      "learning_rate": 7.244673034533431e-06,
      "loss": 0.6236,
      "step": 20933
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.8162448406219482,
      "learning_rate": 7.244034118135642e-06,
      "loss": 0.6652,
      "step": 20934
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4487171173095703,
      "learning_rate": 7.243395201737853e-06,
      "loss": 0.3694,
      "step": 20935
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7703492641448975,
      "learning_rate": 7.2427562853400635e-06,
      "loss": 0.6948,
      "step": 20936
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8064807653427124,
      "learning_rate": 7.242117368942274e-06,
      "loss": 1.177,
      "step": 20937
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8977879285812378,
      "learning_rate": 7.241478452544485e-06,
      "loss": 0.8417,
      "step": 20938
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5106767416000366,
      "learning_rate": 7.2408395361466956e-06,
      "loss": 1.1468,
      "step": 20939
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.271450042724609,
      "learning_rate": 7.240200619748906e-06,
      "loss": 1.0378,
      "step": 20940
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6259759664535522,
      "learning_rate": 7.239561703351117e-06,
      "loss": 0.853,
      "step": 20941
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.106884479522705,
      "learning_rate": 7.238922786953328e-06,
      "loss": 0.9075,
      "step": 20942
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2959926128387451,
      "learning_rate": 7.238283870555538e-06,
      "loss": 0.4874,
      "step": 20943
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6156153678894043,
      "learning_rate": 7.237644954157749e-06,
      "loss": 0.8289,
      "step": 20944
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.9777655601501465,
      "learning_rate": 7.23700603775996e-06,
      "loss": 1.1044,
      "step": 20945
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.566088914871216,
      "learning_rate": 7.23636712136217e-06,
      "loss": 1.015,
      "step": 20946
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8280898332595825,
      "learning_rate": 7.23572820496438e-06,
      "loss": 0.9063,
      "step": 20947
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5369536876678467,
      "learning_rate": 7.235089288566591e-06,
      "loss": 0.8377,
      "step": 20948
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7905850410461426,
      "learning_rate": 7.234450372168802e-06,
      "loss": 0.898,
      "step": 20949
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.226746559143066,
      "learning_rate": 7.233811455771012e-06,
      "loss": 1.5796,
      "step": 20950
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1341025829315186,
      "learning_rate": 7.233172539373223e-06,
      "loss": 1.3763,
      "step": 20951
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7142903804779053,
      "learning_rate": 7.232533622975434e-06,
      "loss": 0.8232,
      "step": 20952
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8736752271652222,
      "learning_rate": 7.2318947065776445e-06,
      "loss": 0.9127,
      "step": 20953
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5582963228225708,
      "learning_rate": 7.231255790179855e-06,
      "loss": 0.8788,
      "step": 20954
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1595795154571533,
      "learning_rate": 7.230616873782066e-06,
      "loss": 1.1275,
      "step": 20955
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8489712476730347,
      "learning_rate": 7.229977957384277e-06,
      "loss": 0.8852,
      "step": 20956
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7880407571792603,
      "learning_rate": 7.229339040986487e-06,
      "loss": 0.8687,
      "step": 20957
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.702965259552002,
      "learning_rate": 7.228700124588698e-06,
      "loss": 0.8269,
      "step": 20958
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.3076300621032715,
      "learning_rate": 7.228061208190909e-06,
      "loss": 0.7477,
      "step": 20959
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6764074563980103,
      "learning_rate": 7.227422291793119e-06,
      "loss": 0.9102,
      "step": 20960
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.591441035270691,
      "learning_rate": 7.22678337539533e-06,
      "loss": 0.9238,
      "step": 20961
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3234775066375732,
      "learning_rate": 7.226144458997541e-06,
      "loss": 0.6771,
      "step": 20962
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.718623638153076,
      "learning_rate": 7.2255055425997515e-06,
      "loss": 1.0971,
      "step": 20963
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9043464660644531,
      "learning_rate": 7.224866626201961e-06,
      "loss": 0.992,
      "step": 20964
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6717644929885864,
      "learning_rate": 7.224227709804172e-06,
      "loss": 0.851,
      "step": 20965
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7515349388122559,
      "learning_rate": 7.223588793406383e-06,
      "loss": 0.8975,
      "step": 20966
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0200235843658447,
      "learning_rate": 7.2229498770085935e-06,
      "loss": 0.9664,
      "step": 20967
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8640695810317993,
      "learning_rate": 7.222310960610804e-06,
      "loss": 1.1312,
      "step": 20968
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0556960105895996,
      "learning_rate": 7.221672044213015e-06,
      "loss": 0.6326,
      "step": 20969
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.531689167022705,
      "learning_rate": 7.2210331278152256e-06,
      "loss": 0.8531,
      "step": 20970
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.678593635559082,
      "learning_rate": 7.220394211417436e-06,
      "loss": 0.8643,
      "step": 20971
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8701897859573364,
      "learning_rate": 7.219755295019647e-06,
      "loss": 1.3003,
      "step": 20972
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.6182847023010254,
      "learning_rate": 7.219116378621858e-06,
      "loss": 1.0155,
      "step": 20973
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9536006450653076,
      "learning_rate": 7.218477462224068e-06,
      "loss": 0.9892,
      "step": 20974
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8288387060165405,
      "learning_rate": 7.217838545826279e-06,
      "loss": 0.7465,
      "step": 20975
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.096705913543701,
      "learning_rate": 7.21719962942849e-06,
      "loss": 0.8344,
      "step": 20976
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.436834335327148,
      "learning_rate": 7.2165607130307005e-06,
      "loss": 1.314,
      "step": 20977
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.929085612297058,
      "learning_rate": 7.215921796632911e-06,
      "loss": 0.9258,
      "step": 20978
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.8004684448242188,
      "learning_rate": 7.215282880235122e-06,
      "loss": 1.2576,
      "step": 20979
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2345376014709473,
      "learning_rate": 7.2146439638373326e-06,
      "loss": 0.9191,
      "step": 20980
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.025191307067871,
      "learning_rate": 7.214005047439543e-06,
      "loss": 0.8297,
      "step": 20981
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.183894157409668,
      "learning_rate": 7.213366131041753e-06,
      "loss": 0.9592,
      "step": 20982
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9012571573257446,
      "learning_rate": 7.212727214643964e-06,
      "loss": 0.8503,
      "step": 20983
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.7404215335845947,
      "learning_rate": 7.2120882982461745e-06,
      "loss": 1.0704,
      "step": 20984
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7326303720474243,
      "learning_rate": 7.211449381848385e-06,
      "loss": 0.8413,
      "step": 20985
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9351797103881836,
      "learning_rate": 7.210810465450596e-06,
      "loss": 0.8512,
      "step": 20986
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.280272960662842,
      "learning_rate": 7.210171549052807e-06,
      "loss": 1.3481,
      "step": 20987
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6678595542907715,
      "learning_rate": 7.209532632655017e-06,
      "loss": 0.9014,
      "step": 20988
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7219361066818237,
      "learning_rate": 7.208893716257228e-06,
      "loss": 0.7631,
      "step": 20989
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.859575867652893,
      "learning_rate": 7.208254799859439e-06,
      "loss": 0.7571,
      "step": 20990
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.2780678272247314,
      "learning_rate": 7.207615883461649e-06,
      "loss": 0.943,
      "step": 20991
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7411905527114868,
      "learning_rate": 7.20697696706386e-06,
      "loss": 0.8296,
      "step": 20992
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6980063915252686,
      "learning_rate": 7.206338050666071e-06,
      "loss": 0.8713,
      "step": 20993
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6328977346420288,
      "learning_rate": 7.2056991342682815e-06,
      "loss": 0.8027,
      "step": 20994
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7943925857543945,
      "learning_rate": 7.205060217870492e-06,
      "loss": 0.8178,
      "step": 20995
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4762167930603027,
      "learning_rate": 7.204421301472703e-06,
      "loss": 0.8631,
      "step": 20996
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.816081166267395,
      "learning_rate": 7.203782385074914e-06,
      "loss": 0.4058,
      "step": 20997
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.6280581951141357,
      "learning_rate": 7.203143468677124e-06,
      "loss": 1.1713,
      "step": 20998
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9307315349578857,
      "learning_rate": 7.202504552279335e-06,
      "loss": 0.846,
      "step": 20999
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7447080612182617,
      "learning_rate": 7.201865635881546e-06,
      "loss": 1.0572,
      "step": 21000
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.5900752544403076,
      "learning_rate": 7.2012267194837556e-06,
      "loss": 0.6437,
      "step": 21001
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5526102781295776,
      "learning_rate": 7.200587803085966e-06,
      "loss": 0.7708,
      "step": 21002
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3756482601165771,
      "learning_rate": 7.199948886688177e-06,
      "loss": 0.3645,
      "step": 21003
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0165300369262695,
      "learning_rate": 7.199309970290388e-06,
      "loss": 0.9593,
      "step": 21004
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.801586627960205,
      "learning_rate": 7.198671053892598e-06,
      "loss": 1.3611,
      "step": 21005
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9209864139556885,
      "learning_rate": 7.198032137494809e-06,
      "loss": 0.816,
      "step": 21006
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.009129047393799,
      "learning_rate": 7.19739322109702e-06,
      "loss": 0.824,
      "step": 21007
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1731953620910645,
      "learning_rate": 7.1967543046992305e-06,
      "loss": 0.8391,
      "step": 21008
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1270411014556885,
      "learning_rate": 7.196115388301441e-06,
      "loss": 0.9173,
      "step": 21009
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.026233673095703,
      "learning_rate": 7.195476471903652e-06,
      "loss": 0.8429,
      "step": 21010
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5887147188186646,
      "learning_rate": 7.1948375555058626e-06,
      "loss": 0.8155,
      "step": 21011
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.947043418884277,
      "learning_rate": 7.194198639108073e-06,
      "loss": 1.6323,
      "step": 21012
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8663660287857056,
      "learning_rate": 7.193559722710284e-06,
      "loss": 0.8712,
      "step": 21013
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7551453113555908,
      "learning_rate": 7.192920806312495e-06,
      "loss": 1.0545,
      "step": 21014
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.489873886108398,
      "learning_rate": 7.192281889914705e-06,
      "loss": 1.5551,
      "step": 21015
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.666021704673767,
      "learning_rate": 7.191642973516916e-06,
      "loss": 0.9595,
      "step": 21016
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.813538670539856,
      "learning_rate": 7.191004057119127e-06,
      "loss": 0.8747,
      "step": 21017
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.590935707092285,
      "learning_rate": 7.1903651407213375e-06,
      "loss": 0.8376,
      "step": 21018
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.400242328643799,
      "learning_rate": 7.189726224323547e-06,
      "loss": 0.9897,
      "step": 21019
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0463874340057373,
      "learning_rate": 7.189087307925758e-06,
      "loss": 0.9119,
      "step": 21020
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1562938690185547,
      "learning_rate": 7.188448391527969e-06,
      "loss": 0.7681,
      "step": 21021
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3569347858428955,
      "learning_rate": 7.187809475130179e-06,
      "loss": 0.8663,
      "step": 21022
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6366100311279297,
      "learning_rate": 7.18717055873239e-06,
      "loss": 0.8503,
      "step": 21023
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.497920274734497,
      "learning_rate": 7.186531642334601e-06,
      "loss": 0.7079,
      "step": 21024
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8386913537979126,
      "learning_rate": 7.1858927259368115e-06,
      "loss": 0.8836,
      "step": 21025
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5469924211502075,
      "learning_rate": 7.185253809539022e-06,
      "loss": 0.8389,
      "step": 21026
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2066752910614014,
      "learning_rate": 7.184614893141233e-06,
      "loss": 0.8834,
      "step": 21027
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.095618963241577,
      "learning_rate": 7.183975976743444e-06,
      "loss": 0.8609,
      "step": 21028
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.595349073410034,
      "learning_rate": 7.183337060345654e-06,
      "loss": 1.1254,
      "step": 21029
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3816674947738647,
      "learning_rate": 7.182698143947865e-06,
      "loss": 0.8016,
      "step": 21030
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8914389610290527,
      "learning_rate": 7.182059227550076e-06,
      "loss": 0.9092,
      "step": 21031
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8452723026275635,
      "learning_rate": 7.181420311152286e-06,
      "loss": 0.8776,
      "step": 21032
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7714025974273682,
      "learning_rate": 7.180781394754497e-06,
      "loss": 0.7279,
      "step": 21033
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0361504554748535,
      "learning_rate": 7.180142478356708e-06,
      "loss": 0.9651,
      "step": 21034
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6302599906921387,
      "learning_rate": 7.1795035619589185e-06,
      "loss": 0.7548,
      "step": 21035
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7255353927612305,
      "learning_rate": 7.178864645561129e-06,
      "loss": 0.94,
      "step": 21036
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8471248149871826,
      "learning_rate": 7.178225729163339e-06,
      "loss": 0.8187,
      "step": 21037
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1846940517425537,
      "learning_rate": 7.17758681276555e-06,
      "loss": 0.791,
      "step": 21038
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.813087821006775,
      "learning_rate": 7.1769478963677605e-06,
      "loss": 1.0037,
      "step": 21039
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.817248582839966,
      "learning_rate": 7.176308979969971e-06,
      "loss": 0.8021,
      "step": 21040
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0852982997894287,
      "learning_rate": 7.175670063572182e-06,
      "loss": 1.089,
      "step": 21041
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.338271379470825,
      "learning_rate": 7.1750311471743926e-06,
      "loss": 0.782,
      "step": 21042
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6656317710876465,
      "learning_rate": 7.174392230776603e-06,
      "loss": 0.9227,
      "step": 21043
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.315624237060547,
      "learning_rate": 7.173753314378814e-06,
      "loss": 0.6311,
      "step": 21044
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5945416688919067,
      "learning_rate": 7.173114397981025e-06,
      "loss": 0.8416,
      "step": 21045
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.4780707359313965,
      "learning_rate": 7.172475481583235e-06,
      "loss": 0.8487,
      "step": 21046
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.182485580444336,
      "learning_rate": 7.171836565185446e-06,
      "loss": 0.7317,
      "step": 21047
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3690522909164429,
      "learning_rate": 7.171197648787657e-06,
      "loss": 0.827,
      "step": 21048
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.8918049335479736,
      "learning_rate": 7.1705587323898675e-06,
      "loss": 0.7719,
      "step": 21049
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1664013862609863,
      "learning_rate": 7.169919815992078e-06,
      "loss": 0.9083,
      "step": 21050
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.9158177375793457,
      "learning_rate": 7.169280899594289e-06,
      "loss": 1.0516,
      "step": 21051
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.955191969871521,
      "learning_rate": 7.1686419831964996e-06,
      "loss": 0.9398,
      "step": 21052
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2359509468078613,
      "learning_rate": 7.16800306679871e-06,
      "loss": 0.6629,
      "step": 21053
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1710219383239746,
      "learning_rate": 7.167364150400921e-06,
      "loss": 0.9215,
      "step": 21054
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.53849196434021,
      "learning_rate": 7.166725234003131e-06,
      "loss": 0.368,
      "step": 21055
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.751427412033081,
      "learning_rate": 7.1660863176053415e-06,
      "loss": 0.8838,
      "step": 21056
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.92805814743042,
      "learning_rate": 7.165447401207552e-06,
      "loss": 0.9029,
      "step": 21057
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.665337324142456,
      "learning_rate": 7.164808484809763e-06,
      "loss": 1.2556,
      "step": 21058
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8247380256652832,
      "learning_rate": 7.164169568411974e-06,
      "loss": 0.8588,
      "step": 21059
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6717222929000854,
      "learning_rate": 7.163530652014184e-06,
      "loss": 0.8907,
      "step": 21060
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.9571735858917236,
      "learning_rate": 7.162891735616395e-06,
      "loss": 0.8518,
      "step": 21061
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.3468737602233887,
      "learning_rate": 7.162252819218606e-06,
      "loss": 0.8314,
      "step": 21062
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.022804021835327,
      "learning_rate": 7.1616139028208164e-06,
      "loss": 1.0511,
      "step": 21063
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.67465341091156,
      "learning_rate": 7.160974986423027e-06,
      "loss": 0.9032,
      "step": 21064
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7358338832855225,
      "learning_rate": 7.160336070025238e-06,
      "loss": 0.8012,
      "step": 21065
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.404097080230713,
      "learning_rate": 7.1596971536274485e-06,
      "loss": 1.2509,
      "step": 21066
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.0389511585235596,
      "learning_rate": 7.159058237229659e-06,
      "loss": 0.6726,
      "step": 21067
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.976181745529175,
      "learning_rate": 7.15841932083187e-06,
      "loss": 1.195,
      "step": 21068
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.841700553894043,
      "learning_rate": 7.157780404434081e-06,
      "loss": 0.8322,
      "step": 21069
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3559417724609375,
      "learning_rate": 7.157141488036291e-06,
      "loss": 0.8254,
      "step": 21070
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8653419017791748,
      "learning_rate": 7.156502571638502e-06,
      "loss": 0.9138,
      "step": 21071
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.7235963344573975,
      "learning_rate": 7.155863655240713e-06,
      "loss": 0.9558,
      "step": 21072
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.251120567321777,
      "learning_rate": 7.1552247388429234e-06,
      "loss": 1.0317,
      "step": 21073
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0174801349639893,
      "learning_rate": 7.154585822445133e-06,
      "loss": 1.0056,
      "step": 21074
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2051842212677,
      "learning_rate": 7.153946906047344e-06,
      "loss": 0.9052,
      "step": 21075
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9898083209991455,
      "learning_rate": 7.153307989649555e-06,
      "loss": 0.8613,
      "step": 21076
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.6844401359558105,
      "learning_rate": 7.152669073251765e-06,
      "loss": 1.086,
      "step": 21077
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.5692811012268066,
      "learning_rate": 7.152030156853976e-06,
      "loss": 1.1633,
      "step": 21078
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6829235553741455,
      "learning_rate": 7.151391240456187e-06,
      "loss": 0.9026,
      "step": 21079
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6241390705108643,
      "learning_rate": 7.1507523240583975e-06,
      "loss": 0.8939,
      "step": 21080
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7084813117980957,
      "learning_rate": 7.150113407660608e-06,
      "loss": 0.9245,
      "step": 21081
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6006110906600952,
      "learning_rate": 7.149474491262819e-06,
      "loss": 0.7422,
      "step": 21082
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7474998235702515,
      "learning_rate": 7.14883557486503e-06,
      "loss": 0.8164,
      "step": 21083
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.8235225677490234,
      "learning_rate": 7.14819665846724e-06,
      "loss": 0.923,
      "step": 21084
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5961018800735474,
      "learning_rate": 7.147557742069451e-06,
      "loss": 0.3745,
      "step": 21085
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2316367626190186,
      "learning_rate": 7.146918825671662e-06,
      "loss": 0.7936,
      "step": 21086
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6075189113616943,
      "learning_rate": 7.146279909273872e-06,
      "loss": 0.7866,
      "step": 21087
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5525493621826172,
      "learning_rate": 7.145640992876083e-06,
      "loss": 0.8615,
      "step": 21088
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.299950361251831,
      "learning_rate": 7.145002076478294e-06,
      "loss": 0.6756,
      "step": 21089
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6066409349441528,
      "learning_rate": 7.1443631600805045e-06,
      "loss": 0.4202,
      "step": 21090
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.754145383834839,
      "learning_rate": 7.143724243682715e-06,
      "loss": 0.9393,
      "step": 21091
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.630595088005066,
      "learning_rate": 7.143085327284925e-06,
      "loss": 0.7962,
      "step": 21092
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.785010814666748,
      "learning_rate": 7.142446410887136e-06,
      "loss": 0.7667,
      "step": 21093
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1477410793304443,
      "learning_rate": 7.1418074944893464e-06,
      "loss": 1.1055,
      "step": 21094
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3909153938293457,
      "learning_rate": 7.141168578091557e-06,
      "loss": 0.9487,
      "step": 21095
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0632333755493164,
      "learning_rate": 7.140529661693768e-06,
      "loss": 0.7534,
      "step": 21096
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2957338094711304,
      "learning_rate": 7.1398907452959785e-06,
      "loss": 0.6985,
      "step": 21097
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9892807006835938,
      "learning_rate": 7.139251828898189e-06,
      "loss": 1.1195,
      "step": 21098
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.608384847640991,
      "learning_rate": 7.1386129125004e-06,
      "loss": 0.7251,
      "step": 21099
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3887431621551514,
      "learning_rate": 7.137973996102611e-06,
      "loss": 0.7149,
      "step": 21100
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6352298259735107,
      "learning_rate": 7.137335079704821e-06,
      "loss": 0.8729,
      "step": 21101
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1570072174072266,
      "learning_rate": 7.136696163307032e-06,
      "loss": 0.9672,
      "step": 21102
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2761120796203613,
      "learning_rate": 7.136057246909243e-06,
      "loss": 0.757,
      "step": 21103
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9011043310165405,
      "learning_rate": 7.1354183305114534e-06,
      "loss": 0.9799,
      "step": 21104
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7276582717895508,
      "learning_rate": 7.134779414113664e-06,
      "loss": 0.8375,
      "step": 21105
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.741287350654602,
      "learning_rate": 7.134140497715875e-06,
      "loss": 0.9518,
      "step": 21106
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.904929518699646,
      "learning_rate": 7.1335015813180855e-06,
      "loss": 1.231,
      "step": 21107
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7667045593261719,
      "learning_rate": 7.132862664920296e-06,
      "loss": 1.0003,
      "step": 21108
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.088587522506714,
      "learning_rate": 7.132223748522507e-06,
      "loss": 0.8727,
      "step": 21109
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.515285849571228,
      "learning_rate": 7.131584832124717e-06,
      "loss": 0.8447,
      "step": 21110
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7478749752044678,
      "learning_rate": 7.1309459157269275e-06,
      "loss": 0.8314,
      "step": 21111
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7535643577575684,
      "learning_rate": 7.130306999329138e-06,
      "loss": 0.8909,
      "step": 21112
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7025374174118042,
      "learning_rate": 7.129668082931349e-06,
      "loss": 0.7833,
      "step": 21113
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9421991109848022,
      "learning_rate": 7.12902916653356e-06,
      "loss": 0.8686,
      "step": 21114
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9683997631072998,
      "learning_rate": 7.12839025013577e-06,
      "loss": 0.9036,
      "step": 21115
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8788325786590576,
      "learning_rate": 7.127751333737981e-06,
      "loss": 1.0757,
      "step": 21116
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6282609701156616,
      "learning_rate": 7.127112417340192e-06,
      "loss": 0.8451,
      "step": 21117
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.859811782836914,
      "learning_rate": 7.126473500942402e-06,
      "loss": 0.7723,
      "step": 21118
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4742363691329956,
      "learning_rate": 7.125834584544613e-06,
      "loss": 0.8697,
      "step": 21119
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3325111865997314,
      "learning_rate": 7.125195668146824e-06,
      "loss": 0.8336,
      "step": 21120
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5895010232925415,
      "learning_rate": 7.1245567517490345e-06,
      "loss": 0.9361,
      "step": 21121
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.582219362258911,
      "learning_rate": 7.123917835351245e-06,
      "loss": 0.9349,
      "step": 21122
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0294108390808105,
      "learning_rate": 7.123278918953456e-06,
      "loss": 0.8261,
      "step": 21123
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4424023628234863,
      "learning_rate": 7.122640002555667e-06,
      "loss": 0.7095,
      "step": 21124
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1711835861206055,
      "learning_rate": 7.122001086157877e-06,
      "loss": 1.0401,
      "step": 21125
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.71786367893219,
      "learning_rate": 7.121362169760088e-06,
      "loss": 0.8716,
      "step": 21126
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7190148830413818,
      "learning_rate": 7.120723253362299e-06,
      "loss": 0.8083,
      "step": 21127
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.61894953250885,
      "learning_rate": 7.1200843369645085e-06,
      "loss": 0.6882,
      "step": 21128
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2738293409347534,
      "learning_rate": 7.119445420566719e-06,
      "loss": 0.4129,
      "step": 21129
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8620851039886475,
      "learning_rate": 7.11880650416893e-06,
      "loss": 0.8437,
      "step": 21130
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.150716543197632,
      "learning_rate": 7.118167587771141e-06,
      "loss": 0.5513,
      "step": 21131
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8586345911026,
      "learning_rate": 7.117528671373351e-06,
      "loss": 0.8839,
      "step": 21132
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0863382816314697,
      "learning_rate": 7.116889754975562e-06,
      "loss": 0.9242,
      "step": 21133
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9202775955200195,
      "learning_rate": 7.116250838577773e-06,
      "loss": 0.94,
      "step": 21134
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5465493202209473,
      "learning_rate": 7.1156119221799834e-06,
      "loss": 0.8322,
      "step": 21135
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.574783444404602,
      "learning_rate": 7.114973005782194e-06,
      "loss": 0.8721,
      "step": 21136
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8226544857025146,
      "learning_rate": 7.114334089384405e-06,
      "loss": 0.8069,
      "step": 21137
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.165644645690918,
      "learning_rate": 7.1136951729866155e-06,
      "loss": 1.212,
      "step": 21138
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5662033557891846,
      "learning_rate": 7.113056256588826e-06,
      "loss": 0.9513,
      "step": 21139
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4540677070617676,
      "learning_rate": 7.112417340191037e-06,
      "loss": 0.8996,
      "step": 21140
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5732765197753906,
      "learning_rate": 7.111778423793248e-06,
      "loss": 0.825,
      "step": 21141
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7888153791427612,
      "learning_rate": 7.111139507395458e-06,
      "loss": 0.9129,
      "step": 21142
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.914238214492798,
      "learning_rate": 7.110500590997669e-06,
      "loss": 1.2484,
      "step": 21143
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.169921398162842,
      "learning_rate": 7.10986167459988e-06,
      "loss": 1.1021,
      "step": 21144
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7436351776123047,
      "learning_rate": 7.1092227582020904e-06,
      "loss": 0.9447,
      "step": 21145
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.7890641689300537,
      "learning_rate": 7.108583841804301e-06,
      "loss": 0.8726,
      "step": 21146
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8261562585830688,
      "learning_rate": 7.107944925406511e-06,
      "loss": 0.8308,
      "step": 21147
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5180189609527588,
      "learning_rate": 7.107306009008722e-06,
      "loss": 0.7376,
      "step": 21148
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.956497311592102,
      "learning_rate": 7.106667092610932e-06,
      "loss": 0.962,
      "step": 21149
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.785385012626648,
      "learning_rate": 7.106028176213143e-06,
      "loss": 0.888,
      "step": 21150
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8620580434799194,
      "learning_rate": 7.105389259815354e-06,
      "loss": 0.8228,
      "step": 21151
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9498344659805298,
      "learning_rate": 7.1047503434175645e-06,
      "loss": 0.879,
      "step": 21152
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6962465047836304,
      "learning_rate": 7.104111427019775e-06,
      "loss": 0.8686,
      "step": 21153
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9179794788360596,
      "learning_rate": 7.103472510621986e-06,
      "loss": 0.9555,
      "step": 21154
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6398968696594238,
      "learning_rate": 7.102833594224197e-06,
      "loss": 0.9872,
      "step": 21155
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9072405099868774,
      "learning_rate": 7.102194677826407e-06,
      "loss": 0.7443,
      "step": 21156
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.2548327445983887,
      "learning_rate": 7.101555761428618e-06,
      "loss": 1.3255,
      "step": 21157
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.833019495010376,
      "learning_rate": 7.100916845030829e-06,
      "loss": 0.8869,
      "step": 21158
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.342606782913208,
      "learning_rate": 7.100277928633039e-06,
      "loss": 0.4048,
      "step": 21159
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5156667232513428,
      "learning_rate": 7.09963901223525e-06,
      "loss": 0.9041,
      "step": 21160
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6480246782302856,
      "learning_rate": 7.099000095837461e-06,
      "loss": 0.8814,
      "step": 21161
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.60443377494812,
      "learning_rate": 7.0983611794396715e-06,
      "loss": 0.8502,
      "step": 21162
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.7101681232452393,
      "learning_rate": 7.097722263041882e-06,
      "loss": 1.1,
      "step": 21163
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.922232985496521,
      "learning_rate": 7.097083346644093e-06,
      "loss": 0.8383,
      "step": 21164
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.606930136680603,
      "learning_rate": 7.096444430246303e-06,
      "loss": 0.8373,
      "step": 21165
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9198626279830933,
      "learning_rate": 7.0958055138485134e-06,
      "loss": 0.8185,
      "step": 21166
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7567899227142334,
      "learning_rate": 7.095166597450724e-06,
      "loss": 0.8473,
      "step": 21167
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7142757177352905,
      "learning_rate": 7.094527681052935e-06,
      "loss": 0.783,
      "step": 21168
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0063424110412598,
      "learning_rate": 7.0938887646551455e-06,
      "loss": 0.9012,
      "step": 21169
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8267462253570557,
      "learning_rate": 7.093249848257356e-06,
      "loss": 0.8709,
      "step": 21170
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.9833731651306152,
      "learning_rate": 7.092610931859567e-06,
      "loss": 0.95,
      "step": 21171
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8541803359985352,
      "learning_rate": 7.091972015461778e-06,
      "loss": 0.8097,
      "step": 21172
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.555354118347168,
      "learning_rate": 7.091333099063988e-06,
      "loss": 0.8466,
      "step": 21173
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3973594903945923,
      "learning_rate": 7.090694182666199e-06,
      "loss": 0.404,
      "step": 21174
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.886035442352295,
      "learning_rate": 7.09005526626841e-06,
      "loss": 0.8668,
      "step": 21175
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.818198561668396,
      "learning_rate": 7.0894163498706204e-06,
      "loss": 0.8826,
      "step": 21176
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.06797456741333,
      "learning_rate": 7.088777433472831e-06,
      "loss": 0.9637,
      "step": 21177
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7526073455810547,
      "learning_rate": 7.088138517075042e-06,
      "loss": 0.741,
      "step": 21178
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.093808174133301,
      "learning_rate": 7.0874996006772525e-06,
      "loss": 0.8726,
      "step": 21179
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6619333028793335,
      "learning_rate": 7.086860684279463e-06,
      "loss": 0.8582,
      "step": 21180
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.359572410583496,
      "learning_rate": 7.086221767881674e-06,
      "loss": 1.0739,
      "step": 21181
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8497697114944458,
      "learning_rate": 7.085582851483885e-06,
      "loss": 0.8499,
      "step": 21182
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6263030767440796,
      "learning_rate": 7.0849439350860945e-06,
      "loss": 0.7878,
      "step": 21183
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.508901596069336,
      "learning_rate": 7.084305018688305e-06,
      "loss": 0.882,
      "step": 21184
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4358834028244019,
      "learning_rate": 7.083666102290516e-06,
      "loss": 0.8369,
      "step": 21185
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5205142498016357,
      "learning_rate": 7.083027185892727e-06,
      "loss": 0.8455,
      "step": 21186
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.4085590839385986,
      "learning_rate": 7.082388269494937e-06,
      "loss": 0.9035,
      "step": 21187
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.90000319480896,
      "learning_rate": 7.081749353097148e-06,
      "loss": 0.8639,
      "step": 21188
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5524245500564575,
      "learning_rate": 7.081110436699359e-06,
      "loss": 0.7759,
      "step": 21189
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0028884410858154,
      "learning_rate": 7.080471520301569e-06,
      "loss": 0.7709,
      "step": 21190
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.1133921146392822,
      "learning_rate": 7.07983260390378e-06,
      "loss": 1.2043,
      "step": 21191
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8079475164413452,
      "learning_rate": 7.079193687505991e-06,
      "loss": 1.1244,
      "step": 21192
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0220420360565186,
      "learning_rate": 7.0785547711082015e-06,
      "loss": 0.9091,
      "step": 21193
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.3983211517333984,
      "learning_rate": 7.077915854710412e-06,
      "loss": 1.1971,
      "step": 21194
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0413589477539062,
      "learning_rate": 7.077276938312623e-06,
      "loss": 0.799,
      "step": 21195
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9427709579467773,
      "learning_rate": 7.076638021914834e-06,
      "loss": 0.8725,
      "step": 21196
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3999658823013306,
      "learning_rate": 7.075999105517044e-06,
      "loss": 0.7473,
      "step": 21197
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6425058841705322,
      "learning_rate": 7.075360189119255e-06,
      "loss": 0.845,
      "step": 21198
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7923450469970703,
      "learning_rate": 7.074721272721466e-06,
      "loss": 0.83,
      "step": 21199
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.924116611480713,
      "learning_rate": 7.074082356323676e-06,
      "loss": 0.8174,
      "step": 21200
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1844725608825684,
      "learning_rate": 7.073443439925886e-06,
      "loss": 0.7197,
      "step": 21201
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.37361216545105,
      "learning_rate": 7.072804523528097e-06,
      "loss": 0.7391,
      "step": 21202
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8999218940734863,
      "learning_rate": 7.072165607130308e-06,
      "loss": 0.8478,
      "step": 21203
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.6610167026519775,
      "learning_rate": 7.071526690732518e-06,
      "loss": 1.0246,
      "step": 21204
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8731510639190674,
      "learning_rate": 7.070887774334729e-06,
      "loss": 0.9504,
      "step": 21205
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8509361743927002,
      "learning_rate": 7.07024885793694e-06,
      "loss": 0.8756,
      "step": 21206
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0143091678619385,
      "learning_rate": 7.0696099415391504e-06,
      "loss": 1.024,
      "step": 21207
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.7438206672668457,
      "learning_rate": 7.068971025141361e-06,
      "loss": 0.7742,
      "step": 21208
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.31585168838501,
      "learning_rate": 7.068332108743572e-06,
      "loss": 1.6571,
      "step": 21209
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9587558507919312,
      "learning_rate": 7.0676931923457825e-06,
      "loss": 1.1032,
      "step": 21210
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7705817222595215,
      "learning_rate": 7.067054275947993e-06,
      "loss": 0.8498,
      "step": 21211
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5551567077636719,
      "learning_rate": 7.066415359550204e-06,
      "loss": 0.7462,
      "step": 21212
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5644984245300293,
      "learning_rate": 7.065776443152415e-06,
      "loss": 0.885,
      "step": 21213
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7887369394302368,
      "learning_rate": 7.065137526754625e-06,
      "loss": 0.8263,
      "step": 21214
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.928518295288086,
      "learning_rate": 7.064498610356836e-06,
      "loss": 0.7044,
      "step": 21215
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.926069259643555,
      "learning_rate": 7.063859693959047e-06,
      "loss": 1.4042,
      "step": 21216
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.09956693649292,
      "learning_rate": 7.0632207775612574e-06,
      "loss": 1.0526,
      "step": 21217
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.8308000564575195,
      "learning_rate": 7.062581861163468e-06,
      "loss": 1.0688,
      "step": 21218
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.831390619277954,
      "learning_rate": 7.061942944765679e-06,
      "loss": 0.889,
      "step": 21219
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2150967121124268,
      "learning_rate": 7.061304028367889e-06,
      "loss": 0.9757,
      "step": 21220
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.713616967201233,
      "learning_rate": 7.060665111970099e-06,
      "loss": 0.8286,
      "step": 21221
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9753303527832031,
      "learning_rate": 7.06002619557231e-06,
      "loss": 0.7985,
      "step": 21222
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.401925563812256,
      "learning_rate": 7.059387279174521e-06,
      "loss": 0.7113,
      "step": 21223
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.0249555110931396,
      "learning_rate": 7.0587483627767315e-06,
      "loss": 0.7236,
      "step": 21224
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.731933832168579,
      "learning_rate": 7.058109446378942e-06,
      "loss": 0.7604,
      "step": 21225
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.692378044128418,
      "learning_rate": 7.057470529981153e-06,
      "loss": 0.9503,
      "step": 21226
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5719594955444336,
      "learning_rate": 7.056831613583364e-06,
      "loss": 0.5771,
      "step": 21227
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0195658206939697,
      "learning_rate": 7.056192697185574e-06,
      "loss": 1.1621,
      "step": 21228
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.439750671386719,
      "learning_rate": 7.055553780787785e-06,
      "loss": 1.8545,
      "step": 21229
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.788175106048584,
      "learning_rate": 7.054914864389996e-06,
      "loss": 0.7969,
      "step": 21230
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.857603549957275,
      "learning_rate": 7.054275947992206e-06,
      "loss": 0.9355,
      "step": 21231
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7860041856765747,
      "learning_rate": 7.053637031594417e-06,
      "loss": 0.9687,
      "step": 21232
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.788835883140564,
      "learning_rate": 7.052998115196628e-06,
      "loss": 0.8842,
      "step": 21233
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.572728633880615,
      "learning_rate": 7.0523591987988385e-06,
      "loss": 1.5342,
      "step": 21234
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4721660614013672,
      "learning_rate": 7.051720282401049e-06,
      "loss": 0.4835,
      "step": 21235
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.219376802444458,
      "learning_rate": 7.05108136600326e-06,
      "loss": 0.9962,
      "step": 21236
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.2404167652130127,
      "learning_rate": 7.050442449605471e-06,
      "loss": 0.9307,
      "step": 21237
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8842899799346924,
      "learning_rate": 7.0498035332076804e-06,
      "loss": 0.8769,
      "step": 21238
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.059859275817871,
      "learning_rate": 7.049164616809891e-06,
      "loss": 0.679,
      "step": 21239
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7051748037338257,
      "learning_rate": 7.048525700412102e-06,
      "loss": 0.8291,
      "step": 21240
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.914114236831665,
      "learning_rate": 7.0478867840143125e-06,
      "loss": 0.8226,
      "step": 21241
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5152502059936523,
      "learning_rate": 7.047247867616523e-06,
      "loss": 0.4349,
      "step": 21242
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.682185411453247,
      "learning_rate": 7.046608951218734e-06,
      "loss": 0.376,
      "step": 21243
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.35194206237793,
      "learning_rate": 7.045970034820945e-06,
      "loss": 1.346,
      "step": 21244
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5371936559677124,
      "learning_rate": 7.045331118423155e-06,
      "loss": 0.7911,
      "step": 21245
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7973600625991821,
      "learning_rate": 7.044692202025366e-06,
      "loss": 0.8599,
      "step": 21246
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.002953052520752,
      "learning_rate": 7.044053285627577e-06,
      "loss": 0.8969,
      "step": 21247
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9178684949874878,
      "learning_rate": 7.0434143692297874e-06,
      "loss": 0.7889,
      "step": 21248
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7919921875,
      "learning_rate": 7.042775452831998e-06,
      "loss": 0.6247,
      "step": 21249
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.170054912567139,
      "learning_rate": 7.042136536434209e-06,
      "loss": 1.489,
      "step": 21250
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6807374954223633,
      "learning_rate": 7.0414976200364195e-06,
      "loss": 0.856,
      "step": 21251
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6833006143569946,
      "learning_rate": 7.04085870363863e-06,
      "loss": 0.8559,
      "step": 21252
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8558343648910522,
      "learning_rate": 7.040219787240841e-06,
      "loss": 0.8037,
      "step": 21253
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.1864311695098877,
      "learning_rate": 7.039580870843052e-06,
      "loss": 0.7325,
      "step": 21254
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.57850182056427,
      "learning_rate": 7.038941954445262e-06,
      "loss": 0.8197,
      "step": 21255
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.785345196723938,
      "learning_rate": 7.038303038047472e-06,
      "loss": 0.8504,
      "step": 21256
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.6727654933929443,
      "learning_rate": 7.037664121649683e-06,
      "loss": 0.7364,
      "step": 21257
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7657631635665894,
      "learning_rate": 7.037025205251893e-06,
      "loss": 0.8388,
      "step": 21258
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.928127408027649,
      "learning_rate": 7.0363862888541035e-06,
      "loss": 0.9356,
      "step": 21259
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.148908853530884,
      "learning_rate": 7.035747372456314e-06,
      "loss": 0.9597,
      "step": 21260
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.7342309951782227,
      "learning_rate": 7.035108456058525e-06,
      "loss": 0.7606,
      "step": 21261
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.21875262260437,
      "learning_rate": 7.0344695396607356e-06,
      "loss": 0.906,
      "step": 21262
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3245240449905396,
      "learning_rate": 7.033830623262946e-06,
      "loss": 0.7452,
      "step": 21263
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.11027455329895,
      "learning_rate": 7.033191706865157e-06,
      "loss": 0.7327,
      "step": 21264
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.461967706680298,
      "learning_rate": 7.032552790467368e-06,
      "loss": 0.9216,
      "step": 21265
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.0524542331695557,
      "learning_rate": 7.031913874069578e-06,
      "loss": 0.9983,
      "step": 21266
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6631494760513306,
      "learning_rate": 7.031274957671789e-06,
      "loss": 0.7695,
      "step": 21267
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6223725080490112,
      "learning_rate": 7.030636041274e-06,
      "loss": 0.8737,
      "step": 21268
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7945103645324707,
      "learning_rate": 7.02999712487621e-06,
      "loss": 0.8112,
      "step": 21269
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8958252668380737,
      "learning_rate": 7.02935820847842e-06,
      "loss": 0.8901,
      "step": 21270
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4799542427062988,
      "learning_rate": 7.028719292080631e-06,
      "loss": 0.492,
      "step": 21271
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6613439321517944,
      "learning_rate": 7.028080375682842e-06,
      "loss": 0.7852,
      "step": 21272
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.790299892425537,
      "learning_rate": 7.027441459285052e-06,
      "loss": 1.3133,
      "step": 21273
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3789758682250977,
      "learning_rate": 7.026802542887263e-06,
      "loss": 0.7796,
      "step": 21274
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9702763557434082,
      "learning_rate": 7.026163626489474e-06,
      "loss": 1.0572,
      "step": 21275
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.085709571838379,
      "learning_rate": 7.0255247100916845e-06,
      "loss": 0.8502,
      "step": 21276
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6596657037734985,
      "learning_rate": 7.024885793693895e-06,
      "loss": 0.8728,
      "step": 21277
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.6525187492370605,
      "learning_rate": 7.024246877296106e-06,
      "loss": 0.5923,
      "step": 21278
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8287453651428223,
      "learning_rate": 7.023607960898317e-06,
      "loss": 0.8061,
      "step": 21279
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.237509727478027,
      "learning_rate": 7.022969044500527e-06,
      "loss": 1.4573,
      "step": 21280
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7171919345855713,
      "learning_rate": 7.022330128102738e-06,
      "loss": 0.8512,
      "step": 21281
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8076361417770386,
      "learning_rate": 7.021691211704949e-06,
      "loss": 0.7922,
      "step": 21282
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.764237880706787,
      "learning_rate": 7.021052295307159e-06,
      "loss": 0.9046,
      "step": 21283
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7002413272857666,
      "learning_rate": 7.02041337890937e-06,
      "loss": 0.8141,
      "step": 21284
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.5033345222473145,
      "learning_rate": 7.019774462511581e-06,
      "loss": 1.7183,
      "step": 21285
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9849587678909302,
      "learning_rate": 7.0191355461137915e-06,
      "loss": 0.824,
      "step": 21286
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.950404405593872,
      "learning_rate": 7.018496629716001e-06,
      "loss": 0.9148,
      "step": 21287
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.138929843902588,
      "learning_rate": 7.017857713318212e-06,
      "loss": 0.7033,
      "step": 21288
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.678806781768799,
      "learning_rate": 7.017218796920423e-06,
      "loss": 0.9949,
      "step": 21289
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7115572690963745,
      "learning_rate": 7.0165798805226335e-06,
      "loss": 0.9935,
      "step": 21290
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.598236083984375,
      "learning_rate": 7.015940964124844e-06,
      "loss": 0.7241,
      "step": 21291
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5920066833496094,
      "learning_rate": 7.015302047727055e-06,
      "loss": 0.8544,
      "step": 21292
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7099961042404175,
      "learning_rate": 7.0146631313292656e-06,
      "loss": 0.8314,
      "step": 21293
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.700137615203857,
      "learning_rate": 7.014024214931476e-06,
      "loss": 1.2375,
      "step": 21294
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9432202577590942,
      "learning_rate": 7.013385298533687e-06,
      "loss": 0.9242,
      "step": 21295
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.718105316162109,
      "learning_rate": 7.012746382135898e-06,
      "loss": 1.1437,
      "step": 21296
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.416529655456543,
      "learning_rate": 7.012107465738108e-06,
      "loss": 1.2473,
      "step": 21297
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.261734962463379,
      "learning_rate": 7.011468549340319e-06,
      "loss": 0.9155,
      "step": 21298
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0037171840667725,
      "learning_rate": 7.01082963294253e-06,
      "loss": 0.9708,
      "step": 21299
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.807011127471924,
      "learning_rate": 7.0101907165447405e-06,
      "loss": 1.201,
      "step": 21300
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7142884731292725,
      "learning_rate": 7.009551800146951e-06,
      "loss": 0.7969,
      "step": 21301
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.911197304725647,
      "learning_rate": 7.008912883749162e-06,
      "loss": 0.8417,
      "step": 21302
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2504031658172607,
      "learning_rate": 7.0082739673513726e-06,
      "loss": 0.7612,
      "step": 21303
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1318612098693848,
      "learning_rate": 7.007635050953583e-06,
      "loss": 0.9364,
      "step": 21304
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.751496434211731,
      "learning_rate": 7.006996134555793e-06,
      "loss": 0.7882,
      "step": 21305
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0322420597076416,
      "learning_rate": 7.006357218158004e-06,
      "loss": 0.8494,
      "step": 21306
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6152201890945435,
      "learning_rate": 7.0057183017602145e-06,
      "loss": 0.8113,
      "step": 21307
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4717780351638794,
      "learning_rate": 7.005079385362425e-06,
      "loss": 0.7786,
      "step": 21308
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.889201283454895,
      "learning_rate": 7.004440468964636e-06,
      "loss": 0.9267,
      "step": 21309
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4811851978302,
      "learning_rate": 7.003801552566847e-06,
      "loss": 0.7635,
      "step": 21310
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2923026084899902,
      "learning_rate": 7.003162636169057e-06,
      "loss": 0.8999,
      "step": 21311
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4037315845489502,
      "learning_rate": 7.002523719771268e-06,
      "loss": 0.9361,
      "step": 21312
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.5357258319854736,
      "learning_rate": 7.001884803373479e-06,
      "loss": 0.8854,
      "step": 21313
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.936041235923767,
      "learning_rate": 7.001245886975689e-06,
      "loss": 0.8459,
      "step": 21314
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2751548290252686,
      "learning_rate": 7.0006069705779e-06,
      "loss": 1.1113,
      "step": 21315
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.992886543273926,
      "learning_rate": 6.999968054180111e-06,
      "loss": 1.1648,
      "step": 21316
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2138917446136475,
      "learning_rate": 6.9993291377823215e-06,
      "loss": 0.8874,
      "step": 21317
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.242870807647705,
      "learning_rate": 6.998690221384532e-06,
      "loss": 0.947,
      "step": 21318
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6767427921295166,
      "learning_rate": 6.998051304986743e-06,
      "loss": 0.8184,
      "step": 21319
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.234614372253418,
      "learning_rate": 6.997412388588954e-06,
      "loss": 0.9641,
      "step": 21320
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5172536373138428,
      "learning_rate": 6.996773472191164e-06,
      "loss": 1.099,
      "step": 21321
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.556413173675537,
      "learning_rate": 6.996134555793375e-06,
      "loss": 0.6995,
      "step": 21322
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.7346765995025635,
      "learning_rate": 6.995495639395586e-06,
      "loss": 0.8549,
      "step": 21323
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.860375165939331,
      "learning_rate": 6.9948567229977956e-06,
      "loss": 1.4092,
      "step": 21324
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.579174757003784,
      "learning_rate": 6.994217806600006e-06,
      "loss": 0.8631,
      "step": 21325
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8675364255905151,
      "learning_rate": 6.993578890202217e-06,
      "loss": 0.7689,
      "step": 21326
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6459596157073975,
      "learning_rate": 6.992939973804428e-06,
      "loss": 0.7743,
      "step": 21327
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.630703330039978,
      "learning_rate": 6.992301057406638e-06,
      "loss": 0.8259,
      "step": 21328
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8626298904418945,
      "learning_rate": 6.991662141008849e-06,
      "loss": 0.828,
      "step": 21329
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.507888317108154,
      "learning_rate": 6.99102322461106e-06,
      "loss": 1.2266,
      "step": 21330
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3253018856048584,
      "learning_rate": 6.9903843082132705e-06,
      "loss": 0.3763,
      "step": 21331
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7295513153076172,
      "learning_rate": 6.989745391815481e-06,
      "loss": 0.8587,
      "step": 21332
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7597887516021729,
      "learning_rate": 6.989106475417692e-06,
      "loss": 0.8984,
      "step": 21333
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4796032905578613,
      "learning_rate": 6.9884675590199026e-06,
      "loss": 0.8129,
      "step": 21334
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0255825519561768,
      "learning_rate": 6.987828642622113e-06,
      "loss": 0.8174,
      "step": 21335
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8279039859771729,
      "learning_rate": 6.987189726224324e-06,
      "loss": 0.8672,
      "step": 21336
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4827380180358887,
      "learning_rate": 6.986550809826535e-06,
      "loss": 1.0015,
      "step": 21337
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.873033046722412,
      "learning_rate": 6.985911893428745e-06,
      "loss": 0.9138,
      "step": 21338
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7459443807601929,
      "learning_rate": 6.985272977030956e-06,
      "loss": 0.845,
      "step": 21339
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8238452672958374,
      "learning_rate": 6.984634060633167e-06,
      "loss": 0.9339,
      "step": 21340
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.669109582901001,
      "learning_rate": 6.9839951442353775e-06,
      "loss": 1.0575,
      "step": 21341
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.604114055633545,
      "learning_rate": 6.983356227837587e-06,
      "loss": 0.7932,
      "step": 21342
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.6074180603027344,
      "learning_rate": 6.982717311439798e-06,
      "loss": 0.9,
      "step": 21343
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7978897094726562,
      "learning_rate": 6.982078395042009e-06,
      "loss": 0.7183,
      "step": 21344
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8709945678710938,
      "learning_rate": 6.981439478644219e-06,
      "loss": 1.0009,
      "step": 21345
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3952136039733887,
      "learning_rate": 6.98080056224643e-06,
      "loss": 1.0478,
      "step": 21346
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1794238090515137,
      "learning_rate": 6.980161645848641e-06,
      "loss": 0.7826,
      "step": 21347
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7363641262054443,
      "learning_rate": 6.9795227294508515e-06,
      "loss": 0.8374,
      "step": 21348
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5058705806732178,
      "learning_rate": 6.978883813053062e-06,
      "loss": 0.8001,
      "step": 21349
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5577951669692993,
      "learning_rate": 6.978244896655273e-06,
      "loss": 0.836,
      "step": 21350
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.238741397857666,
      "learning_rate": 6.977605980257484e-06,
      "loss": 0.8696,
      "step": 21351
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4299710988998413,
      "learning_rate": 6.976967063859694e-06,
      "loss": 0.7809,
      "step": 21352
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.861332654953003,
      "learning_rate": 6.976328147461905e-06,
      "loss": 0.9355,
      "step": 21353
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5841426849365234,
      "learning_rate": 6.975689231064116e-06,
      "loss": 0.8178,
      "step": 21354
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.559701919555664,
      "learning_rate": 6.975050314666326e-06,
      "loss": 0.6549,
      "step": 21355
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.8563451766967773,
      "learning_rate": 6.974411398268537e-06,
      "loss": 1.1724,
      "step": 21356
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1075387001037598,
      "learning_rate": 6.973772481870748e-06,
      "loss": 0.928,
      "step": 21357
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.8190598487854,
      "learning_rate": 6.9731335654729585e-06,
      "loss": 1.4015,
      "step": 21358
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5900744199752808,
      "learning_rate": 6.972494649075169e-06,
      "loss": 0.8709,
      "step": 21359
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7588554620742798,
      "learning_rate": 6.971855732677379e-06,
      "loss": 0.8455,
      "step": 21360
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.406954288482666,
      "learning_rate": 6.97121681627959e-06,
      "loss": 1.0408,
      "step": 21361
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3280906677246094,
      "learning_rate": 6.9705778998818005e-06,
      "loss": 0.7656,
      "step": 21362
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.612154245376587,
      "learning_rate": 6.969938983484011e-06,
      "loss": 1.4273,
      "step": 21363
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.167346239089966,
      "learning_rate": 6.969300067086222e-06,
      "loss": 0.8998,
      "step": 21364
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2129592895507812,
      "learning_rate": 6.9686611506884326e-06,
      "loss": 0.8733,
      "step": 21365
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6953052282333374,
      "learning_rate": 6.968022234290643e-06,
      "loss": 0.7728,
      "step": 21366
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9079594612121582,
      "learning_rate": 6.967383317892854e-06,
      "loss": 1.1214,
      "step": 21367
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6735531091690063,
      "learning_rate": 6.966744401495065e-06,
      "loss": 0.8338,
      "step": 21368
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.917229413986206,
      "learning_rate": 6.966105485097275e-06,
      "loss": 0.9191,
      "step": 21369
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0550525188446045,
      "learning_rate": 6.965466568699486e-06,
      "loss": 0.8508,
      "step": 21370
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.227330446243286,
      "learning_rate": 6.964827652301697e-06,
      "loss": 1.1181,
      "step": 21371
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6514607667922974,
      "learning_rate": 6.9641887359039075e-06,
      "loss": 0.8325,
      "step": 21372
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.080258369445801,
      "learning_rate": 6.963549819506118e-06,
      "loss": 0.8314,
      "step": 21373
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6635750532150269,
      "learning_rate": 6.962910903108329e-06,
      "loss": 0.763,
      "step": 21374
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2405917644500732,
      "learning_rate": 6.9622719867105396e-06,
      "loss": 0.8373,
      "step": 21375
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.717029094696045,
      "learning_rate": 6.96163307031275e-06,
      "loss": 0.9202,
      "step": 21376
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.190493583679199,
      "learning_rate": 6.960994153914961e-06,
      "loss": 1.1176,
      "step": 21377
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0731160640716553,
      "learning_rate": 6.960355237517171e-06,
      "loss": 0.9484,
      "step": 21378
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.8237481117248535,
      "learning_rate": 6.9597163211193815e-06,
      "loss": 0.8084,
      "step": 21379
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5416178703308105,
      "learning_rate": 6.959077404721592e-06,
      "loss": 0.8001,
      "step": 21380
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.851693868637085,
      "learning_rate": 6.958438488323803e-06,
      "loss": 0.7619,
      "step": 21381
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.030116319656372,
      "learning_rate": 6.957799571926014e-06,
      "loss": 0.8386,
      "step": 21382
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5089147090911865,
      "learning_rate": 6.957160655528224e-06,
      "loss": 0.7439,
      "step": 21383
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.514646291732788,
      "learning_rate": 6.956521739130435e-06,
      "loss": 0.8723,
      "step": 21384
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9643818140029907,
      "learning_rate": 6.955882822732646e-06,
      "loss": 0.8248,
      "step": 21385
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5699775218963623,
      "learning_rate": 6.955243906334856e-06,
      "loss": 0.8024,
      "step": 21386
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.046518087387085,
      "learning_rate": 6.954604989937067e-06,
      "loss": 0.8422,
      "step": 21387
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.843667984008789,
      "learning_rate": 6.953966073539278e-06,
      "loss": 1.2596,
      "step": 21388
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1660356521606445,
      "learning_rate": 6.9533271571414885e-06,
      "loss": 0.9205,
      "step": 21389
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.198002576828003,
      "learning_rate": 6.952688240743699e-06,
      "loss": 0.5502,
      "step": 21390
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.4745588302612305,
      "learning_rate": 6.95204932434591e-06,
      "loss": 1.4041,
      "step": 21391
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.7838521003723145,
      "learning_rate": 6.951410407948121e-06,
      "loss": 0.991,
      "step": 21392
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4581856727600098,
      "learning_rate": 6.950771491550331e-06,
      "loss": 0.7846,
      "step": 21393
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8412339687347412,
      "learning_rate": 6.950132575152542e-06,
      "loss": 0.8544,
      "step": 21394
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.696877121925354,
      "learning_rate": 6.949493658754753e-06,
      "loss": 0.7654,
      "step": 21395
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4607162475585938,
      "learning_rate": 6.948854742356963e-06,
      "loss": 0.4893,
      "step": 21396
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.03233003616333,
      "learning_rate": 6.948215825959173e-06,
      "loss": 0.8184,
      "step": 21397
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9169081449508667,
      "learning_rate": 6.947576909561384e-06,
      "loss": 0.8036,
      "step": 21398
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8906244039535522,
      "learning_rate": 6.946937993163595e-06,
      "loss": 0.9538,
      "step": 21399
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.958767294883728,
      "learning_rate": 6.946299076765805e-06,
      "loss": 0.9477,
      "step": 21400
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7100329399108887,
      "learning_rate": 6.945660160368016e-06,
      "loss": 0.7684,
      "step": 21401
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.556849002838135,
      "learning_rate": 6.945021243970227e-06,
      "loss": 1.2134,
      "step": 21402
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0841691493988037,
      "learning_rate": 6.9443823275724375e-06,
      "loss": 0.8728,
      "step": 21403
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6906921863555908,
      "learning_rate": 6.943743411174648e-06,
      "loss": 0.7552,
      "step": 21404
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.7444467544555664,
      "learning_rate": 6.943104494776859e-06,
      "loss": 0.5163,
      "step": 21405
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5391194820404053,
      "learning_rate": 6.9424655783790696e-06,
      "loss": 0.8537,
      "step": 21406
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0691933631896973,
      "learning_rate": 6.94182666198128e-06,
      "loss": 0.8376,
      "step": 21407
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.4183549880981445,
      "learning_rate": 6.941187745583491e-06,
      "loss": 0.792,
      "step": 21408
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9660977125167847,
      "learning_rate": 6.940548829185702e-06,
      "loss": 0.7016,
      "step": 21409
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3579792976379395,
      "learning_rate": 6.939909912787912e-06,
      "loss": 0.4777,
      "step": 21410
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7799954414367676,
      "learning_rate": 6.939270996390123e-06,
      "loss": 0.8392,
      "step": 21411
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7225278615951538,
      "learning_rate": 6.938632079992334e-06,
      "loss": 0.8729,
      "step": 21412
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.893749713897705,
      "learning_rate": 6.9379931635945445e-06,
      "loss": 1.3598,
      "step": 21413
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.069390296936035,
      "learning_rate": 6.937354247196755e-06,
      "loss": 0.8208,
      "step": 21414
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8073254823684692,
      "learning_rate": 6.936715330798965e-06,
      "loss": 1.101,
      "step": 21415
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6741960048675537,
      "learning_rate": 6.936076414401176e-06,
      "loss": 0.7981,
      "step": 21416
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6583926677703857,
      "learning_rate": 6.9354374980033864e-06,
      "loss": 0.8455,
      "step": 21417
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.850468158721924,
      "learning_rate": 6.934798581605597e-06,
      "loss": 0.7667,
      "step": 21418
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7397822141647339,
      "learning_rate": 6.934159665207808e-06,
      "loss": 0.8155,
      "step": 21419
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.693509101867676,
      "learning_rate": 6.9335207488100185e-06,
      "loss": 0.76,
      "step": 21420
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.321444511413574,
      "learning_rate": 6.932881832412229e-06,
      "loss": 1.2619,
      "step": 21421
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3624496459960938,
      "learning_rate": 6.93224291601444e-06,
      "loss": 0.9276,
      "step": 21422
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6551965475082397,
      "learning_rate": 6.931603999616651e-06,
      "loss": 0.8771,
      "step": 21423
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4419194459915161,
      "learning_rate": 6.930965083218861e-06,
      "loss": 0.7668,
      "step": 21424
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.210064172744751,
      "learning_rate": 6.930326166821072e-06,
      "loss": 0.934,
      "step": 21425
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9353224039077759,
      "learning_rate": 6.929687250423283e-06,
      "loss": 0.9687,
      "step": 21426
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.838183045387268,
      "learning_rate": 6.9290483340254934e-06,
      "loss": 0.8222,
      "step": 21427
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6868773698806763,
      "learning_rate": 6.928409417627704e-06,
      "loss": 0.8191,
      "step": 21428
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.353208541870117,
      "learning_rate": 6.927770501229915e-06,
      "loss": 1.08,
      "step": 21429
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.857135534286499,
      "learning_rate": 6.9271315848321255e-06,
      "loss": 0.8989,
      "step": 21430
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7437483072280884,
      "learning_rate": 6.926492668434336e-06,
      "loss": 0.8535,
      "step": 21431
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7361787557601929,
      "learning_rate": 6.925853752036547e-06,
      "loss": 0.802,
      "step": 21432
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5173048973083496,
      "learning_rate": 6.925214835638757e-06,
      "loss": 1.0177,
      "step": 21433
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.182856798171997,
      "learning_rate": 6.9245759192409675e-06,
      "loss": 0.85,
      "step": 21434
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4584931135177612,
      "learning_rate": 6.923937002843178e-06,
      "loss": 0.3807,
      "step": 21435
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.785371780395508,
      "learning_rate": 6.923298086445389e-06,
      "loss": 1.1611,
      "step": 21436
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.7418506145477295,
      "learning_rate": 6.9226591700476e-06,
      "loss": 0.9262,
      "step": 21437
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8608208894729614,
      "learning_rate": 6.92202025364981e-06,
      "loss": 0.8726,
      "step": 21438
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7159968614578247,
      "learning_rate": 6.921381337252021e-06,
      "loss": 0.866,
      "step": 21439
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0492942333221436,
      "learning_rate": 6.920742420854232e-06,
      "loss": 0.9343,
      "step": 21440
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7291463613510132,
      "learning_rate": 6.920103504456442e-06,
      "loss": 0.8392,
      "step": 21441
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.391729474067688,
      "learning_rate": 6.919464588058653e-06,
      "loss": 0.7839,
      "step": 21442
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5440285205841064,
      "learning_rate": 6.918825671660864e-06,
      "loss": 0.7857,
      "step": 21443
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7526895999908447,
      "learning_rate": 6.9181867552630745e-06,
      "loss": 0.8993,
      "step": 21444
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.9270474910736084,
      "learning_rate": 6.917547838865285e-06,
      "loss": 0.9639,
      "step": 21445
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9062976837158203,
      "learning_rate": 6.916908922467496e-06,
      "loss": 1.1324,
      "step": 21446
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1479785442352295,
      "learning_rate": 6.916270006069707e-06,
      "loss": 0.7003,
      "step": 21447
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.8467326164245605,
      "learning_rate": 6.915631089671917e-06,
      "loss": 0.7909,
      "step": 21448
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.052459716796875,
      "learning_rate": 6.914992173274128e-06,
      "loss": 0.8438,
      "step": 21449
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5497095584869385,
      "learning_rate": 6.914353256876339e-06,
      "loss": 0.7474,
      "step": 21450
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.7435429096221924,
      "learning_rate": 6.9137143404785485e-06,
      "loss": 0.6312,
      "step": 21451
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.673509359359741,
      "learning_rate": 6.913075424080759e-06,
      "loss": 0.9191,
      "step": 21452
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.056363582611084,
      "learning_rate": 6.91243650768297e-06,
      "loss": 0.9441,
      "step": 21453
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.764418363571167,
      "learning_rate": 6.911797591285181e-06,
      "loss": 1.0535,
      "step": 21454
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5787678956985474,
      "learning_rate": 6.911158674887391e-06,
      "loss": 0.82,
      "step": 21455
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.758366584777832,
      "learning_rate": 6.910519758489602e-06,
      "loss": 0.874,
      "step": 21456
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9411437511444092,
      "learning_rate": 6.909880842091813e-06,
      "loss": 0.7605,
      "step": 21457
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5088629722595215,
      "learning_rate": 6.9092419256940234e-06,
      "loss": 0.4281,
      "step": 21458
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7556378841400146,
      "learning_rate": 6.908603009296234e-06,
      "loss": 0.8856,
      "step": 21459
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7752829790115356,
      "learning_rate": 6.907964092898445e-06,
      "loss": 0.9104,
      "step": 21460
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7611643075942993,
      "learning_rate": 6.9073251765006555e-06,
      "loss": 0.8863,
      "step": 21461
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5466461181640625,
      "learning_rate": 6.906686260102866e-06,
      "loss": 0.7696,
      "step": 21462
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.756656289100647,
      "learning_rate": 6.906047343705077e-06,
      "loss": 0.8328,
      "step": 21463
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.067485809326172,
      "learning_rate": 6.905408427307288e-06,
      "loss": 0.8778,
      "step": 21464
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5999939441680908,
      "learning_rate": 6.904769510909498e-06,
      "loss": 0.8382,
      "step": 21465
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9046663045883179,
      "learning_rate": 6.904130594511709e-06,
      "loss": 0.8441,
      "step": 21466
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.6761257648468018,
      "learning_rate": 6.90349167811392e-06,
      "loss": 0.8717,
      "step": 21467
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0464839935302734,
      "learning_rate": 6.9028527617161304e-06,
      "loss": 1.1961,
      "step": 21468
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4705368280410767,
      "learning_rate": 6.902213845318341e-06,
      "loss": 0.8665,
      "step": 21469
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3209165334701538,
      "learning_rate": 6.901574928920551e-06,
      "loss": 0.4288,
      "step": 21470
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.120002508163452,
      "learning_rate": 6.900936012522762e-06,
      "loss": 0.8846,
      "step": 21471
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7128568887710571,
      "learning_rate": 6.900297096124972e-06,
      "loss": 0.904,
      "step": 21472
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.670576572418213,
      "learning_rate": 6.899658179727183e-06,
      "loss": 0.8026,
      "step": 21473
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.687949538230896,
      "learning_rate": 6.899019263329394e-06,
      "loss": 0.9229,
      "step": 21474
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6360150575637817,
      "learning_rate": 6.8983803469316045e-06,
      "loss": 0.7748,
      "step": 21475
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5841697454452515,
      "learning_rate": 6.897741430533815e-06,
      "loss": 0.8959,
      "step": 21476
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.603148341178894,
      "learning_rate": 6.897102514136026e-06,
      "loss": 0.7569,
      "step": 21477
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.525066614151001,
      "learning_rate": 6.896463597738237e-06,
      "loss": 0.9507,
      "step": 21478
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7637054920196533,
      "learning_rate": 6.895824681340447e-06,
      "loss": 0.8724,
      "step": 21479
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8145185708999634,
      "learning_rate": 6.895185764942658e-06,
      "loss": 0.7337,
      "step": 21480
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0013926029205322,
      "learning_rate": 6.894546848544869e-06,
      "loss": 0.8637,
      "step": 21481
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9241539239883423,
      "learning_rate": 6.893907932147079e-06,
      "loss": 0.799,
      "step": 21482
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8524348735809326,
      "learning_rate": 6.89326901574929e-06,
      "loss": 0.8407,
      "step": 21483
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8702383041381836,
      "learning_rate": 6.892630099351501e-06,
      "loss": 0.9048,
      "step": 21484
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.081085205078125,
      "learning_rate": 6.8919911829537115e-06,
      "loss": 1.1745,
      "step": 21485
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.900707721710205,
      "learning_rate": 6.891352266555922e-06,
      "loss": 0.8841,
      "step": 21486
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.776965618133545,
      "learning_rate": 6.890713350158133e-06,
      "loss": 0.89,
      "step": 21487
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0226478576660156,
      "learning_rate": 6.890074433760343e-06,
      "loss": 1.2214,
      "step": 21488
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.236635684967041,
      "learning_rate": 6.8894355173625534e-06,
      "loss": 0.9932,
      "step": 21489
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9523166418075562,
      "learning_rate": 6.888796600964764e-06,
      "loss": 0.9385,
      "step": 21490
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4454615116119385,
      "learning_rate": 6.888157684566975e-06,
      "loss": 0.8095,
      "step": 21491
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9153671264648438,
      "learning_rate": 6.8875187681691855e-06,
      "loss": 1.0631,
      "step": 21492
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5849272012710571,
      "learning_rate": 6.886879851771396e-06,
      "loss": 0.7502,
      "step": 21493
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.9061686992645264,
      "learning_rate": 6.886240935373607e-06,
      "loss": 0.7281,
      "step": 21494
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.1385161876678467,
      "learning_rate": 6.885602018975818e-06,
      "loss": 1.3597,
      "step": 21495
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4299947023391724,
      "learning_rate": 6.884963102578028e-06,
      "loss": 0.7948,
      "step": 21496
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.027674436569214,
      "learning_rate": 6.884324186180239e-06,
      "loss": 0.9433,
      "step": 21497
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.980520486831665,
      "learning_rate": 6.88368526978245e-06,
      "loss": 0.9371,
      "step": 21498
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.722759485244751,
      "learning_rate": 6.8830463533846604e-06,
      "loss": 1.0197,
      "step": 21499
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6624836921691895,
      "learning_rate": 6.882407436986871e-06,
      "loss": 0.8086,
      "step": 21500
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.63968825340271,
      "learning_rate": 6.881768520589082e-06,
      "loss": 0.858,
      "step": 21501
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.051772356033325,
      "learning_rate": 6.8811296041912925e-06,
      "loss": 0.8563,
      "step": 21502
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0744407176971436,
      "learning_rate": 6.880490687793503e-06,
      "loss": 0.9577,
      "step": 21503
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5483198165893555,
      "learning_rate": 6.879851771395714e-06,
      "loss": 0.8699,
      "step": 21504
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0862464904785156,
      "learning_rate": 6.879212854997925e-06,
      "loss": 1.1035,
      "step": 21505
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7079803943634033,
      "learning_rate": 6.8785739386001345e-06,
      "loss": 1.0133,
      "step": 21506
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.009920358657837,
      "learning_rate": 6.877935022202345e-06,
      "loss": 1.232,
      "step": 21507
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.670312762260437,
      "learning_rate": 6.877296105804556e-06,
      "loss": 0.8686,
      "step": 21508
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.99039626121521,
      "learning_rate": 6.876657189406767e-06,
      "loss": 1.0535,
      "step": 21509
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.410275936126709,
      "learning_rate": 6.876018273008977e-06,
      "loss": 1.1039,
      "step": 21510
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.537461757659912,
      "learning_rate": 6.875379356611188e-06,
      "loss": 0.9047,
      "step": 21511
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.761028528213501,
      "learning_rate": 6.874740440213399e-06,
      "loss": 0.9982,
      "step": 21512
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.939395546913147,
      "learning_rate": 6.874101523815609e-06,
      "loss": 0.921,
      "step": 21513
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.460103154182434,
      "learning_rate": 6.87346260741782e-06,
      "loss": 0.7895,
      "step": 21514
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4654239416122437,
      "learning_rate": 6.872823691020031e-06,
      "loss": 0.8609,
      "step": 21515
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9798322916030884,
      "learning_rate": 6.8721847746222415e-06,
      "loss": 0.894,
      "step": 21516
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.73425555229187,
      "learning_rate": 6.871545858224452e-06,
      "loss": 0.7625,
      "step": 21517
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.68375825881958,
      "learning_rate": 6.870906941826663e-06,
      "loss": 0.8057,
      "step": 21518
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.158158779144287,
      "learning_rate": 6.870268025428874e-06,
      "loss": 1.365,
      "step": 21519
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9708532094955444,
      "learning_rate": 6.869629109031084e-06,
      "loss": 0.7751,
      "step": 21520
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.456847906112671,
      "learning_rate": 6.868990192633295e-06,
      "loss": 0.4211,
      "step": 21521
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4641802310943604,
      "learning_rate": 6.868351276235506e-06,
      "loss": 0.4277,
      "step": 21522
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.000978708267212,
      "learning_rate": 6.867712359837716e-06,
      "loss": 0.9003,
      "step": 21523
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6823066473007202,
      "learning_rate": 6.867073443439926e-06,
      "loss": 0.4692,
      "step": 21524
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6614779233932495,
      "learning_rate": 6.866434527042137e-06,
      "loss": 0.8573,
      "step": 21525
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1070914268493652,
      "learning_rate": 6.865795610644348e-06,
      "loss": 1.2209,
      "step": 21526
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0807998180389404,
      "learning_rate": 6.865156694246558e-06,
      "loss": 0.8618,
      "step": 21527
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9127253293991089,
      "learning_rate": 6.864517777848769e-06,
      "loss": 1.1024,
      "step": 21528
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.6059608459472656,
      "learning_rate": 6.86387886145098e-06,
      "loss": 1.0668,
      "step": 21529
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.696223735809326,
      "learning_rate": 6.8632399450531904e-06,
      "loss": 0.9981,
      "step": 21530
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5898371934890747,
      "learning_rate": 6.862601028655401e-06,
      "loss": 0.8604,
      "step": 21531
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4421945810317993,
      "learning_rate": 6.861962112257612e-06,
      "loss": 0.787,
      "step": 21532
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.465430736541748,
      "learning_rate": 6.8613231958598225e-06,
      "loss": 0.8996,
      "step": 21533
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.579563856124878,
      "learning_rate": 6.860684279462033e-06,
      "loss": 0.8397,
      "step": 21534
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.264741897583008,
      "learning_rate": 6.860045363064244e-06,
      "loss": 0.9199,
      "step": 21535
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.649728536605835,
      "learning_rate": 6.859406446666455e-06,
      "loss": 0.7935,
      "step": 21536
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4272942543029785,
      "learning_rate": 6.858767530268665e-06,
      "loss": 0.8567,
      "step": 21537
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.315593719482422,
      "learning_rate": 6.858128613870876e-06,
      "loss": 0.8843,
      "step": 21538
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.005656719207764,
      "learning_rate": 6.857489697473087e-06,
      "loss": 1.0911,
      "step": 21539
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.8351171016693115,
      "learning_rate": 6.8568507810752974e-06,
      "loss": 0.6744,
      "step": 21540
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1878435611724854,
      "learning_rate": 6.856211864677508e-06,
      "loss": 1.1364,
      "step": 21541
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.18782901763916,
      "learning_rate": 6.855572948279719e-06,
      "loss": 0.9504,
      "step": 21542
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0734105110168457,
      "learning_rate": 6.854934031881929e-06,
      "loss": 0.8215,
      "step": 21543
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6744974851608276,
      "learning_rate": 6.854295115484139e-06,
      "loss": 0.8096,
      "step": 21544
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.474209785461426,
      "learning_rate": 6.85365619908635e-06,
      "loss": 0.9053,
      "step": 21545
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.658182978630066,
      "learning_rate": 6.853017282688561e-06,
      "loss": 0.8379,
      "step": 21546
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5693347454071045,
      "learning_rate": 6.8523783662907715e-06,
      "loss": 0.3677,
      "step": 21547
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.9180755615234375,
      "learning_rate": 6.851739449892982e-06,
      "loss": 1.5617,
      "step": 21548
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7690019607543945,
      "learning_rate": 6.851100533495193e-06,
      "loss": 0.9119,
      "step": 21549
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2382001876831055,
      "learning_rate": 6.850461617097404e-06,
      "loss": 0.7545,
      "step": 21550
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9316822290420532,
      "learning_rate": 6.849822700699614e-06,
      "loss": 0.9393,
      "step": 21551
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6307404041290283,
      "learning_rate": 6.849183784301825e-06,
      "loss": 0.7866,
      "step": 21552
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3797322511672974,
      "learning_rate": 6.848544867904036e-06,
      "loss": 0.3849,
      "step": 21553
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3917587995529175,
      "learning_rate": 6.847905951506246e-06,
      "loss": 0.7349,
      "step": 21554
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9111067056655884,
      "learning_rate": 6.847267035108457e-06,
      "loss": 0.9855,
      "step": 21555
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.799484372138977,
      "learning_rate": 6.846628118710668e-06,
      "loss": 0.9171,
      "step": 21556
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6782373189926147,
      "learning_rate": 6.8459892023128785e-06,
      "loss": 0.8706,
      "step": 21557
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.54144287109375,
      "learning_rate": 6.845350285915089e-06,
      "loss": 0.9227,
      "step": 21558
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8513731956481934,
      "learning_rate": 6.8447113695173e-06,
      "loss": 0.8922,
      "step": 21559
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4343478679656982,
      "learning_rate": 6.844072453119511e-06,
      "loss": 0.9183,
      "step": 21560
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3061357736587524,
      "learning_rate": 6.8434335367217204e-06,
      "loss": 0.3779,
      "step": 21561
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2069876194000244,
      "learning_rate": 6.842794620323931e-06,
      "loss": 0.9298,
      "step": 21562
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6447118520736694,
      "learning_rate": 6.842155703926142e-06,
      "loss": 0.9663,
      "step": 21563
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.7089946269989014,
      "learning_rate": 6.8415167875283525e-06,
      "loss": 0.8752,
      "step": 21564
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2799744606018066,
      "learning_rate": 6.840877871130563e-06,
      "loss": 0.781,
      "step": 21565
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9367170333862305,
      "learning_rate": 6.840238954732774e-06,
      "loss": 0.8873,
      "step": 21566
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.345501661300659,
      "learning_rate": 6.839600038334985e-06,
      "loss": 1.1101,
      "step": 21567
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2907761335372925,
      "learning_rate": 6.838961121937195e-06,
      "loss": 0.3341,
      "step": 21568
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.535014271736145,
      "learning_rate": 6.838322205539406e-06,
      "loss": 0.7997,
      "step": 21569
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.415255308151245,
      "learning_rate": 6.837683289141617e-06,
      "loss": 0.6788,
      "step": 21570
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.530620574951172,
      "learning_rate": 6.8370443727438274e-06,
      "loss": 1.0367,
      "step": 21571
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.114234685897827,
      "learning_rate": 6.836405456346038e-06,
      "loss": 1.0057,
      "step": 21572
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.0254149436950684,
      "learning_rate": 6.835766539948249e-06,
      "loss": 0.7659,
      "step": 21573
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5284286737442017,
      "learning_rate": 6.8351276235504595e-06,
      "loss": 0.6646,
      "step": 21574
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.767467737197876,
      "learning_rate": 6.83448870715267e-06,
      "loss": 1.0236,
      "step": 21575
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7405858039855957,
      "learning_rate": 6.833849790754881e-06,
      "loss": 1.0509,
      "step": 21576
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.998518705368042,
      "learning_rate": 6.833210874357092e-06,
      "loss": 1.2011,
      "step": 21577
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.184441328048706,
      "learning_rate": 6.832571957959302e-06,
      "loss": 0.8972,
      "step": 21578
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7967443466186523,
      "learning_rate": 6.831933041561512e-06,
      "loss": 0.8917,
      "step": 21579
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0726442337036133,
      "learning_rate": 6.831294125163723e-06,
      "loss": 0.8858,
      "step": 21580
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.849814534187317,
      "learning_rate": 6.830655208765934e-06,
      "loss": 0.9159,
      "step": 21581
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9049450159072876,
      "learning_rate": 6.830016292368144e-06,
      "loss": 0.8261,
      "step": 21582
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.608687162399292,
      "learning_rate": 6.829377375970355e-06,
      "loss": 0.962,
      "step": 21583
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8898638486862183,
      "learning_rate": 6.828738459572566e-06,
      "loss": 0.9988,
      "step": 21584
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.773061990737915,
      "learning_rate": 6.828099543174776e-06,
      "loss": 0.8911,
      "step": 21585
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.91504967212677,
      "learning_rate": 6.827460626776987e-06,
      "loss": 0.8132,
      "step": 21586
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.11832332611084,
      "learning_rate": 6.826821710379198e-06,
      "loss": 0.9101,
      "step": 21587
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5206797122955322,
      "learning_rate": 6.8261827939814085e-06,
      "loss": 0.8466,
      "step": 21588
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6758657693862915,
      "learning_rate": 6.825543877583619e-06,
      "loss": 0.7344,
      "step": 21589
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.367141246795654,
      "learning_rate": 6.82490496118583e-06,
      "loss": 1.7156,
      "step": 21590
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.017303943634033,
      "learning_rate": 6.824266044788041e-06,
      "loss": 0.9524,
      "step": 21591
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.039258003234863,
      "learning_rate": 6.823627128390251e-06,
      "loss": 1.3436,
      "step": 21592
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0471761226654053,
      "learning_rate": 6.822988211992462e-06,
      "loss": 0.8244,
      "step": 21593
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6564732789993286,
      "learning_rate": 6.822349295594673e-06,
      "loss": 0.8952,
      "step": 21594
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7024800777435303,
      "learning_rate": 6.821710379196883e-06,
      "loss": 0.8351,
      "step": 21595
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8861613273620605,
      "learning_rate": 6.821071462799094e-06,
      "loss": 0.7885,
      "step": 21596
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4770102500915527,
      "learning_rate": 6.820432546401304e-06,
      "loss": 1.0365,
      "step": 21597
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9427767992019653,
      "learning_rate": 6.819793630003515e-06,
      "loss": 0.9346,
      "step": 21598
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.583190679550171,
      "learning_rate": 6.819154713605725e-06,
      "loss": 0.8786,
      "step": 21599
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5053030252456665,
      "learning_rate": 6.818515797207936e-06,
      "loss": 0.8186,
      "step": 21600
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2136740684509277,
      "learning_rate": 6.817876880810147e-06,
      "loss": 0.9526,
      "step": 21601
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.520965576171875,
      "learning_rate": 6.8172379644123574e-06,
      "loss": 0.6293,
      "step": 21602
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4402265548706055,
      "learning_rate": 6.816599048014568e-06,
      "loss": 0.8182,
      "step": 21603
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.815735936164856,
      "learning_rate": 6.815960131616779e-06,
      "loss": 0.8378,
      "step": 21604
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.8371996879577637,
      "learning_rate": 6.8153212152189895e-06,
      "loss": 0.8489,
      "step": 21605
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4333146810531616,
      "learning_rate": 6.8146822988212e-06,
      "loss": 0.8286,
      "step": 21606
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.0103249549865723,
      "learning_rate": 6.814043382423411e-06,
      "loss": 0.7546,
      "step": 21607
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0625548362731934,
      "learning_rate": 6.813404466025622e-06,
      "loss": 0.8557,
      "step": 21608
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.6765389442443848,
      "learning_rate": 6.812765549627832e-06,
      "loss": 0.9766,
      "step": 21609
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0453717708587646,
      "learning_rate": 6.812126633230043e-06,
      "loss": 0.8118,
      "step": 21610
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.070008754730225,
      "learning_rate": 6.811487716832254e-06,
      "loss": 1.1963,
      "step": 21611
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.995327115058899,
      "learning_rate": 6.8108488004344644e-06,
      "loss": 1.2529,
      "step": 21612
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8943122625350952,
      "learning_rate": 6.810209884036675e-06,
      "loss": 1.122,
      "step": 21613
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9803593158721924,
      "learning_rate": 6.809570967638886e-06,
      "loss": 0.8248,
      "step": 21614
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.028989791870117,
      "learning_rate": 6.8089320512410965e-06,
      "loss": 0.7694,
      "step": 21615
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4600753784179688,
      "learning_rate": 6.808293134843306e-06,
      "loss": 1.1688,
      "step": 21616
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2114455699920654,
      "learning_rate": 6.807654218445517e-06,
      "loss": 0.9094,
      "step": 21617
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.358061671257019,
      "learning_rate": 6.807015302047728e-06,
      "loss": 0.7603,
      "step": 21618
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.203359842300415,
      "learning_rate": 6.8063763856499385e-06,
      "loss": 0.8747,
      "step": 21619
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5762200355529785,
      "learning_rate": 6.805737469252149e-06,
      "loss": 0.8343,
      "step": 21620
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8060383796691895,
      "learning_rate": 6.80509855285436e-06,
      "loss": 0.9236,
      "step": 21621
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0595180988311768,
      "learning_rate": 6.804459636456571e-06,
      "loss": 1.2,
      "step": 21622
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5987203121185303,
      "learning_rate": 6.803820720058781e-06,
      "loss": 0.9009,
      "step": 21623
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6104799509048462,
      "learning_rate": 6.803181803660992e-06,
      "loss": 0.9014,
      "step": 21624
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.379255771636963,
      "learning_rate": 6.802542887263203e-06,
      "loss": 0.7854,
      "step": 21625
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.2977206707000732,
      "learning_rate": 6.801903970865413e-06,
      "loss": 1.046,
      "step": 21626
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.726327419281006,
      "learning_rate": 6.801265054467624e-06,
      "loss": 1.5029,
      "step": 21627
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5865908861160278,
      "learning_rate": 6.800626138069835e-06,
      "loss": 0.8774,
      "step": 21628
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.995970606803894,
      "learning_rate": 6.7999872216720455e-06,
      "loss": 0.89,
      "step": 21629
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.883912444114685,
      "learning_rate": 6.799348305274256e-06,
      "loss": 0.8947,
      "step": 21630
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5102272033691406,
      "learning_rate": 6.798709388876467e-06,
      "loss": 0.8439,
      "step": 21631
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.51411509513855,
      "learning_rate": 6.798070472478678e-06,
      "loss": 0.8279,
      "step": 21632
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.961413860321045,
      "learning_rate": 6.797431556080888e-06,
      "loss": 1.559,
      "step": 21633
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7153559923171997,
      "learning_rate": 6.796792639683098e-06,
      "loss": 0.8349,
      "step": 21634
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.0857295989990234,
      "learning_rate": 6.796153723285309e-06,
      "loss": 1.2036,
      "step": 21635
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9089974164962769,
      "learning_rate": 6.7955148068875196e-06,
      "loss": 0.7636,
      "step": 21636
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7975013256072998,
      "learning_rate": 6.79487589048973e-06,
      "loss": 0.8764,
      "step": 21637
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.069136142730713,
      "learning_rate": 6.794236974091941e-06,
      "loss": 1.4716,
      "step": 21638
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5044028759002686,
      "learning_rate": 6.793598057694152e-06,
      "loss": 0.8198,
      "step": 21639
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.082988739013672,
      "learning_rate": 6.792959141296362e-06,
      "loss": 0.7606,
      "step": 21640
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.602649450302124,
      "learning_rate": 6.792320224898573e-06,
      "loss": 0.6925,
      "step": 21641
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2121827602386475,
      "learning_rate": 6.791681308500784e-06,
      "loss": 0.6617,
      "step": 21642
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9103379249572754,
      "learning_rate": 6.7910423921029945e-06,
      "loss": 0.9491,
      "step": 21643
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6902744770050049,
      "learning_rate": 6.790403475705205e-06,
      "loss": 0.8031,
      "step": 21644
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5029404163360596,
      "learning_rate": 6.789764559307415e-06,
      "loss": 0.8101,
      "step": 21645
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4132137298583984,
      "learning_rate": 6.789125642909625e-06,
      "loss": 0.7975,
      "step": 21646
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5770752429962158,
      "learning_rate": 6.7884867265118356e-06,
      "loss": 0.8298,
      "step": 21647
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3792378902435303,
      "learning_rate": 6.787847810114046e-06,
      "loss": 0.7067,
      "step": 21648
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.616462469100952,
      "learning_rate": 6.787208893716257e-06,
      "loss": 1.1931,
      "step": 21649
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.104729175567627,
      "learning_rate": 6.786569977318468e-06,
      "loss": 0.8746,
      "step": 21650
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6315327882766724,
      "learning_rate": 6.785931060920678e-06,
      "loss": 0.4073,
      "step": 21651
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7833032608032227,
      "learning_rate": 6.785292144522889e-06,
      "loss": 0.875,
      "step": 21652
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.645603895187378,
      "learning_rate": 6.7846532281251e-06,
      "loss": 0.8384,
      "step": 21653
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7789885997772217,
      "learning_rate": 6.7840143117273105e-06,
      "loss": 0.8464,
      "step": 21654
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3668144941329956,
      "learning_rate": 6.783375395329521e-06,
      "loss": 0.3578,
      "step": 21655
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7840770483016968,
      "learning_rate": 6.782736478931732e-06,
      "loss": 0.9457,
      "step": 21656
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.582923173904419,
      "learning_rate": 6.7820975625339426e-06,
      "loss": 1.081,
      "step": 21657
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5404154062271118,
      "learning_rate": 6.781458646136153e-06,
      "loss": 1.1533,
      "step": 21658
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8324918746948242,
      "learning_rate": 6.780819729738364e-06,
      "loss": 0.8377,
      "step": 21659
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7634780406951904,
      "learning_rate": 6.780180813340575e-06,
      "loss": 0.9742,
      "step": 21660
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7402453422546387,
      "learning_rate": 6.779541896942785e-06,
      "loss": 0.5272,
      "step": 21661
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8649283647537231,
      "learning_rate": 6.778902980544996e-06,
      "loss": 0.8122,
      "step": 21662
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9269676208496094,
      "learning_rate": 6.778264064147207e-06,
      "loss": 0.8417,
      "step": 21663
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.5028152465820312,
      "learning_rate": 6.7776251477494175e-06,
      "loss": 1.1064,
      "step": 21664
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6856096982955933,
      "learning_rate": 6.776986231351627e-06,
      "loss": 0.8475,
      "step": 21665
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.83797287940979,
      "learning_rate": 6.776347314953838e-06,
      "loss": 1.0177,
      "step": 21666
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.998023509979248,
      "learning_rate": 6.775708398556049e-06,
      "loss": 0.9254,
      "step": 21667
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.381197214126587,
      "learning_rate": 6.775069482158259e-06,
      "loss": 0.9256,
      "step": 21668
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8820527791976929,
      "learning_rate": 6.77443056576047e-06,
      "loss": 0.9854,
      "step": 21669
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.142791271209717,
      "learning_rate": 6.773791649362681e-06,
      "loss": 0.9354,
      "step": 21670
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5868436098098755,
      "learning_rate": 6.7731527329648915e-06,
      "loss": 0.888,
      "step": 21671
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6335248947143555,
      "learning_rate": 6.772513816567102e-06,
      "loss": 0.7983,
      "step": 21672
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.167879581451416,
      "learning_rate": 6.771874900169313e-06,
      "loss": 1.5857,
      "step": 21673
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.2780232429504395,
      "learning_rate": 6.771235983771524e-06,
      "loss": 0.9952,
      "step": 21674
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.270678758621216,
      "learning_rate": 6.770597067373734e-06,
      "loss": 0.9438,
      "step": 21675
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.607334017753601,
      "learning_rate": 6.769958150975945e-06,
      "loss": 0.9217,
      "step": 21676
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8358454704284668,
      "learning_rate": 6.769319234578156e-06,
      "loss": 1.0715,
      "step": 21677
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8437824249267578,
      "learning_rate": 6.768680318180366e-06,
      "loss": 0.9751,
      "step": 21678
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7019201517105103,
      "learning_rate": 6.768041401782577e-06,
      "loss": 0.9001,
      "step": 21679
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6437036991119385,
      "learning_rate": 6.767402485384788e-06,
      "loss": 0.8977,
      "step": 21680
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.798262596130371,
      "learning_rate": 6.7667635689869985e-06,
      "loss": 0.9801,
      "step": 21681
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9749717712402344,
      "learning_rate": 6.766124652589209e-06,
      "loss": 0.7466,
      "step": 21682
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8340436220169067,
      "learning_rate": 6.765485736191419e-06,
      "loss": 0.8801,
      "step": 21683
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5597018003463745,
      "learning_rate": 6.76484681979363e-06,
      "loss": 0.834,
      "step": 21684
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5973886251449585,
      "learning_rate": 6.7642079033958405e-06,
      "loss": 0.8514,
      "step": 21685
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8328838348388672,
      "learning_rate": 6.763568986998051e-06,
      "loss": 0.8784,
      "step": 21686
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.169402837753296,
      "learning_rate": 6.762930070600262e-06,
      "loss": 0.9555,
      "step": 21687
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.870679497718811,
      "learning_rate": 6.7622911542024726e-06,
      "loss": 0.5865,
      "step": 21688
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.893442153930664,
      "learning_rate": 6.761652237804683e-06,
      "loss": 1.7199,
      "step": 21689
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9121530055999756,
      "learning_rate": 6.761013321406894e-06,
      "loss": 0.983,
      "step": 21690
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.775688409805298,
      "learning_rate": 6.760374405009105e-06,
      "loss": 1.0368,
      "step": 21691
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.741660714149475,
      "learning_rate": 6.759735488611315e-06,
      "loss": 0.8935,
      "step": 21692
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3256460428237915,
      "learning_rate": 6.759096572213526e-06,
      "loss": 0.7251,
      "step": 21693
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7077890634536743,
      "learning_rate": 6.758457655815737e-06,
      "loss": 0.878,
      "step": 21694
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7450129985809326,
      "learning_rate": 6.7578187394179475e-06,
      "loss": 0.7958,
      "step": 21695
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5247970819473267,
      "learning_rate": 6.757179823020158e-06,
      "loss": 0.8066,
      "step": 21696
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.168747425079346,
      "learning_rate": 6.756540906622369e-06,
      "loss": 1.4062,
      "step": 21697
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.999915361404419,
      "learning_rate": 6.7559019902245796e-06,
      "loss": 0.822,
      "step": 21698
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8073370456695557,
      "learning_rate": 6.75526307382679e-06,
      "loss": 0.9568,
      "step": 21699
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4653371572494507,
      "learning_rate": 6.754624157429001e-06,
      "loss": 0.7789,
      "step": 21700
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.85687255859375,
      "learning_rate": 6.753985241031211e-06,
      "loss": 0.861,
      "step": 21701
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5947946310043335,
      "learning_rate": 6.7533463246334215e-06,
      "loss": 0.8113,
      "step": 21702
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8332328796386719,
      "learning_rate": 6.752707408235632e-06,
      "loss": 1.1897,
      "step": 21703
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.583378553390503,
      "learning_rate": 6.752068491837843e-06,
      "loss": 0.7483,
      "step": 21704
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3163912296295166,
      "learning_rate": 6.751429575440054e-06,
      "loss": 0.912,
      "step": 21705
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5195330381393433,
      "learning_rate": 6.750790659042264e-06,
      "loss": 0.8026,
      "step": 21706
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6729239225387573,
      "learning_rate": 6.750151742644475e-06,
      "loss": 0.8974,
      "step": 21707
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.771008014678955,
      "learning_rate": 6.749512826246686e-06,
      "loss": 0.8606,
      "step": 21708
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0684525966644287,
      "learning_rate": 6.748873909848896e-06,
      "loss": 0.9632,
      "step": 21709
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.408051490783691,
      "learning_rate": 6.748234993451107e-06,
      "loss": 1.1886,
      "step": 21710
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.9330384731292725,
      "learning_rate": 6.747596077053318e-06,
      "loss": 0.8121,
      "step": 21711
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9162251949310303,
      "learning_rate": 6.7469571606555285e-06,
      "loss": 0.9694,
      "step": 21712
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.2181596755981445,
      "learning_rate": 6.746318244257739e-06,
      "loss": 1.1999,
      "step": 21713
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7229317426681519,
      "learning_rate": 6.74567932785995e-06,
      "loss": 0.8637,
      "step": 21714
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2696433067321777,
      "learning_rate": 6.745040411462161e-06,
      "loss": 0.8948,
      "step": 21715
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.775371789932251,
      "learning_rate": 6.744401495064371e-06,
      "loss": 0.9305,
      "step": 21716
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8114123344421387,
      "learning_rate": 6.743762578666582e-06,
      "loss": 1.1314,
      "step": 21717
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0530622005462646,
      "learning_rate": 6.743123662268793e-06,
      "loss": 0.7363,
      "step": 21718
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.4640631675720215,
      "learning_rate": 6.742484745871003e-06,
      "loss": 1.1829,
      "step": 21719
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.122323989868164,
      "learning_rate": 6.741845829473213e-06,
      "loss": 1.1542,
      "step": 21720
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5412908792495728,
      "learning_rate": 6.741206913075424e-06,
      "loss": 0.5814,
      "step": 21721
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.144911527633667,
      "learning_rate": 6.740567996677635e-06,
      "loss": 0.9074,
      "step": 21722
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.987684726715088,
      "learning_rate": 6.739929080279845e-06,
      "loss": 0.6979,
      "step": 21723
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.816062927246094,
      "learning_rate": 6.739290163882056e-06,
      "loss": 1.2349,
      "step": 21724
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.727652549743652,
      "learning_rate": 6.738651247484267e-06,
      "loss": 1.5825,
      "step": 21725
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1459805965423584,
      "learning_rate": 6.7380123310864775e-06,
      "loss": 0.9643,
      "step": 21726
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7587684392929077,
      "learning_rate": 6.737373414688688e-06,
      "loss": 0.9491,
      "step": 21727
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.20590877532959,
      "learning_rate": 6.736734498290899e-06,
      "loss": 0.9548,
      "step": 21728
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.1666786670684814,
      "learning_rate": 6.7360955818931096e-06,
      "loss": 0.9145,
      "step": 21729
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5156387090682983,
      "learning_rate": 6.73545666549532e-06,
      "loss": 0.8095,
      "step": 21730
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9466253519058228,
      "learning_rate": 6.734817749097531e-06,
      "loss": 1.1282,
      "step": 21731
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2625608444213867,
      "learning_rate": 6.734178832699742e-06,
      "loss": 0.9918,
      "step": 21732
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7536691427230835,
      "learning_rate": 6.733539916301952e-06,
      "loss": 0.8207,
      "step": 21733
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.1494674682617188,
      "learning_rate": 6.732900999904163e-06,
      "loss": 1.2418,
      "step": 21734
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.615989327430725,
      "learning_rate": 6.732262083506374e-06,
      "loss": 0.7872,
      "step": 21735
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.7532739639282227,
      "learning_rate": 6.7316231671085845e-06,
      "loss": 0.8539,
      "step": 21736
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3033409118652344,
      "learning_rate": 6.730984250710795e-06,
      "loss": 0.7149,
      "step": 21737
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4201951026916504,
      "learning_rate": 6.730345334313005e-06,
      "loss": 0.8452,
      "step": 21738
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5920647382736206,
      "learning_rate": 6.729706417915216e-06,
      "loss": 0.7998,
      "step": 21739
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.450140476226807,
      "learning_rate": 6.729067501517426e-06,
      "loss": 0.8273,
      "step": 21740
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6488748788833618,
      "learning_rate": 6.728428585119637e-06,
      "loss": 0.7892,
      "step": 21741
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6924406290054321,
      "learning_rate": 6.727789668721848e-06,
      "loss": 0.9199,
      "step": 21742
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9437674283981323,
      "learning_rate": 6.7271507523240585e-06,
      "loss": 1.0613,
      "step": 21743
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5722671747207642,
      "learning_rate": 6.726511835926269e-06,
      "loss": 0.9142,
      "step": 21744
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9446818828582764,
      "learning_rate": 6.72587291952848e-06,
      "loss": 0.9335,
      "step": 21745
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8301825523376465,
      "learning_rate": 6.725234003130691e-06,
      "loss": 0.8767,
      "step": 21746
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6471682786941528,
      "learning_rate": 6.724595086732901e-06,
      "loss": 0.7836,
      "step": 21747
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.0178911685943604,
      "learning_rate": 6.723956170335112e-06,
      "loss": 0.9958,
      "step": 21748
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.30013370513916,
      "learning_rate": 6.723317253937323e-06,
      "loss": 1.0788,
      "step": 21749
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.958381414413452,
      "learning_rate": 6.722678337539533e-06,
      "loss": 1.068,
      "step": 21750
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6672065258026123,
      "learning_rate": 6.722039421141744e-06,
      "loss": 0.7927,
      "step": 21751
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4972589015960693,
      "learning_rate": 6.721400504743955e-06,
      "loss": 0.8772,
      "step": 21752
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7346714735031128,
      "learning_rate": 6.7207615883461655e-06,
      "loss": 0.8271,
      "step": 21753
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7742564678192139,
      "learning_rate": 6.720122671948376e-06,
      "loss": 0.8846,
      "step": 21754
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.439910888671875,
      "learning_rate": 6.719483755550587e-06,
      "loss": 1.0723,
      "step": 21755
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.650139570236206,
      "learning_rate": 6.718844839152797e-06,
      "loss": 0.8593,
      "step": 21756
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.7120351791381836,
      "learning_rate": 6.7182059227550075e-06,
      "loss": 0.9498,
      "step": 21757
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7400516271591187,
      "learning_rate": 6.717567006357218e-06,
      "loss": 0.8336,
      "step": 21758
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.363410472869873,
      "learning_rate": 6.716928089959429e-06,
      "loss": 0.8323,
      "step": 21759
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.0330135822296143,
      "learning_rate": 6.7162891735616396e-06,
      "loss": 0.7883,
      "step": 21760
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.537131905555725,
      "learning_rate": 6.71565025716385e-06,
      "loss": 0.9189,
      "step": 21761
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.2890474796295166,
      "learning_rate": 6.715011340766061e-06,
      "loss": 0.9028,
      "step": 21762
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9135969877243042,
      "learning_rate": 6.714372424368272e-06,
      "loss": 0.9156,
      "step": 21763
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.769575357437134,
      "learning_rate": 6.713733507970482e-06,
      "loss": 0.7876,
      "step": 21764
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.398516297340393,
      "learning_rate": 6.713094591572693e-06,
      "loss": 0.4203,
      "step": 21765
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2175655364990234,
      "learning_rate": 6.712455675174904e-06,
      "loss": 0.8924,
      "step": 21766
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.6129729747772217,
      "learning_rate": 6.7118167587771145e-06,
      "loss": 0.7493,
      "step": 21767
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.292592525482178,
      "learning_rate": 6.711177842379325e-06,
      "loss": 1.3825,
      "step": 21768
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.8865692615509033,
      "learning_rate": 6.710538925981536e-06,
      "loss": 0.6928,
      "step": 21769
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.290191173553467,
      "learning_rate": 6.7099000095837466e-06,
      "loss": 1.0686,
      "step": 21770
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.023130178451538,
      "learning_rate": 6.709261093185957e-06,
      "loss": 0.8486,
      "step": 21771
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.7928643226623535,
      "learning_rate": 6.708622176788168e-06,
      "loss": 1.033,
      "step": 21772
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1614818572998047,
      "learning_rate": 6.707983260390379e-06,
      "loss": 0.8033,
      "step": 21773
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.939951777458191,
      "learning_rate": 6.7073443439925885e-06,
      "loss": 0.9096,
      "step": 21774
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1003201007843018,
      "learning_rate": 6.706705427594799e-06,
      "loss": 0.9196,
      "step": 21775
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6772162914276123,
      "learning_rate": 6.70606651119701e-06,
      "loss": 0.868,
      "step": 21776
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.866848349571228,
      "learning_rate": 6.705427594799221e-06,
      "loss": 0.8065,
      "step": 21777
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.603448510169983,
      "learning_rate": 6.704788678401431e-06,
      "loss": 0.8567,
      "step": 21778
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.592398762702942,
      "learning_rate": 6.704149762003642e-06,
      "loss": 0.8141,
      "step": 21779
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4828526973724365,
      "learning_rate": 6.703510845605853e-06,
      "loss": 0.4041,
      "step": 21780
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.181128978729248,
      "learning_rate": 6.702871929208063e-06,
      "loss": 1.2417,
      "step": 21781
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.019525527954102,
      "learning_rate": 6.702233012810274e-06,
      "loss": 1.0185,
      "step": 21782
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8917139768600464,
      "learning_rate": 6.701594096412485e-06,
      "loss": 0.876,
      "step": 21783
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.380203247070312,
      "learning_rate": 6.7009551800146955e-06,
      "loss": 1.5802,
      "step": 21784
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.2717490196228027,
      "learning_rate": 6.700316263616906e-06,
      "loss": 0.6636,
      "step": 21785
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1480507850646973,
      "learning_rate": 6.699677347219117e-06,
      "loss": 0.967,
      "step": 21786
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.6448941230773926,
      "learning_rate": 6.699038430821328e-06,
      "loss": 1.1741,
      "step": 21787
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7153180837631226,
      "learning_rate": 6.698399514423538e-06,
      "loss": 0.8344,
      "step": 21788
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.559534072875977,
      "learning_rate": 6.697760598025749e-06,
      "loss": 1.1732,
      "step": 21789
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6498558521270752,
      "learning_rate": 6.69712168162796e-06,
      "loss": 0.8341,
      "step": 21790
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9623945951461792,
      "learning_rate": 6.69648276523017e-06,
      "loss": 0.8027,
      "step": 21791
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0920042991638184,
      "learning_rate": 6.695843848832381e-06,
      "loss": 1.071,
      "step": 21792
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.0948030948638916,
      "learning_rate": 6.695204932434591e-06,
      "loss": 0.9238,
      "step": 21793
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.594960331916809,
      "learning_rate": 6.694566016036802e-06,
      "loss": 0.8453,
      "step": 21794
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8110038042068481,
      "learning_rate": 6.693927099639012e-06,
      "loss": 0.8633,
      "step": 21795
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9267795085906982,
      "learning_rate": 6.693288183241223e-06,
      "loss": 0.858,
      "step": 21796
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8455692529678345,
      "learning_rate": 6.692649266843434e-06,
      "loss": 0.7946,
      "step": 21797
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3794257640838623,
      "learning_rate": 6.6920103504456445e-06,
      "loss": 0.6743,
      "step": 21798
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6610592603683472,
      "learning_rate": 6.691371434047855e-06,
      "loss": 0.9549,
      "step": 21799
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6840261220932007,
      "learning_rate": 6.690732517650066e-06,
      "loss": 0.8111,
      "step": 21800
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.171766519546509,
      "learning_rate": 6.6900936012522766e-06,
      "loss": 1.217,
      "step": 21801
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5987253189086914,
      "learning_rate": 6.689454684854487e-06,
      "loss": 0.8435,
      "step": 21802
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.141136884689331,
      "learning_rate": 6.688815768456698e-06,
      "loss": 0.8264,
      "step": 21803
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1246213912963867,
      "learning_rate": 6.688176852058909e-06,
      "loss": 0.8273,
      "step": 21804
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.5490379333496094,
      "learning_rate": 6.687537935661119e-06,
      "loss": 1.1563,
      "step": 21805
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1046245098114014,
      "learning_rate": 6.68689901926333e-06,
      "loss": 0.9281,
      "step": 21806
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8865115642547607,
      "learning_rate": 6.686260102865541e-06,
      "loss": 0.8512,
      "step": 21807
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1425375938415527,
      "learning_rate": 6.6856211864677515e-06,
      "loss": 0.8288,
      "step": 21808
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.730969190597534,
      "learning_rate": 6.684982270069962e-06,
      "loss": 0.9819,
      "step": 21809
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.633451223373413,
      "learning_rate": 6.684343353672173e-06,
      "loss": 0.827,
      "step": 21810
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.73872971534729,
      "learning_rate": 6.683704437274383e-06,
      "loss": 1.0881,
      "step": 21811
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.861298084259033,
      "learning_rate": 6.6830655208765934e-06,
      "loss": 0.9541,
      "step": 21812
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.04192590713501,
      "learning_rate": 6.682426604478804e-06,
      "loss": 0.9892,
      "step": 21813
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8686161041259766,
      "learning_rate": 6.681787688081015e-06,
      "loss": 0.9299,
      "step": 21814
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2995471954345703,
      "learning_rate": 6.6811487716832255e-06,
      "loss": 0.4071,
      "step": 21815
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6931532621383667,
      "learning_rate": 6.680509855285436e-06,
      "loss": 0.7071,
      "step": 21816
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.147996187210083,
      "learning_rate": 6.679870938887647e-06,
      "loss": 1.0517,
      "step": 21817
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.27996563911438,
      "learning_rate": 6.679232022489858e-06,
      "loss": 0.6473,
      "step": 21818
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.302726745605469,
      "learning_rate": 6.678593106092068e-06,
      "loss": 1.2806,
      "step": 21819
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.607569932937622,
      "learning_rate": 6.677954189694279e-06,
      "loss": 0.6321,
      "step": 21820
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7781195640563965,
      "learning_rate": 6.67731527329649e-06,
      "loss": 0.9359,
      "step": 21821
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.184347629547119,
      "learning_rate": 6.6766763568987004e-06,
      "loss": 0.7571,
      "step": 21822
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7922821044921875,
      "learning_rate": 6.676037440500911e-06,
      "loss": 0.7763,
      "step": 21823
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8768055438995361,
      "learning_rate": 6.675398524103122e-06,
      "loss": 0.8694,
      "step": 21824
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7230850458145142,
      "learning_rate": 6.6747596077053325e-06,
      "loss": 0.8146,
      "step": 21825
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9087563753128052,
      "learning_rate": 6.674120691307543e-06,
      "loss": 0.845,
      "step": 21826
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8490904569625854,
      "learning_rate": 6.673481774909754e-06,
      "loss": 0.802,
      "step": 21827
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.96813702583313,
      "learning_rate": 6.672842858511965e-06,
      "loss": 0.3931,
      "step": 21828
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3385772705078125,
      "learning_rate": 6.6722039421141745e-06,
      "loss": 1.0092,
      "step": 21829
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5882484912872314,
      "learning_rate": 6.671565025716385e-06,
      "loss": 0.9747,
      "step": 21830
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4662697315216064,
      "learning_rate": 6.670926109318596e-06,
      "loss": 1.0105,
      "step": 21831
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.475867986679077,
      "learning_rate": 6.670287192920807e-06,
      "loss": 1.093,
      "step": 21832
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8794584274291992,
      "learning_rate": 6.669648276523017e-06,
      "loss": 0.8645,
      "step": 21833
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.87777578830719,
      "learning_rate": 6.669009360125228e-06,
      "loss": 1.1048,
      "step": 21834
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.861159086227417,
      "learning_rate": 6.668370443727439e-06,
      "loss": 0.9422,
      "step": 21835
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7878541946411133,
      "learning_rate": 6.667731527329649e-06,
      "loss": 0.8256,
      "step": 21836
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3786650896072388,
      "learning_rate": 6.66709261093186e-06,
      "loss": 0.7099,
      "step": 21837
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.536847710609436,
      "learning_rate": 6.666453694534071e-06,
      "loss": 0.9819,
      "step": 21838
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7550272941589355,
      "learning_rate": 6.6658147781362815e-06,
      "loss": 0.7201,
      "step": 21839
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.034741163253784,
      "learning_rate": 6.665175861738492e-06,
      "loss": 0.8271,
      "step": 21840
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8489028215408325,
      "learning_rate": 6.664536945340703e-06,
      "loss": 0.9487,
      "step": 21841
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4780445098876953,
      "learning_rate": 6.663898028942914e-06,
      "loss": 1.3594,
      "step": 21842
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.036531448364258,
      "learning_rate": 6.663259112545124e-06,
      "loss": 0.9327,
      "step": 21843
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0200998783111572,
      "learning_rate": 6.662620196147335e-06,
      "loss": 0.8014,
      "step": 21844
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.304774522781372,
      "learning_rate": 6.661981279749546e-06,
      "loss": 0.8288,
      "step": 21845
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.017733573913574,
      "learning_rate": 6.661342363351756e-06,
      "loss": 1.434,
      "step": 21846
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6466597318649292,
      "learning_rate": 6.660703446953966e-06,
      "loss": 0.8515,
      "step": 21847
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0238630771636963,
      "learning_rate": 6.660064530556177e-06,
      "loss": 0.9458,
      "step": 21848
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.817356586456299,
      "learning_rate": 6.659425614158388e-06,
      "loss": 1.0273,
      "step": 21849
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8462796211242676,
      "learning_rate": 6.658786697760598e-06,
      "loss": 0.8888,
      "step": 21850
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7694485187530518,
      "learning_rate": 6.658147781362809e-06,
      "loss": 0.8716,
      "step": 21851
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.461672306060791,
      "learning_rate": 6.65750886496502e-06,
      "loss": 0.8886,
      "step": 21852
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.7608642578125,
      "learning_rate": 6.6568699485672304e-06,
      "loss": 1.3588,
      "step": 21853
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7124791145324707,
      "learning_rate": 6.656231032169441e-06,
      "loss": 0.814,
      "step": 21854
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8659534454345703,
      "learning_rate": 6.655592115771652e-06,
      "loss": 0.8165,
      "step": 21855
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9039658308029175,
      "learning_rate": 6.6549531993738625e-06,
      "loss": 0.9566,
      "step": 21856
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7378196716308594,
      "learning_rate": 6.654314282976073e-06,
      "loss": 0.8052,
      "step": 21857
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.812791109085083,
      "learning_rate": 6.653675366578284e-06,
      "loss": 0.7962,
      "step": 21858
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7171142101287842,
      "learning_rate": 6.653036450180495e-06,
      "loss": 0.9168,
      "step": 21859
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.8931705951690674,
      "learning_rate": 6.652397533782705e-06,
      "loss": 1.0226,
      "step": 21860
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0906572341918945,
      "learning_rate": 6.651758617384916e-06,
      "loss": 0.8441,
      "step": 21861
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.167462110519409,
      "learning_rate": 6.651119700987127e-06,
      "loss": 0.9778,
      "step": 21862
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5629953145980835,
      "learning_rate": 6.6504807845893374e-06,
      "loss": 0.7786,
      "step": 21863
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8701716661453247,
      "learning_rate": 6.649841868191548e-06,
      "loss": 0.817,
      "step": 21864
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5870236158370972,
      "learning_rate": 6.649202951793759e-06,
      "loss": 0.806,
      "step": 21865
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.522330403327942,
      "learning_rate": 6.648564035395969e-06,
      "loss": 0.8611,
      "step": 21866
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9991470575332642,
      "learning_rate": 6.647925118998179e-06,
      "loss": 0.8654,
      "step": 21867
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.028001308441162,
      "learning_rate": 6.64728620260039e-06,
      "loss": 1.1738,
      "step": 21868
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7813056707382202,
      "learning_rate": 6.646647286202601e-06,
      "loss": 0.8651,
      "step": 21869
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8038862943649292,
      "learning_rate": 6.6460083698048115e-06,
      "loss": 0.9809,
      "step": 21870
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.5039913654327393,
      "learning_rate": 6.645369453407022e-06,
      "loss": 1.02,
      "step": 21871
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.6017873287200928,
      "learning_rate": 6.644730537009233e-06,
      "loss": 1.2173,
      "step": 21872
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.9755847454071045,
      "learning_rate": 6.644091620611444e-06,
      "loss": 0.9233,
      "step": 21873
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7478066682815552,
      "learning_rate": 6.643452704213654e-06,
      "loss": 0.8168,
      "step": 21874
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5940954685211182,
      "learning_rate": 6.642813787815865e-06,
      "loss": 0.7972,
      "step": 21875
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8799643516540527,
      "learning_rate": 6.642174871418076e-06,
      "loss": 1.0289,
      "step": 21876
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1488895416259766,
      "learning_rate": 6.641535955020286e-06,
      "loss": 0.9694,
      "step": 21877
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.3701257705688477,
      "learning_rate": 6.640897038622497e-06,
      "loss": 0.8483,
      "step": 21878
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9487251043319702,
      "learning_rate": 6.640258122224708e-06,
      "loss": 0.8828,
      "step": 21879
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.294957160949707,
      "learning_rate": 6.6396192058269185e-06,
      "loss": 0.9826,
      "step": 21880
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7338050603866577,
      "learning_rate": 6.638980289429129e-06,
      "loss": 0.7736,
      "step": 21881
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.51418924331665,
      "learning_rate": 6.63834137303134e-06,
      "loss": 1.3792,
      "step": 21882
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4287666082382202,
      "learning_rate": 6.637702456633551e-06,
      "loss": 0.4518,
      "step": 21883
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9179216623306274,
      "learning_rate": 6.6370635402357604e-06,
      "loss": 0.9084,
      "step": 21884
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.490378975868225,
      "learning_rate": 6.636424623837971e-06,
      "loss": 0.4091,
      "step": 21885
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9419444799423218,
      "learning_rate": 6.635785707440182e-06,
      "loss": 0.8078,
      "step": 21886
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3255819082260132,
      "learning_rate": 6.6351467910423925e-06,
      "loss": 0.4169,
      "step": 21887
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.923292398452759,
      "learning_rate": 6.634507874644603e-06,
      "loss": 1.0539,
      "step": 21888
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4628134965896606,
      "learning_rate": 6.633868958246814e-06,
      "loss": 0.8831,
      "step": 21889
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.971315860748291,
      "learning_rate": 6.633230041849025e-06,
      "loss": 1.5315,
      "step": 21890
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.384779930114746,
      "learning_rate": 6.632591125451235e-06,
      "loss": 0.6754,
      "step": 21891
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8495084047317505,
      "learning_rate": 6.631952209053446e-06,
      "loss": 0.7321,
      "step": 21892
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5953545570373535,
      "learning_rate": 6.631313292655657e-06,
      "loss": 0.8479,
      "step": 21893
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9895329475402832,
      "learning_rate": 6.6306743762578674e-06,
      "loss": 0.9896,
      "step": 21894
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4834978580474854,
      "learning_rate": 6.630035459860078e-06,
      "loss": 1.0668,
      "step": 21895
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.215976238250732,
      "learning_rate": 6.629396543462289e-06,
      "loss": 0.99,
      "step": 21896
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5388615131378174,
      "learning_rate": 6.6287576270644995e-06,
      "loss": 0.5964,
      "step": 21897
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7524687051773071,
      "learning_rate": 6.62811871066671e-06,
      "loss": 0.8523,
      "step": 21898
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0153212547302246,
      "learning_rate": 6.627479794268921e-06,
      "loss": 0.6952,
      "step": 21899
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7005360126495361,
      "learning_rate": 6.626840877871132e-06,
      "loss": 0.9043,
      "step": 21900
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.8220324516296387,
      "learning_rate": 6.626201961473342e-06,
      "loss": 1.2614,
      "step": 21901
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7353945970535278,
      "learning_rate": 6.625563045075552e-06,
      "loss": 0.868,
      "step": 21902
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.781667709350586,
      "learning_rate": 6.624924128677763e-06,
      "loss": 1.1813,
      "step": 21903
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3702369928359985,
      "learning_rate": 6.624285212279974e-06,
      "loss": 0.7287,
      "step": 21904
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.223958969116211,
      "learning_rate": 6.623646295882184e-06,
      "loss": 0.8694,
      "step": 21905
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.776888370513916,
      "learning_rate": 6.623007379484395e-06,
      "loss": 1.129,
      "step": 21906
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8287140130996704,
      "learning_rate": 6.622368463086606e-06,
      "loss": 0.8112,
      "step": 21907
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7280337810516357,
      "learning_rate": 6.621729546688816e-06,
      "loss": 1.1434,
      "step": 21908
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5784456729888916,
      "learning_rate": 6.621090630291027e-06,
      "loss": 0.9233,
      "step": 21909
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5278948545455933,
      "learning_rate": 6.620451713893238e-06,
      "loss": 0.8274,
      "step": 21910
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5343666076660156,
      "learning_rate": 6.6198127974954485e-06,
      "loss": 0.9809,
      "step": 21911
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.499484539031982,
      "learning_rate": 6.619173881097659e-06,
      "loss": 1.2383,
      "step": 21912
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.235380172729492,
      "learning_rate": 6.61853496469987e-06,
      "loss": 1.3526,
      "step": 21913
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8181023597717285,
      "learning_rate": 6.617896048302081e-06,
      "loss": 0.8482,
      "step": 21914
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5858932733535767,
      "learning_rate": 6.617257131904291e-06,
      "loss": 0.8378,
      "step": 21915
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.884332180023193,
      "learning_rate": 6.616618215506502e-06,
      "loss": 1.578,
      "step": 21916
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6104364395141602,
      "learning_rate": 6.615979299108713e-06,
      "loss": 0.8226,
      "step": 21917
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.164085865020752,
      "learning_rate": 6.615340382710923e-06,
      "loss": 1.0951,
      "step": 21918
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7775731086730957,
      "learning_rate": 6.614701466313134e-06,
      "loss": 0.815,
      "step": 21919
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5083999633789062,
      "learning_rate": 6.614062549915344e-06,
      "loss": 0.8302,
      "step": 21920
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.364525079727173,
      "learning_rate": 6.613423633517555e-06,
      "loss": 0.7251,
      "step": 21921
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2586586475372314,
      "learning_rate": 6.612784717119765e-06,
      "loss": 0.6795,
      "step": 21922
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.704655408859253,
      "learning_rate": 6.612145800721976e-06,
      "loss": 0.9242,
      "step": 21923
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5370161533355713,
      "learning_rate": 6.611506884324187e-06,
      "loss": 0.3921,
      "step": 21924
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1086487770080566,
      "learning_rate": 6.6108679679263974e-06,
      "loss": 0.825,
      "step": 21925
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.633952260017395,
      "learning_rate": 6.610229051528608e-06,
      "loss": 0.9296,
      "step": 21926
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.7069218158721924,
      "learning_rate": 6.609590135130819e-06,
      "loss": 0.8422,
      "step": 21927
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.269221544265747,
      "learning_rate": 6.6089512187330295e-06,
      "loss": 0.9754,
      "step": 21928
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.740215301513672,
      "learning_rate": 6.60831230233524e-06,
      "loss": 0.8243,
      "step": 21929
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8898056745529175,
      "learning_rate": 6.607673385937451e-06,
      "loss": 0.922,
      "step": 21930
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8591021299362183,
      "learning_rate": 6.607034469539662e-06,
      "loss": 0.5384,
      "step": 21931
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6888208389282227,
      "learning_rate": 6.606395553141872e-06,
      "loss": 1.0741,
      "step": 21932
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8707367181777954,
      "learning_rate": 6.605756636744083e-06,
      "loss": 0.9031,
      "step": 21933
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5106011629104614,
      "learning_rate": 6.605117720346294e-06,
      "loss": 0.8627,
      "step": 21934
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9539116621017456,
      "learning_rate": 6.6044788039485044e-06,
      "loss": 0.7268,
      "step": 21935
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2635881900787354,
      "learning_rate": 6.603839887550715e-06,
      "loss": 0.8557,
      "step": 21936
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.165130615234375,
      "learning_rate": 6.603200971152926e-06,
      "loss": 0.889,
      "step": 21937
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9787131547927856,
      "learning_rate": 6.6025620547551365e-06,
      "loss": 0.7782,
      "step": 21938
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5619484186172485,
      "learning_rate": 6.601923138357346e-06,
      "loss": 0.7131,
      "step": 21939
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.8092586994171143,
      "learning_rate": 6.601284221959557e-06,
      "loss": 1.1116,
      "step": 21940
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.7107205390930176,
      "learning_rate": 6.600645305561768e-06,
      "loss": 1.0724,
      "step": 21941
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9376848936080933,
      "learning_rate": 6.6000063891639785e-06,
      "loss": 0.7972,
      "step": 21942
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4227023124694824,
      "learning_rate": 6.599367472766189e-06,
      "loss": 0.8713,
      "step": 21943
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.327099084854126,
      "learning_rate": 6.5987285563684e-06,
      "loss": 0.8779,
      "step": 21944
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4479377269744873,
      "learning_rate": 6.598089639970611e-06,
      "loss": 0.8346,
      "step": 21945
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7979642152786255,
      "learning_rate": 6.597450723572821e-06,
      "loss": 0.8796,
      "step": 21946
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9911928176879883,
      "learning_rate": 6.596811807175032e-06,
      "loss": 0.9168,
      "step": 21947
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7398576736450195,
      "learning_rate": 6.596172890777243e-06,
      "loss": 0.8083,
      "step": 21948
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.787360668182373,
      "learning_rate": 6.595533974379453e-06,
      "loss": 0.9092,
      "step": 21949
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6798568964004517,
      "learning_rate": 6.594895057981664e-06,
      "loss": 0.7888,
      "step": 21950
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.98441481590271,
      "learning_rate": 6.594256141583875e-06,
      "loss": 0.7911,
      "step": 21951
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6354734897613525,
      "learning_rate": 6.5936172251860855e-06,
      "loss": 0.8244,
      "step": 21952
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9719871282577515,
      "learning_rate": 6.592978308788296e-06,
      "loss": 0.8108,
      "step": 21953
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.866663932800293,
      "learning_rate": 6.592339392390507e-06,
      "loss": 0.8927,
      "step": 21954
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6824148893356323,
      "learning_rate": 6.591700475992718e-06,
      "loss": 0.8097,
      "step": 21955
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.251166582107544,
      "learning_rate": 6.591061559594928e-06,
      "loss": 0.916,
      "step": 21956
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8857897520065308,
      "learning_rate": 6.590422643197138e-06,
      "loss": 0.8764,
      "step": 21957
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.174602746963501,
      "learning_rate": 6.589783726799349e-06,
      "loss": 0.8819,
      "step": 21958
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0786702632904053,
      "learning_rate": 6.5891448104015595e-06,
      "loss": 0.8524,
      "step": 21959
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0349082946777344,
      "learning_rate": 6.58850589400377e-06,
      "loss": 0.6986,
      "step": 21960
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.12404203414917,
      "learning_rate": 6.587866977605981e-06,
      "loss": 0.7215,
      "step": 21961
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8993943929672241,
      "learning_rate": 6.587228061208192e-06,
      "loss": 0.91,
      "step": 21962
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.813322067260742,
      "learning_rate": 6.586589144810402e-06,
      "loss": 0.8748,
      "step": 21963
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.041189670562744,
      "learning_rate": 6.585950228412613e-06,
      "loss": 1.144,
      "step": 21964
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.511780738830566,
      "learning_rate": 6.585311312014824e-06,
      "loss": 0.9164,
      "step": 21965
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1054608821868896,
      "learning_rate": 6.5846723956170344e-06,
      "loss": 0.8738,
      "step": 21966
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7618964910507202,
      "learning_rate": 6.584033479219245e-06,
      "loss": 0.9192,
      "step": 21967
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.622232437133789,
      "learning_rate": 6.583394562821456e-06,
      "loss": 1.2464,
      "step": 21968
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.0096540451049805,
      "learning_rate": 6.5827556464236665e-06,
      "loss": 0.6397,
      "step": 21969
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.122105598449707,
      "learning_rate": 6.582116730025877e-06,
      "loss": 1.4817,
      "step": 21970
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9269016981124878,
      "learning_rate": 6.581477813628088e-06,
      "loss": 0.9088,
      "step": 21971
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.741755723953247,
      "learning_rate": 6.580838897230299e-06,
      "loss": 0.9798,
      "step": 21972
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.8245959281921387,
      "learning_rate": 6.580199980832509e-06,
      "loss": 1.1652,
      "step": 21973
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.456768751144409,
      "learning_rate": 6.57956106443472e-06,
      "loss": 0.8827,
      "step": 21974
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.799317479133606,
      "learning_rate": 6.57892214803693e-06,
      "loss": 0.8456,
      "step": 21975
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6541889905929565,
      "learning_rate": 6.578283231639141e-06,
      "loss": 0.8297,
      "step": 21976
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.8907711505889893,
      "learning_rate": 6.577644315241351e-06,
      "loss": 0.9206,
      "step": 21977
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8497309684753418,
      "learning_rate": 6.577005398843562e-06,
      "loss": 0.8984,
      "step": 21978
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1473915576934814,
      "learning_rate": 6.576366482445773e-06,
      "loss": 0.9571,
      "step": 21979
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7121285200119019,
      "learning_rate": 6.575727566047983e-06,
      "loss": 0.8925,
      "step": 21980
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.847399950027466,
      "learning_rate": 6.575088649650194e-06,
      "loss": 1.3664,
      "step": 21981
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.476949691772461,
      "learning_rate": 6.574449733252405e-06,
      "loss": 0.3772,
      "step": 21982
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.587294340133667,
      "learning_rate": 6.5738108168546155e-06,
      "loss": 0.7233,
      "step": 21983
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.429685354232788,
      "learning_rate": 6.573171900456826e-06,
      "loss": 0.9574,
      "step": 21984
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.1515116691589355,
      "learning_rate": 6.572532984059037e-06,
      "loss": 0.9278,
      "step": 21985
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8100817203521729,
      "learning_rate": 6.571894067661248e-06,
      "loss": 0.9068,
      "step": 21986
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.630829930305481,
      "learning_rate": 6.571255151263458e-06,
      "loss": 0.7651,
      "step": 21987
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.771660327911377,
      "learning_rate": 6.570616234865669e-06,
      "loss": 0.848,
      "step": 21988
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5325596332550049,
      "learning_rate": 6.56997731846788e-06,
      "loss": 0.877,
      "step": 21989
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8725870847702026,
      "learning_rate": 6.56933840207009e-06,
      "loss": 0.8495,
      "step": 21990
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8167802095413208,
      "learning_rate": 6.568699485672301e-06,
      "loss": 0.8056,
      "step": 21991
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.643756628036499,
      "learning_rate": 6.568060569274512e-06,
      "loss": 0.7145,
      "step": 21992
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.10823917388916,
      "learning_rate": 6.567421652876722e-06,
      "loss": 0.8451,
      "step": 21993
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9729442596435547,
      "learning_rate": 6.566782736478932e-06,
      "loss": 0.8168,
      "step": 21994
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5337319374084473,
      "learning_rate": 6.566143820081143e-06,
      "loss": 0.8454,
      "step": 21995
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3936684131622314,
      "learning_rate": 6.565504903683354e-06,
      "loss": 0.9498,
      "step": 21996
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4665030241012573,
      "learning_rate": 6.5648659872855644e-06,
      "loss": 0.8718,
      "step": 21997
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9139330387115479,
      "learning_rate": 6.564227070887775e-06,
      "loss": 0.9199,
      "step": 21998
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9489833116531372,
      "learning_rate": 6.563588154489986e-06,
      "loss": 0.8623,
      "step": 21999
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4434196949005127,
      "learning_rate": 6.5629492380921965e-06,
      "loss": 0.9058,
      "step": 22000
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6734853982925415,
      "learning_rate": 6.562310321694407e-06,
      "loss": 0.8616,
      "step": 22001
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.023738145828247,
      "learning_rate": 6.561671405296618e-06,
      "loss": 0.9644,
      "step": 22002
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6540160179138184,
      "learning_rate": 6.561032488898829e-06,
      "loss": 0.8531,
      "step": 22003
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6499764919281006,
      "learning_rate": 6.560393572501039e-06,
      "loss": 0.818,
      "step": 22004
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.2288641929626465,
      "learning_rate": 6.55975465610325e-06,
      "loss": 0.8849,
      "step": 22005
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3837947845458984,
      "learning_rate": 6.559115739705461e-06,
      "loss": 0.7278,
      "step": 22006
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.487351655960083,
      "learning_rate": 6.5584768233076714e-06,
      "loss": 0.9612,
      "step": 22007
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.277000904083252,
      "learning_rate": 6.557837906909882e-06,
      "loss": 0.908,
      "step": 22008
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.700065016746521,
      "learning_rate": 6.557198990512093e-06,
      "loss": 0.6339,
      "step": 22009
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1438443660736084,
      "learning_rate": 6.5565600741143035e-06,
      "loss": 1.0105,
      "step": 22010
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.663065195083618,
      "learning_rate": 6.555921157716514e-06,
      "loss": 1.3228,
      "step": 22011
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.561049461364746,
      "learning_rate": 6.555282241318724e-06,
      "loss": 0.9117,
      "step": 22012
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.727640986442566,
      "learning_rate": 6.554643324920935e-06,
      "loss": 0.9061,
      "step": 22013
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4053025245666504,
      "learning_rate": 6.5540044085231455e-06,
      "loss": 0.7021,
      "step": 22014
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7313743829727173,
      "learning_rate": 6.553365492125356e-06,
      "loss": 0.909,
      "step": 22015
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.6084320545196533,
      "learning_rate": 6.552726575727567e-06,
      "loss": 0.4888,
      "step": 22016
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.892096519470215,
      "learning_rate": 6.552087659329778e-06,
      "loss": 0.9925,
      "step": 22017
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5551998615264893,
      "learning_rate": 6.551448742931988e-06,
      "loss": 0.975,
      "step": 22018
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5905978679656982,
      "learning_rate": 6.550809826534199e-06,
      "loss": 0.8287,
      "step": 22019
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6598259210586548,
      "learning_rate": 6.55017091013641e-06,
      "loss": 0.7713,
      "step": 22020
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3159899711608887,
      "learning_rate": 6.54953199373862e-06,
      "loss": 0.9299,
      "step": 22021
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0265254974365234,
      "learning_rate": 6.548893077340831e-06,
      "loss": 0.9106,
      "step": 22022
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.884520173072815,
      "learning_rate": 6.548254160943042e-06,
      "loss": 0.8043,
      "step": 22023
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.727530479431152,
      "learning_rate": 6.5476152445452525e-06,
      "loss": 1.3107,
      "step": 22024
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5357893705368042,
      "learning_rate": 6.546976328147463e-06,
      "loss": 0.8861,
      "step": 22025
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8951563835144043,
      "learning_rate": 6.546337411749674e-06,
      "loss": 0.9181,
      "step": 22026
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3159677982330322,
      "learning_rate": 6.545698495351885e-06,
      "loss": 0.7946,
      "step": 22027
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6311836242675781,
      "learning_rate": 6.545059578954095e-06,
      "loss": 0.8753,
      "step": 22028
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.296638250350952,
      "learning_rate": 6.544420662556306e-06,
      "loss": 0.7876,
      "step": 22029
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.324251174926758,
      "learning_rate": 6.543781746158516e-06,
      "loss": 1.055,
      "step": 22030
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2354238033294678,
      "learning_rate": 6.543142829760726e-06,
      "loss": 0.9155,
      "step": 22031
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9693435430526733,
      "learning_rate": 6.542503913362936e-06,
      "loss": 0.8455,
      "step": 22032
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.866569995880127,
      "learning_rate": 6.541864996965147e-06,
      "loss": 0.9231,
      "step": 22033
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7658649682998657,
      "learning_rate": 6.541226080567358e-06,
      "loss": 0.7906,
      "step": 22034
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2686240673065186,
      "learning_rate": 6.5405871641695685e-06,
      "loss": 1.1397,
      "step": 22035
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9600815773010254,
      "learning_rate": 6.539948247771779e-06,
      "loss": 0.9401,
      "step": 22036
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7660962343215942,
      "learning_rate": 6.53930933137399e-06,
      "loss": 0.8868,
      "step": 22037
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2092761993408203,
      "learning_rate": 6.538670414976201e-06,
      "loss": 1.115,
      "step": 22038
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9977524280548096,
      "learning_rate": 6.538031498578411e-06,
      "loss": 0.8744,
      "step": 22039
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.521748423576355,
      "learning_rate": 6.537392582180622e-06,
      "loss": 0.8568,
      "step": 22040
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1710307598114014,
      "learning_rate": 6.536753665782833e-06,
      "loss": 0.8764,
      "step": 22041
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.972406268119812,
      "learning_rate": 6.5361147493850426e-06,
      "loss": 0.8701,
      "step": 22042
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.9855217933654785,
      "learning_rate": 6.535475832987253e-06,
      "loss": 1.2446,
      "step": 22043
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.056016683578491,
      "learning_rate": 6.534836916589464e-06,
      "loss": 1.158,
      "step": 22044
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.939039707183838,
      "learning_rate": 6.534198000191675e-06,
      "loss": 0.9965,
      "step": 22045
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6445437669754028,
      "learning_rate": 6.533559083793885e-06,
      "loss": 0.8818,
      "step": 22046
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1605710983276367,
      "learning_rate": 6.532920167396096e-06,
      "loss": 1.0039,
      "step": 22047
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6115713119506836,
      "learning_rate": 6.532281250998307e-06,
      "loss": 0.8135,
      "step": 22048
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.616098403930664,
      "learning_rate": 6.5316423346005175e-06,
      "loss": 0.7624,
      "step": 22049
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.424283504486084,
      "learning_rate": 6.531003418202728e-06,
      "loss": 1.4332,
      "step": 22050
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.905626058578491,
      "learning_rate": 6.530364501804939e-06,
      "loss": 0.7068,
      "step": 22051
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.920522928237915,
      "learning_rate": 6.5297255854071496e-06,
      "loss": 1.1399,
      "step": 22052
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4807437658309937,
      "learning_rate": 6.52908666900936e-06,
      "loss": 0.8315,
      "step": 22053
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.761299729347229,
      "learning_rate": 6.528447752611571e-06,
      "loss": 0.774,
      "step": 22054
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.314262628555298,
      "learning_rate": 6.527808836213782e-06,
      "loss": 0.9001,
      "step": 22055
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.23016095161438,
      "learning_rate": 6.527169919815992e-06,
      "loss": 0.9884,
      "step": 22056
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.50557541847229,
      "learning_rate": 6.526531003418203e-06,
      "loss": 0.4652,
      "step": 22057
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.90445077419281,
      "learning_rate": 6.525892087020414e-06,
      "loss": 0.7966,
      "step": 22058
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.2113423347473145,
      "learning_rate": 6.5252531706226245e-06,
      "loss": 1.0043,
      "step": 22059
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4071738719940186,
      "learning_rate": 6.524614254224835e-06,
      "loss": 0.937,
      "step": 22060
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1952905654907227,
      "learning_rate": 6.523975337827045e-06,
      "loss": 0.7164,
      "step": 22061
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7786600589752197,
      "learning_rate": 6.523336421429256e-06,
      "loss": 0.8912,
      "step": 22062
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.353389263153076,
      "learning_rate": 6.522697505031466e-06,
      "loss": 0.9688,
      "step": 22063
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.782394289970398,
      "learning_rate": 6.522058588633677e-06,
      "loss": 0.9172,
      "step": 22064
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1170787811279297,
      "learning_rate": 6.521419672235888e-06,
      "loss": 0.8399,
      "step": 22065
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.2436673641204834,
      "learning_rate": 6.5207807558380985e-06,
      "loss": 0.7317,
      "step": 22066
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6149396896362305,
      "learning_rate": 6.520141839440309e-06,
      "loss": 0.8379,
      "step": 22067
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7356857061386108,
      "learning_rate": 6.51950292304252e-06,
      "loss": 0.9,
      "step": 22068
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6154468059539795,
      "learning_rate": 6.518864006644731e-06,
      "loss": 0.885,
      "step": 22069
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2936763763427734,
      "learning_rate": 6.518225090246941e-06,
      "loss": 0.9429,
      "step": 22070
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0940117835998535,
      "learning_rate": 6.517586173849152e-06,
      "loss": 1.0397,
      "step": 22071
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9047633409500122,
      "learning_rate": 6.516947257451363e-06,
      "loss": 0.7643,
      "step": 22072
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9793535470962524,
      "learning_rate": 6.516308341053573e-06,
      "loss": 1.0065,
      "step": 22073
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9136921167373657,
      "learning_rate": 6.515669424655784e-06,
      "loss": 0.9183,
      "step": 22074
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.25394344329834,
      "learning_rate": 6.515030508257995e-06,
      "loss": 1.2305,
      "step": 22075
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.115483283996582,
      "learning_rate": 6.5143915918602055e-06,
      "loss": 0.9777,
      "step": 22076
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9782778024673462,
      "learning_rate": 6.513752675462416e-06,
      "loss": 0.8482,
      "step": 22077
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.1691181659698486,
      "learning_rate": 6.513113759064627e-06,
      "loss": 0.9135,
      "step": 22078
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.7268762588500977,
      "learning_rate": 6.512474842666837e-06,
      "loss": 0.789,
      "step": 22079
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.7146072387695312,
      "learning_rate": 6.5118359262690475e-06,
      "loss": 0.6026,
      "step": 22080
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4507704973220825,
      "learning_rate": 6.511197009871258e-06,
      "loss": 0.8949,
      "step": 22081
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.8194644451141357,
      "learning_rate": 6.510558093473469e-06,
      "loss": 1.0451,
      "step": 22082
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.5897231101989746,
      "learning_rate": 6.5099191770756796e-06,
      "loss": 0.7536,
      "step": 22083
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7770094871520996,
      "learning_rate": 6.50928026067789e-06,
      "loss": 0.8518,
      "step": 22084
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.889215350151062,
      "learning_rate": 6.508641344280101e-06,
      "loss": 0.8934,
      "step": 22085
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.706583857536316,
      "learning_rate": 6.508002427882312e-06,
      "loss": 0.9033,
      "step": 22086
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7355018854141235,
      "learning_rate": 6.507363511484522e-06,
      "loss": 0.7732,
      "step": 22087
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.0223703384399414,
      "learning_rate": 6.506724595086733e-06,
      "loss": 0.8899,
      "step": 22088
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.212310791015625,
      "learning_rate": 6.506085678688944e-06,
      "loss": 0.9335,
      "step": 22089
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5926591157913208,
      "learning_rate": 6.5054467622911545e-06,
      "loss": 0.8775,
      "step": 22090
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.603169322013855,
      "learning_rate": 6.504807845893365e-06,
      "loss": 0.868,
      "step": 22091
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.8224475383758545,
      "learning_rate": 6.504168929495576e-06,
      "loss": 1.2284,
      "step": 22092
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3948085308074951,
      "learning_rate": 6.5035300130977866e-06,
      "loss": 0.7993,
      "step": 22093
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1453049182891846,
      "learning_rate": 6.502891096699997e-06,
      "loss": 0.9782,
      "step": 22094
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9956953525543213,
      "learning_rate": 6.502252180302208e-06,
      "loss": 0.9249,
      "step": 22095
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7615113258361816,
      "learning_rate": 6.501613263904419e-06,
      "loss": 0.8306,
      "step": 22096
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.349552869796753,
      "learning_rate": 6.5009743475066285e-06,
      "loss": 0.9307,
      "step": 22097
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3147902488708496,
      "learning_rate": 6.500335431108839e-06,
      "loss": 0.9299,
      "step": 22098
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1110284328460693,
      "learning_rate": 6.49969651471105e-06,
      "loss": 0.8734,
      "step": 22099
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.115530967712402,
      "learning_rate": 6.499057598313261e-06,
      "loss": 1.0304,
      "step": 22100
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.213963270187378,
      "learning_rate": 6.498418681915471e-06,
      "loss": 0.526,
      "step": 22101
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.7747244834899902,
      "learning_rate": 6.497779765517682e-06,
      "loss": 0.9346,
      "step": 22102
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.9939703941345215,
      "learning_rate": 6.497140849119893e-06,
      "loss": 1.4486,
      "step": 22103
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8429279327392578,
      "learning_rate": 6.496501932722103e-06,
      "loss": 0.7131,
      "step": 22104
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2366349697113037,
      "learning_rate": 6.495863016324314e-06,
      "loss": 1.0229,
      "step": 22105
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.228814125061035,
      "learning_rate": 6.495224099926525e-06,
      "loss": 1.1717,
      "step": 22106
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6323566436767578,
      "learning_rate": 6.4945851835287355e-06,
      "loss": 0.8222,
      "step": 22107
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7399013042449951,
      "learning_rate": 6.493946267130946e-06,
      "loss": 0.9606,
      "step": 22108
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0136866569519043,
      "learning_rate": 6.493307350733157e-06,
      "loss": 0.9181,
      "step": 22109
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.462179183959961,
      "learning_rate": 6.492668434335368e-06,
      "loss": 0.9545,
      "step": 22110
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3511440753936768,
      "learning_rate": 6.492029517937578e-06,
      "loss": 0.8393,
      "step": 22111
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.18894362449646,
      "learning_rate": 6.491390601539789e-06,
      "loss": 0.8089,
      "step": 22112
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3884097337722778,
      "learning_rate": 6.490751685142e-06,
      "loss": 0.4623,
      "step": 22113
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5215160846710205,
      "learning_rate": 6.49011276874421e-06,
      "loss": 0.6232,
      "step": 22114
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5408610105514526,
      "learning_rate": 6.48947385234642e-06,
      "loss": 0.8517,
      "step": 22115
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.5912513732910156,
      "learning_rate": 6.488834935948631e-06,
      "loss": 1.0146,
      "step": 22116
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6152925491333008,
      "learning_rate": 6.488196019550842e-06,
      "loss": 0.8489,
      "step": 22117
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.783666968345642,
      "learning_rate": 6.487557103153052e-06,
      "loss": 0.9127,
      "step": 22118
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9345049858093262,
      "learning_rate": 6.486918186755263e-06,
      "loss": 0.8728,
      "step": 22119
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.496798038482666,
      "learning_rate": 6.486279270357474e-06,
      "loss": 1.2019,
      "step": 22120
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5664021968841553,
      "learning_rate": 6.4856403539596845e-06,
      "loss": 0.8046,
      "step": 22121
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.655230164527893,
      "learning_rate": 6.485001437561895e-06,
      "loss": 0.8978,
      "step": 22122
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.712090253829956,
      "learning_rate": 6.484362521164106e-06,
      "loss": 0.8667,
      "step": 22123
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8267288208007812,
      "learning_rate": 6.4837236047663166e-06,
      "loss": 0.8959,
      "step": 22124
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.595515489578247,
      "learning_rate": 6.483084688368527e-06,
      "loss": 0.8408,
      "step": 22125
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8215962648391724,
      "learning_rate": 6.482445771970738e-06,
      "loss": 0.7523,
      "step": 22126
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.7918734550476074,
      "learning_rate": 6.481806855572949e-06,
      "loss": 1.2934,
      "step": 22127
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.878171682357788,
      "learning_rate": 6.481167939175159e-06,
      "loss": 1.1474,
      "step": 22128
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9314333200454712,
      "learning_rate": 6.48052902277737e-06,
      "loss": 1.0138,
      "step": 22129
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.492436170578003,
      "learning_rate": 6.479890106379581e-06,
      "loss": 0.7629,
      "step": 22130
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.77890682220459,
      "learning_rate": 6.4792511899817915e-06,
      "loss": 0.6741,
      "step": 22131
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.456857681274414,
      "learning_rate": 6.478612273584002e-06,
      "loss": 0.9162,
      "step": 22132
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.896485686302185,
      "learning_rate": 6.477973357186213e-06,
      "loss": 1.1437,
      "step": 22133
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9230180978775024,
      "learning_rate": 6.477334440788423e-06,
      "loss": 0.9276,
      "step": 22134
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.306997776031494,
      "learning_rate": 6.476695524390633e-06,
      "loss": 1.4128,
      "step": 22135
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.803596019744873,
      "learning_rate": 6.476056607992844e-06,
      "loss": 0.8782,
      "step": 22136
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.69184410572052,
      "learning_rate": 6.475417691595055e-06,
      "loss": 0.8048,
      "step": 22137
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7665046453475952,
      "learning_rate": 6.4747787751972655e-06,
      "loss": 0.7967,
      "step": 22138
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9900518655776978,
      "learning_rate": 6.474139858799476e-06,
      "loss": 0.8888,
      "step": 22139
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6832077503204346,
      "learning_rate": 6.473500942401687e-06,
      "loss": 0.8511,
      "step": 22140
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8497620820999146,
      "learning_rate": 6.472862026003898e-06,
      "loss": 0.974,
      "step": 22141
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5049452781677246,
      "learning_rate": 6.472223109606108e-06,
      "loss": 0.8974,
      "step": 22142
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4331114292144775,
      "learning_rate": 6.471584193208319e-06,
      "loss": 0.9719,
      "step": 22143
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7259279489517212,
      "learning_rate": 6.47094527681053e-06,
      "loss": 0.8792,
      "step": 22144
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0350229740142822,
      "learning_rate": 6.47030636041274e-06,
      "loss": 0.8976,
      "step": 22145
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7488216161727905,
      "learning_rate": 6.469667444014951e-06,
      "loss": 0.9137,
      "step": 22146
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.287008762359619,
      "learning_rate": 6.469028527617162e-06,
      "loss": 0.8755,
      "step": 22147
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.781375765800476,
      "learning_rate": 6.4683896112193725e-06,
      "loss": 0.8892,
      "step": 22148
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9662082195281982,
      "learning_rate": 6.467750694821583e-06,
      "loss": 0.9603,
      "step": 22149
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0328307151794434,
      "learning_rate": 6.467111778423794e-06,
      "loss": 0.9308,
      "step": 22150
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.167396306991577,
      "learning_rate": 6.466472862026005e-06,
      "loss": 0.9787,
      "step": 22151
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.437602996826172,
      "learning_rate": 6.4658339456282145e-06,
      "loss": 0.8874,
      "step": 22152
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.559149980545044,
      "learning_rate": 6.465195029230425e-06,
      "loss": 0.856,
      "step": 22153
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.650132179260254,
      "learning_rate": 6.464556112832636e-06,
      "loss": 0.9266,
      "step": 22154
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8914988040924072,
      "learning_rate": 6.4639171964348466e-06,
      "loss": 0.8332,
      "step": 22155
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6181259155273438,
      "learning_rate": 6.463278280037057e-06,
      "loss": 0.802,
      "step": 22156
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6551793813705444,
      "learning_rate": 6.462639363639268e-06,
      "loss": 0.9379,
      "step": 22157
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9203459024429321,
      "learning_rate": 6.462000447241479e-06,
      "loss": 0.8275,
      "step": 22158
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.340122938156128,
      "learning_rate": 6.461361530843689e-06,
      "loss": 1.275,
      "step": 22159
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6643922328948975,
      "learning_rate": 6.4607226144459e-06,
      "loss": 0.8259,
      "step": 22160
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7221709489822388,
      "learning_rate": 6.460083698048111e-06,
      "loss": 0.8709,
      "step": 22161
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8393930196762085,
      "learning_rate": 6.4594447816503215e-06,
      "loss": 0.8558,
      "step": 22162
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8932569026947021,
      "learning_rate": 6.458805865252532e-06,
      "loss": 0.9249,
      "step": 22163
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6920539140701294,
      "learning_rate": 6.458166948854743e-06,
      "loss": 0.9112,
      "step": 22164
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0805914402008057,
      "learning_rate": 6.4575280324569536e-06,
      "loss": 0.7723,
      "step": 22165
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1589980125427246,
      "learning_rate": 6.456889116059164e-06,
      "loss": 0.845,
      "step": 22166
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8956663608551025,
      "learning_rate": 6.456250199661375e-06,
      "loss": 1.0022,
      "step": 22167
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7827080488204956,
      "learning_rate": 6.455611283263586e-06,
      "loss": 0.8462,
      "step": 22168
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8223655223846436,
      "learning_rate": 6.454972366865796e-06,
      "loss": 0.5269,
      "step": 22169
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.818152666091919,
      "learning_rate": 6.454333450468006e-06,
      "loss": 0.8146,
      "step": 22170
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1280081272125244,
      "learning_rate": 6.453694534070217e-06,
      "loss": 0.9406,
      "step": 22171
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9065465927124023,
      "learning_rate": 6.453055617672428e-06,
      "loss": 0.7014,
      "step": 22172
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7498654127120972,
      "learning_rate": 6.452416701274638e-06,
      "loss": 0.8031,
      "step": 22173
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3696303367614746,
      "learning_rate": 6.451777784876849e-06,
      "loss": 0.4486,
      "step": 22174
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.070924997329712,
      "learning_rate": 6.45113886847906e-06,
      "loss": 0.7841,
      "step": 22175
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6985974311828613,
      "learning_rate": 6.4504999520812704e-06,
      "loss": 0.5209,
      "step": 22176
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.585538387298584,
      "learning_rate": 6.449861035683481e-06,
      "loss": 0.8097,
      "step": 22177
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0338926315307617,
      "learning_rate": 6.449222119285692e-06,
      "loss": 0.8777,
      "step": 22178
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7344403266906738,
      "learning_rate": 6.4485832028879025e-06,
      "loss": 0.7618,
      "step": 22179
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2638572454452515,
      "learning_rate": 6.447944286490113e-06,
      "loss": 0.4269,
      "step": 22180
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5094630718231201,
      "learning_rate": 6.447305370092324e-06,
      "loss": 0.7766,
      "step": 22181
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8137495517730713,
      "learning_rate": 6.446666453694535e-06,
      "loss": 0.8365,
      "step": 22182
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8004060983657837,
      "learning_rate": 6.446027537296745e-06,
      "loss": 1.146,
      "step": 22183
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.80838680267334,
      "learning_rate": 6.445388620898956e-06,
      "loss": 0.9309,
      "step": 22184
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5151000022888184,
      "learning_rate": 6.444749704501167e-06,
      "loss": 0.7302,
      "step": 22185
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4271236658096313,
      "learning_rate": 6.4441107881033774e-06,
      "loss": 0.521,
      "step": 22186
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2970664501190186,
      "learning_rate": 6.443471871705588e-06,
      "loss": 0.8956,
      "step": 22187
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.939873218536377,
      "learning_rate": 6.442832955307798e-06,
      "loss": 0.9336,
      "step": 22188
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.547585964202881,
      "learning_rate": 6.442194038910009e-06,
      "loss": 0.9903,
      "step": 22189
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.9863152503967285,
      "learning_rate": 6.441555122512219e-06,
      "loss": 1.4142,
      "step": 22190
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6256861686706543,
      "learning_rate": 6.44091620611443e-06,
      "loss": 0.8275,
      "step": 22191
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.415363311767578,
      "learning_rate": 6.440277289716641e-06,
      "loss": 0.8231,
      "step": 22192
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8338497877120972,
      "learning_rate": 6.4396383733188515e-06,
      "loss": 0.7896,
      "step": 22193
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.300137758255005,
      "learning_rate": 6.438999456921062e-06,
      "loss": 0.8827,
      "step": 22194
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.002486228942871,
      "learning_rate": 6.438360540523273e-06,
      "loss": 0.8048,
      "step": 22195
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.898289442062378,
      "learning_rate": 6.4377216241254836e-06,
      "loss": 0.7879,
      "step": 22196
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6902679204940796,
      "learning_rate": 6.437082707727694e-06,
      "loss": 0.6946,
      "step": 22197
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.812718391418457,
      "learning_rate": 6.436443791329905e-06,
      "loss": 0.7846,
      "step": 22198
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.9613993167877197,
      "learning_rate": 6.435804874932116e-06,
      "loss": 0.7312,
      "step": 22199
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.27364444732666,
      "learning_rate": 6.435165958534326e-06,
      "loss": 1.1541,
      "step": 22200
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7063966989517212,
      "learning_rate": 6.434527042136537e-06,
      "loss": 0.9446,
      "step": 22201
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0582878589630127,
      "learning_rate": 6.433888125738748e-06,
      "loss": 0.8154,
      "step": 22202
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7804211378097534,
      "learning_rate": 6.4332492093409585e-06,
      "loss": 0.8683,
      "step": 22203
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.821420669555664,
      "learning_rate": 6.432610292943169e-06,
      "loss": 0.9155,
      "step": 22204
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8594006299972534,
      "learning_rate": 6.43197137654538e-06,
      "loss": 0.9353,
      "step": 22205
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.440876841545105,
      "learning_rate": 6.4313324601475906e-06,
      "loss": 0.8251,
      "step": 22206
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.8291852474212646,
      "learning_rate": 6.4306935437498004e-06,
      "loss": 0.7965,
      "step": 22207
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2506778240203857,
      "learning_rate": 6.430054627352011e-06,
      "loss": 0.8745,
      "step": 22208
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.3275980949401855,
      "learning_rate": 6.429415710954222e-06,
      "loss": 1.1869,
      "step": 22209
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6714410781860352,
      "learning_rate": 6.4287767945564325e-06,
      "loss": 0.9227,
      "step": 22210
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2289397716522217,
      "learning_rate": 6.428137878158643e-06,
      "loss": 0.9029,
      "step": 22211
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7457332611083984,
      "learning_rate": 6.427498961760854e-06,
      "loss": 0.8659,
      "step": 22212
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4627268314361572,
      "learning_rate": 6.426860045363065e-06,
      "loss": 1.3195,
      "step": 22213
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9838054180145264,
      "learning_rate": 6.426221128965275e-06,
      "loss": 0.8007,
      "step": 22214
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.184436559677124,
      "learning_rate": 6.425582212567486e-06,
      "loss": 0.9244,
      "step": 22215
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7872511148452759,
      "learning_rate": 6.424943296169697e-06,
      "loss": 0.7833,
      "step": 22216
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9960269927978516,
      "learning_rate": 6.4243043797719074e-06,
      "loss": 0.6592,
      "step": 22217
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6987594366073608,
      "learning_rate": 6.423665463374118e-06,
      "loss": 0.8764,
      "step": 22218
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7269680500030518,
      "learning_rate": 6.423026546976329e-06,
      "loss": 0.8424,
      "step": 22219
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7220430374145508,
      "learning_rate": 6.4223876305785395e-06,
      "loss": 1.0973,
      "step": 22220
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.9425253868103027,
      "learning_rate": 6.42174871418075e-06,
      "loss": 1.0103,
      "step": 22221
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0073225498199463,
      "learning_rate": 6.421109797782961e-06,
      "loss": 0.8992,
      "step": 22222
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.851578712463379,
      "learning_rate": 6.420470881385172e-06,
      "loss": 0.9717,
      "step": 22223
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.952863335609436,
      "learning_rate": 6.419831964987382e-06,
      "loss": 0.8245,
      "step": 22224
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5977163314819336,
      "learning_rate": 6.419193048589592e-06,
      "loss": 0.4117,
      "step": 22225
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.924845814704895,
      "learning_rate": 6.418554132191803e-06,
      "loss": 0.8594,
      "step": 22226
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6299656629562378,
      "learning_rate": 6.417915215794014e-06,
      "loss": 0.8132,
      "step": 22227
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.790084719657898,
      "learning_rate": 6.417276299396224e-06,
      "loss": 1.1708,
      "step": 22228
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0190036296844482,
      "learning_rate": 6.416637382998435e-06,
      "loss": 0.7922,
      "step": 22229
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.9546003341674805,
      "learning_rate": 6.415998466600646e-06,
      "loss": 1.2256,
      "step": 22230
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.170367240905762,
      "learning_rate": 6.415359550202856e-06,
      "loss": 1.2249,
      "step": 22231
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7930364608764648,
      "learning_rate": 6.414720633805067e-06,
      "loss": 0.8702,
      "step": 22232
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4385135173797607,
      "learning_rate": 6.414081717407278e-06,
      "loss": 0.832,
      "step": 22233
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9040882587432861,
      "learning_rate": 6.4134428010094885e-06,
      "loss": 1.0116,
      "step": 22234
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8278357982635498,
      "learning_rate": 6.412803884611699e-06,
      "loss": 0.8028,
      "step": 22235
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.707521677017212,
      "learning_rate": 6.41216496821391e-06,
      "loss": 0.8914,
      "step": 22236
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3392534255981445,
      "learning_rate": 6.411526051816121e-06,
      "loss": 0.809,
      "step": 22237
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.726271629333496,
      "learning_rate": 6.410887135418331e-06,
      "loss": 0.8751,
      "step": 22238
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.469022274017334,
      "learning_rate": 6.410248219020542e-06,
      "loss": 0.7788,
      "step": 22239
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5161523818969727,
      "learning_rate": 6.409609302622753e-06,
      "loss": 0.9799,
      "step": 22240
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4025774002075195,
      "learning_rate": 6.408970386224963e-06,
      "loss": 0.829,
      "step": 22241
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5995962619781494,
      "learning_rate": 6.408331469827174e-06,
      "loss": 0.8362,
      "step": 22242
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.7439823150634766,
      "learning_rate": 6.407692553429384e-06,
      "loss": 0.737,
      "step": 22243
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9973087310791016,
      "learning_rate": 6.407053637031595e-06,
      "loss": 0.968,
      "step": 22244
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7126315832138062,
      "learning_rate": 6.406414720633805e-06,
      "loss": 0.8141,
      "step": 22245
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6481620073318481,
      "learning_rate": 6.405775804236016e-06,
      "loss": 0.8177,
      "step": 22246
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.671473503112793,
      "learning_rate": 6.405136887838227e-06,
      "loss": 1.1366,
      "step": 22247
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6087549924850464,
      "learning_rate": 6.4044979714404374e-06,
      "loss": 0.8677,
      "step": 22248
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4539990425109863,
      "learning_rate": 6.403859055042648e-06,
      "loss": 0.9341,
      "step": 22249
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.764719843864441,
      "learning_rate": 6.403220138644859e-06,
      "loss": 0.9661,
      "step": 22250
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9240479469299316,
      "learning_rate": 6.4025812222470695e-06,
      "loss": 0.9434,
      "step": 22251
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.645986557006836,
      "learning_rate": 6.40194230584928e-06,
      "loss": 0.8889,
      "step": 22252
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.339682102203369,
      "learning_rate": 6.401303389451491e-06,
      "loss": 0.8366,
      "step": 22253
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.797585964202881,
      "learning_rate": 6.400664473053702e-06,
      "loss": 1.1837,
      "step": 22254
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.250868320465088,
      "learning_rate": 6.400025556655912e-06,
      "loss": 1.0063,
      "step": 22255
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7883340120315552,
      "learning_rate": 6.399386640258123e-06,
      "loss": 0.9218,
      "step": 22256
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7814298868179321,
      "learning_rate": 6.398747723860334e-06,
      "loss": 0.7869,
      "step": 22257
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.9414212703704834,
      "learning_rate": 6.3981088074625444e-06,
      "loss": 0.9476,
      "step": 22258
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.848677635192871,
      "learning_rate": 6.397469891064755e-06,
      "loss": 0.8296,
      "step": 22259
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.588490962982178,
      "learning_rate": 6.396830974666966e-06,
      "loss": 1.3404,
      "step": 22260
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.942992687225342,
      "learning_rate": 6.396192058269176e-06,
      "loss": 1.0037,
      "step": 22261
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6465930938720703,
      "learning_rate": 6.395553141871386e-06,
      "loss": 1.2127,
      "step": 22262
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8055087327957153,
      "learning_rate": 6.394914225473597e-06,
      "loss": 0.8715,
      "step": 22263
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1842894554138184,
      "learning_rate": 6.394275309075808e-06,
      "loss": 0.5194,
      "step": 22264
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7157022953033447,
      "learning_rate": 6.3936363926780185e-06,
      "loss": 0.7859,
      "step": 22265
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7885807752609253,
      "learning_rate": 6.392997476280229e-06,
      "loss": 0.861,
      "step": 22266
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0891594886779785,
      "learning_rate": 6.39235855988244e-06,
      "loss": 0.9445,
      "step": 22267
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5301789045333862,
      "learning_rate": 6.391719643484651e-06,
      "loss": 0.83,
      "step": 22268
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2276721000671387,
      "learning_rate": 6.391080727086861e-06,
      "loss": 0.6919,
      "step": 22269
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5868475437164307,
      "learning_rate": 6.390441810689072e-06,
      "loss": 0.7734,
      "step": 22270
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9101603031158447,
      "learning_rate": 6.389802894291283e-06,
      "loss": 0.8765,
      "step": 22271
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6936450004577637,
      "learning_rate": 6.389163977893493e-06,
      "loss": 0.9049,
      "step": 22272
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8196632862091064,
      "learning_rate": 6.388525061495704e-06,
      "loss": 0.83,
      "step": 22273
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5676392316818237,
      "learning_rate": 6.387886145097915e-06,
      "loss": 0.8389,
      "step": 22274
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.762821078300476,
      "learning_rate": 6.3872472287001255e-06,
      "loss": 0.9386,
      "step": 22275
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5393551588058472,
      "learning_rate": 6.386608312302336e-06,
      "loss": 0.3334,
      "step": 22276
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4706480503082275,
      "learning_rate": 6.385969395904547e-06,
      "loss": 0.7691,
      "step": 22277
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.912287712097168,
      "learning_rate": 6.385330479506758e-06,
      "loss": 0.8605,
      "step": 22278
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.116887092590332,
      "learning_rate": 6.384691563108968e-06,
      "loss": 0.991,
      "step": 22279
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.6345198154449463,
      "learning_rate": 6.384052646711178e-06,
      "loss": 1.2588,
      "step": 22280
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7803153991699219,
      "learning_rate": 6.383413730313389e-06,
      "loss": 0.9426,
      "step": 22281
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.03232479095459,
      "learning_rate": 6.3827748139155995e-06,
      "loss": 1.313,
      "step": 22282
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6958531141281128,
      "learning_rate": 6.38213589751781e-06,
      "loss": 0.8619,
      "step": 22283
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5471560955047607,
      "learning_rate": 6.381496981120021e-06,
      "loss": 0.8269,
      "step": 22284
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.2627170085906982,
      "learning_rate": 6.380858064722232e-06,
      "loss": 0.9243,
      "step": 22285
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8526902198791504,
      "learning_rate": 6.380219148324442e-06,
      "loss": 0.7916,
      "step": 22286
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.109225273132324,
      "learning_rate": 6.379580231926653e-06,
      "loss": 0.8676,
      "step": 22287
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.150120735168457,
      "learning_rate": 6.378941315528864e-06,
      "loss": 0.808,
      "step": 22288
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.829665184020996,
      "learning_rate": 6.3783023991310744e-06,
      "loss": 0.6075,
      "step": 22289
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.936954379081726,
      "learning_rate": 6.377663482733285e-06,
      "loss": 0.9307,
      "step": 22290
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.182039976119995,
      "learning_rate": 6.377024566335496e-06,
      "loss": 1.0641,
      "step": 22291
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.7139840126037598,
      "learning_rate": 6.3763856499377065e-06,
      "loss": 0.6148,
      "step": 22292
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2886428833007812,
      "learning_rate": 6.375746733539917e-06,
      "loss": 0.9704,
      "step": 22293
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.884452819824219,
      "learning_rate": 6.375107817142128e-06,
      "loss": 0.942,
      "step": 22294
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0214016437530518,
      "learning_rate": 6.374468900744339e-06,
      "loss": 0.8893,
      "step": 22295
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.198021650314331,
      "learning_rate": 6.373829984346549e-06,
      "loss": 0.9028,
      "step": 22296
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5675740242004395,
      "learning_rate": 6.37319106794876e-06,
      "loss": 0.8439,
      "step": 22297
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5647205114364624,
      "learning_rate": 6.37255215155097e-06,
      "loss": 0.8639,
      "step": 22298
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5685758590698242,
      "learning_rate": 6.371913235153181e-06,
      "loss": 0.874,
      "step": 22299
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.8695006370544434,
      "learning_rate": 6.371274318755391e-06,
      "loss": 0.872,
      "step": 22300
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6107559204101562,
      "learning_rate": 6.370635402357602e-06,
      "loss": 0.7615,
      "step": 22301
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.222419261932373,
      "learning_rate": 6.369996485959813e-06,
      "loss": 0.7905,
      "step": 22302
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0894854068756104,
      "learning_rate": 6.369357569562023e-06,
      "loss": 1.1566,
      "step": 22303
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.523206114768982,
      "learning_rate": 6.368718653164234e-06,
      "loss": 0.8583,
      "step": 22304
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.108499765396118,
      "learning_rate": 6.368079736766445e-06,
      "loss": 0.8764,
      "step": 22305
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.100450038909912,
      "learning_rate": 6.3674408203686555e-06,
      "loss": 0.8858,
      "step": 22306
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5648932456970215,
      "learning_rate": 6.366801903970866e-06,
      "loss": 0.7715,
      "step": 22307
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.017010450363159,
      "learning_rate": 6.366162987573077e-06,
      "loss": 0.8101,
      "step": 22308
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7248868942260742,
      "learning_rate": 6.365524071175288e-06,
      "loss": 0.8633,
      "step": 22309
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6313527822494507,
      "learning_rate": 6.364885154777498e-06,
      "loss": 0.8181,
      "step": 22310
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5521471500396729,
      "learning_rate": 6.364246238379709e-06,
      "loss": 0.8454,
      "step": 22311
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.028989315032959,
      "learning_rate": 6.36360732198192e-06,
      "loss": 1.2021,
      "step": 22312
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9061633348464966,
      "learning_rate": 6.36296840558413e-06,
      "loss": 0.7601,
      "step": 22313
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.997318983078003,
      "learning_rate": 6.362329489186341e-06,
      "loss": 1.1839,
      "step": 22314
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3268933296203613,
      "learning_rate": 6.361690572788552e-06,
      "loss": 0.75,
      "step": 22315
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.137148380279541,
      "learning_rate": 6.361051656390762e-06,
      "loss": 1.2508,
      "step": 22316
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.599738359451294,
      "learning_rate": 6.360412739992972e-06,
      "loss": 0.8752,
      "step": 22317
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5336382389068604,
      "learning_rate": 6.359773823595183e-06,
      "loss": 0.7722,
      "step": 22318
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6635690927505493,
      "learning_rate": 6.359134907197394e-06,
      "loss": 0.8332,
      "step": 22319
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0742807388305664,
      "learning_rate": 6.3584959907996044e-06,
      "loss": 0.9374,
      "step": 22320
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9552801847457886,
      "learning_rate": 6.357857074401815e-06,
      "loss": 1.1812,
      "step": 22321
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2718942165374756,
      "learning_rate": 6.357218158004026e-06,
      "loss": 1.0176,
      "step": 22322
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4999568462371826,
      "learning_rate": 6.3565792416062365e-06,
      "loss": 0.8791,
      "step": 22323
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5995779037475586,
      "learning_rate": 6.355940325208447e-06,
      "loss": 0.7673,
      "step": 22324
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.105567932128906,
      "learning_rate": 6.355301408810658e-06,
      "loss": 1.8178,
      "step": 22325
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.8414268493652344,
      "learning_rate": 6.354662492412869e-06,
      "loss": 0.7534,
      "step": 22326
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1634230613708496,
      "learning_rate": 6.354023576015079e-06,
      "loss": 0.864,
      "step": 22327
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6513515710830688,
      "learning_rate": 6.35338465961729e-06,
      "loss": 1.0714,
      "step": 22328
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.638889312744141,
      "learning_rate": 6.352745743219501e-06,
      "loss": 0.8269,
      "step": 22329
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3633859157562256,
      "learning_rate": 6.3521068268217114e-06,
      "loss": 0.817,
      "step": 22330
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.715214729309082,
      "learning_rate": 6.351467910423922e-06,
      "loss": 0.8942,
      "step": 22331
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.8612897396087646,
      "learning_rate": 6.350828994026133e-06,
      "loss": 1.0059,
      "step": 22332
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.905255913734436,
      "learning_rate": 6.3501900776283435e-06,
      "loss": 0.9663,
      "step": 22333
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.5663235187530518,
      "learning_rate": 6.349551161230553e-06,
      "loss": 1.1311,
      "step": 22334
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.263573408126831,
      "learning_rate": 6.348912244832764e-06,
      "loss": 0.5294,
      "step": 22335
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.748109221458435,
      "learning_rate": 6.348273328434975e-06,
      "loss": 0.8173,
      "step": 22336
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.4329802989959717,
      "learning_rate": 6.3476344120371855e-06,
      "loss": 1.3295,
      "step": 22337
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2198870182037354,
      "learning_rate": 6.346995495639396e-06,
      "loss": 0.8761,
      "step": 22338
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.797773003578186,
      "learning_rate": 6.346356579241607e-06,
      "loss": 0.783,
      "step": 22339
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6544744968414307,
      "learning_rate": 6.345717662843818e-06,
      "loss": 0.8582,
      "step": 22340
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.010622262954712,
      "learning_rate": 6.345078746446028e-06,
      "loss": 0.8453,
      "step": 22341
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6665276288986206,
      "learning_rate": 6.344439830048239e-06,
      "loss": 0.4712,
      "step": 22342
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.628373384475708,
      "learning_rate": 6.34380091365045e-06,
      "loss": 0.8326,
      "step": 22343
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.103313446044922,
      "learning_rate": 6.34316199725266e-06,
      "loss": 0.9455,
      "step": 22344
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.141390800476074,
      "learning_rate": 6.342523080854871e-06,
      "loss": 0.9377,
      "step": 22345
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3376357555389404,
      "learning_rate": 6.341884164457082e-06,
      "loss": 0.7771,
      "step": 22346
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.752997636795044,
      "learning_rate": 6.3412452480592925e-06,
      "loss": 0.884,
      "step": 22347
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9711296558380127,
      "learning_rate": 6.340606331661503e-06,
      "loss": 0.8089,
      "step": 22348
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.850959300994873,
      "learning_rate": 6.339967415263714e-06,
      "loss": 0.9645,
      "step": 22349
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2462685108184814,
      "learning_rate": 6.339328498865925e-06,
      "loss": 0.9525,
      "step": 22350
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.117150068283081,
      "learning_rate": 6.338689582468135e-06,
      "loss": 1.2395,
      "step": 22351
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.9660987854003906,
      "learning_rate": 6.338050666070346e-06,
      "loss": 0.824,
      "step": 22352
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9156711101531982,
      "learning_rate": 6.337411749672556e-06,
      "loss": 0.8458,
      "step": 22353
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5661938190460205,
      "learning_rate": 6.3367728332747665e-06,
      "loss": 0.8869,
      "step": 22354
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1428520679473877,
      "learning_rate": 6.336133916876977e-06,
      "loss": 0.8701,
      "step": 22355
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7643646001815796,
      "learning_rate": 6.335495000479188e-06,
      "loss": 0.7867,
      "step": 22356
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.756767988204956,
      "learning_rate": 6.334856084081399e-06,
      "loss": 1.1095,
      "step": 22357
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0557167530059814,
      "learning_rate": 6.334217167683609e-06,
      "loss": 1.0164,
      "step": 22358
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3055706024169922,
      "learning_rate": 6.33357825128582e-06,
      "loss": 0.7575,
      "step": 22359
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7130602598190308,
      "learning_rate": 6.332939334888031e-06,
      "loss": 0.7469,
      "step": 22360
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5462207794189453,
      "learning_rate": 6.3323004184902414e-06,
      "loss": 0.8336,
      "step": 22361
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8082919120788574,
      "learning_rate": 6.331661502092452e-06,
      "loss": 0.9075,
      "step": 22362
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5700520277023315,
      "learning_rate": 6.331022585694663e-06,
      "loss": 0.8472,
      "step": 22363
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6244356632232666,
      "learning_rate": 6.3303836692968735e-06,
      "loss": 1.011,
      "step": 22364
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.001311302185059,
      "learning_rate": 6.329744752899084e-06,
      "loss": 1.3202,
      "step": 22365
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1092123985290527,
      "learning_rate": 6.329105836501295e-06,
      "loss": 1.0745,
      "step": 22366
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8137681484222412,
      "learning_rate": 6.328466920103506e-06,
      "loss": 0.8586,
      "step": 22367
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3068132400512695,
      "learning_rate": 6.327828003705716e-06,
      "loss": 0.711,
      "step": 22368
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.749798774719238,
      "learning_rate": 6.327189087307927e-06,
      "loss": 1.6055,
      "step": 22369
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0024328231811523,
      "learning_rate": 6.326550170910138e-06,
      "loss": 0.76,
      "step": 22370
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.304457187652588,
      "learning_rate": 6.325911254512348e-06,
      "loss": 1.6765,
      "step": 22371
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.413846015930176,
      "learning_rate": 6.325272338114558e-06,
      "loss": 0.9677,
      "step": 22372
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.960605263710022,
      "learning_rate": 6.324633421716769e-06,
      "loss": 0.8089,
      "step": 22373
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.098564863204956,
      "learning_rate": 6.32399450531898e-06,
      "loss": 1.1279,
      "step": 22374
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2560572624206543,
      "learning_rate": 6.32335558892119e-06,
      "loss": 0.9584,
      "step": 22375
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.804349422454834,
      "learning_rate": 6.322716672523401e-06,
      "loss": 0.7966,
      "step": 22376
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0304055213928223,
      "learning_rate": 6.322077756125612e-06,
      "loss": 1.1801,
      "step": 22377
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5569432973861694,
      "learning_rate": 6.3214388397278225e-06,
      "loss": 0.8049,
      "step": 22378
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.801705241203308,
      "learning_rate": 6.320799923330033e-06,
      "loss": 0.8828,
      "step": 22379
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.220531463623047,
      "learning_rate": 6.320161006932244e-06,
      "loss": 0.7952,
      "step": 22380
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.368190288543701,
      "learning_rate": 6.319522090534455e-06,
      "loss": 0.7758,
      "step": 22381
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9915122985839844,
      "learning_rate": 6.318883174136665e-06,
      "loss": 0.9709,
      "step": 22382
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.309701442718506,
      "learning_rate": 6.318244257738876e-06,
      "loss": 1.0378,
      "step": 22383
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.739022970199585,
      "learning_rate": 6.317605341341087e-06,
      "loss": 0.2863,
      "step": 22384
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8857064247131348,
      "learning_rate": 6.316966424943297e-06,
      "loss": 0.8825,
      "step": 22385
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0561230182647705,
      "learning_rate": 6.316327508545508e-06,
      "loss": 0.8501,
      "step": 22386
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.73268985748291,
      "learning_rate": 6.315688592147719e-06,
      "loss": 0.9878,
      "step": 22387
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.623164176940918,
      "learning_rate": 6.3150496757499295e-06,
      "loss": 0.6663,
      "step": 22388
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.882206439971924,
      "learning_rate": 6.314410759352139e-06,
      "loss": 0.8914,
      "step": 22389
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7089545726776123,
      "learning_rate": 6.31377184295435e-06,
      "loss": 0.9599,
      "step": 22390
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4052422046661377,
      "learning_rate": 6.313132926556561e-06,
      "loss": 0.4604,
      "step": 22391
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0223097801208496,
      "learning_rate": 6.3124940101587715e-06,
      "loss": 0.9798,
      "step": 22392
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8543875217437744,
      "learning_rate": 6.311855093760982e-06,
      "loss": 0.901,
      "step": 22393
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.157107353210449,
      "learning_rate": 6.311216177363193e-06,
      "loss": 0.9796,
      "step": 22394
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0081703662872314,
      "learning_rate": 6.3105772609654036e-06,
      "loss": 0.6965,
      "step": 22395
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.23689341545105,
      "learning_rate": 6.309938344567614e-06,
      "loss": 0.8627,
      "step": 22396
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9196065664291382,
      "learning_rate": 6.309299428169825e-06,
      "loss": 1.1326,
      "step": 22397
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0423851013183594,
      "learning_rate": 6.308660511772036e-06,
      "loss": 0.8103,
      "step": 22398
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3904025554656982,
      "learning_rate": 6.308021595374246e-06,
      "loss": 0.8655,
      "step": 22399
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0376038551330566,
      "learning_rate": 6.307382678976457e-06,
      "loss": 0.6786,
      "step": 22400
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8222757577896118,
      "learning_rate": 6.306743762578668e-06,
      "loss": 0.8878,
      "step": 22401
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8543574810028076,
      "learning_rate": 6.3061048461808785e-06,
      "loss": 0.9413,
      "step": 22402
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8811894655227661,
      "learning_rate": 6.305465929783089e-06,
      "loss": 0.9214,
      "step": 22403
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.882808208465576,
      "learning_rate": 6.3048270133853e-06,
      "loss": 1.2595,
      "step": 22404
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.762835144996643,
      "learning_rate": 6.3041880969875106e-06,
      "loss": 0.827,
      "step": 22405
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1266674995422363,
      "learning_rate": 6.303549180589721e-06,
      "loss": 0.9013,
      "step": 22406
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.282029628753662,
      "learning_rate": 6.302910264191932e-06,
      "loss": 1.2122,
      "step": 22407
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.2233684062957764,
      "learning_rate": 6.302271347794142e-06,
      "loss": 1.1526,
      "step": 22408
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4122182130813599,
      "learning_rate": 6.3016324313963525e-06,
      "loss": 0.3528,
      "step": 22409
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.680109977722168,
      "learning_rate": 6.300993514998563e-06,
      "loss": 0.9143,
      "step": 22410
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.3166921138763428,
      "learning_rate": 6.300354598600774e-06,
      "loss": 1.0158,
      "step": 22411
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5189125537872314,
      "learning_rate": 6.299715682202985e-06,
      "loss": 0.8555,
      "step": 22412
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9536106586456299,
      "learning_rate": 6.299076765805195e-06,
      "loss": 0.8783,
      "step": 22413
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6830992698669434,
      "learning_rate": 6.298437849407406e-06,
      "loss": 0.9176,
      "step": 22414
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7713769674301147,
      "learning_rate": 6.297798933009617e-06,
      "loss": 0.8555,
      "step": 22415
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3441718816757202,
      "learning_rate": 6.297160016611827e-06,
      "loss": 0.3524,
      "step": 22416
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7923778295516968,
      "learning_rate": 6.296521100214038e-06,
      "loss": 0.8547,
      "step": 22417
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0609729290008545,
      "learning_rate": 6.295882183816248e-06,
      "loss": 0.8669,
      "step": 22418
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7595852613449097,
      "learning_rate": 6.295243267418459e-06,
      "loss": 0.8159,
      "step": 22419
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.572816252708435,
      "learning_rate": 6.2946043510206685e-06,
      "loss": 0.7762,
      "step": 22420
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.388868808746338,
      "learning_rate": 6.293965434622879e-06,
      "loss": 1.8897,
      "step": 22421
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7714096307754517,
      "learning_rate": 6.29332651822509e-06,
      "loss": 0.7924,
      "step": 22422
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.677846670150757,
      "learning_rate": 6.292687601827301e-06,
      "loss": 1.297,
      "step": 22423
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5824912786483765,
      "learning_rate": 6.292048685429511e-06,
      "loss": 0.8134,
      "step": 22424
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.208798408508301,
      "learning_rate": 6.291409769031722e-06,
      "loss": 1.0254,
      "step": 22425
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5755540132522583,
      "learning_rate": 6.290770852633933e-06,
      "loss": 0.3341,
      "step": 22426
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.840770721435547,
      "learning_rate": 6.290131936236143e-06,
      "loss": 1.0075,
      "step": 22427
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6363556385040283,
      "learning_rate": 6.289493019838354e-06,
      "loss": 0.7695,
      "step": 22428
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8407566547393799,
      "learning_rate": 6.288854103440565e-06,
      "loss": 0.8981,
      "step": 22429
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.921779632568359,
      "learning_rate": 6.2882151870427755e-06,
      "loss": 1.8419,
      "step": 22430
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.70073664188385,
      "learning_rate": 6.287576270644986e-06,
      "loss": 0.8795,
      "step": 22431
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9082412719726562,
      "learning_rate": 6.286937354247197e-06,
      "loss": 0.6432,
      "step": 22432
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0929012298583984,
      "learning_rate": 6.286298437849408e-06,
      "loss": 0.8821,
      "step": 22433
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8682512044906616,
      "learning_rate": 6.285659521451618e-06,
      "loss": 0.8788,
      "step": 22434
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.243727445602417,
      "learning_rate": 6.285020605053829e-06,
      "loss": 1.204,
      "step": 22435
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4845331907272339,
      "learning_rate": 6.28438168865604e-06,
      "loss": 0.6914,
      "step": 22436
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.998936414718628,
      "learning_rate": 6.28374277225825e-06,
      "loss": 0.8331,
      "step": 22437
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1212515830993652,
      "learning_rate": 6.28310385586046e-06,
      "loss": 1.1689,
      "step": 22438
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.028916120529175,
      "learning_rate": 6.282464939462671e-06,
      "loss": 0.827,
      "step": 22439
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.222472906112671,
      "learning_rate": 6.281826023064882e-06,
      "loss": 0.8565,
      "step": 22440
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9912660121917725,
      "learning_rate": 6.281187106667092e-06,
      "loss": 1.0014,
      "step": 22441
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6661614179611206,
      "learning_rate": 6.280548190269303e-06,
      "loss": 0.8591,
      "step": 22442
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8496756553649902,
      "learning_rate": 6.279909273871514e-06,
      "loss": 0.8511,
      "step": 22443
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.358794689178467,
      "learning_rate": 6.2792703574737245e-06,
      "loss": 0.9704,
      "step": 22444
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8115525245666504,
      "learning_rate": 6.278631441075935e-06,
      "loss": 0.8283,
      "step": 22445
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.624300956726074,
      "learning_rate": 6.277992524678146e-06,
      "loss": 1.6253,
      "step": 22446
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.864927887916565,
      "learning_rate": 6.2773536082803566e-06,
      "loss": 0.8495,
      "step": 22447
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.05167293548584,
      "learning_rate": 6.276714691882567e-06,
      "loss": 0.8353,
      "step": 22448
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.370719075202942,
      "learning_rate": 6.276075775484778e-06,
      "loss": 0.4326,
      "step": 22449
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.575594186782837,
      "learning_rate": 6.275436859086989e-06,
      "loss": 1.1889,
      "step": 22450
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7027068138122559,
      "learning_rate": 6.274797942689199e-06,
      "loss": 0.8759,
      "step": 22451
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.854940176010132,
      "learning_rate": 6.27415902629141e-06,
      "loss": 0.928,
      "step": 22452
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2126383781433105,
      "learning_rate": 6.273520109893621e-06,
      "loss": 0.7025,
      "step": 22453
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9232690334320068,
      "learning_rate": 6.2728811934958315e-06,
      "loss": 0.7938,
      "step": 22454
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.1427009105682373,
      "learning_rate": 6.272242277098042e-06,
      "loss": 1.2112,
      "step": 22455
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.5048041343688965,
      "learning_rate": 6.271603360700253e-06,
      "loss": 0.9013,
      "step": 22456
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5447514057159424,
      "learning_rate": 6.270964444302463e-06,
      "loss": 0.9593,
      "step": 22457
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.69413423538208,
      "learning_rate": 6.270325527904673e-06,
      "loss": 0.7885,
      "step": 22458
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.783274173736572,
      "learning_rate": 6.269686611506884e-06,
      "loss": 1.2835,
      "step": 22459
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9225130081176758,
      "learning_rate": 6.269047695109095e-06,
      "loss": 1.2287,
      "step": 22460
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.165517568588257,
      "learning_rate": 6.2684087787113055e-06,
      "loss": 0.9559,
      "step": 22461
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7773902416229248,
      "learning_rate": 6.267769862313516e-06,
      "loss": 0.7742,
      "step": 22462
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1728086471557617,
      "learning_rate": 6.267130945915727e-06,
      "loss": 0.77,
      "step": 22463
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.285646677017212,
      "learning_rate": 6.266492029517938e-06,
      "loss": 1.0329,
      "step": 22464
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.702952265739441,
      "learning_rate": 6.265853113120148e-06,
      "loss": 0.9059,
      "step": 22465
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0503792762756348,
      "learning_rate": 6.265214196722359e-06,
      "loss": 0.929,
      "step": 22466
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5593351125717163,
      "learning_rate": 6.26457528032457e-06,
      "loss": 0.8527,
      "step": 22467
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8278378248214722,
      "learning_rate": 6.26393636392678e-06,
      "loss": 0.8362,
      "step": 22468
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.786917805671692,
      "learning_rate": 6.263297447528991e-06,
      "loss": 0.8075,
      "step": 22469
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6749277114868164,
      "learning_rate": 6.262658531131202e-06,
      "loss": 0.8519,
      "step": 22470
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8381298780441284,
      "learning_rate": 6.2620196147334125e-06,
      "loss": 0.9842,
      "step": 22471
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.282362699508667,
      "learning_rate": 6.261380698335623e-06,
      "loss": 0.7031,
      "step": 22472
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.053890228271484,
      "learning_rate": 6.260741781937834e-06,
      "loss": 1.2311,
      "step": 22473
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3507654666900635,
      "learning_rate": 6.260102865540045e-06,
      "loss": 0.81,
      "step": 22474
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5850119590759277,
      "learning_rate": 6.2594639491422545e-06,
      "loss": 0.8349,
      "step": 22475
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.361177444458008,
      "learning_rate": 6.258825032744465e-06,
      "loss": 2.0036,
      "step": 22476
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3079850673675537,
      "learning_rate": 6.258186116346676e-06,
      "loss": 0.9128,
      "step": 22477
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6262627840042114,
      "learning_rate": 6.2575471999488866e-06,
      "loss": 0.8223,
      "step": 22478
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0516374111175537,
      "learning_rate": 6.256908283551097e-06,
      "loss": 0.8253,
      "step": 22479
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7506358623504639,
      "learning_rate": 6.256269367153308e-06,
      "loss": 0.8358,
      "step": 22480
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.200087308883667,
      "learning_rate": 6.255630450755519e-06,
      "loss": 0.8872,
      "step": 22481
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7501214742660522,
      "learning_rate": 6.254991534357729e-06,
      "loss": 0.7934,
      "step": 22482
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.14862585067749,
      "learning_rate": 6.25435261795994e-06,
      "loss": 1.2015,
      "step": 22483
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9641802310943604,
      "learning_rate": 6.253713701562151e-06,
      "loss": 0.8616,
      "step": 22484
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.915992259979248,
      "learning_rate": 6.2530747851643615e-06,
      "loss": 0.9166,
      "step": 22485
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2851595878601074,
      "learning_rate": 6.252435868766572e-06,
      "loss": 0.4368,
      "step": 22486
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4752752780914307,
      "learning_rate": 6.251796952368783e-06,
      "loss": 0.8989,
      "step": 22487
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.825321078300476,
      "learning_rate": 6.2511580359709936e-06,
      "loss": 0.7499,
      "step": 22488
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8042852878570557,
      "learning_rate": 6.250519119573204e-06,
      "loss": 0.8482,
      "step": 22489
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.172773599624634,
      "learning_rate": 6.249880203175415e-06,
      "loss": 1.111,
      "step": 22490
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.058129072189331,
      "learning_rate": 6.249241286777626e-06,
      "loss": 0.789,
      "step": 22491
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5082015991210938,
      "learning_rate": 6.248602370379836e-06,
      "loss": 0.8058,
      "step": 22492
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.578449249267578,
      "learning_rate": 6.247963453982046e-06,
      "loss": 1.2554,
      "step": 22493
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3904242515563965,
      "learning_rate": 6.247324537584257e-06,
      "loss": 0.7284,
      "step": 22494
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.061302661895752,
      "learning_rate": 6.246685621186468e-06,
      "loss": 0.7454,
      "step": 22495
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.435757875442505,
      "learning_rate": 6.246046704788678e-06,
      "loss": 0.8238,
      "step": 22496
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.320338726043701,
      "learning_rate": 6.245407788390889e-06,
      "loss": 0.6784,
      "step": 22497
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8717137575149536,
      "learning_rate": 6.2447688719931e-06,
      "loss": 0.9366,
      "step": 22498
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6304593086242676,
      "learning_rate": 6.24412995559531e-06,
      "loss": 0.9205,
      "step": 22499
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.4503865242004395,
      "learning_rate": 6.243491039197521e-06,
      "loss": 1.1153,
      "step": 22500
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6614789962768555,
      "learning_rate": 6.242852122799732e-06,
      "loss": 0.8481,
      "step": 22501
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.17305326461792,
      "learning_rate": 6.2422132064019425e-06,
      "loss": 0.829,
      "step": 22502
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.101484775543213,
      "learning_rate": 6.241574290004153e-06,
      "loss": 1.1967,
      "step": 22503
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.537492275238037,
      "learning_rate": 6.240935373606364e-06,
      "loss": 0.7869,
      "step": 22504
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7101356983184814,
      "learning_rate": 6.240296457208575e-06,
      "loss": 0.9546,
      "step": 22505
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8602734804153442,
      "learning_rate": 6.239657540810785e-06,
      "loss": 0.8455,
      "step": 22506
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0617740154266357,
      "learning_rate": 6.239018624412996e-06,
      "loss": 0.9131,
      "step": 22507
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.227382659912109,
      "learning_rate": 6.238379708015207e-06,
      "loss": 1.3516,
      "step": 22508
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7402067184448242,
      "learning_rate": 6.237740791617417e-06,
      "loss": 0.9315,
      "step": 22509
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.229428291320801,
      "learning_rate": 6.237101875219628e-06,
      "loss": 0.9205,
      "step": 22510
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0188865661621094,
      "learning_rate": 6.236462958821838e-06,
      "loss": 0.8659,
      "step": 22511
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.004181146621704,
      "learning_rate": 6.235824042424049e-06,
      "loss": 1.165,
      "step": 22512
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5362670421600342,
      "learning_rate": 6.235185126026259e-06,
      "loss": 0.9219,
      "step": 22513
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.18853759765625,
      "learning_rate": 6.23454620962847e-06,
      "loss": 0.9261,
      "step": 22514
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.722937822341919,
      "learning_rate": 6.233907293230681e-06,
      "loss": 0.777,
      "step": 22515
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6429052352905273,
      "learning_rate": 6.2332683768328915e-06,
      "loss": 0.9139,
      "step": 22516
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6633378267288208,
      "learning_rate": 6.232629460435102e-06,
      "loss": 0.8036,
      "step": 22517
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7732174396514893,
      "learning_rate": 6.231990544037313e-06,
      "loss": 0.8842,
      "step": 22518
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.858730673789978,
      "learning_rate": 6.2313516276395236e-06,
      "loss": 0.9434,
      "step": 22519
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3528051376342773,
      "learning_rate": 6.230712711241734e-06,
      "loss": 0.8794,
      "step": 22520
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.20576548576355,
      "learning_rate": 6.230073794843945e-06,
      "loss": 0.7886,
      "step": 22521
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6186909675598145,
      "learning_rate": 6.229434878446156e-06,
      "loss": 0.8606,
      "step": 22522
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.70578134059906,
      "learning_rate": 6.228795962048366e-06,
      "loss": 0.8851,
      "step": 22523
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.2476348876953125,
      "learning_rate": 6.228157045650577e-06,
      "loss": 1.1315,
      "step": 22524
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9873985052108765,
      "learning_rate": 6.227518129252788e-06,
      "loss": 1.0166,
      "step": 22525
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6556998491287231,
      "learning_rate": 6.2268792128549985e-06,
      "loss": 0.6148,
      "step": 22526
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.836174249649048,
      "learning_rate": 6.226240296457209e-06,
      "loss": 0.8561,
      "step": 22527
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.165477991104126,
      "learning_rate": 6.22560138005942e-06,
      "loss": 0.9287,
      "step": 22528
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4574939012527466,
      "learning_rate": 6.2249624636616306e-06,
      "loss": 0.795,
      "step": 22529
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4362173080444336,
      "learning_rate": 6.22432354726384e-06,
      "loss": 0.5874,
      "step": 22530
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8767733573913574,
      "learning_rate": 6.223684630866051e-06,
      "loss": 0.8957,
      "step": 22531
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2107608318328857,
      "learning_rate": 6.223045714468262e-06,
      "loss": 0.5106,
      "step": 22532
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8328170776367188,
      "learning_rate": 6.2224067980704725e-06,
      "loss": 0.9112,
      "step": 22533
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.06055736541748,
      "learning_rate": 6.221767881672683e-06,
      "loss": 1.4455,
      "step": 22534
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7851250171661377,
      "learning_rate": 6.221128965274894e-06,
      "loss": 0.8924,
      "step": 22535
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4280167818069458,
      "learning_rate": 6.220490048877105e-06,
      "loss": 0.7225,
      "step": 22536
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8639564514160156,
      "learning_rate": 6.219851132479315e-06,
      "loss": 0.7544,
      "step": 22537
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.051546812057495,
      "learning_rate": 6.219212216081526e-06,
      "loss": 0.7347,
      "step": 22538
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.149016857147217,
      "learning_rate": 6.218573299683737e-06,
      "loss": 0.9098,
      "step": 22539
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9552462100982666,
      "learning_rate": 6.217934383285947e-06,
      "loss": 0.9864,
      "step": 22540
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.130143165588379,
      "learning_rate": 6.217295466888158e-06,
      "loss": 1.3132,
      "step": 22541
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6585729122161865,
      "learning_rate": 6.216656550490369e-06,
      "loss": 0.4113,
      "step": 22542
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5805963277816772,
      "learning_rate": 6.2160176340925795e-06,
      "loss": 0.7913,
      "step": 22543
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0769786834716797,
      "learning_rate": 6.21537871769479e-06,
      "loss": 0.8762,
      "step": 22544
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0239384174346924,
      "learning_rate": 6.214739801297001e-06,
      "loss": 0.8872,
      "step": 22545
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5515812635421753,
      "learning_rate": 6.214100884899212e-06,
      "loss": 0.8886,
      "step": 22546
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.983903169631958,
      "learning_rate": 6.213461968501422e-06,
      "loss": 0.9292,
      "step": 22547
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.566663146018982,
      "learning_rate": 6.212823052103632e-06,
      "loss": 0.8639,
      "step": 22548
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9319641590118408,
      "learning_rate": 6.212184135705843e-06,
      "loss": 0.7655,
      "step": 22549
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4459929466247559,
      "learning_rate": 6.2115452193080536e-06,
      "loss": 0.7562,
      "step": 22550
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.964571475982666,
      "learning_rate": 6.210906302910264e-06,
      "loss": 0.8542,
      "step": 22551
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5837466716766357,
      "learning_rate": 6.210267386512475e-06,
      "loss": 0.8997,
      "step": 22552
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.196443557739258,
      "learning_rate": 6.209628470114686e-06,
      "loss": 0.4604,
      "step": 22553
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8580405712127686,
      "learning_rate": 6.208989553716896e-06,
      "loss": 0.8604,
      "step": 22554
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9736368656158447,
      "learning_rate": 6.208350637319107e-06,
      "loss": 1.0528,
      "step": 22555
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5644209384918213,
      "learning_rate": 6.207711720921318e-06,
      "loss": 0.7731,
      "step": 22556
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.965569257736206,
      "learning_rate": 6.2070728045235285e-06,
      "loss": 0.9069,
      "step": 22557
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6918737888336182,
      "learning_rate": 6.206433888125739e-06,
      "loss": 0.8569,
      "step": 22558
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.725780963897705,
      "learning_rate": 6.20579497172795e-06,
      "loss": 0.8982,
      "step": 22559
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8034429550170898,
      "learning_rate": 6.2051560553301606e-06,
      "loss": 0.8806,
      "step": 22560
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7404500246047974,
      "learning_rate": 6.204517138932371e-06,
      "loss": 0.8306,
      "step": 22561
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8647242784500122,
      "learning_rate": 6.203878222534582e-06,
      "loss": 1.1702,
      "step": 22562
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2243545055389404,
      "learning_rate": 6.203239306136793e-06,
      "loss": 0.8923,
      "step": 22563
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.166162014007568,
      "learning_rate": 6.202600389739003e-06,
      "loss": 0.718,
      "step": 22564
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9263322353363037,
      "learning_rate": 6.201961473341214e-06,
      "loss": 0.759,
      "step": 22565
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.512773275375366,
      "learning_rate": 6.201322556943424e-06,
      "loss": 1.2305,
      "step": 22566
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7698110342025757,
      "learning_rate": 6.200683640545635e-06,
      "loss": 0.9501,
      "step": 22567
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6277942657470703,
      "learning_rate": 6.200044724147845e-06,
      "loss": 1.1238,
      "step": 22568
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8260773420333862,
      "learning_rate": 6.199405807750056e-06,
      "loss": 0.9466,
      "step": 22569
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7474918365478516,
      "learning_rate": 6.198766891352267e-06,
      "loss": 1.0652,
      "step": 22570
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.7768313884735107,
      "learning_rate": 6.1981279749544774e-06,
      "loss": 0.6222,
      "step": 22571
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9439200162887573,
      "learning_rate": 6.197489058556688e-06,
      "loss": 0.9639,
      "step": 22572
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8998749256134033,
      "learning_rate": 6.196850142158899e-06,
      "loss": 0.9727,
      "step": 22573
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.756805419921875,
      "learning_rate": 6.1962112257611095e-06,
      "loss": 0.9356,
      "step": 22574
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.5889644622802734,
      "learning_rate": 6.19557230936332e-06,
      "loss": 0.8166,
      "step": 22575
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.487797737121582,
      "learning_rate": 6.194933392965531e-06,
      "loss": 0.923,
      "step": 22576
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7513132095336914,
      "learning_rate": 6.194294476567742e-06,
      "loss": 0.9324,
      "step": 22577
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0928220748901367,
      "learning_rate": 6.193655560169952e-06,
      "loss": 0.7857,
      "step": 22578
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.084353446960449,
      "learning_rate": 6.193016643772163e-06,
      "loss": 0.953,
      "step": 22579
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.63762092590332,
      "learning_rate": 6.192377727374374e-06,
      "loss": 0.9143,
      "step": 22580
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7497634887695312,
      "learning_rate": 6.1917388109765844e-06,
      "loss": 1.1932,
      "step": 22581
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.373152017593384,
      "learning_rate": 6.191099894578795e-06,
      "loss": 1.1625,
      "step": 22582
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8792890310287476,
      "learning_rate": 6.190460978181006e-06,
      "loss": 0.9059,
      "step": 22583
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.798040747642517,
      "learning_rate": 6.189822061783216e-06,
      "loss": 0.8438,
      "step": 22584
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8777425289154053,
      "learning_rate": 6.189183145385426e-06,
      "loss": 0.4842,
      "step": 22585
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.614222764968872,
      "learning_rate": 6.188544228987637e-06,
      "loss": 0.9106,
      "step": 22586
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4349682331085205,
      "learning_rate": 6.187905312589848e-06,
      "loss": 0.9854,
      "step": 22587
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4869056940078735,
      "learning_rate": 6.1872663961920585e-06,
      "loss": 0.818,
      "step": 22588
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.6757736206054688,
      "learning_rate": 6.186627479794269e-06,
      "loss": 1.0294,
      "step": 22589
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8981566429138184,
      "learning_rate": 6.18598856339648e-06,
      "loss": 0.8421,
      "step": 22590
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.8235015869140625,
      "learning_rate": 6.185349646998691e-06,
      "loss": 1.1104,
      "step": 22591
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5056017637252808,
      "learning_rate": 6.184710730600901e-06,
      "loss": 0.4205,
      "step": 22592
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1260111331939697,
      "learning_rate": 6.184071814203112e-06,
      "loss": 0.9573,
      "step": 22593
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.288620710372925,
      "learning_rate": 6.183432897805323e-06,
      "loss": 1.0936,
      "step": 22594
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.649011492729187,
      "learning_rate": 6.182793981407533e-06,
      "loss": 0.9118,
      "step": 22595
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0493006706237793,
      "learning_rate": 6.182155065009744e-06,
      "loss": 0.858,
      "step": 22596
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.024522066116333,
      "learning_rate": 6.181516148611955e-06,
      "loss": 0.9301,
      "step": 22597
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7318229675292969,
      "learning_rate": 6.1808772322141655e-06,
      "loss": 1.0022,
      "step": 22598
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4639341831207275,
      "learning_rate": 6.180238315816376e-06,
      "loss": 0.9751,
      "step": 22599
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3683388233184814,
      "learning_rate": 6.179599399418587e-06,
      "loss": 1.0602,
      "step": 22600
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8811321258544922,
      "learning_rate": 6.178960483020798e-06,
      "loss": 0.8732,
      "step": 22601
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5419604778289795,
      "learning_rate": 6.178321566623008e-06,
      "loss": 0.9301,
      "step": 22602
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2300703525543213,
      "learning_rate": 6.177682650225218e-06,
      "loss": 0.8977,
      "step": 22603
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.3489458560943604,
      "learning_rate": 6.177043733827429e-06,
      "loss": 0.951,
      "step": 22604
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.975450873374939,
      "learning_rate": 6.1764048174296395e-06,
      "loss": 1.0749,
      "step": 22605
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.669119358062744,
      "learning_rate": 6.17576590103185e-06,
      "loss": 0.7895,
      "step": 22606
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6183576583862305,
      "learning_rate": 6.175126984634061e-06,
      "loss": 1.012,
      "step": 22607
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.276376724243164,
      "learning_rate": 6.174488068236272e-06,
      "loss": 0.9154,
      "step": 22608
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0134167671203613,
      "learning_rate": 6.173849151838482e-06,
      "loss": 1.1172,
      "step": 22609
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.5879335403442383,
      "learning_rate": 6.173210235440693e-06,
      "loss": 0.8751,
      "step": 22610
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.999464750289917,
      "learning_rate": 6.172571319042904e-06,
      "loss": 0.8839,
      "step": 22611
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7256957292556763,
      "learning_rate": 6.1719324026451144e-06,
      "loss": 0.8484,
      "step": 22612
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3306353092193604,
      "learning_rate": 6.171293486247325e-06,
      "loss": 0.7791,
      "step": 22613
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7211426496505737,
      "learning_rate": 6.170654569849536e-06,
      "loss": 0.7943,
      "step": 22614
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.160912990570068,
      "learning_rate": 6.1700156534517465e-06,
      "loss": 0.8459,
      "step": 22615
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8033965826034546,
      "learning_rate": 6.169376737053957e-06,
      "loss": 0.8369,
      "step": 22616
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.03832745552063,
      "learning_rate": 6.168737820656168e-06,
      "loss": 0.9639,
      "step": 22617
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9331116676330566,
      "learning_rate": 6.168098904258379e-06,
      "loss": 0.826,
      "step": 22618
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8073667287826538,
      "learning_rate": 6.167459987860589e-06,
      "loss": 0.8816,
      "step": 22619
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7172459363937378,
      "learning_rate": 6.1668210714628e-06,
      "loss": 0.8221,
      "step": 22620
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.063416004180908,
      "learning_rate": 6.16618215506501e-06,
      "loss": 1.2191,
      "step": 22621
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.784902572631836,
      "learning_rate": 6.165543238667221e-06,
      "loss": 0.8639,
      "step": 22622
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.028693199157715,
      "learning_rate": 6.164904322269431e-06,
      "loss": 0.9229,
      "step": 22623
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.101564884185791,
      "learning_rate": 6.164265405871642e-06,
      "loss": 1.0314,
      "step": 22624
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0550858974456787,
      "learning_rate": 6.163626489473853e-06,
      "loss": 0.8167,
      "step": 22625
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.1684110164642334,
      "learning_rate": 6.162987573076063e-06,
      "loss": 0.8983,
      "step": 22626
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.555976390838623,
      "learning_rate": 6.162348656678274e-06,
      "loss": 0.8823,
      "step": 22627
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8474342823028564,
      "learning_rate": 6.161709740280485e-06,
      "loss": 0.8401,
      "step": 22628
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6856045722961426,
      "learning_rate": 6.1610708238826955e-06,
      "loss": 0.7991,
      "step": 22629
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5526869297027588,
      "learning_rate": 6.160431907484906e-06,
      "loss": 0.7172,
      "step": 22630
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.030128240585327,
      "learning_rate": 6.159792991087117e-06,
      "loss": 0.9339,
      "step": 22631
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5358991622924805,
      "learning_rate": 6.159154074689328e-06,
      "loss": 0.8645,
      "step": 22632
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8114304542541504,
      "learning_rate": 6.158515158291538e-06,
      "loss": 0.962,
      "step": 22633
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0373411178588867,
      "learning_rate": 6.157876241893749e-06,
      "loss": 0.8651,
      "step": 22634
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0842251777648926,
      "learning_rate": 6.15723732549596e-06,
      "loss": 0.9766,
      "step": 22635
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6997199058532715,
      "learning_rate": 6.15659840909817e-06,
      "loss": 0.965,
      "step": 22636
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6564373970031738,
      "learning_rate": 6.155959492700381e-06,
      "loss": 0.8272,
      "step": 22637
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.703044056892395,
      "learning_rate": 6.155320576302592e-06,
      "loss": 0.8582,
      "step": 22638
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0332682132720947,
      "learning_rate": 6.154681659904802e-06,
      "loss": 0.856,
      "step": 22639
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9591922760009766,
      "learning_rate": 6.154042743507012e-06,
      "loss": 1.0694,
      "step": 22640
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8294215202331543,
      "learning_rate": 6.153403827109223e-06,
      "loss": 0.8982,
      "step": 22641
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.230705499649048,
      "learning_rate": 6.152764910711434e-06,
      "loss": 0.9231,
      "step": 22642
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.0856542587280273,
      "learning_rate": 6.1521259943136444e-06,
      "loss": 0.8245,
      "step": 22643
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.764914870262146,
      "learning_rate": 6.151487077915855e-06,
      "loss": 1.0746,
      "step": 22644
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8679313659667969,
      "learning_rate": 6.150848161518066e-06,
      "loss": 1.1198,
      "step": 22645
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.153884172439575,
      "learning_rate": 6.1502092451202765e-06,
      "loss": 0.9117,
      "step": 22646
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5963512659072876,
      "learning_rate": 6.149570328722487e-06,
      "loss": 0.9055,
      "step": 22647
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.533498287200928,
      "learning_rate": 6.148931412324698e-06,
      "loss": 0.6861,
      "step": 22648
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5379575490951538,
      "learning_rate": 6.148292495926909e-06,
      "loss": 0.8081,
      "step": 22649
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6562381982803345,
      "learning_rate": 6.147653579529119e-06,
      "loss": 0.8755,
      "step": 22650
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.574295163154602,
      "learning_rate": 6.14701466313133e-06,
      "loss": 0.8369,
      "step": 22651
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.385512828826904,
      "learning_rate": 6.146375746733541e-06,
      "loss": 1.1161,
      "step": 22652
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.854697585105896,
      "learning_rate": 6.1457368303357514e-06,
      "loss": 0.9013,
      "step": 22653
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1660778522491455,
      "learning_rate": 6.145097913937962e-06,
      "loss": 0.8271,
      "step": 22654
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7233986854553223,
      "learning_rate": 6.144458997540173e-06,
      "loss": 0.6497,
      "step": 22655
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.990364670753479,
      "learning_rate": 6.1438200811423835e-06,
      "loss": 1.2478,
      "step": 22656
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8895601034164429,
      "learning_rate": 6.143181164744593e-06,
      "loss": 0.8381,
      "step": 22657
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1037983894348145,
      "learning_rate": 6.142542248346804e-06,
      "loss": 0.8736,
      "step": 22658
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.489801406860352,
      "learning_rate": 6.141903331949015e-06,
      "loss": 1.2944,
      "step": 22659
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8992960453033447,
      "learning_rate": 6.1412644155512255e-06,
      "loss": 0.8399,
      "step": 22660
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7246901988983154,
      "learning_rate": 6.140625499153436e-06,
      "loss": 0.839,
      "step": 22661
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5433242321014404,
      "learning_rate": 6.139986582755647e-06,
      "loss": 0.7594,
      "step": 22662
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.836829423904419,
      "learning_rate": 6.139347666357858e-06,
      "loss": 0.9722,
      "step": 22663
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0366058349609375,
      "learning_rate": 6.138708749960068e-06,
      "loss": 0.855,
      "step": 22664
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7402753829956055,
      "learning_rate": 6.138069833562279e-06,
      "loss": 0.9275,
      "step": 22665
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.3100874423980713,
      "learning_rate": 6.13743091716449e-06,
      "loss": 0.9155,
      "step": 22666
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4220244884490967,
      "learning_rate": 6.1367920007667e-06,
      "loss": 0.4661,
      "step": 22667
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9750934839248657,
      "learning_rate": 6.136153084368911e-06,
      "loss": 0.8678,
      "step": 22668
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8233413696289062,
      "learning_rate": 6.135514167971122e-06,
      "loss": 0.6169,
      "step": 22669
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5895631313323975,
      "learning_rate": 6.1348752515733325e-06,
      "loss": 0.8305,
      "step": 22670
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.361131429672241,
      "learning_rate": 6.134236335175543e-06,
      "loss": 0.7139,
      "step": 22671
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8313106298446655,
      "learning_rate": 6.133597418777754e-06,
      "loss": 0.8255,
      "step": 22672
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3348703384399414,
      "learning_rate": 6.132958502379965e-06,
      "loss": 0.8183,
      "step": 22673
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.393932819366455,
      "learning_rate": 6.132319585982175e-06,
      "loss": 1.236,
      "step": 22674
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1875224113464355,
      "learning_rate": 6.131680669584386e-06,
      "loss": 0.7681,
      "step": 22675
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6919896602630615,
      "learning_rate": 6.131041753186596e-06,
      "loss": 0.8005,
      "step": 22676
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.242554187774658,
      "learning_rate": 6.1304028367888065e-06,
      "loss": 0.7165,
      "step": 22677
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.338043212890625,
      "learning_rate": 6.129763920391017e-06,
      "loss": 1.0913,
      "step": 22678
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.6373863220214844,
      "learning_rate": 6.129125003993228e-06,
      "loss": 1.0204,
      "step": 22679
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.252091407775879,
      "learning_rate": 6.128486087595439e-06,
      "loss": 0.7764,
      "step": 22680
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.50620698928833,
      "learning_rate": 6.127847171197649e-06,
      "loss": 0.7254,
      "step": 22681
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4641947746276855,
      "learning_rate": 6.12720825479986e-06,
      "loss": 0.7442,
      "step": 22682
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9656705856323242,
      "learning_rate": 6.126569338402071e-06,
      "loss": 0.9516,
      "step": 22683
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7817491292953491,
      "learning_rate": 6.1259304220042814e-06,
      "loss": 0.9157,
      "step": 22684
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9369961023330688,
      "learning_rate": 6.125291505606492e-06,
      "loss": 0.7943,
      "step": 22685
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7901368141174316,
      "learning_rate": 6.124652589208703e-06,
      "loss": 0.7787,
      "step": 22686
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0967230796813965,
      "learning_rate": 6.1240136728109135e-06,
      "loss": 0.7866,
      "step": 22687
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5429062843322754,
      "learning_rate": 6.123374756413124e-06,
      "loss": 0.8161,
      "step": 22688
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7491799592971802,
      "learning_rate": 6.122735840015335e-06,
      "loss": 0.7576,
      "step": 22689
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6325105428695679,
      "learning_rate": 6.122096923617546e-06,
      "loss": 0.7816,
      "step": 22690
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8377292156219482,
      "learning_rate": 6.121458007219756e-06,
      "loss": 0.8603,
      "step": 22691
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6777969598770142,
      "learning_rate": 6.120819090821967e-06,
      "loss": 1.2164,
      "step": 22692
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5850489139556885,
      "learning_rate": 6.120180174424178e-06,
      "loss": 0.8428,
      "step": 22693
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9488067626953125,
      "learning_rate": 6.119541258026388e-06,
      "loss": 0.887,
      "step": 22694
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.766548752784729,
      "learning_rate": 6.118902341628598e-06,
      "loss": 0.7819,
      "step": 22695
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.599948763847351,
      "learning_rate": 6.118263425230809e-06,
      "loss": 0.8651,
      "step": 22696
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.373368740081787,
      "learning_rate": 6.11762450883302e-06,
      "loss": 0.9036,
      "step": 22697
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7936280965805054,
      "learning_rate": 6.11698559243523e-06,
      "loss": 1.1324,
      "step": 22698
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2700490951538086,
      "learning_rate": 6.116346676037441e-06,
      "loss": 0.9807,
      "step": 22699
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.890772819519043,
      "learning_rate": 6.115707759639652e-06,
      "loss": 0.8149,
      "step": 22700
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8061286211013794,
      "learning_rate": 6.1150688432418625e-06,
      "loss": 0.7538,
      "step": 22701
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7910946607589722,
      "learning_rate": 6.114429926844073e-06,
      "loss": 0.7525,
      "step": 22702
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.6655640602111816,
      "learning_rate": 6.113791010446284e-06,
      "loss": 1.2774,
      "step": 22703
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6847972869873047,
      "learning_rate": 6.113152094048495e-06,
      "loss": 1.0031,
      "step": 22704
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6443870067596436,
      "learning_rate": 6.112513177650705e-06,
      "loss": 0.9522,
      "step": 22705
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9650516510009766,
      "learning_rate": 6.111874261252916e-06,
      "loss": 0.8192,
      "step": 22706
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6505365371704102,
      "learning_rate": 6.111235344855127e-06,
      "loss": 0.8805,
      "step": 22707
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9292876720428467,
      "learning_rate": 6.110596428457337e-06,
      "loss": 0.7108,
      "step": 22708
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4220805168151855,
      "learning_rate": 6.109957512059548e-06,
      "loss": 0.9959,
      "step": 22709
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.367124319076538,
      "learning_rate": 6.109318595661759e-06,
      "loss": 0.8817,
      "step": 22710
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.266462564468384,
      "learning_rate": 6.1086796792639695e-06,
      "loss": 1.1031,
      "step": 22711
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6096140146255493,
      "learning_rate": 6.108040762866179e-06,
      "loss": 0.7721,
      "step": 22712
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.870026588439941,
      "learning_rate": 6.10740184646839e-06,
      "loss": 1.2006,
      "step": 22713
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.342414379119873,
      "learning_rate": 6.106762930070601e-06,
      "loss": 0.8158,
      "step": 22714
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.394298553466797,
      "learning_rate": 6.1061240136728114e-06,
      "loss": 0.8564,
      "step": 22715
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.3192968368530273,
      "learning_rate": 6.105485097275022e-06,
      "loss": 0.8372,
      "step": 22716
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.519904613494873,
      "learning_rate": 6.104846180877233e-06,
      "loss": 1.0569,
      "step": 22717
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1380934715270996,
      "learning_rate": 6.1042072644794435e-06,
      "loss": 1.2118,
      "step": 22718
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0759894847869873,
      "learning_rate": 6.103568348081654e-06,
      "loss": 0.9035,
      "step": 22719
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.6303319931030273,
      "learning_rate": 6.102929431683865e-06,
      "loss": 0.7932,
      "step": 22720
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7567312717437744,
      "learning_rate": 6.102290515286076e-06,
      "loss": 0.8182,
      "step": 22721
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.130270004272461,
      "learning_rate": 6.101651598888286e-06,
      "loss": 0.9785,
      "step": 22722
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7907946109771729,
      "learning_rate": 6.101012682490497e-06,
      "loss": 0.8624,
      "step": 22723
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0594825744628906,
      "learning_rate": 6.100373766092708e-06,
      "loss": 0.9001,
      "step": 22724
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4405910968780518,
      "learning_rate": 6.0997348496949184e-06,
      "loss": 0.83,
      "step": 22725
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1771912574768066,
      "learning_rate": 6.099095933297129e-06,
      "loss": 0.7669,
      "step": 22726
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.746697187423706,
      "learning_rate": 6.09845701689934e-06,
      "loss": 0.8415,
      "step": 22727
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9621964693069458,
      "learning_rate": 6.0978181005015505e-06,
      "loss": 0.8658,
      "step": 22728
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4528576135635376,
      "learning_rate": 6.097179184103761e-06,
      "loss": 0.7506,
      "step": 22729
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0752787590026855,
      "learning_rate": 6.096540267705971e-06,
      "loss": 0.9869,
      "step": 22730
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9382247924804688,
      "learning_rate": 6.095901351308182e-06,
      "loss": 0.7499,
      "step": 22731
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3722548484802246,
      "learning_rate": 6.0952624349103925e-06,
      "loss": 0.7775,
      "step": 22732
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.22993540763855,
      "learning_rate": 6.094623518512603e-06,
      "loss": 0.9854,
      "step": 22733
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.094386100769043,
      "learning_rate": 6.093984602114814e-06,
      "loss": 0.7448,
      "step": 22734
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8723138570785522,
      "learning_rate": 6.093345685717025e-06,
      "loss": 0.8192,
      "step": 22735
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.451479196548462,
      "learning_rate": 6.092706769319235e-06,
      "loss": 1.023,
      "step": 22736
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8439079523086548,
      "learning_rate": 6.092067852921446e-06,
      "loss": 0.9182,
      "step": 22737
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.201249599456787,
      "learning_rate": 6.091428936523657e-06,
      "loss": 0.9138,
      "step": 22738
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5681021213531494,
      "learning_rate": 6.090790020125867e-06,
      "loss": 0.7336,
      "step": 22739
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.2722530364990234,
      "learning_rate": 6.090151103728078e-06,
      "loss": 0.8408,
      "step": 22740
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.745521903038025,
      "learning_rate": 6.089512187330289e-06,
      "loss": 0.8841,
      "step": 22741
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.416348457336426,
      "learning_rate": 6.0888732709324995e-06,
      "loss": 1.1257,
      "step": 22742
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.463627338409424,
      "learning_rate": 6.08823435453471e-06,
      "loss": 1.0174,
      "step": 22743
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.008843898773193,
      "learning_rate": 6.087595438136921e-06,
      "loss": 1.6956,
      "step": 22744
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.487146019935608,
      "learning_rate": 6.086956521739132e-06,
      "loss": 0.4487,
      "step": 22745
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.652264356613159,
      "learning_rate": 6.086317605341342e-06,
      "loss": 0.5659,
      "step": 22746
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.96024751663208,
      "learning_rate": 6.085678688943553e-06,
      "loss": 0.8566,
      "step": 22747
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5692623853683472,
      "learning_rate": 6.085039772545764e-06,
      "loss": 0.8058,
      "step": 22748
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7734742164611816,
      "learning_rate": 6.0844008561479735e-06,
      "loss": 1.1028,
      "step": 22749
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7526931762695312,
      "learning_rate": 6.083761939750184e-06,
      "loss": 0.946,
      "step": 22750
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8551185131073,
      "learning_rate": 6.083123023352395e-06,
      "loss": 0.8817,
      "step": 22751
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6949021816253662,
      "learning_rate": 6.082484106954606e-06,
      "loss": 0.8482,
      "step": 22752
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9655319452285767,
      "learning_rate": 6.081845190556816e-06,
      "loss": 0.9682,
      "step": 22753
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.794028639793396,
      "learning_rate": 6.081206274159027e-06,
      "loss": 0.8923,
      "step": 22754
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1398024559020996,
      "learning_rate": 6.080567357761238e-06,
      "loss": 0.9149,
      "step": 22755
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9114609956741333,
      "learning_rate": 6.0799284413634484e-06,
      "loss": 0.9219,
      "step": 22756
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.7446348667144775,
      "learning_rate": 6.079289524965659e-06,
      "loss": 0.6383,
      "step": 22757
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9392062425613403,
      "learning_rate": 6.07865060856787e-06,
      "loss": 0.8944,
      "step": 22758
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.700001835823059,
      "learning_rate": 6.0780116921700805e-06,
      "loss": 0.8125,
      "step": 22759
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.566732406616211,
      "learning_rate": 6.077372775772291e-06,
      "loss": 0.8029,
      "step": 22760
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7080107927322388,
      "learning_rate": 6.076733859374502e-06,
      "loss": 0.859,
      "step": 22761
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8615000247955322,
      "learning_rate": 6.076094942976713e-06,
      "loss": 0.7325,
      "step": 22762
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6385836601257324,
      "learning_rate": 6.075456026578923e-06,
      "loss": 0.7703,
      "step": 22763
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7088396549224854,
      "learning_rate": 6.074817110181134e-06,
      "loss": 0.9013,
      "step": 22764
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8748266696929932,
      "learning_rate": 6.074178193783345e-06,
      "loss": 0.9281,
      "step": 22765
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6466259956359863,
      "learning_rate": 6.0735392773855554e-06,
      "loss": 0.8903,
      "step": 22766
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.555416226387024,
      "learning_rate": 6.072900360987765e-06,
      "loss": 0.3814,
      "step": 22767
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9998596906661987,
      "learning_rate": 6.072261444589976e-06,
      "loss": 0.8757,
      "step": 22768
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7773311138153076,
      "learning_rate": 6.071622528192187e-06,
      "loss": 0.829,
      "step": 22769
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2023096084594727,
      "learning_rate": 6.070983611794397e-06,
      "loss": 0.8748,
      "step": 22770
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.211401462554932,
      "learning_rate": 6.070344695396608e-06,
      "loss": 1.3139,
      "step": 22771
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7474535703659058,
      "learning_rate": 6.069705778998819e-06,
      "loss": 0.9544,
      "step": 22772
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.6976466178894043,
      "learning_rate": 6.0690668626010295e-06,
      "loss": 0.9358,
      "step": 22773
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6994813680648804,
      "learning_rate": 6.06842794620324e-06,
      "loss": 0.7017,
      "step": 22774
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9393651485443115,
      "learning_rate": 6.067789029805451e-06,
      "loss": 0.7758,
      "step": 22775
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4337348937988281,
      "learning_rate": 6.067150113407662e-06,
      "loss": 1.1452,
      "step": 22776
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.913656711578369,
      "learning_rate": 6.066511197009872e-06,
      "loss": 1.7916,
      "step": 22777
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.00553560256958,
      "learning_rate": 6.065872280612083e-06,
      "loss": 0.9403,
      "step": 22778
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5671263933181763,
      "learning_rate": 6.065233364214294e-06,
      "loss": 0.8104,
      "step": 22779
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4383658170700073,
      "learning_rate": 6.064594447816504e-06,
      "loss": 1.0651,
      "step": 22780
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.054835557937622,
      "learning_rate": 6.063955531418715e-06,
      "loss": 0.874,
      "step": 22781
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8588998317718506,
      "learning_rate": 6.063316615020926e-06,
      "loss": 0.8584,
      "step": 22782
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.4095470905303955,
      "learning_rate": 6.0626776986231365e-06,
      "loss": 0.8789,
      "step": 22783
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.043896436691284,
      "learning_rate": 6.062038782225347e-06,
      "loss": 0.933,
      "step": 22784
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9464941024780273,
      "learning_rate": 6.061399865827557e-06,
      "loss": 0.8672,
      "step": 22785
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5804179906845093,
      "learning_rate": 6.060760949429768e-06,
      "loss": 0.788,
      "step": 22786
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.035982370376587,
      "learning_rate": 6.0601220330319785e-06,
      "loss": 0.8793,
      "step": 22787
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.753805160522461,
      "learning_rate": 6.059483116634189e-06,
      "loss": 0.9208,
      "step": 22788
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.471764087677002,
      "learning_rate": 6.0588442002364e-06,
      "loss": 1.1045,
      "step": 22789
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4803974628448486,
      "learning_rate": 6.0582052838386106e-06,
      "loss": 0.7926,
      "step": 22790
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8779631853103638,
      "learning_rate": 6.057566367440821e-06,
      "loss": 0.8385,
      "step": 22791
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.402746558189392,
      "learning_rate": 6.056927451043032e-06,
      "loss": 0.3857,
      "step": 22792
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.010145664215088,
      "learning_rate": 6.056288534645243e-06,
      "loss": 1.1331,
      "step": 22793
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8254492282867432,
      "learning_rate": 6.055649618247453e-06,
      "loss": 0.8157,
      "step": 22794
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8627526760101318,
      "learning_rate": 6.055010701849664e-06,
      "loss": 0.8832,
      "step": 22795
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.558974027633667,
      "learning_rate": 6.054371785451875e-06,
      "loss": 0.781,
      "step": 22796
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.761962652206421,
      "learning_rate": 6.0537328690540855e-06,
      "loss": 0.9114,
      "step": 22797
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1398353576660156,
      "learning_rate": 6.053093952656296e-06,
      "loss": 0.9751,
      "step": 22798
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.793406367301941,
      "learning_rate": 6.052455036258507e-06,
      "loss": 0.8088,
      "step": 22799
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6979281902313232,
      "learning_rate": 6.0518161198607176e-06,
      "loss": 0.8497,
      "step": 22800
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9360151290893555,
      "learning_rate": 6.051177203462928e-06,
      "loss": 0.8082,
      "step": 22801
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8697161674499512,
      "learning_rate": 6.050538287065139e-06,
      "loss": 1.165,
      "step": 22802
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7287551164627075,
      "learning_rate": 6.049899370667349e-06,
      "loss": 0.7992,
      "step": 22803
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0742883682250977,
      "learning_rate": 6.049260454269559e-06,
      "loss": 0.7251,
      "step": 22804
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6484297513961792,
      "learning_rate": 6.048621537871769e-06,
      "loss": 0.8167,
      "step": 22805
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7627702951431274,
      "learning_rate": 6.04798262147398e-06,
      "loss": 0.8474,
      "step": 22806
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0492372512817383,
      "learning_rate": 6.047343705076191e-06,
      "loss": 0.8248,
      "step": 22807
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.611206531524658,
      "learning_rate": 6.0467047886784015e-06,
      "loss": 1.0183,
      "step": 22808
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2022218704223633,
      "learning_rate": 6.046065872280612e-06,
      "loss": 0.8668,
      "step": 22809
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7385497093200684,
      "learning_rate": 6.045426955882823e-06,
      "loss": 0.9007,
      "step": 22810
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7990831136703491,
      "learning_rate": 6.0447880394850336e-06,
      "loss": 0.8878,
      "step": 22811
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5817657709121704,
      "learning_rate": 6.044149123087244e-06,
      "loss": 0.8588,
      "step": 22812
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.9833879470825195,
      "learning_rate": 6.043510206689455e-06,
      "loss": 1.0662,
      "step": 22813
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6067862510681152,
      "learning_rate": 6.042871290291666e-06,
      "loss": 0.7205,
      "step": 22814
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.7096664905548096,
      "learning_rate": 6.042232373893876e-06,
      "loss": 1.1169,
      "step": 22815
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.125727653503418,
      "learning_rate": 6.041593457496086e-06,
      "loss": 1.3225,
      "step": 22816
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.518576741218567,
      "learning_rate": 6.040954541098297e-06,
      "loss": 0.7169,
      "step": 22817
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.73619544506073,
      "learning_rate": 6.040315624700508e-06,
      "loss": 0.9546,
      "step": 22818
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.827238917350769,
      "learning_rate": 6.039676708302718e-06,
      "loss": 0.8484,
      "step": 22819
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.5410079956054688,
      "learning_rate": 6.039037791904929e-06,
      "loss": 1.0043,
      "step": 22820
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.498242735862732,
      "learning_rate": 6.03839887550714e-06,
      "loss": 0.4259,
      "step": 22821
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.072146892547607,
      "learning_rate": 6.03775995910935e-06,
      "loss": 0.6966,
      "step": 22822
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7107683420181274,
      "learning_rate": 6.037121042711561e-06,
      "loss": 0.8657,
      "step": 22823
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2035534381866455,
      "learning_rate": 6.036482126313772e-06,
      "loss": 0.8858,
      "step": 22824
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.546190857887268,
      "learning_rate": 6.0358432099159825e-06,
      "loss": 0.7997,
      "step": 22825
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.334137201309204,
      "learning_rate": 6.035204293518193e-06,
      "loss": 0.4008,
      "step": 22826
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6025190353393555,
      "learning_rate": 6.034565377120404e-06,
      "loss": 0.7693,
      "step": 22827
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6419782638549805,
      "learning_rate": 6.033926460722615e-06,
      "loss": 0.8706,
      "step": 22828
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.267621040344238,
      "learning_rate": 6.033287544324825e-06,
      "loss": 1.387,
      "step": 22829
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2455759048461914,
      "learning_rate": 6.032648627927036e-06,
      "loss": 0.7642,
      "step": 22830
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.562069296836853,
      "learning_rate": 6.032009711529247e-06,
      "loss": 0.8475,
      "step": 22831
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8005183935165405,
      "learning_rate": 6.031370795131457e-06,
      "loss": 0.8522,
      "step": 22832
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.144270658493042,
      "learning_rate": 6.030731878733668e-06,
      "loss": 0.803,
      "step": 22833
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.885736107826233,
      "learning_rate": 6.030092962335878e-06,
      "loss": 1.1091,
      "step": 22834
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.0135953426361084,
      "learning_rate": 6.029454045938089e-06,
      "loss": 1.05,
      "step": 22835
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.021172046661377,
      "learning_rate": 6.028815129540299e-06,
      "loss": 1.1728,
      "step": 22836
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8059816360473633,
      "learning_rate": 6.02817621314251e-06,
      "loss": 0.9219,
      "step": 22837
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6768908500671387,
      "learning_rate": 6.027537296744721e-06,
      "loss": 0.8591,
      "step": 22838
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1011714935302734,
      "learning_rate": 6.0268983803469315e-06,
      "loss": 0.7503,
      "step": 22839
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.659706711769104,
      "learning_rate": 6.026259463949142e-06,
      "loss": 0.733,
      "step": 22840
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.185638904571533,
      "learning_rate": 6.025620547551353e-06,
      "loss": 0.9801,
      "step": 22841
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7622638940811157,
      "learning_rate": 6.0249816311535636e-06,
      "loss": 0.853,
      "step": 22842
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.9221765995025635,
      "learning_rate": 6.024342714755774e-06,
      "loss": 1.1561,
      "step": 22843
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8045755624771118,
      "learning_rate": 6.023703798357985e-06,
      "loss": 0.9574,
      "step": 22844
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9082378149032593,
      "learning_rate": 6.023064881960196e-06,
      "loss": 0.878,
      "step": 22845
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4351511001586914,
      "learning_rate": 6.022425965562406e-06,
      "loss": 0.7829,
      "step": 22846
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.013148307800293,
      "learning_rate": 6.021787049164617e-06,
      "loss": 0.9181,
      "step": 22847
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.556673288345337,
      "learning_rate": 6.021148132766828e-06,
      "loss": 0.706,
      "step": 22848
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.5160467624664307,
      "learning_rate": 6.0205092163690385e-06,
      "loss": 1.0036,
      "step": 22849
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.483532905578613,
      "learning_rate": 6.019870299971249e-06,
      "loss": 1.2772,
      "step": 22850
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8620882034301758,
      "learning_rate": 6.01923138357346e-06,
      "loss": 0.8149,
      "step": 22851
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.393967390060425,
      "learning_rate": 6.0185924671756706e-06,
      "loss": 1.0116,
      "step": 22852
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2203433513641357,
      "learning_rate": 6.01795355077788e-06,
      "loss": 0.4103,
      "step": 22853
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6704180240631104,
      "learning_rate": 6.017314634380091e-06,
      "loss": 0.7491,
      "step": 22854
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.402324676513672,
      "learning_rate": 6.016675717982302e-06,
      "loss": 1.6464,
      "step": 22855
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8939156532287598,
      "learning_rate": 6.0160368015845125e-06,
      "loss": 1.1473,
      "step": 22856
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6507432460784912,
      "learning_rate": 6.015397885186723e-06,
      "loss": 0.6243,
      "step": 22857
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5107626914978027,
      "learning_rate": 6.014758968788934e-06,
      "loss": 0.8147,
      "step": 22858
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4577354192733765,
      "learning_rate": 6.014120052391145e-06,
      "loss": 0.4817,
      "step": 22859
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.946352243423462,
      "learning_rate": 6.013481135993355e-06,
      "loss": 0.9088,
      "step": 22860
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.406064510345459,
      "learning_rate": 6.012842219595566e-06,
      "loss": 0.7051,
      "step": 22861
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2573673725128174,
      "learning_rate": 6.012203303197777e-06,
      "loss": 1.0098,
      "step": 22862
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9039208889007568,
      "learning_rate": 6.011564386799987e-06,
      "loss": 1.0531,
      "step": 22863
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.497035264968872,
      "learning_rate": 6.010925470402198e-06,
      "loss": 0.3575,
      "step": 22864
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.281579971313477,
      "learning_rate": 6.010286554004409e-06,
      "loss": 1.1436,
      "step": 22865
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0987730026245117,
      "learning_rate": 6.0096476376066195e-06,
      "loss": 0.6886,
      "step": 22866
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6819727420806885,
      "learning_rate": 6.00900872120883e-06,
      "loss": 0.8883,
      "step": 22867
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.347721576690674,
      "learning_rate": 6.008369804811041e-06,
      "loss": 0.8957,
      "step": 22868
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.3054873943328857,
      "learning_rate": 6.007730888413252e-06,
      "loss": 0.6896,
      "step": 22869
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6900004148483276,
      "learning_rate": 6.007091972015462e-06,
      "loss": 0.8901,
      "step": 22870
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.382901191711426,
      "learning_rate": 6.006453055617672e-06,
      "loss": 2.0652,
      "step": 22871
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.770977735519409,
      "learning_rate": 6.005814139219883e-06,
      "loss": 0.9807,
      "step": 22872
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.976597547531128,
      "learning_rate": 6.0051752228220936e-06,
      "loss": 0.8681,
      "step": 22873
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6831104755401611,
      "learning_rate": 6.004536306424304e-06,
      "loss": 0.8305,
      "step": 22874
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5918426513671875,
      "learning_rate": 6.003897390026515e-06,
      "loss": 1.0879,
      "step": 22875
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.587090253829956,
      "learning_rate": 6.003258473628726e-06,
      "loss": 0.854,
      "step": 22876
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.331221580505371,
      "learning_rate": 6.002619557230936e-06,
      "loss": 0.8342,
      "step": 22877
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6625667810440063,
      "learning_rate": 6.001980640833147e-06,
      "loss": 0.8018,
      "step": 22878
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7967561483383179,
      "learning_rate": 6.001341724435358e-06,
      "loss": 0.9649,
      "step": 22879
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4829275608062744,
      "learning_rate": 6.0007028080375685e-06,
      "loss": 0.8686,
      "step": 22880
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.981467604637146,
      "learning_rate": 6.000063891639779e-06,
      "loss": 0.9008,
      "step": 22881
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.3565430641174316,
      "learning_rate": 5.99942497524199e-06,
      "loss": 0.8212,
      "step": 22882
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8525193929672241,
      "learning_rate": 5.9987860588442006e-06,
      "loss": 0.8905,
      "step": 22883
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8097277879714966,
      "learning_rate": 5.998147142446411e-06,
      "loss": 0.929,
      "step": 22884
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7028576135635376,
      "learning_rate": 5.997508226048622e-06,
      "loss": 0.9231,
      "step": 22885
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5826938152313232,
      "learning_rate": 5.996869309650833e-06,
      "loss": 0.8578,
      "step": 22886
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9290556907653809,
      "learning_rate": 5.996230393253043e-06,
      "loss": 1.0908,
      "step": 22887
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.257744073867798,
      "learning_rate": 5.995591476855254e-06,
      "loss": 1.118,
      "step": 22888
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.12367844581604,
      "learning_rate": 5.994952560457464e-06,
      "loss": 0.9209,
      "step": 22889
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.520042061805725,
      "learning_rate": 5.994313644059675e-06,
      "loss": 1.0792,
      "step": 22890
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2219958305358887,
      "learning_rate": 5.993674727661885e-06,
      "loss": 0.8222,
      "step": 22891
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.818164587020874,
      "learning_rate": 5.993035811264096e-06,
      "loss": 0.7599,
      "step": 22892
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2736977338790894,
      "learning_rate": 5.992396894866307e-06,
      "loss": 0.8128,
      "step": 22893
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9334514141082764,
      "learning_rate": 5.991757978468517e-06,
      "loss": 0.8103,
      "step": 22894
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7351163625717163,
      "learning_rate": 5.991119062070728e-06,
      "loss": 0.9166,
      "step": 22895
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3401981592178345,
      "learning_rate": 5.990480145672939e-06,
      "loss": 0.3506,
      "step": 22896
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.731396198272705,
      "learning_rate": 5.9898412292751495e-06,
      "loss": 0.8917,
      "step": 22897
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.913295030593872,
      "learning_rate": 5.98920231287736e-06,
      "loss": 0.9682,
      "step": 22898
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6275995969772339,
      "learning_rate": 5.988563396479571e-06,
      "loss": 0.7903,
      "step": 22899
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2566063404083252,
      "learning_rate": 5.987924480081782e-06,
      "loss": 0.414,
      "step": 22900
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.637538194656372,
      "learning_rate": 5.987285563683992e-06,
      "loss": 0.8469,
      "step": 22901
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6829233169555664,
      "learning_rate": 5.986646647286203e-06,
      "loss": 0.7835,
      "step": 22902
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9332891702651978,
      "learning_rate": 5.986007730888414e-06,
      "loss": 0.8155,
      "step": 22903
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.914564609527588,
      "learning_rate": 5.985368814490624e-06,
      "loss": 0.8008,
      "step": 22904
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6830030679702759,
      "learning_rate": 5.984729898092835e-06,
      "loss": 0.7332,
      "step": 22905
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5657832622528076,
      "learning_rate": 5.984090981695046e-06,
      "loss": 0.8554,
      "step": 22906
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.092906951904297,
      "learning_rate": 5.983452065297256e-06,
      "loss": 1.0232,
      "step": 22907
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.752751111984253,
      "learning_rate": 5.982813148899466e-06,
      "loss": 0.9215,
      "step": 22908
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.762816071510315,
      "learning_rate": 5.982174232501677e-06,
      "loss": 0.87,
      "step": 22909
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9245595932006836,
      "learning_rate": 5.981535316103888e-06,
      "loss": 0.3826,
      "step": 22910
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7187740802764893,
      "learning_rate": 5.9808963997060985e-06,
      "loss": 0.8686,
      "step": 22911
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.676676630973816,
      "learning_rate": 5.980257483308309e-06,
      "loss": 0.8995,
      "step": 22912
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9644460678100586,
      "learning_rate": 5.97961856691052e-06,
      "loss": 0.7143,
      "step": 22913
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8432871103286743,
      "learning_rate": 5.9789796505127306e-06,
      "loss": 0.9189,
      "step": 22914
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.728292465209961,
      "learning_rate": 5.978340734114941e-06,
      "loss": 0.7881,
      "step": 22915
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.625320911407471,
      "learning_rate": 5.977701817717152e-06,
      "loss": 1.4765,
      "step": 22916
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9184037446975708,
      "learning_rate": 5.977062901319363e-06,
      "loss": 0.7579,
      "step": 22917
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.970905303955078,
      "learning_rate": 5.976423984921573e-06,
      "loss": 0.8039,
      "step": 22918
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7737847566604614,
      "learning_rate": 5.975785068523784e-06,
      "loss": 0.8708,
      "step": 22919
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.613528847694397,
      "learning_rate": 5.975146152125995e-06,
      "loss": 0.9066,
      "step": 22920
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.215028762817383,
      "learning_rate": 5.9745072357282055e-06,
      "loss": 0.7765,
      "step": 22921
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9721431732177734,
      "learning_rate": 5.973868319330416e-06,
      "loss": 0.8642,
      "step": 22922
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.321205735206604,
      "learning_rate": 5.973229402932627e-06,
      "loss": 0.4089,
      "step": 22923
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8089747428894043,
      "learning_rate": 5.9725904865348376e-06,
      "loss": 0.9385,
      "step": 22924
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7941051721572876,
      "learning_rate": 5.971951570137048e-06,
      "loss": 0.8121,
      "step": 22925
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5416665077209473,
      "learning_rate": 5.971312653739258e-06,
      "loss": 0.9031,
      "step": 22926
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.854743480682373,
      "learning_rate": 5.970673737341469e-06,
      "loss": 0.5649,
      "step": 22927
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8985811471939087,
      "learning_rate": 5.9700348209436795e-06,
      "loss": 0.8425,
      "step": 22928
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8739253282546997,
      "learning_rate": 5.96939590454589e-06,
      "loss": 0.9255,
      "step": 22929
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.851594090461731,
      "learning_rate": 5.968756988148101e-06,
      "loss": 0.8752,
      "step": 22930
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.4731597900390625,
      "learning_rate": 5.968118071750312e-06,
      "loss": 1.5367,
      "step": 22931
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8166559934616089,
      "learning_rate": 5.967479155352522e-06,
      "loss": 0.7556,
      "step": 22932
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.686835765838623,
      "learning_rate": 5.966840238954733e-06,
      "loss": 1.3666,
      "step": 22933
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.176856279373169,
      "learning_rate": 5.966201322556944e-06,
      "loss": 1.1908,
      "step": 22934
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.08247971534729,
      "learning_rate": 5.965562406159154e-06,
      "loss": 0.7675,
      "step": 22935
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.257861375808716,
      "learning_rate": 5.964923489761365e-06,
      "loss": 0.8281,
      "step": 22936
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9825265407562256,
      "learning_rate": 5.964284573363576e-06,
      "loss": 1.0135,
      "step": 22937
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6370207071304321,
      "learning_rate": 5.9636456569657865e-06,
      "loss": 0.8803,
      "step": 22938
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8383891582489014,
      "learning_rate": 5.963006740567997e-06,
      "loss": 0.9187,
      "step": 22939
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6017780303955078,
      "learning_rate": 5.962367824170208e-06,
      "loss": 0.8245,
      "step": 22940
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.668184518814087,
      "learning_rate": 5.961728907772419e-06,
      "loss": 1.1878,
      "step": 22941
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.1917667388916016,
      "learning_rate": 5.961089991374629e-06,
      "loss": 0.9839,
      "step": 22942
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.711890459060669,
      "learning_rate": 5.96045107497684e-06,
      "loss": 1.0561,
      "step": 22943
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.886785864830017,
      "learning_rate": 5.95981215857905e-06,
      "loss": 1.1118,
      "step": 22944
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.128730535507202,
      "learning_rate": 5.9591732421812606e-06,
      "loss": 1.1543,
      "step": 22945
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7089275121688843,
      "learning_rate": 5.958534325783471e-06,
      "loss": 0.9401,
      "step": 22946
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.885830283164978,
      "learning_rate": 5.957895409385682e-06,
      "loss": 0.9031,
      "step": 22947
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8137519359588623,
      "learning_rate": 5.957256492987893e-06,
      "loss": 0.7991,
      "step": 22948
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.228909015655518,
      "learning_rate": 5.956617576590103e-06,
      "loss": 1.414,
      "step": 22949
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.892435550689697,
      "learning_rate": 5.955978660192314e-06,
      "loss": 1.4397,
      "step": 22950
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3707932233810425,
      "learning_rate": 5.955339743794525e-06,
      "loss": 0.4425,
      "step": 22951
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4375747442245483,
      "learning_rate": 5.9547008273967355e-06,
      "loss": 0.8434,
      "step": 22952
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8379262685775757,
      "learning_rate": 5.954061910998946e-06,
      "loss": 1.105,
      "step": 22953
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.254689931869507,
      "learning_rate": 5.953422994601157e-06,
      "loss": 0.9275,
      "step": 22954
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.749161720275879,
      "learning_rate": 5.9527840782033676e-06,
      "loss": 0.7412,
      "step": 22955
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.8486106395721436,
      "learning_rate": 5.952145161805578e-06,
      "loss": 0.8429,
      "step": 22956
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8987805843353271,
      "learning_rate": 5.951506245407789e-06,
      "loss": 0.8714,
      "step": 22957
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.504364013671875,
      "learning_rate": 5.95086732901e-06,
      "loss": 0.7412,
      "step": 22958
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8938157558441162,
      "learning_rate": 5.95022841261221e-06,
      "loss": 0.6464,
      "step": 22959
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.456984043121338,
      "learning_rate": 5.949589496214421e-06,
      "loss": 1.2647,
      "step": 22960
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0299720764160156,
      "learning_rate": 5.948950579816632e-06,
      "loss": 0.9556,
      "step": 22961
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6039202213287354,
      "learning_rate": 5.948311663418842e-06,
      "loss": 0.801,
      "step": 22962
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.691280841827393,
      "learning_rate": 5.947672747021052e-06,
      "loss": 1.5617,
      "step": 22963
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.153017044067383,
      "learning_rate": 5.947033830623263e-06,
      "loss": 0.8039,
      "step": 22964
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7775168418884277,
      "learning_rate": 5.946394914225474e-06,
      "loss": 0.8295,
      "step": 22965
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6104085445404053,
      "learning_rate": 5.9457559978276844e-06,
      "loss": 0.8236,
      "step": 22966
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0250473022460938,
      "learning_rate": 5.945117081429895e-06,
      "loss": 0.7083,
      "step": 22967
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.099750518798828,
      "learning_rate": 5.944478165032106e-06,
      "loss": 0.9208,
      "step": 22968
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.101334810256958,
      "learning_rate": 5.9438392486343165e-06,
      "loss": 0.8991,
      "step": 22969
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.7600924968719482,
      "learning_rate": 5.943200332236527e-06,
      "loss": 0.7691,
      "step": 22970
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.181842088699341,
      "learning_rate": 5.942561415838738e-06,
      "loss": 0.7845,
      "step": 22971
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1757969856262207,
      "learning_rate": 5.941922499440949e-06,
      "loss": 0.8531,
      "step": 22972
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3664637804031372,
      "learning_rate": 5.941283583043159e-06,
      "loss": 0.7357,
      "step": 22973
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6123441457748413,
      "learning_rate": 5.94064466664537e-06,
      "loss": 0.7675,
      "step": 22974
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6411902904510498,
      "learning_rate": 5.940005750247581e-06,
      "loss": 0.8234,
      "step": 22975
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.207149505615234,
      "learning_rate": 5.9393668338497914e-06,
      "loss": 0.7158,
      "step": 22976
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7626458406448364,
      "learning_rate": 5.938727917452002e-06,
      "loss": 0.8225,
      "step": 22977
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2078421115875244,
      "learning_rate": 5.938089001054213e-06,
      "loss": 0.8423,
      "step": 22978
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.726516246795654,
      "learning_rate": 5.9374500846564235e-06,
      "loss": 1.0106,
      "step": 22979
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.37813401222229,
      "learning_rate": 5.936811168258633e-06,
      "loss": 0.9277,
      "step": 22980
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4948055744171143,
      "learning_rate": 5.936172251860844e-06,
      "loss": 0.3662,
      "step": 22981
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.530991554260254,
      "learning_rate": 5.935533335463055e-06,
      "loss": 1.458,
      "step": 22982
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.653846740722656,
      "learning_rate": 5.9348944190652655e-06,
      "loss": 1.3468,
      "step": 22983
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.693809151649475,
      "learning_rate": 5.934255502667476e-06,
      "loss": 0.8895,
      "step": 22984
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.814560055732727,
      "learning_rate": 5.933616586269687e-06,
      "loss": 0.8842,
      "step": 22985
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6440297365188599,
      "learning_rate": 5.932977669871898e-06,
      "loss": 0.8398,
      "step": 22986
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.876737356185913,
      "learning_rate": 5.932338753474108e-06,
      "loss": 0.9209,
      "step": 22987
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.6485466957092285,
      "learning_rate": 5.931699837076319e-06,
      "loss": 1.0458,
      "step": 22988
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.519323706626892,
      "learning_rate": 5.93106092067853e-06,
      "loss": 0.8183,
      "step": 22989
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8997222185134888,
      "learning_rate": 5.93042200428074e-06,
      "loss": 0.9773,
      "step": 22990
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.752436876296997,
      "learning_rate": 5.929783087882951e-06,
      "loss": 0.7361,
      "step": 22991
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9998526573181152,
      "learning_rate": 5.929144171485162e-06,
      "loss": 0.9239,
      "step": 22992
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8990113735198975,
      "learning_rate": 5.9285052550873725e-06,
      "loss": 0.8298,
      "step": 22993
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1616785526275635,
      "learning_rate": 5.927866338689583e-06,
      "loss": 0.9484,
      "step": 22994
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3765804767608643,
      "learning_rate": 5.927227422291794e-06,
      "loss": 0.3755,
      "step": 22995
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4197510480880737,
      "learning_rate": 5.926588505894005e-06,
      "loss": 0.8634,
      "step": 22996
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8519116640090942,
      "learning_rate": 5.925949589496215e-06,
      "loss": 0.8676,
      "step": 22997
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9854224920272827,
      "learning_rate": 5.925310673098426e-06,
      "loss": 1.1205,
      "step": 22998
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.14093542098999,
      "learning_rate": 5.924671756700636e-06,
      "loss": 0.7828,
      "step": 22999
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.650665760040283,
      "learning_rate": 5.9240328403028465e-06,
      "loss": 0.627,
      "step": 23000
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.190326690673828,
      "learning_rate": 5.923393923905057e-06,
      "loss": 0.8892,
      "step": 23001
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.253973960876465,
      "learning_rate": 5.922755007507268e-06,
      "loss": 0.9761,
      "step": 23002
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2148170471191406,
      "learning_rate": 5.922116091109479e-06,
      "loss": 1.2958,
      "step": 23003
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.8861594200134277,
      "learning_rate": 5.921477174711689e-06,
      "loss": 1.1801,
      "step": 23004
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.643316388130188,
      "learning_rate": 5.9208382583139e-06,
      "loss": 0.912,
      "step": 23005
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.723800778388977,
      "learning_rate": 5.920199341916111e-06,
      "loss": 0.8202,
      "step": 23006
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7252616882324219,
      "learning_rate": 5.9195604255183214e-06,
      "loss": 0.8759,
      "step": 23007
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6846661567687988,
      "learning_rate": 5.918921509120532e-06,
      "loss": 0.8623,
      "step": 23008
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9054392576217651,
      "learning_rate": 5.918282592722743e-06,
      "loss": 0.8466,
      "step": 23009
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5012388229370117,
      "learning_rate": 5.9176436763249535e-06,
      "loss": 0.8192,
      "step": 23010
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5778456926345825,
      "learning_rate": 5.917004759927164e-06,
      "loss": 0.9155,
      "step": 23011
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.691472053527832,
      "learning_rate": 5.916365843529375e-06,
      "loss": 0.793,
      "step": 23012
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6135807037353516,
      "learning_rate": 5.915726927131586e-06,
      "loss": 0.7939,
      "step": 23013
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.279450416564941,
      "learning_rate": 5.915088010733796e-06,
      "loss": 0.7468,
      "step": 23014
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.212538242340088,
      "learning_rate": 5.914449094336007e-06,
      "loss": 0.7484,
      "step": 23015
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.972407817840576,
      "learning_rate": 5.913810177938218e-06,
      "loss": 1.6097,
      "step": 23016
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.651353359222412,
      "learning_rate": 5.913171261540428e-06,
      "loss": 0.8714,
      "step": 23017
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8529322147369385,
      "learning_rate": 5.912532345142638e-06,
      "loss": 0.8698,
      "step": 23018
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1001532077789307,
      "learning_rate": 5.911893428744849e-06,
      "loss": 0.8585,
      "step": 23019
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2861149311065674,
      "learning_rate": 5.91125451234706e-06,
      "loss": 1.0495,
      "step": 23020
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.13460636138916,
      "learning_rate": 5.91061559594927e-06,
      "loss": 0.837,
      "step": 23021
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.384995460510254,
      "learning_rate": 5.909976679551481e-06,
      "loss": 0.6016,
      "step": 23022
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5081084966659546,
      "learning_rate": 5.909337763153692e-06,
      "loss": 0.8477,
      "step": 23023
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.9945225715637207,
      "learning_rate": 5.9086988467559025e-06,
      "loss": 1.1966,
      "step": 23024
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9650037288665771,
      "learning_rate": 5.908059930358113e-06,
      "loss": 0.8728,
      "step": 23025
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6525002717971802,
      "learning_rate": 5.907421013960324e-06,
      "loss": 0.8944,
      "step": 23026
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2095718383789062,
      "learning_rate": 5.906782097562535e-06,
      "loss": 0.8794,
      "step": 23027
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7770073413848877,
      "learning_rate": 5.906143181164745e-06,
      "loss": 0.6544,
      "step": 23028
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6745444536209106,
      "learning_rate": 5.905504264766956e-06,
      "loss": 0.7698,
      "step": 23029
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2518162727355957,
      "learning_rate": 5.904865348369167e-06,
      "loss": 0.9347,
      "step": 23030
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.183542728424072,
      "learning_rate": 5.904226431971377e-06,
      "loss": 1.2294,
      "step": 23031
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.602738857269287,
      "learning_rate": 5.903587515573588e-06,
      "loss": 1.413,
      "step": 23032
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9903720617294312,
      "learning_rate": 5.902948599175799e-06,
      "loss": 0.9475,
      "step": 23033
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6089544296264648,
      "learning_rate": 5.9023096827780095e-06,
      "loss": 0.8634,
      "step": 23034
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.6276254653930664,
      "learning_rate": 5.901670766380219e-06,
      "loss": 0.993,
      "step": 23035
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8284302949905396,
      "learning_rate": 5.90103184998243e-06,
      "loss": 0.8897,
      "step": 23036
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3448034524917603,
      "learning_rate": 5.900392933584641e-06,
      "loss": 0.3811,
      "step": 23037
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7245882749557495,
      "learning_rate": 5.8997540171868514e-06,
      "loss": 0.8864,
      "step": 23038
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4643335342407227,
      "learning_rate": 5.899115100789062e-06,
      "loss": 0.8334,
      "step": 23039
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6029040813446045,
      "learning_rate": 5.898476184391273e-06,
      "loss": 0.7356,
      "step": 23040
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1153931617736816,
      "learning_rate": 5.8978372679934835e-06,
      "loss": 0.9179,
      "step": 23041
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.3438327312469482,
      "learning_rate": 5.897198351595694e-06,
      "loss": 0.9328,
      "step": 23042
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8549394607543945,
      "learning_rate": 5.896559435197905e-06,
      "loss": 0.8912,
      "step": 23043
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.103442430496216,
      "learning_rate": 5.895920518800116e-06,
      "loss": 0.8811,
      "step": 23044
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.252939224243164,
      "learning_rate": 5.895281602402326e-06,
      "loss": 1.5875,
      "step": 23045
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.060879707336426,
      "learning_rate": 5.894642686004537e-06,
      "loss": 1.1374,
      "step": 23046
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6957716941833496,
      "learning_rate": 5.894003769606748e-06,
      "loss": 0.9256,
      "step": 23047
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.088324069976807,
      "learning_rate": 5.8933648532089584e-06,
      "loss": 1.0101,
      "step": 23048
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.243360757827759,
      "learning_rate": 5.892725936811169e-06,
      "loss": 0.7299,
      "step": 23049
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.95067298412323,
      "learning_rate": 5.89208702041338e-06,
      "loss": 0.847,
      "step": 23050
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9777677059173584,
      "learning_rate": 5.8914481040155905e-06,
      "loss": 0.9646,
      "step": 23051
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1299190521240234,
      "learning_rate": 5.890809187617801e-06,
      "loss": 0.9958,
      "step": 23052
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.573857069015503,
      "learning_rate": 5.890170271220011e-06,
      "loss": 0.7806,
      "step": 23053
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.503452181816101,
      "learning_rate": 5.889531354822222e-06,
      "loss": 0.8384,
      "step": 23054
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5951502323150635,
      "learning_rate": 5.8888924384244325e-06,
      "loss": 0.7422,
      "step": 23055
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.390988349914551,
      "learning_rate": 5.888253522026643e-06,
      "loss": 0.956,
      "step": 23056
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.428023338317871,
      "learning_rate": 5.887614605628854e-06,
      "loss": 0.7987,
      "step": 23057
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7881193161010742,
      "learning_rate": 5.886975689231065e-06,
      "loss": 0.7808,
      "step": 23058
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.7618300914764404,
      "learning_rate": 5.886336772833275e-06,
      "loss": 0.9034,
      "step": 23059
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.119910717010498,
      "learning_rate": 5.885697856435486e-06,
      "loss": 0.9956,
      "step": 23060
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.903833270072937,
      "learning_rate": 5.885058940037697e-06,
      "loss": 0.511,
      "step": 23061
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2537341117858887,
      "learning_rate": 5.884420023639907e-06,
      "loss": 0.9161,
      "step": 23062
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.940927267074585,
      "learning_rate": 5.883781107242118e-06,
      "loss": 0.9138,
      "step": 23063
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2108163833618164,
      "learning_rate": 5.883142190844329e-06,
      "loss": 0.9365,
      "step": 23064
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.01074481010437,
      "learning_rate": 5.8825032744465395e-06,
      "loss": 0.736,
      "step": 23065
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8398512601852417,
      "learning_rate": 5.88186435804875e-06,
      "loss": 1.2278,
      "step": 23066
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.013134956359863,
      "learning_rate": 5.881225441650961e-06,
      "loss": 1.2669,
      "step": 23067
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9817099571228027,
      "learning_rate": 5.880586525253172e-06,
      "loss": 0.9006,
      "step": 23068
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.32938289642334,
      "learning_rate": 5.879947608855382e-06,
      "loss": 0.6838,
      "step": 23069
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0100162029266357,
      "learning_rate": 5.879308692457593e-06,
      "loss": 0.91,
      "step": 23070
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9108291864395142,
      "learning_rate": 5.878669776059804e-06,
      "loss": 0.7964,
      "step": 23071
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7677581310272217,
      "learning_rate": 5.8780308596620135e-06,
      "loss": 0.7941,
      "step": 23072
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7596800327301025,
      "learning_rate": 5.877391943264224e-06,
      "loss": 0.4914,
      "step": 23073
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2674355506896973,
      "learning_rate": 5.876753026866435e-06,
      "loss": 0.9385,
      "step": 23074
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8570870161056519,
      "learning_rate": 5.876114110468646e-06,
      "loss": 0.8999,
      "step": 23075
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5755553245544434,
      "learning_rate": 5.875475194070856e-06,
      "loss": 0.9063,
      "step": 23076
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4972186088562012,
      "learning_rate": 5.874836277673067e-06,
      "loss": 0.7902,
      "step": 23077
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.3557262420654297,
      "learning_rate": 5.874197361275278e-06,
      "loss": 0.7776,
      "step": 23078
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8642899990081787,
      "learning_rate": 5.8735584448774884e-06,
      "loss": 1.1399,
      "step": 23079
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.365095615386963,
      "learning_rate": 5.872919528479699e-06,
      "loss": 1.0929,
      "step": 23080
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3669066429138184,
      "learning_rate": 5.87228061208191e-06,
      "loss": 0.6378,
      "step": 23081
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0392467975616455,
      "learning_rate": 5.8716416956841205e-06,
      "loss": 0.8309,
      "step": 23082
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.636575222015381,
      "learning_rate": 5.871002779286331e-06,
      "loss": 1.3341,
      "step": 23083
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.2835378646850586,
      "learning_rate": 5.870363862888542e-06,
      "loss": 1.112,
      "step": 23084
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.426660060882568,
      "learning_rate": 5.869724946490753e-06,
      "loss": 1.6263,
      "step": 23085
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.2676665782928467,
      "learning_rate": 5.869086030092963e-06,
      "loss": 1.0505,
      "step": 23086
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4871355295181274,
      "learning_rate": 5.868447113695174e-06,
      "loss": 0.3868,
      "step": 23087
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.1317379474639893,
      "learning_rate": 5.867808197297385e-06,
      "loss": 0.7326,
      "step": 23088
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.785341739654541,
      "learning_rate": 5.8671692808995954e-06,
      "loss": 0.7956,
      "step": 23089
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.11517333984375,
      "learning_rate": 5.866530364501805e-06,
      "loss": 0.9276,
      "step": 23090
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4824482202529907,
      "learning_rate": 5.865891448104016e-06,
      "loss": 0.8754,
      "step": 23091
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6898759603500366,
      "learning_rate": 5.865252531706227e-06,
      "loss": 0.8182,
      "step": 23092
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5168569087982178,
      "learning_rate": 5.864613615308437e-06,
      "loss": 0.8046,
      "step": 23093
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6327340602874756,
      "learning_rate": 5.863974698910648e-06,
      "loss": 0.8832,
      "step": 23094
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.026972532272339,
      "learning_rate": 5.863335782512859e-06,
      "loss": 0.8511,
      "step": 23095
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8949331045150757,
      "learning_rate": 5.8626968661150695e-06,
      "loss": 0.7838,
      "step": 23096
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8416162729263306,
      "learning_rate": 5.86205794971728e-06,
      "loss": 0.8315,
      "step": 23097
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.013787031173706,
      "learning_rate": 5.861419033319491e-06,
      "loss": 0.8809,
      "step": 23098
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9140372276306152,
      "learning_rate": 5.860780116921702e-06,
      "loss": 0.8213,
      "step": 23099
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.1329450607299805,
      "learning_rate": 5.860141200523912e-06,
      "loss": 0.9638,
      "step": 23100
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.64824914932251,
      "learning_rate": 5.859502284126123e-06,
      "loss": 1.3433,
      "step": 23101
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.46117103099823,
      "learning_rate": 5.858863367728334e-06,
      "loss": 0.7883,
      "step": 23102
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.3146281242370605,
      "learning_rate": 5.858224451330544e-06,
      "loss": 1.1558,
      "step": 23103
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.652004599571228,
      "learning_rate": 5.857585534932755e-06,
      "loss": 0.7975,
      "step": 23104
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.6465063095092773,
      "learning_rate": 5.856946618534966e-06,
      "loss": 1.1719,
      "step": 23105
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6559752225875854,
      "learning_rate": 5.8563077021371765e-06,
      "loss": 0.8104,
      "step": 23106
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7058950662612915,
      "learning_rate": 5.855668785739387e-06,
      "loss": 0.89,
      "step": 23107
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.2159366607666016,
      "learning_rate": 5.855029869341597e-06,
      "loss": 1.0113,
      "step": 23108
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8698841333389282,
      "learning_rate": 5.854390952943808e-06,
      "loss": 0.9346,
      "step": 23109
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8234381675720215,
      "learning_rate": 5.8537520365460184e-06,
      "loss": 0.7722,
      "step": 23110
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4085164070129395,
      "learning_rate": 5.853113120148229e-06,
      "loss": 0.877,
      "step": 23111
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.809372067451477,
      "learning_rate": 5.85247420375044e-06,
      "loss": 0.8511,
      "step": 23112
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.235220432281494,
      "learning_rate": 5.8518352873526505e-06,
      "loss": 0.9144,
      "step": 23113
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.100972652435303,
      "learning_rate": 5.851196370954861e-06,
      "loss": 0.9119,
      "step": 23114
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.074713706970215,
      "learning_rate": 5.850557454557072e-06,
      "loss": 0.8365,
      "step": 23115
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.040034770965576,
      "learning_rate": 5.849918538159283e-06,
      "loss": 0.8513,
      "step": 23116
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.733201503753662,
      "learning_rate": 5.849279621761493e-06,
      "loss": 1.2557,
      "step": 23117
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6871992349624634,
      "learning_rate": 5.848640705363704e-06,
      "loss": 0.8459,
      "step": 23118
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8998339176177979,
      "learning_rate": 5.848001788965915e-06,
      "loss": 0.8696,
      "step": 23119
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.834639310836792,
      "learning_rate": 5.8473628725681254e-06,
      "loss": 0.8561,
      "step": 23120
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0139620304107666,
      "learning_rate": 5.846723956170336e-06,
      "loss": 1.1907,
      "step": 23121
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.022953987121582,
      "learning_rate": 5.846085039772547e-06,
      "loss": 0.8726,
      "step": 23122
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.2080953121185303,
      "learning_rate": 5.8454461233747575e-06,
      "loss": 0.8556,
      "step": 23123
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.786416530609131,
      "learning_rate": 5.844807206976968e-06,
      "loss": 0.9322,
      "step": 23124
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0033466815948486,
      "learning_rate": 5.844168290579179e-06,
      "loss": 0.8805,
      "step": 23125
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.460676193237305,
      "learning_rate": 5.843529374181389e-06,
      "loss": 1.0527,
      "step": 23126
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0443170070648193,
      "learning_rate": 5.8428904577835995e-06,
      "loss": 0.8269,
      "step": 23127
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5589975118637085,
      "learning_rate": 5.84225154138581e-06,
      "loss": 0.4319,
      "step": 23128
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.5743443965911865,
      "learning_rate": 5.841612624988021e-06,
      "loss": 1.0662,
      "step": 23129
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9120122194290161,
      "learning_rate": 5.840973708590232e-06,
      "loss": 0.8177,
      "step": 23130
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.144238471984863,
      "learning_rate": 5.840334792192442e-06,
      "loss": 1.3446,
      "step": 23131
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2245516777038574,
      "learning_rate": 5.839695875794653e-06,
      "loss": 0.8034,
      "step": 23132
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8119200468063354,
      "learning_rate": 5.839056959396864e-06,
      "loss": 0.7568,
      "step": 23133
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1085150241851807,
      "learning_rate": 5.838418042999074e-06,
      "loss": 0.7112,
      "step": 23134
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8560798168182373,
      "learning_rate": 5.837779126601285e-06,
      "loss": 0.9473,
      "step": 23135
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.277569532394409,
      "learning_rate": 5.837140210203496e-06,
      "loss": 0.9297,
      "step": 23136
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.900674819946289,
      "learning_rate": 5.8365012938057065e-06,
      "loss": 0.9208,
      "step": 23137
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.547443389892578,
      "learning_rate": 5.835862377407917e-06,
      "loss": 1.04,
      "step": 23138
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6186121702194214,
      "learning_rate": 5.835223461010128e-06,
      "loss": 0.904,
      "step": 23139
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.7974071502685547,
      "learning_rate": 5.834584544612339e-06,
      "loss": 0.5589,
      "step": 23140
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.496010661125183,
      "learning_rate": 5.833945628214549e-06,
      "loss": 0.8579,
      "step": 23141
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9579213857650757,
      "learning_rate": 5.83330671181676e-06,
      "loss": 1.228,
      "step": 23142
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.502323865890503,
      "learning_rate": 5.832667795418971e-06,
      "loss": 0.801,
      "step": 23143
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9805572032928467,
      "learning_rate": 5.832028879021181e-06,
      "loss": 0.878,
      "step": 23144
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8509600162506104,
      "learning_rate": 5.831389962623391e-06,
      "loss": 0.9437,
      "step": 23145
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.978760004043579,
      "learning_rate": 5.830751046225602e-06,
      "loss": 0.8272,
      "step": 23146
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7546099424362183,
      "learning_rate": 5.830112129827813e-06,
      "loss": 0.9002,
      "step": 23147
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1477065086364746,
      "learning_rate": 5.829473213430023e-06,
      "loss": 1.03,
      "step": 23148
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.3305177688598633,
      "learning_rate": 5.828834297032234e-06,
      "loss": 1.1445,
      "step": 23149
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6352717876434326,
      "learning_rate": 5.828195380634445e-06,
      "loss": 0.4698,
      "step": 23150
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0628843307495117,
      "learning_rate": 5.8275564642366555e-06,
      "loss": 0.7927,
      "step": 23151
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9828873872756958,
      "learning_rate": 5.826917547838866e-06,
      "loss": 0.974,
      "step": 23152
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.092036485671997,
      "learning_rate": 5.826278631441077e-06,
      "loss": 0.9796,
      "step": 23153
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9324108362197876,
      "learning_rate": 5.8256397150432875e-06,
      "loss": 0.9907,
      "step": 23154
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5084658861160278,
      "learning_rate": 5.825000798645498e-06,
      "loss": 0.4166,
      "step": 23155
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.159395217895508,
      "learning_rate": 5.824361882247709e-06,
      "loss": 0.7875,
      "step": 23156
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4752169847488403,
      "learning_rate": 5.82372296584992e-06,
      "loss": 0.8329,
      "step": 23157
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.4727606773376465,
      "learning_rate": 5.82308404945213e-06,
      "loss": 1.1752,
      "step": 23158
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.04093861579895,
      "learning_rate": 5.822445133054341e-06,
      "loss": 0.8296,
      "step": 23159
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.067122220993042,
      "learning_rate": 5.821806216656552e-06,
      "loss": 0.9871,
      "step": 23160
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3320024013519287,
      "learning_rate": 5.8211673002587624e-06,
      "loss": 0.3761,
      "step": 23161
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.979811191558838,
      "learning_rate": 5.820528383860973e-06,
      "loss": 1.09,
      "step": 23162
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7969764471054077,
      "learning_rate": 5.819889467463183e-06,
      "loss": 0.8209,
      "step": 23163
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9310426712036133,
      "learning_rate": 5.819250551065394e-06,
      "loss": 0.6613,
      "step": 23164
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.1870877742767334,
      "learning_rate": 5.818611634667604e-06,
      "loss": 0.6048,
      "step": 23165
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6689671277999878,
      "learning_rate": 5.817972718269815e-06,
      "loss": 0.8964,
      "step": 23166
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.957411050796509,
      "learning_rate": 5.817333801872026e-06,
      "loss": 1.1972,
      "step": 23167
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.822756290435791,
      "learning_rate": 5.8166948854742365e-06,
      "loss": 1.5749,
      "step": 23168
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8137043714523315,
      "learning_rate": 5.816055969076447e-06,
      "loss": 0.8964,
      "step": 23169
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7476247549057007,
      "learning_rate": 5.815417052678658e-06,
      "loss": 0.9566,
      "step": 23170
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1081228256225586,
      "learning_rate": 5.814778136280869e-06,
      "loss": 0.7594,
      "step": 23171
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.4569039344787598,
      "learning_rate": 5.814139219883079e-06,
      "loss": 0.8524,
      "step": 23172
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5614324808120728,
      "learning_rate": 5.81350030348529e-06,
      "loss": 0.9119,
      "step": 23173
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8723982572555542,
      "learning_rate": 5.812861387087501e-06,
      "loss": 0.8549,
      "step": 23174
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9136922359466553,
      "learning_rate": 5.812222470689711e-06,
      "loss": 0.8718,
      "step": 23175
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.135840892791748,
      "learning_rate": 5.811583554291922e-06,
      "loss": 0.8374,
      "step": 23176
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.749299049377441,
      "learning_rate": 5.810944637894133e-06,
      "loss": 0.9772,
      "step": 23177
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.6875617504119873,
      "learning_rate": 5.8103057214963435e-06,
      "loss": 1.0713,
      "step": 23178
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5776318311691284,
      "learning_rate": 5.809666805098554e-06,
      "loss": 0.7824,
      "step": 23179
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.517268657684326,
      "learning_rate": 5.809027888700765e-06,
      "loss": 1.1183,
      "step": 23180
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8952703475952148,
      "learning_rate": 5.808388972302975e-06,
      "loss": 0.8484,
      "step": 23181
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.5496432781219482,
      "learning_rate": 5.8077500559051855e-06,
      "loss": 1.0795,
      "step": 23182
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6928616762161255,
      "learning_rate": 5.807111139507396e-06,
      "loss": 0.4514,
      "step": 23183
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0704915523529053,
      "learning_rate": 5.806472223109607e-06,
      "loss": 0.7854,
      "step": 23184
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8049172163009644,
      "learning_rate": 5.8058333067118176e-06,
      "loss": 0.7718,
      "step": 23185
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.250864267349243,
      "learning_rate": 5.805194390314028e-06,
      "loss": 1.2065,
      "step": 23186
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.874323844909668,
      "learning_rate": 5.804555473916239e-06,
      "loss": 1.2186,
      "step": 23187
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6468462944030762,
      "learning_rate": 5.80391655751845e-06,
      "loss": 1.0369,
      "step": 23188
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2920753955841064,
      "learning_rate": 5.80327764112066e-06,
      "loss": 1.2172,
      "step": 23189
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0742266178131104,
      "learning_rate": 5.802638724722871e-06,
      "loss": 0.6536,
      "step": 23190
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.8017637729644775,
      "learning_rate": 5.801999808325081e-06,
      "loss": 1.073,
      "step": 23191
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1274993419647217,
      "learning_rate": 5.801360891927292e-06,
      "loss": 0.9429,
      "step": 23192
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4427589178085327,
      "learning_rate": 5.800721975529502e-06,
      "loss": 0.4486,
      "step": 23193
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0052201747894287,
      "learning_rate": 5.800083059131712e-06,
      "loss": 0.8914,
      "step": 23194
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.110652446746826,
      "learning_rate": 5.799444142733923e-06,
      "loss": 1.2588,
      "step": 23195
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1422529220581055,
      "learning_rate": 5.7988052263361336e-06,
      "loss": 0.9255,
      "step": 23196
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.188595771789551,
      "learning_rate": 5.798166309938344e-06,
      "loss": 1.1318,
      "step": 23197
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2703046798706055,
      "learning_rate": 5.797527393540555e-06,
      "loss": 0.8291,
      "step": 23198
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.661625862121582,
      "learning_rate": 5.796888477142766e-06,
      "loss": 0.9478,
      "step": 23199
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5289487838745117,
      "learning_rate": 5.796249560744976e-06,
      "loss": 0.9559,
      "step": 23200
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1479275226593018,
      "learning_rate": 5.795610644347187e-06,
      "loss": 0.7904,
      "step": 23201
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1957974433898926,
      "learning_rate": 5.794971727949398e-06,
      "loss": 0.8145,
      "step": 23202
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7462362051010132,
      "learning_rate": 5.7943328115516085e-06,
      "loss": 0.8615,
      "step": 23203
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1538047790527344,
      "learning_rate": 5.793693895153819e-06,
      "loss": 0.7004,
      "step": 23204
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5907621383666992,
      "learning_rate": 5.79305497875603e-06,
      "loss": 0.7993,
      "step": 23205
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9125865697860718,
      "learning_rate": 5.7924160623582406e-06,
      "loss": 0.8656,
      "step": 23206
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.563952684402466,
      "learning_rate": 5.791777145960451e-06,
      "loss": 0.8737,
      "step": 23207
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5315463542938232,
      "learning_rate": 5.791138229562662e-06,
      "loss": 0.7076,
      "step": 23208
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.747934103012085,
      "learning_rate": 5.790499313164873e-06,
      "loss": 0.9089,
      "step": 23209
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.5187253952026367,
      "learning_rate": 5.789860396767083e-06,
      "loss": 1.0732,
      "step": 23210
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.6960010528564453,
      "learning_rate": 5.789221480369294e-06,
      "loss": 1.0407,
      "step": 23211
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6989803314208984,
      "learning_rate": 5.788582563971504e-06,
      "loss": 0.7703,
      "step": 23212
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.921417236328125,
      "learning_rate": 5.787943647573715e-06,
      "loss": 0.8931,
      "step": 23213
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9105255603790283,
      "learning_rate": 5.787304731175925e-06,
      "loss": 0.8626,
      "step": 23214
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.922611713409424,
      "learning_rate": 5.786665814778136e-06,
      "loss": 1.3702,
      "step": 23215
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.467673897743225,
      "learning_rate": 5.786026898380347e-06,
      "loss": 0.779,
      "step": 23216
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.104734420776367,
      "learning_rate": 5.785387981982557e-06,
      "loss": 1.1399,
      "step": 23217
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8321748971939087,
      "learning_rate": 5.784749065584768e-06,
      "loss": 0.9017,
      "step": 23218
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.9957354068756104,
      "learning_rate": 5.784110149186979e-06,
      "loss": 1.1308,
      "step": 23219
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.3727705478668213,
      "learning_rate": 5.7834712327891895e-06,
      "loss": 0.7936,
      "step": 23220
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8847525119781494,
      "learning_rate": 5.7828323163914e-06,
      "loss": 0.7677,
      "step": 23221
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.225741863250732,
      "learning_rate": 5.782193399993611e-06,
      "loss": 1.0837,
      "step": 23222
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2500414848327637,
      "learning_rate": 5.781554483595822e-06,
      "loss": 0.9056,
      "step": 23223
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.381817102432251,
      "learning_rate": 5.780915567198032e-06,
      "loss": 0.8114,
      "step": 23224
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8824068307876587,
      "learning_rate": 5.780276650800243e-06,
      "loss": 0.9054,
      "step": 23225
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6608127355575562,
      "learning_rate": 5.779637734402454e-06,
      "loss": 0.8684,
      "step": 23226
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8582648038864136,
      "learning_rate": 5.778998818004664e-06,
      "loss": 1.0845,
      "step": 23227
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.318458080291748,
      "learning_rate": 5.778359901606875e-06,
      "loss": 1.0067,
      "step": 23228
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7893757820129395,
      "learning_rate": 5.777720985209086e-06,
      "loss": 1.0221,
      "step": 23229
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.924691677093506,
      "learning_rate": 5.777082068811296e-06,
      "loss": 0.9255,
      "step": 23230
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8103114366531372,
      "learning_rate": 5.776443152413506e-06,
      "loss": 0.7775,
      "step": 23231
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6212528944015503,
      "learning_rate": 5.775804236015717e-06,
      "loss": 0.8481,
      "step": 23232
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3898885250091553,
      "learning_rate": 5.775165319617928e-06,
      "loss": 0.5347,
      "step": 23233
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2956297397613525,
      "learning_rate": 5.7745264032201385e-06,
      "loss": 0.9529,
      "step": 23234
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0789785385131836,
      "learning_rate": 5.773887486822349e-06,
      "loss": 0.7116,
      "step": 23235
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5207120180130005,
      "learning_rate": 5.77324857042456e-06,
      "loss": 0.7287,
      "step": 23236
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9956110715866089,
      "learning_rate": 5.7726096540267706e-06,
      "loss": 0.8659,
      "step": 23237
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.354538083076477,
      "learning_rate": 5.771970737628981e-06,
      "loss": 0.7159,
      "step": 23238
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.388139486312866,
      "learning_rate": 5.771331821231192e-06,
      "loss": 1.0181,
      "step": 23239
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9784374237060547,
      "learning_rate": 5.770692904833403e-06,
      "loss": 0.9761,
      "step": 23240
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9940264225006104,
      "learning_rate": 5.770053988435613e-06,
      "loss": 0.8007,
      "step": 23241
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3064522743225098,
      "learning_rate": 5.769415072037824e-06,
      "loss": 0.7061,
      "step": 23242
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.802340269088745,
      "learning_rate": 5.768776155640035e-06,
      "loss": 0.636,
      "step": 23243
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8483223915100098,
      "learning_rate": 5.7681372392422455e-06,
      "loss": 0.9341,
      "step": 23244
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.6087517738342285,
      "learning_rate": 5.767498322844456e-06,
      "loss": 0.6597,
      "step": 23245
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.213496446609497,
      "learning_rate": 5.766859406446667e-06,
      "loss": 0.8951,
      "step": 23246
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.17378306388855,
      "learning_rate": 5.7662204900488776e-06,
      "loss": 0.9066,
      "step": 23247
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9509860277175903,
      "learning_rate": 5.765581573651087e-06,
      "loss": 0.8027,
      "step": 23248
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9481674432754517,
      "learning_rate": 5.764942657253298e-06,
      "loss": 0.8616,
      "step": 23249
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8267128467559814,
      "learning_rate": 5.764303740855509e-06,
      "loss": 0.8175,
      "step": 23250
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.008108139038086,
      "learning_rate": 5.7636648244577195e-06,
      "loss": 1.4242,
      "step": 23251
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.758538842201233,
      "learning_rate": 5.76302590805993e-06,
      "loss": 0.9313,
      "step": 23252
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6446102857589722,
      "learning_rate": 5.762386991662141e-06,
      "loss": 0.7849,
      "step": 23253
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9107097387313843,
      "learning_rate": 5.761748075264352e-06,
      "loss": 0.9111,
      "step": 23254
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.6866583824157715,
      "learning_rate": 5.761109158866562e-06,
      "loss": 0.8465,
      "step": 23255
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.724895715713501,
      "learning_rate": 5.760470242468773e-06,
      "loss": 0.8167,
      "step": 23256
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8451075553894043,
      "learning_rate": 5.759831326070984e-06,
      "loss": 0.9356,
      "step": 23257
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3273522853851318,
      "learning_rate": 5.759192409673194e-06,
      "loss": 0.4289,
      "step": 23258
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.532397747039795,
      "learning_rate": 5.758553493275405e-06,
      "loss": 1.1931,
      "step": 23259
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.443578243255615,
      "learning_rate": 5.757914576877616e-06,
      "loss": 1.2668,
      "step": 23260
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.838498830795288,
      "learning_rate": 5.7572756604798265e-06,
      "loss": 0.9002,
      "step": 23261
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9597171545028687,
      "learning_rate": 5.756636744082037e-06,
      "loss": 1.0906,
      "step": 23262
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1685056686401367,
      "learning_rate": 5.755997827684248e-06,
      "loss": 0.887,
      "step": 23263
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8980581760406494,
      "learning_rate": 5.755358911286459e-06,
      "loss": 0.9239,
      "step": 23264
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.594547748565674,
      "learning_rate": 5.754719994888669e-06,
      "loss": 0.8371,
      "step": 23265
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1788816452026367,
      "learning_rate": 5.75408107849088e-06,
      "loss": 0.9237,
      "step": 23266
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.576839804649353,
      "learning_rate": 5.75344216209309e-06,
      "loss": 0.8284,
      "step": 23267
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.767061233520508,
      "learning_rate": 5.7528032456953006e-06,
      "loss": 1.5015,
      "step": 23268
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5050146579742432,
      "learning_rate": 5.752164329297511e-06,
      "loss": 0.8674,
      "step": 23269
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9181668758392334,
      "learning_rate": 5.751525412899722e-06,
      "loss": 0.869,
      "step": 23270
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7247766256332397,
      "learning_rate": 5.750886496501933e-06,
      "loss": 0.8322,
      "step": 23271
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.178086757659912,
      "learning_rate": 5.750247580104143e-06,
      "loss": 0.9356,
      "step": 23272
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.16605544090271,
      "learning_rate": 5.749608663706354e-06,
      "loss": 0.9746,
      "step": 23273
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.046739101409912,
      "learning_rate": 5.748969747308565e-06,
      "loss": 0.884,
      "step": 23274
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.962667465209961,
      "learning_rate": 5.7483308309107755e-06,
      "loss": 0.9202,
      "step": 23275
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6553317308425903,
      "learning_rate": 5.747691914512986e-06,
      "loss": 0.8406,
      "step": 23276
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5042216777801514,
      "learning_rate": 5.747052998115197e-06,
      "loss": 0.5761,
      "step": 23277
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1274428367614746,
      "learning_rate": 5.7464140817174076e-06,
      "loss": 0.8067,
      "step": 23278
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.014165163040161,
      "learning_rate": 5.745775165319618e-06,
      "loss": 1.0087,
      "step": 23279
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0093600749969482,
      "learning_rate": 5.745136248921829e-06,
      "loss": 0.5114,
      "step": 23280
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6222823858261108,
      "learning_rate": 5.74449733252404e-06,
      "loss": 0.6841,
      "step": 23281
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4882687330245972,
      "learning_rate": 5.74385841612625e-06,
      "loss": 0.8668,
      "step": 23282
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.29810094833374,
      "learning_rate": 5.743219499728461e-06,
      "loss": 1.3514,
      "step": 23283
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.563689947128296,
      "learning_rate": 5.742580583330672e-06,
      "loss": 0.9577,
      "step": 23284
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9955177307128906,
      "learning_rate": 5.741941666932882e-06,
      "loss": 0.9287,
      "step": 23285
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.0737063884735107,
      "learning_rate": 5.741302750535092e-06,
      "loss": 0.8613,
      "step": 23286
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.524710774421692,
      "learning_rate": 5.740663834137303e-06,
      "loss": 0.7784,
      "step": 23287
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.375537395477295,
      "learning_rate": 5.740024917739514e-06,
      "loss": 0.923,
      "step": 23288
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.110011100769043,
      "learning_rate": 5.739386001341724e-06,
      "loss": 0.8907,
      "step": 23289
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7694092988967896,
      "learning_rate": 5.738747084943935e-06,
      "loss": 0.8238,
      "step": 23290
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.845059633255005,
      "learning_rate": 5.738108168546146e-06,
      "loss": 0.5371,
      "step": 23291
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7000596523284912,
      "learning_rate": 5.7374692521483565e-06,
      "loss": 0.7915,
      "step": 23292
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6131361722946167,
      "learning_rate": 5.736830335750567e-06,
      "loss": 0.8483,
      "step": 23293
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1628081798553467,
      "learning_rate": 5.736191419352778e-06,
      "loss": 0.9391,
      "step": 23294
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9902129173278809,
      "learning_rate": 5.735552502954989e-06,
      "loss": 0.8724,
      "step": 23295
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9023675918579102,
      "learning_rate": 5.734913586557199e-06,
      "loss": 0.7671,
      "step": 23296
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.239011287689209,
      "learning_rate": 5.73427467015941e-06,
      "loss": 0.7234,
      "step": 23297
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6833171844482422,
      "learning_rate": 5.733635753761621e-06,
      "loss": 0.5055,
      "step": 23298
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8000630140304565,
      "learning_rate": 5.732996837363831e-06,
      "loss": 0.942,
      "step": 23299
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.016059160232544,
      "learning_rate": 5.732357920966042e-06,
      "loss": 0.7061,
      "step": 23300
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.1418352127075195,
      "learning_rate": 5.731719004568253e-06,
      "loss": 0.9007,
      "step": 23301
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.888026714324951,
      "learning_rate": 5.7310800881704635e-06,
      "loss": 1.1387,
      "step": 23302
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0480668544769287,
      "learning_rate": 5.730441171772673e-06,
      "loss": 0.9666,
      "step": 23303
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5390052795410156,
      "learning_rate": 5.729802255374884e-06,
      "loss": 0.855,
      "step": 23304
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3250479698181152,
      "learning_rate": 5.729163338977095e-06,
      "loss": 0.8305,
      "step": 23305
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7997435331344604,
      "learning_rate": 5.7285244225793055e-06,
      "loss": 0.8701,
      "step": 23306
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9159200191497803,
      "learning_rate": 5.727885506181516e-06,
      "loss": 0.9776,
      "step": 23307
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8699696063995361,
      "learning_rate": 5.727246589783727e-06,
      "loss": 1.1003,
      "step": 23308
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5929337739944458,
      "learning_rate": 5.7266076733859376e-06,
      "loss": 0.8406,
      "step": 23309
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.707885503768921,
      "learning_rate": 5.725968756988148e-06,
      "loss": 0.8413,
      "step": 23310
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6240432262420654,
      "learning_rate": 5.725329840590359e-06,
      "loss": 0.8064,
      "step": 23311
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7063913345336914,
      "learning_rate": 5.72469092419257e-06,
      "loss": 0.8676,
      "step": 23312
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5192428827285767,
      "learning_rate": 5.72405200779478e-06,
      "loss": 0.3559,
      "step": 23313
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.5433692932128906,
      "learning_rate": 5.723413091396991e-06,
      "loss": 1.0614,
      "step": 23314
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6065572500228882,
      "learning_rate": 5.722774174999202e-06,
      "loss": 0.9177,
      "step": 23315
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.687812328338623,
      "learning_rate": 5.7221352586014125e-06,
      "loss": 0.6213,
      "step": 23316
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.521673321723938,
      "learning_rate": 5.721496342203623e-06,
      "loss": 0.8283,
      "step": 23317
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.972489595413208,
      "learning_rate": 5.720857425805834e-06,
      "loss": 0.8134,
      "step": 23318
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.214394569396973,
      "learning_rate": 5.7202185094080446e-06,
      "loss": 0.9357,
      "step": 23319
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3492399454116821,
      "learning_rate": 5.719579593010255e-06,
      "loss": 0.5013,
      "step": 23320
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.670945405960083,
      "learning_rate": 5.718940676612465e-06,
      "loss": 1.0395,
      "step": 23321
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0725350379943848,
      "learning_rate": 5.718301760214676e-06,
      "loss": 0.7722,
      "step": 23322
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0870678424835205,
      "learning_rate": 5.7176628438168865e-06,
      "loss": 0.8547,
      "step": 23323
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8284022808074951,
      "learning_rate": 5.717023927419097e-06,
      "loss": 0.9116,
      "step": 23324
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.450930118560791,
      "learning_rate": 5.716385011021308e-06,
      "loss": 0.9326,
      "step": 23325
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9927867650985718,
      "learning_rate": 5.715746094623519e-06,
      "loss": 0.901,
      "step": 23326
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.469758152961731,
      "learning_rate": 5.715107178225729e-06,
      "loss": 0.83,
      "step": 23327
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6296452283859253,
      "learning_rate": 5.71446826182794e-06,
      "loss": 0.8635,
      "step": 23328
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0043299198150635,
      "learning_rate": 5.713829345430151e-06,
      "loss": 0.8577,
      "step": 23329
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3297152519226074,
      "learning_rate": 5.7131904290323614e-06,
      "loss": 0.7564,
      "step": 23330
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.674329161643982,
      "learning_rate": 5.712551512634572e-06,
      "loss": 0.9105,
      "step": 23331
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7048250436782837,
      "learning_rate": 5.711912596236783e-06,
      "loss": 0.8973,
      "step": 23332
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.2406773567199707,
      "learning_rate": 5.7112736798389935e-06,
      "loss": 0.9092,
      "step": 23333
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0245862007141113,
      "learning_rate": 5.710634763441204e-06,
      "loss": 0.891,
      "step": 23334
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9196549654006958,
      "learning_rate": 5.709995847043415e-06,
      "loss": 0.9283,
      "step": 23335
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6810874938964844,
      "learning_rate": 5.709356930645626e-06,
      "loss": 0.8615,
      "step": 23336
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.4196362495422363,
      "learning_rate": 5.708718014247836e-06,
      "loss": 1.0297,
      "step": 23337
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.0878982543945312,
      "learning_rate": 5.708079097850047e-06,
      "loss": 0.9483,
      "step": 23338
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.6314425468444824,
      "learning_rate": 5.707440181452258e-06,
      "loss": 0.6435,
      "step": 23339
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0366761684417725,
      "learning_rate": 5.706801265054468e-06,
      "loss": 1.0026,
      "step": 23340
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.774116039276123,
      "learning_rate": 5.706162348656678e-06,
      "loss": 0.8263,
      "step": 23341
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8404275178909302,
      "learning_rate": 5.705523432258889e-06,
      "loss": 0.7615,
      "step": 23342
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0102925300598145,
      "learning_rate": 5.7048845158611e-06,
      "loss": 0.772,
      "step": 23343
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9751206636428833,
      "learning_rate": 5.70424559946331e-06,
      "loss": 0.8738,
      "step": 23344
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.43804931640625,
      "learning_rate": 5.703606683065521e-06,
      "loss": 0.4595,
      "step": 23345
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1579158306121826,
      "learning_rate": 5.702967766667732e-06,
      "loss": 0.8754,
      "step": 23346
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.922486424446106,
      "learning_rate": 5.7023288502699425e-06,
      "loss": 0.8637,
      "step": 23347
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7160152196884155,
      "learning_rate": 5.701689933872153e-06,
      "loss": 0.8718,
      "step": 23348
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2413818836212158,
      "learning_rate": 5.701051017474364e-06,
      "loss": 0.3984,
      "step": 23349
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.809791922569275,
      "learning_rate": 5.700412101076575e-06,
      "loss": 0.9185,
      "step": 23350
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4420185089111328,
      "learning_rate": 5.699773184678785e-06,
      "loss": 0.7617,
      "step": 23351
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.446397066116333,
      "learning_rate": 5.699134268280996e-06,
      "loss": 1.0246,
      "step": 23352
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8585368394851685,
      "learning_rate": 5.698495351883207e-06,
      "loss": 0.9996,
      "step": 23353
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5759997367858887,
      "learning_rate": 5.697856435485417e-06,
      "loss": 0.7403,
      "step": 23354
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5987211465835571,
      "learning_rate": 5.697217519087628e-06,
      "loss": 0.7909,
      "step": 23355
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7757991552352905,
      "learning_rate": 5.696578602689839e-06,
      "loss": 0.928,
      "step": 23356
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1398861408233643,
      "learning_rate": 5.6959396862920495e-06,
      "loss": 0.8337,
      "step": 23357
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8832341432571411,
      "learning_rate": 5.695300769894259e-06,
      "loss": 0.8674,
      "step": 23358
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.7261111736297607,
      "learning_rate": 5.69466185349647e-06,
      "loss": 0.7253,
      "step": 23359
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0249521732330322,
      "learning_rate": 5.694022937098681e-06,
      "loss": 1.0639,
      "step": 23360
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.086503267288208,
      "learning_rate": 5.6933840207008914e-06,
      "loss": 1.2392,
      "step": 23361
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.75473952293396,
      "learning_rate": 5.692745104303102e-06,
      "loss": 0.849,
      "step": 23362
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.270075559616089,
      "learning_rate": 5.692106187905313e-06,
      "loss": 0.7514,
      "step": 23363
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8861736059188843,
      "learning_rate": 5.6914672715075235e-06,
      "loss": 0.6868,
      "step": 23364
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9412646293640137,
      "learning_rate": 5.690828355109734e-06,
      "loss": 1.0764,
      "step": 23365
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.04109525680542,
      "learning_rate": 5.690189438711945e-06,
      "loss": 0.9768,
      "step": 23366
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9223718643188477,
      "learning_rate": 5.689550522314156e-06,
      "loss": 0.8861,
      "step": 23367
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.751908779144287,
      "learning_rate": 5.688911605916366e-06,
      "loss": 1.3148,
      "step": 23368
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1679253578186035,
      "learning_rate": 5.688272689518577e-06,
      "loss": 0.8869,
      "step": 23369
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3604416847229004,
      "learning_rate": 5.687633773120788e-06,
      "loss": 0.37,
      "step": 23370
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.631181716918945,
      "learning_rate": 5.6869948567229984e-06,
      "loss": 1.0183,
      "step": 23371
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.466404438018799,
      "learning_rate": 5.686355940325209e-06,
      "loss": 0.7485,
      "step": 23372
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8854955434799194,
      "learning_rate": 5.68571702392742e-06,
      "loss": 0.7515,
      "step": 23373
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9735498428344727,
      "learning_rate": 5.6850781075296305e-06,
      "loss": 0.9087,
      "step": 23374
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5993967056274414,
      "learning_rate": 5.684439191131841e-06,
      "loss": 0.7996,
      "step": 23375
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5944364070892334,
      "learning_rate": 5.683800274734051e-06,
      "loss": 0.8266,
      "step": 23376
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.182984352111816,
      "learning_rate": 5.683161358336262e-06,
      "loss": 1.0649,
      "step": 23377
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8999762535095215,
      "learning_rate": 5.6825224419384725e-06,
      "loss": 0.8715,
      "step": 23378
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2400755882263184,
      "learning_rate": 5.681883525540683e-06,
      "loss": 0.4198,
      "step": 23379
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0012433528900146,
      "learning_rate": 5.681244609142894e-06,
      "loss": 0.9114,
      "step": 23380
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.796952724456787,
      "learning_rate": 5.680605692745105e-06,
      "loss": 0.8229,
      "step": 23381
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7914806604385376,
      "learning_rate": 5.679966776347315e-06,
      "loss": 0.85,
      "step": 23382
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.8716022968292236,
      "learning_rate": 5.679327859949526e-06,
      "loss": 1.374,
      "step": 23383
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4299215078353882,
      "learning_rate": 5.678688943551737e-06,
      "loss": 0.4741,
      "step": 23384
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.8359334468841553,
      "learning_rate": 5.678050027153947e-06,
      "loss": 1.1347,
      "step": 23385
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.359933853149414,
      "learning_rate": 5.677411110756158e-06,
      "loss": 0.9357,
      "step": 23386
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9192497730255127,
      "learning_rate": 5.676772194358369e-06,
      "loss": 0.9216,
      "step": 23387
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7726606130599976,
      "learning_rate": 5.6761332779605795e-06,
      "loss": 0.9567,
      "step": 23388
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8642029762268066,
      "learning_rate": 5.67549436156279e-06,
      "loss": 1.0205,
      "step": 23389
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.626756191253662,
      "learning_rate": 5.674855445165001e-06,
      "loss": 0.8463,
      "step": 23390
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.077798843383789,
      "learning_rate": 5.674216528767212e-06,
      "loss": 0.9787,
      "step": 23391
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5617338418960571,
      "learning_rate": 5.673577612369422e-06,
      "loss": 0.8027,
      "step": 23392
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7517521381378174,
      "learning_rate": 5.672938695971633e-06,
      "loss": 0.894,
      "step": 23393
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6267651319503784,
      "learning_rate": 5.672299779573843e-06,
      "loss": 0.7836,
      "step": 23394
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.1518399715423584,
      "learning_rate": 5.6716608631760535e-06,
      "loss": 0.9617,
      "step": 23395
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.7174251079559326,
      "learning_rate": 5.671021946778264e-06,
      "loss": 0.9077,
      "step": 23396
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9846248626708984,
      "learning_rate": 5.670383030380475e-06,
      "loss": 1.1443,
      "step": 23397
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.200892448425293,
      "learning_rate": 5.669744113982686e-06,
      "loss": 0.8035,
      "step": 23398
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1153557300567627,
      "learning_rate": 5.669105197584896e-06,
      "loss": 1.1255,
      "step": 23399
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5975594520568848,
      "learning_rate": 5.668466281187107e-06,
      "loss": 0.7742,
      "step": 23400
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.4509966373443604,
      "learning_rate": 5.667827364789318e-06,
      "loss": 1.1393,
      "step": 23401
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.906315565109253,
      "learning_rate": 5.6671884483915284e-06,
      "loss": 1.0166,
      "step": 23402
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.4691479206085205,
      "learning_rate": 5.666549531993739e-06,
      "loss": 0.8085,
      "step": 23403
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.88916015625,
      "learning_rate": 5.66591061559595e-06,
      "loss": 0.8678,
      "step": 23404
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.43881893157959,
      "learning_rate": 5.6652716991981605e-06,
      "loss": 0.7872,
      "step": 23405
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.6336214542388916,
      "learning_rate": 5.664632782800371e-06,
      "loss": 0.8717,
      "step": 23406
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.173757553100586,
      "learning_rate": 5.663993866402582e-06,
      "loss": 0.7915,
      "step": 23407
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.4337189197540283,
      "learning_rate": 5.663354950004793e-06,
      "loss": 0.8594,
      "step": 23408
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.121366500854492,
      "learning_rate": 5.662716033607003e-06,
      "loss": 0.749,
      "step": 23409
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6470073461532593,
      "learning_rate": 5.662077117209214e-06,
      "loss": 0.8374,
      "step": 23410
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.99992299079895,
      "learning_rate": 5.661438200811425e-06,
      "loss": 0.8863,
      "step": 23411
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.634011745452881,
      "learning_rate": 5.6607992844136354e-06,
      "loss": 1.252,
      "step": 23412
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0092525482177734,
      "learning_rate": 5.660160368015845e-06,
      "loss": 0.8608,
      "step": 23413
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7601637840270996,
      "learning_rate": 5.659521451618056e-06,
      "loss": 0.7586,
      "step": 23414
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1066200733184814,
      "learning_rate": 5.658882535220267e-06,
      "loss": 1.0629,
      "step": 23415
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.748365044593811,
      "learning_rate": 5.658243618822477e-06,
      "loss": 0.7992,
      "step": 23416
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5272561311721802,
      "learning_rate": 5.657604702424688e-06,
      "loss": 0.7341,
      "step": 23417
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8225075006484985,
      "learning_rate": 5.656965786026899e-06,
      "loss": 0.9148,
      "step": 23418
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.472069263458252,
      "learning_rate": 5.6563268696291095e-06,
      "loss": 1.1563,
      "step": 23419
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.787636399269104,
      "learning_rate": 5.65568795323132e-06,
      "loss": 0.8782,
      "step": 23420
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.4889914989471436,
      "learning_rate": 5.655049036833531e-06,
      "loss": 0.7635,
      "step": 23421
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.476243019104004,
      "learning_rate": 5.654410120435742e-06,
      "loss": 0.8201,
      "step": 23422
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9002474546432495,
      "learning_rate": 5.653771204037952e-06,
      "loss": 0.8682,
      "step": 23423
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.6970438957214355,
      "learning_rate": 5.653132287640163e-06,
      "loss": 1.0272,
      "step": 23424
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.565290927886963,
      "learning_rate": 5.652493371242374e-06,
      "loss": 0.8311,
      "step": 23425
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9284024238586426,
      "learning_rate": 5.651854454844584e-06,
      "loss": 0.6757,
      "step": 23426
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8848381042480469,
      "learning_rate": 5.651215538446795e-06,
      "loss": 0.8455,
      "step": 23427
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7961599826812744,
      "learning_rate": 5.650576622049006e-06,
      "loss": 0.8077,
      "step": 23428
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8220667839050293,
      "learning_rate": 5.6499377056512165e-06,
      "loss": 0.8308,
      "step": 23429
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7054022550582886,
      "learning_rate": 5.649298789253427e-06,
      "loss": 0.8334,
      "step": 23430
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.518054246902466,
      "learning_rate": 5.648659872855637e-06,
      "loss": 1.7972,
      "step": 23431
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5723110437393188,
      "learning_rate": 5.648020956457848e-06,
      "loss": 0.8284,
      "step": 23432
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7821815013885498,
      "learning_rate": 5.6473820400600584e-06,
      "loss": 0.8704,
      "step": 23433
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8184561729431152,
      "learning_rate": 5.646743123662269e-06,
      "loss": 1.0689,
      "step": 23434
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5131075382232666,
      "learning_rate": 5.64610420726448e-06,
      "loss": 0.868,
      "step": 23435
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.085742950439453,
      "learning_rate": 5.6454652908666905e-06,
      "loss": 0.9277,
      "step": 23436
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2823891639709473,
      "learning_rate": 5.644826374468901e-06,
      "loss": 0.9985,
      "step": 23437
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.9954562187194824,
      "learning_rate": 5.644187458071112e-06,
      "loss": 1.2263,
      "step": 23438
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.05964994430542,
      "learning_rate": 5.643548541673323e-06,
      "loss": 0.9967,
      "step": 23439
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.742425799369812,
      "learning_rate": 5.642909625275533e-06,
      "loss": 0.846,
      "step": 23440
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.091320514678955,
      "learning_rate": 5.642270708877744e-06,
      "loss": 1.4145,
      "step": 23441
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6910943984985352,
      "learning_rate": 5.641631792479955e-06,
      "loss": 0.8981,
      "step": 23442
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5569508075714111,
      "learning_rate": 5.6409928760821654e-06,
      "loss": 0.8446,
      "step": 23443
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.051340341567993,
      "learning_rate": 5.640353959684376e-06,
      "loss": 0.7787,
      "step": 23444
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5713536739349365,
      "learning_rate": 5.639715043286587e-06,
      "loss": 0.8567,
      "step": 23445
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7401291131973267,
      "learning_rate": 5.6390761268887975e-06,
      "loss": 0.8877,
      "step": 23446
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5983293056488037,
      "learning_rate": 5.638437210491008e-06,
      "loss": 0.6546,
      "step": 23447
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8339602947235107,
      "learning_rate": 5.637798294093219e-06,
      "loss": 0.7532,
      "step": 23448
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9188998937606812,
      "learning_rate": 5.637159377695429e-06,
      "loss": 0.9854,
      "step": 23449
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5390383005142212,
      "learning_rate": 5.6365204612976395e-06,
      "loss": 0.8389,
      "step": 23450
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.23486065864563,
      "learning_rate": 5.63588154489985e-06,
      "loss": 1.2228,
      "step": 23451
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.4934206008911133,
      "learning_rate": 5.635242628502061e-06,
      "loss": 1.1084,
      "step": 23452
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9746114015579224,
      "learning_rate": 5.634603712104272e-06,
      "loss": 0.7956,
      "step": 23453
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3440093994140625,
      "learning_rate": 5.633964795706482e-06,
      "loss": 0.9358,
      "step": 23454
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1747188568115234,
      "learning_rate": 5.633325879308693e-06,
      "loss": 0.8962,
      "step": 23455
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.149265766143799,
      "learning_rate": 5.632686962910904e-06,
      "loss": 1.3342,
      "step": 23456
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1285459995269775,
      "learning_rate": 5.632048046513114e-06,
      "loss": 1.1583,
      "step": 23457
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.480541467666626,
      "learning_rate": 5.631409130115325e-06,
      "loss": 1.0464,
      "step": 23458
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2505271434783936,
      "learning_rate": 5.630770213717536e-06,
      "loss": 0.7928,
      "step": 23459
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4764232635498047,
      "learning_rate": 5.6301312973197465e-06,
      "loss": 0.8601,
      "step": 23460
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.9143147468566895,
      "learning_rate": 5.629492380921957e-06,
      "loss": 0.8746,
      "step": 23461
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6182218790054321,
      "learning_rate": 5.628853464524168e-06,
      "loss": 0.4989,
      "step": 23462
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.3054556846618652,
      "learning_rate": 5.628214548126379e-06,
      "loss": 0.8671,
      "step": 23463
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.589116096496582,
      "learning_rate": 5.627575631728589e-06,
      "loss": 0.6092,
      "step": 23464
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7333869934082031,
      "learning_rate": 5.6269367153308e-06,
      "loss": 0.7706,
      "step": 23465
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4162300825119019,
      "learning_rate": 5.626297798933011e-06,
      "loss": 0.4581,
      "step": 23466
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.022216320037842,
      "learning_rate": 5.6256588825352205e-06,
      "loss": 1.7576,
      "step": 23467
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2946724891662598,
      "learning_rate": 5.625019966137431e-06,
      "loss": 0.8306,
      "step": 23468
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3735781908035278,
      "learning_rate": 5.624381049739642e-06,
      "loss": 0.4754,
      "step": 23469
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9407479763031006,
      "learning_rate": 5.623742133341853e-06,
      "loss": 0.9335,
      "step": 23470
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.5695018768310547,
      "learning_rate": 5.623103216944063e-06,
      "loss": 0.9645,
      "step": 23471
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2620790004730225,
      "learning_rate": 5.622464300546274e-06,
      "loss": 0.6431,
      "step": 23472
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.807761549949646,
      "learning_rate": 5.621825384148485e-06,
      "loss": 0.8524,
      "step": 23473
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1240792274475098,
      "learning_rate": 5.6211864677506954e-06,
      "loss": 0.969,
      "step": 23474
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2648844718933105,
      "learning_rate": 5.620547551352906e-06,
      "loss": 0.8339,
      "step": 23475
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5913766622543335,
      "learning_rate": 5.619908634955117e-06,
      "loss": 0.7786,
      "step": 23476
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5271672010421753,
      "learning_rate": 5.6192697185573275e-06,
      "loss": 0.8809,
      "step": 23477
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.271066665649414,
      "learning_rate": 5.618630802159538e-06,
      "loss": 0.8794,
      "step": 23478
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.975048780441284,
      "learning_rate": 5.617991885761749e-06,
      "loss": 0.5751,
      "step": 23479
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7408674955368042,
      "learning_rate": 5.61735296936396e-06,
      "loss": 0.9527,
      "step": 23480
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.599395751953125,
      "learning_rate": 5.61671405296617e-06,
      "loss": 0.3839,
      "step": 23481
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8766717910766602,
      "learning_rate": 5.616075136568381e-06,
      "loss": 0.7896,
      "step": 23482
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4280407428741455,
      "learning_rate": 5.615436220170592e-06,
      "loss": 0.3909,
      "step": 23483
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7859828472137451,
      "learning_rate": 5.6147973037728024e-06,
      "loss": 0.8529,
      "step": 23484
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.8152737617492676,
      "learning_rate": 5.614158387375013e-06,
      "loss": 1.1902,
      "step": 23485
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9244221448898315,
      "learning_rate": 5.613519470977223e-06,
      "loss": 0.9551,
      "step": 23486
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.5370588302612305,
      "learning_rate": 5.612880554579434e-06,
      "loss": 0.7006,
      "step": 23487
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.61738920211792,
      "learning_rate": 5.612241638181644e-06,
      "loss": 0.8511,
      "step": 23488
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8796650171279907,
      "learning_rate": 5.611602721783855e-06,
      "loss": 0.9159,
      "step": 23489
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7507535219192505,
      "learning_rate": 5.610963805386066e-06,
      "loss": 0.8637,
      "step": 23490
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.664327621459961,
      "learning_rate": 5.6103248889882765e-06,
      "loss": 1.2011,
      "step": 23491
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.403106451034546,
      "learning_rate": 5.609685972590487e-06,
      "loss": 0.9648,
      "step": 23492
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7273279428482056,
      "learning_rate": 5.609047056192698e-06,
      "loss": 1.0636,
      "step": 23493
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3506851196289062,
      "learning_rate": 5.608408139794909e-06,
      "loss": 0.8437,
      "step": 23494
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8104180097579956,
      "learning_rate": 5.607769223397119e-06,
      "loss": 0.8342,
      "step": 23495
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5423377752304077,
      "learning_rate": 5.60713030699933e-06,
      "loss": 0.7825,
      "step": 23496
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6326017379760742,
      "learning_rate": 5.606491390601541e-06,
      "loss": 0.7988,
      "step": 23497
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.753657579421997,
      "learning_rate": 5.605852474203751e-06,
      "loss": 0.7527,
      "step": 23498
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.6142258644104,
      "learning_rate": 5.605213557805962e-06,
      "loss": 1.245,
      "step": 23499
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0859756469726562,
      "learning_rate": 5.604574641408173e-06,
      "loss": 0.8555,
      "step": 23500
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2788925170898438,
      "learning_rate": 5.6039357250103835e-06,
      "loss": 0.8961,
      "step": 23501
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.4691076278686523,
      "learning_rate": 5.603296808612594e-06,
      "loss": 0.5044,
      "step": 23502
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.718616008758545,
      "learning_rate": 5.602657892214805e-06,
      "loss": 1.4317,
      "step": 23503
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.7694554328918457,
      "learning_rate": 5.602018975817015e-06,
      "loss": 1.17,
      "step": 23504
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.071485996246338,
      "learning_rate": 5.6013800594192254e-06,
      "loss": 0.8318,
      "step": 23505
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.606410264968872,
      "learning_rate": 5.600741143021436e-06,
      "loss": 0.8913,
      "step": 23506
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9235845804214478,
      "learning_rate": 5.600102226623647e-06,
      "loss": 0.7451,
      "step": 23507
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9306676387786865,
      "learning_rate": 5.5994633102258575e-06,
      "loss": 1.0804,
      "step": 23508
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.047207832336426,
      "learning_rate": 5.598824393828068e-06,
      "loss": 1.1224,
      "step": 23509
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7539541721343994,
      "learning_rate": 5.598185477430279e-06,
      "loss": 0.7323,
      "step": 23510
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.852881669998169,
      "learning_rate": 5.59754656103249e-06,
      "loss": 0.9978,
      "step": 23511
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1312620639801025,
      "learning_rate": 5.5969076446347e-06,
      "loss": 1.0117,
      "step": 23512
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7378087043762207,
      "learning_rate": 5.596268728236911e-06,
      "loss": 0.8021,
      "step": 23513
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9540120363235474,
      "learning_rate": 5.595629811839122e-06,
      "loss": 0.8425,
      "step": 23514
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.122860908508301,
      "learning_rate": 5.5949908954413324e-06,
      "loss": 0.9196,
      "step": 23515
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9338220357894897,
      "learning_rate": 5.594351979043543e-06,
      "loss": 0.9467,
      "step": 23516
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.187523365020752,
      "learning_rate": 5.593713062645754e-06,
      "loss": 0.7218,
      "step": 23517
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7094979286193848,
      "learning_rate": 5.5930741462479645e-06,
      "loss": 0.8415,
      "step": 23518
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.876773476600647,
      "learning_rate": 5.592435229850175e-06,
      "loss": 0.7778,
      "step": 23519
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.689193606376648,
      "learning_rate": 5.591796313452386e-06,
      "loss": 0.8308,
      "step": 23520
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7922250032424927,
      "learning_rate": 5.591157397054597e-06,
      "loss": 0.79,
      "step": 23521
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1321535110473633,
      "learning_rate": 5.5905184806568065e-06,
      "loss": 1.0341,
      "step": 23522
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.607693076133728,
      "learning_rate": 5.589879564259017e-06,
      "loss": 0.8236,
      "step": 23523
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.286297559738159,
      "learning_rate": 5.589240647861228e-06,
      "loss": 1.0311,
      "step": 23524
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7906534671783447,
      "learning_rate": 5.588601731463439e-06,
      "loss": 0.871,
      "step": 23525
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8681821823120117,
      "learning_rate": 5.587962815065649e-06,
      "loss": 0.8884,
      "step": 23526
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.91240656375885,
      "learning_rate": 5.58732389866786e-06,
      "loss": 0.9735,
      "step": 23527
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.5736541748046875,
      "learning_rate": 5.586684982270071e-06,
      "loss": 0.7977,
      "step": 23528
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.631774663925171,
      "learning_rate": 5.586046065872281e-06,
      "loss": 0.7454,
      "step": 23529
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8866137266159058,
      "learning_rate": 5.585407149474492e-06,
      "loss": 0.9542,
      "step": 23530
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.107956647872925,
      "learning_rate": 5.584768233076703e-06,
      "loss": 1.1833,
      "step": 23531
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.185112714767456,
      "learning_rate": 5.5841293166789135e-06,
      "loss": 0.7431,
      "step": 23532
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.622326612472534,
      "learning_rate": 5.583490400281124e-06,
      "loss": 1.1567,
      "step": 23533
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.051060676574707,
      "learning_rate": 5.582851483883335e-06,
      "loss": 0.8677,
      "step": 23534
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1802196502685547,
      "learning_rate": 5.582212567485546e-06,
      "loss": 0.8255,
      "step": 23535
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.841986894607544,
      "learning_rate": 5.581573651087756e-06,
      "loss": 0.6702,
      "step": 23536
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5333974361419678,
      "learning_rate": 5.580934734689967e-06,
      "loss": 0.873,
      "step": 23537
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8065402507781982,
      "learning_rate": 5.580295818292178e-06,
      "loss": 0.8958,
      "step": 23538
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.798200249671936,
      "learning_rate": 5.579656901894388e-06,
      "loss": 0.8468,
      "step": 23539
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5646617412567139,
      "learning_rate": 5.579017985496599e-06,
      "loss": 0.8338,
      "step": 23540
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.036600112915039,
      "learning_rate": 5.578379069098809e-06,
      "loss": 0.8916,
      "step": 23541
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.105958938598633,
      "learning_rate": 5.57774015270102e-06,
      "loss": 1.1842,
      "step": 23542
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4089767932891846,
      "learning_rate": 5.57710123630323e-06,
      "loss": 0.4695,
      "step": 23543
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.563230037689209,
      "learning_rate": 5.576462319905441e-06,
      "loss": 1.1626,
      "step": 23544
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.491487503051758,
      "learning_rate": 5.575823403507652e-06,
      "loss": 0.8697,
      "step": 23545
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2995808124542236,
      "learning_rate": 5.5751844871098625e-06,
      "loss": 0.7841,
      "step": 23546
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.413738489151001,
      "learning_rate": 5.574545570712073e-06,
      "loss": 0.7102,
      "step": 23547
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.46346116065979,
      "learning_rate": 5.573906654314284e-06,
      "loss": 0.8284,
      "step": 23548
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.002113103866577,
      "learning_rate": 5.5732677379164946e-06,
      "loss": 0.7994,
      "step": 23549
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.8894200325012207,
      "learning_rate": 5.572628821518705e-06,
      "loss": 1.0426,
      "step": 23550
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.013422012329102,
      "learning_rate": 5.571989905120916e-06,
      "loss": 0.5875,
      "step": 23551
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5616320371627808,
      "learning_rate": 5.571350988723127e-06,
      "loss": 0.8596,
      "step": 23552
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0706799030303955,
      "learning_rate": 5.570712072325337e-06,
      "loss": 0.8403,
      "step": 23553
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1796071529388428,
      "learning_rate": 5.570073155927548e-06,
      "loss": 0.878,
      "step": 23554
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.789738416671753,
      "learning_rate": 5.569434239529759e-06,
      "loss": 0.9482,
      "step": 23555
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.4982805252075195,
      "learning_rate": 5.5687953231319695e-06,
      "loss": 0.9752,
      "step": 23556
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.593228578567505,
      "learning_rate": 5.56815640673418e-06,
      "loss": 0.9787,
      "step": 23557
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0573086738586426,
      "learning_rate": 5.567517490336391e-06,
      "loss": 0.9247,
      "step": 23558
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7261004447937012,
      "learning_rate": 5.566878573938601e-06,
      "loss": 0.8356,
      "step": 23559
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8525733947753906,
      "learning_rate": 5.566239657540811e-06,
      "loss": 0.8027,
      "step": 23560
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.037856340408325,
      "learning_rate": 5.565600741143022e-06,
      "loss": 0.9097,
      "step": 23561
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.245427370071411,
      "learning_rate": 5.564961824745233e-06,
      "loss": 0.9905,
      "step": 23562
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.820740818977356,
      "learning_rate": 5.5643229083474435e-06,
      "loss": 0.7789,
      "step": 23563
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2984585762023926,
      "learning_rate": 5.563683991949654e-06,
      "loss": 0.8151,
      "step": 23564
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8027251958847046,
      "learning_rate": 5.563045075551865e-06,
      "loss": 0.8507,
      "step": 23565
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6897900104522705,
      "learning_rate": 5.562406159154076e-06,
      "loss": 0.8961,
      "step": 23566
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.147240161895752,
      "learning_rate": 5.561767242756286e-06,
      "loss": 0.8631,
      "step": 23567
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.536613702774048,
      "learning_rate": 5.561128326358497e-06,
      "loss": 0.9075,
      "step": 23568
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.302821636199951,
      "learning_rate": 5.560489409960708e-06,
      "loss": 0.9732,
      "step": 23569
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3217302560806274,
      "learning_rate": 5.559850493562918e-06,
      "loss": 0.3887,
      "step": 23570
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6309518814086914,
      "learning_rate": 5.559211577165129e-06,
      "loss": 0.984,
      "step": 23571
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9138058423995972,
      "learning_rate": 5.55857266076734e-06,
      "loss": 0.7942,
      "step": 23572
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9348686933517456,
      "learning_rate": 5.5579337443695505e-06,
      "loss": 0.8599,
      "step": 23573
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.9717354774475098,
      "learning_rate": 5.557294827971761e-06,
      "loss": 1.086,
      "step": 23574
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.300628423690796,
      "learning_rate": 5.556655911573972e-06,
      "loss": 0.8872,
      "step": 23575
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5171676874160767,
      "learning_rate": 5.556016995176183e-06,
      "loss": 0.7473,
      "step": 23576
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7499555349349976,
      "learning_rate": 5.555378078778392e-06,
      "loss": 0.8153,
      "step": 23577
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.244570255279541,
      "learning_rate": 5.554739162380602e-06,
      "loss": 1.1703,
      "step": 23578
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.125922203063965,
      "learning_rate": 5.554100245982813e-06,
      "loss": 1.317,
      "step": 23579
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4175894260406494,
      "learning_rate": 5.553461329585024e-06,
      "loss": 0.4097,
      "step": 23580
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.619220495223999,
      "learning_rate": 5.552822413187234e-06,
      "loss": 0.8983,
      "step": 23581
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6736373901367188,
      "learning_rate": 5.552183496789445e-06,
      "loss": 0.8087,
      "step": 23582
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4140254259109497,
      "learning_rate": 5.551544580391656e-06,
      "loss": 0.3886,
      "step": 23583
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0388526916503906,
      "learning_rate": 5.5509056639938665e-06,
      "loss": 0.8613,
      "step": 23584
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8327757120132446,
      "learning_rate": 5.550266747596077e-06,
      "loss": 0.9305,
      "step": 23585
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8828909397125244,
      "learning_rate": 5.549627831198288e-06,
      "loss": 0.6831,
      "step": 23586
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3975332975387573,
      "learning_rate": 5.548988914800499e-06,
      "loss": 0.7303,
      "step": 23587
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7548353672027588,
      "learning_rate": 5.548349998402709e-06,
      "loss": 0.9003,
      "step": 23588
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.68715500831604,
      "learning_rate": 5.54771108200492e-06,
      "loss": 0.7788,
      "step": 23589
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4367761611938477,
      "learning_rate": 5.54707216560713e-06,
      "loss": 0.7796,
      "step": 23590
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.244431257247925,
      "learning_rate": 5.5464332492093406e-06,
      "loss": 0.8253,
      "step": 23591
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.639385938644409,
      "learning_rate": 5.545794332811551e-06,
      "loss": 0.5909,
      "step": 23592
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7752838134765625,
      "learning_rate": 5.545155416413762e-06,
      "loss": 0.7403,
      "step": 23593
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7939565181732178,
      "learning_rate": 5.544516500015973e-06,
      "loss": 0.9066,
      "step": 23594
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.094978094100952,
      "learning_rate": 5.543877583618183e-06,
      "loss": 0.8482,
      "step": 23595
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.4176573753356934,
      "learning_rate": 5.543238667220394e-06,
      "loss": 0.9261,
      "step": 23596
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.6870675086975098,
      "learning_rate": 5.542599750822605e-06,
      "loss": 1.0839,
      "step": 23597
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.258094072341919,
      "learning_rate": 5.5419608344248155e-06,
      "loss": 0.5288,
      "step": 23598
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7774895429611206,
      "learning_rate": 5.541321918027026e-06,
      "loss": 0.7805,
      "step": 23599
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.660546898841858,
      "learning_rate": 5.540683001629237e-06,
      "loss": 0.4242,
      "step": 23600
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.244683027267456,
      "learning_rate": 5.5400440852314476e-06,
      "loss": 0.4991,
      "step": 23601
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9785206317901611,
      "learning_rate": 5.539405168833658e-06,
      "loss": 0.8723,
      "step": 23602
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.9950010776519775,
      "learning_rate": 5.538766252435869e-06,
      "loss": 1.1653,
      "step": 23603
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.989065170288086,
      "learning_rate": 5.53812733603808e-06,
      "loss": 0.7552,
      "step": 23604
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8464655876159668,
      "learning_rate": 5.53748841964029e-06,
      "loss": 0.8842,
      "step": 23605
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7997002601623535,
      "learning_rate": 5.536849503242501e-06,
      "loss": 0.9172,
      "step": 23606
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.399437427520752,
      "learning_rate": 5.536210586844712e-06,
      "loss": 0.8217,
      "step": 23607
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8748619556427002,
      "learning_rate": 5.535571670446922e-06,
      "loss": 0.9659,
      "step": 23608
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.606688380241394,
      "learning_rate": 5.534932754049132e-06,
      "loss": 0.8751,
      "step": 23609
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.4611265659332275,
      "learning_rate": 5.534293837651343e-06,
      "loss": 0.8617,
      "step": 23610
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7791924476623535,
      "learning_rate": 5.533654921253554e-06,
      "loss": 0.9059,
      "step": 23611
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.720880389213562,
      "learning_rate": 5.533016004855764e-06,
      "loss": 0.8032,
      "step": 23612
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.7283694744110107,
      "learning_rate": 5.532377088457975e-06,
      "loss": 0.9546,
      "step": 23613
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.0885095596313477,
      "learning_rate": 5.531738172060186e-06,
      "loss": 0.8842,
      "step": 23614
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4503467082977295,
      "learning_rate": 5.5310992556623965e-06,
      "loss": 0.7751,
      "step": 23615
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.99620521068573,
      "learning_rate": 5.530460339264607e-06,
      "loss": 0.8721,
      "step": 23616
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.948439121246338,
      "learning_rate": 5.529821422866818e-06,
      "loss": 1.0864,
      "step": 23617
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.392054557800293,
      "learning_rate": 5.529182506469029e-06,
      "loss": 1.0089,
      "step": 23618
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.8253402709960938,
      "learning_rate": 5.528543590071239e-06,
      "loss": 1.3077,
      "step": 23619
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.001383066177368,
      "learning_rate": 5.52790467367345e-06,
      "loss": 1.1353,
      "step": 23620
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.873257875442505,
      "learning_rate": 5.527265757275661e-06,
      "loss": 1.0055,
      "step": 23621
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8640843629837036,
      "learning_rate": 5.526626840877871e-06,
      "loss": 0.9577,
      "step": 23622
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6436759233474731,
      "learning_rate": 5.525987924480082e-06,
      "loss": 0.8747,
      "step": 23623
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.127229928970337,
      "learning_rate": 5.525349008082293e-06,
      "loss": 1.2699,
      "step": 23624
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7537869215011597,
      "learning_rate": 5.5247100916845035e-06,
      "loss": 0.8822,
      "step": 23625
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9349660873413086,
      "learning_rate": 5.524071175286713e-06,
      "loss": 1.2455,
      "step": 23626
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8876851797103882,
      "learning_rate": 5.523432258888924e-06,
      "loss": 0.9292,
      "step": 23627
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9689702987670898,
      "learning_rate": 5.522793342491135e-06,
      "loss": 0.7871,
      "step": 23628
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6099915504455566,
      "learning_rate": 5.5221544260933455e-06,
      "loss": 0.822,
      "step": 23629
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3761801719665527,
      "learning_rate": 5.521515509695556e-06,
      "loss": 0.7717,
      "step": 23630
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1059467792510986,
      "learning_rate": 5.520876593297767e-06,
      "loss": 0.8452,
      "step": 23631
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.146536350250244,
      "learning_rate": 5.5202376768999776e-06,
      "loss": 0.8868,
      "step": 23632
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.6794700622558594,
      "learning_rate": 5.519598760502188e-06,
      "loss": 0.7981,
      "step": 23633
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7895110845565796,
      "learning_rate": 5.518959844104399e-06,
      "loss": 0.786,
      "step": 23634
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.727716088294983,
      "learning_rate": 5.51832092770661e-06,
      "loss": 0.9127,
      "step": 23635
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5456151962280273,
      "learning_rate": 5.51768201130882e-06,
      "loss": 0.8708,
      "step": 23636
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.995627522468567,
      "learning_rate": 5.517043094911031e-06,
      "loss": 1.2206,
      "step": 23637
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4830942153930664,
      "learning_rate": 5.516404178513242e-06,
      "loss": 0.928,
      "step": 23638
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1753458976745605,
      "learning_rate": 5.5157652621154525e-06,
      "loss": 0.8468,
      "step": 23639
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.3787851333618164,
      "learning_rate": 5.515126345717663e-06,
      "loss": 1.0368,
      "step": 23640
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8137800693511963,
      "learning_rate": 5.514487429319874e-06,
      "loss": 0.783,
      "step": 23641
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.177947521209717,
      "learning_rate": 5.5138485129220846e-06,
      "loss": 0.8582,
      "step": 23642
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.066180944442749,
      "learning_rate": 5.513209596524295e-06,
      "loss": 0.7193,
      "step": 23643
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4064360857009888,
      "learning_rate": 5.512570680126505e-06,
      "loss": 0.3152,
      "step": 23644
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.439720630645752,
      "learning_rate": 5.511931763728716e-06,
      "loss": 1.18,
      "step": 23645
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7056742906570435,
      "learning_rate": 5.5112928473309265e-06,
      "loss": 0.9441,
      "step": 23646
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6315523386001587,
      "learning_rate": 5.510653930933137e-06,
      "loss": 0.8538,
      "step": 23647
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9771116971969604,
      "learning_rate": 5.510015014535348e-06,
      "loss": 0.8092,
      "step": 23648
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.052134037017822,
      "learning_rate": 5.509376098137559e-06,
      "loss": 1.5649,
      "step": 23649
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8112845420837402,
      "learning_rate": 5.508737181739769e-06,
      "loss": 1.1444,
      "step": 23650
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6666743755340576,
      "learning_rate": 5.50809826534198e-06,
      "loss": 0.8102,
      "step": 23651
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6002304553985596,
      "learning_rate": 5.507459348944191e-06,
      "loss": 0.404,
      "step": 23652
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.432098865509033,
      "learning_rate": 5.506820432546401e-06,
      "loss": 1.0068,
      "step": 23653
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9722274541854858,
      "learning_rate": 5.506181516148612e-06,
      "loss": 0.9198,
      "step": 23654
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1191320419311523,
      "learning_rate": 5.505542599750823e-06,
      "loss": 0.8429,
      "step": 23655
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7750797271728516,
      "learning_rate": 5.5049036833530335e-06,
      "loss": 0.8332,
      "step": 23656
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.446581840515137,
      "learning_rate": 5.504264766955244e-06,
      "loss": 1.0889,
      "step": 23657
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3655591011047363,
      "learning_rate": 5.503625850557455e-06,
      "loss": 0.9031,
      "step": 23658
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.898569107055664,
      "learning_rate": 5.502986934159666e-06,
      "loss": 0.8614,
      "step": 23659
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.412774682044983,
      "learning_rate": 5.502348017761876e-06,
      "loss": 0.7657,
      "step": 23660
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5652104616165161,
      "learning_rate": 5.501709101364087e-06,
      "loss": 0.789,
      "step": 23661
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1573667526245117,
      "learning_rate": 5.501070184966298e-06,
      "loss": 0.8981,
      "step": 23662
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0362472534179688,
      "learning_rate": 5.5004312685685076e-06,
      "loss": 0.8347,
      "step": 23663
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8600637912750244,
      "learning_rate": 5.499792352170718e-06,
      "loss": 1.1455,
      "step": 23664
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.5896430015563965,
      "learning_rate": 5.499153435772929e-06,
      "loss": 1.4903,
      "step": 23665
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1984519958496094,
      "learning_rate": 5.49851451937514e-06,
      "loss": 1.0107,
      "step": 23666
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.644697666168213,
      "learning_rate": 5.49787560297735e-06,
      "loss": 0.4646,
      "step": 23667
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6865389347076416,
      "learning_rate": 5.497236686579561e-06,
      "loss": 0.8396,
      "step": 23668
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.870925784111023,
      "learning_rate": 5.496597770181772e-06,
      "loss": 0.7637,
      "step": 23669
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.978469729423523,
      "learning_rate": 5.4959588537839825e-06,
      "loss": 0.9109,
      "step": 23670
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.975080132484436,
      "learning_rate": 5.495319937386193e-06,
      "loss": 0.8767,
      "step": 23671
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9068466424942017,
      "learning_rate": 5.494681020988404e-06,
      "loss": 0.935,
      "step": 23672
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7241227626800537,
      "learning_rate": 5.4940421045906146e-06,
      "loss": 0.7524,
      "step": 23673
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4871622323989868,
      "learning_rate": 5.493403188192825e-06,
      "loss": 0.8361,
      "step": 23674
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9035128355026245,
      "learning_rate": 5.492764271795036e-06,
      "loss": 1.0066,
      "step": 23675
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.017817735671997,
      "learning_rate": 5.492125355397247e-06,
      "loss": 0.8998,
      "step": 23676
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4228547811508179,
      "learning_rate": 5.491486438999457e-06,
      "loss": 0.8276,
      "step": 23677
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6623152494430542,
      "learning_rate": 5.490847522601668e-06,
      "loss": 0.7876,
      "step": 23678
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8796006441116333,
      "learning_rate": 5.490208606203879e-06,
      "loss": 0.7635,
      "step": 23679
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.4428646564483643,
      "learning_rate": 5.4895696898060895e-06,
      "loss": 0.8936,
      "step": 23680
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6595067977905273,
      "learning_rate": 5.488930773408299e-06,
      "loss": 1.2073,
      "step": 23681
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.766230821609497,
      "learning_rate": 5.48829185701051e-06,
      "loss": 0.7475,
      "step": 23682
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5250375270843506,
      "learning_rate": 5.487652940612721e-06,
      "loss": 0.8452,
      "step": 23683
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7588955163955688,
      "learning_rate": 5.487014024214931e-06,
      "loss": 0.8248,
      "step": 23684
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.002476692199707,
      "learning_rate": 5.486375107817142e-06,
      "loss": 0.8447,
      "step": 23685
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6615564823150635,
      "learning_rate": 5.485736191419353e-06,
      "loss": 0.8083,
      "step": 23686
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8387514352798462,
      "learning_rate": 5.4850972750215635e-06,
      "loss": 0.7748,
      "step": 23687
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0549588203430176,
      "learning_rate": 5.484458358623774e-06,
      "loss": 0.9352,
      "step": 23688
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0348563194274902,
      "learning_rate": 5.483819442225985e-06,
      "loss": 0.8782,
      "step": 23689
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1457278728485107,
      "learning_rate": 5.483180525828196e-06,
      "loss": 0.8913,
      "step": 23690
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.3536932468414307,
      "learning_rate": 5.482541609430406e-06,
      "loss": 1.1156,
      "step": 23691
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6498228311538696,
      "learning_rate": 5.481902693032617e-06,
      "loss": 0.4433,
      "step": 23692
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1205668449401855,
      "learning_rate": 5.481263776634828e-06,
      "loss": 0.9159,
      "step": 23693
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.8712668418884277,
      "learning_rate": 5.480624860237038e-06,
      "loss": 0.7279,
      "step": 23694
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.660529613494873,
      "learning_rate": 5.479985943839249e-06,
      "loss": 1.1514,
      "step": 23695
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.468467712402344,
      "learning_rate": 5.47934702744146e-06,
      "loss": 1.2009,
      "step": 23696
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4978246688842773,
      "learning_rate": 5.4787081110436705e-06,
      "loss": 0.7171,
      "step": 23697
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.631539821624756,
      "learning_rate": 5.478069194645881e-06,
      "loss": 0.9055,
      "step": 23698
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.09738302230835,
      "learning_rate": 5.477430278248091e-06,
      "loss": 1.2376,
      "step": 23699
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3522083759307861,
      "learning_rate": 5.476791361850302e-06,
      "loss": 0.8164,
      "step": 23700
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.062047958374023,
      "learning_rate": 5.4761524454525125e-06,
      "loss": 1.19,
      "step": 23701
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.581980586051941,
      "learning_rate": 5.475513529054723e-06,
      "loss": 0.3538,
      "step": 23702
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4333772659301758,
      "learning_rate": 5.474874612656934e-06,
      "loss": 0.7716,
      "step": 23703
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2869932651519775,
      "learning_rate": 5.4742356962591446e-06,
      "loss": 0.8209,
      "step": 23704
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1232407093048096,
      "learning_rate": 5.473596779861355e-06,
      "loss": 0.8469,
      "step": 23705
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.249624013900757,
      "learning_rate": 5.472957863463566e-06,
      "loss": 0.8458,
      "step": 23706
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.422628402709961,
      "learning_rate": 5.472318947065777e-06,
      "loss": 0.8038,
      "step": 23707
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0759217739105225,
      "learning_rate": 5.471680030667987e-06,
      "loss": 0.8689,
      "step": 23708
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.357457160949707,
      "learning_rate": 5.471041114270198e-06,
      "loss": 0.825,
      "step": 23709
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2498176097869873,
      "learning_rate": 5.470402197872409e-06,
      "loss": 0.9608,
      "step": 23710
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.401066780090332,
      "learning_rate": 5.4697632814746195e-06,
      "loss": 0.953,
      "step": 23711
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.633772373199463,
      "learning_rate": 5.46912436507683e-06,
      "loss": 0.8898,
      "step": 23712
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3167446851730347,
      "learning_rate": 5.468485448679041e-06,
      "loss": 0.8624,
      "step": 23713
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6134206056594849,
      "learning_rate": 5.4678465322812516e-06,
      "loss": 0.7536,
      "step": 23714
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9620672464370728,
      "learning_rate": 5.467207615883462e-06,
      "loss": 0.9138,
      "step": 23715
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7963168621063232,
      "learning_rate": 5.466568699485673e-06,
      "loss": 0.8477,
      "step": 23716
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8706992864608765,
      "learning_rate": 5.465929783087883e-06,
      "loss": 0.9265,
      "step": 23717
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6040968894958496,
      "learning_rate": 5.4652908666900935e-06,
      "loss": 0.7262,
      "step": 23718
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.74294114112854,
      "learning_rate": 5.464651950292304e-06,
      "loss": 1.3984,
      "step": 23719
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4618271589279175,
      "learning_rate": 5.464013033894515e-06,
      "loss": 0.8249,
      "step": 23720
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5915112495422363,
      "learning_rate": 5.463374117496726e-06,
      "loss": 0.7485,
      "step": 23721
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7020949125289917,
      "learning_rate": 5.462735201098936e-06,
      "loss": 0.8697,
      "step": 23722
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.629406929016113,
      "learning_rate": 5.462096284701147e-06,
      "loss": 0.8951,
      "step": 23723
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.482933521270752,
      "learning_rate": 5.461457368303358e-06,
      "loss": 0.7998,
      "step": 23724
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4025993347167969,
      "learning_rate": 5.4608184519055684e-06,
      "loss": 0.4358,
      "step": 23725
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.440247893333435,
      "learning_rate": 5.460179535507779e-06,
      "loss": 0.3405,
      "step": 23726
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1330833435058594,
      "learning_rate": 5.45954061910999e-06,
      "loss": 0.8731,
      "step": 23727
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.513889789581299,
      "learning_rate": 5.4589017027122005e-06,
      "loss": 1.7059,
      "step": 23728
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.6795361042022705,
      "learning_rate": 5.458262786314411e-06,
      "loss": 0.9487,
      "step": 23729
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.567786455154419,
      "learning_rate": 5.457623869916622e-06,
      "loss": 0.8096,
      "step": 23730
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4982714653015137,
      "learning_rate": 5.456984953518833e-06,
      "loss": 0.8498,
      "step": 23731
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.4025063514709473,
      "learning_rate": 5.456346037121043e-06,
      "loss": 1.0346,
      "step": 23732
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4903916120529175,
      "learning_rate": 5.455707120723254e-06,
      "loss": 0.3768,
      "step": 23733
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7341605424880981,
      "learning_rate": 5.455068204325465e-06,
      "loss": 0.8781,
      "step": 23734
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6533315181732178,
      "learning_rate": 5.4544292879276754e-06,
      "loss": 0.884,
      "step": 23735
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.89125919342041,
      "learning_rate": 5.453790371529885e-06,
      "loss": 0.9077,
      "step": 23736
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.271878957748413,
      "learning_rate": 5.453151455132096e-06,
      "loss": 0.3834,
      "step": 23737
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.986359715461731,
      "learning_rate": 5.452512538734307e-06,
      "loss": 0.8134,
      "step": 23738
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.574027419090271,
      "learning_rate": 5.451873622336517e-06,
      "loss": 0.7543,
      "step": 23739
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5772819519042969,
      "learning_rate": 5.451234705938728e-06,
      "loss": 0.894,
      "step": 23740
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6164991855621338,
      "learning_rate": 5.450595789540939e-06,
      "loss": 0.8391,
      "step": 23741
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4711120128631592,
      "learning_rate": 5.4499568731431495e-06,
      "loss": 0.8115,
      "step": 23742
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.818422794342041,
      "learning_rate": 5.44931795674536e-06,
      "loss": 0.9165,
      "step": 23743
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4880481958389282,
      "learning_rate": 5.448679040347571e-06,
      "loss": 0.807,
      "step": 23744
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.224440336227417,
      "learning_rate": 5.448040123949782e-06,
      "loss": 0.8106,
      "step": 23745
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8200464248657227,
      "learning_rate": 5.447401207551992e-06,
      "loss": 0.8673,
      "step": 23746
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4741795063018799,
      "learning_rate": 5.446762291154203e-06,
      "loss": 0.9121,
      "step": 23747
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0645196437835693,
      "learning_rate": 5.446123374756414e-06,
      "loss": 0.8773,
      "step": 23748
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7902854681015015,
      "learning_rate": 5.445484458358624e-06,
      "loss": 0.8545,
      "step": 23749
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3408315181732178,
      "learning_rate": 5.444845541960835e-06,
      "loss": 0.4745,
      "step": 23750
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.703235149383545,
      "learning_rate": 5.444206625563046e-06,
      "loss": 0.6712,
      "step": 23751
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5557092428207397,
      "learning_rate": 5.4435677091652565e-06,
      "loss": 0.4848,
      "step": 23752
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.76970636844635,
      "learning_rate": 5.442928792767467e-06,
      "loss": 0.7375,
      "step": 23753
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.719429612159729,
      "learning_rate": 5.442289876369677e-06,
      "loss": 0.8361,
      "step": 23754
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7958980798721313,
      "learning_rate": 5.441650959971888e-06,
      "loss": 0.7477,
      "step": 23755
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9638665914535522,
      "learning_rate": 5.4410120435740984e-06,
      "loss": 0.8215,
      "step": 23756
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.235931873321533,
      "learning_rate": 5.440373127176309e-06,
      "loss": 1.0485,
      "step": 23757
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1688084602355957,
      "learning_rate": 5.43973421077852e-06,
      "loss": 0.784,
      "step": 23758
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.071106433868408,
      "learning_rate": 5.4390952943807305e-06,
      "loss": 0.884,
      "step": 23759
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.991411566734314,
      "learning_rate": 5.438456377982941e-06,
      "loss": 0.8756,
      "step": 23760
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8744866847991943,
      "learning_rate": 5.437817461585152e-06,
      "loss": 0.8426,
      "step": 23761
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.064589738845825,
      "learning_rate": 5.437178545187363e-06,
      "loss": 0.9007,
      "step": 23762
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.867255926132202,
      "learning_rate": 5.436539628789573e-06,
      "loss": 0.9752,
      "step": 23763
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.740410566329956,
      "learning_rate": 5.435900712391784e-06,
      "loss": 0.8993,
      "step": 23764
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0103952884674072,
      "learning_rate": 5.435261795993995e-06,
      "loss": 0.8816,
      "step": 23765
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8197435140609741,
      "learning_rate": 5.4346228795962054e-06,
      "loss": 0.7707,
      "step": 23766
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.850313663482666,
      "learning_rate": 5.433983963198416e-06,
      "loss": 0.8913,
      "step": 23767
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5505729913711548,
      "learning_rate": 5.433345046800627e-06,
      "loss": 0.823,
      "step": 23768
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3766937255859375,
      "learning_rate": 5.4327061304028375e-06,
      "loss": 0.929,
      "step": 23769
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.2560243606567383,
      "learning_rate": 5.432067214005048e-06,
      "loss": 0.8779,
      "step": 23770
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.627835988998413,
      "learning_rate": 5.431428297607259e-06,
      "loss": 0.7725,
      "step": 23771
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8477946519851685,
      "learning_rate": 5.430789381209469e-06,
      "loss": 0.7689,
      "step": 23772
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6145508289337158,
      "learning_rate": 5.4301504648116795e-06,
      "loss": 0.7453,
      "step": 23773
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.8760132789611816,
      "learning_rate": 5.42951154841389e-06,
      "loss": 0.6681,
      "step": 23774
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7995576858520508,
      "learning_rate": 5.428872632016101e-06,
      "loss": 0.952,
      "step": 23775
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7130823135375977,
      "learning_rate": 5.428233715618312e-06,
      "loss": 0.918,
      "step": 23776
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.090541124343872,
      "learning_rate": 5.427594799220522e-06,
      "loss": 1.3795,
      "step": 23777
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.125694751739502,
      "learning_rate": 5.426955882822733e-06,
      "loss": 1.0168,
      "step": 23778
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.5878162384033203,
      "learning_rate": 5.426316966424944e-06,
      "loss": 0.7236,
      "step": 23779
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1657564640045166,
      "learning_rate": 5.425678050027154e-06,
      "loss": 0.7836,
      "step": 23780
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9123384952545166,
      "learning_rate": 5.425039133629365e-06,
      "loss": 0.7384,
      "step": 23781
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.4266321659088135,
      "learning_rate": 5.424400217231576e-06,
      "loss": 1.1876,
      "step": 23782
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.021183729171753,
      "learning_rate": 5.4237613008337865e-06,
      "loss": 0.9423,
      "step": 23783
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6029067039489746,
      "learning_rate": 5.423122384435997e-06,
      "loss": 0.451,
      "step": 23784
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6647608280181885,
      "learning_rate": 5.422483468038208e-06,
      "loss": 0.8385,
      "step": 23785
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.379372239112854,
      "learning_rate": 5.421844551640419e-06,
      "loss": 0.7578,
      "step": 23786
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.267413854598999,
      "learning_rate": 5.421205635242629e-06,
      "loss": 0.8033,
      "step": 23787
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.822951078414917,
      "learning_rate": 5.42056671884484e-06,
      "loss": 0.8447,
      "step": 23788
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.237524509429932,
      "learning_rate": 5.419927802447051e-06,
      "loss": 1.1844,
      "step": 23789
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.031219482421875,
      "learning_rate": 5.4192888860492605e-06,
      "loss": 0.9211,
      "step": 23790
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8601094484329224,
      "learning_rate": 5.418649969651471e-06,
      "loss": 0.8405,
      "step": 23791
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1953611373901367,
      "learning_rate": 5.418011053253682e-06,
      "loss": 0.7246,
      "step": 23792
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3255674839019775,
      "learning_rate": 5.417372136855893e-06,
      "loss": 0.9138,
      "step": 23793
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6326885223388672,
      "learning_rate": 5.416733220458103e-06,
      "loss": 0.7778,
      "step": 23794
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7973612546920776,
      "learning_rate": 5.416094304060314e-06,
      "loss": 0.9621,
      "step": 23795
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.294247627258301,
      "learning_rate": 5.415455387662525e-06,
      "loss": 1.0828,
      "step": 23796
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.38497257232666,
      "learning_rate": 5.4148164712647354e-06,
      "loss": 1.0577,
      "step": 23797
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9510397911071777,
      "learning_rate": 5.414177554866946e-06,
      "loss": 0.8821,
      "step": 23798
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.1556239128112793,
      "learning_rate": 5.413538638469157e-06,
      "loss": 1.2107,
      "step": 23799
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.322835445404053,
      "learning_rate": 5.4128997220713675e-06,
      "loss": 1.1787,
      "step": 23800
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.8887124061584473,
      "learning_rate": 5.412260805673578e-06,
      "loss": 0.9244,
      "step": 23801
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.1768248081207275,
      "learning_rate": 5.411621889275789e-06,
      "loss": 1.042,
      "step": 23802
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6578396558761597,
      "learning_rate": 5.410982972878e-06,
      "loss": 0.7038,
      "step": 23803
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0857126712799072,
      "learning_rate": 5.41034405648021e-06,
      "loss": 0.8718,
      "step": 23804
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7153109312057495,
      "learning_rate": 5.409705140082421e-06,
      "loss": 0.9134,
      "step": 23805
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.569941520690918,
      "learning_rate": 5.409066223684632e-06,
      "loss": 0.7942,
      "step": 23806
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9538114070892334,
      "learning_rate": 5.4084273072868424e-06,
      "loss": 0.9332,
      "step": 23807
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8410619497299194,
      "learning_rate": 5.407788390889053e-06,
      "loss": 1.196,
      "step": 23808
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7756078243255615,
      "learning_rate": 5.407149474491263e-06,
      "loss": 1.1064,
      "step": 23809
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3333933353424072,
      "learning_rate": 5.406510558093474e-06,
      "loss": 0.4028,
      "step": 23810
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9494116306304932,
      "learning_rate": 5.405871641695684e-06,
      "loss": 0.7735,
      "step": 23811
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8643341064453125,
      "learning_rate": 5.405232725297895e-06,
      "loss": 0.8702,
      "step": 23812
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.969224452972412,
      "learning_rate": 5.404593808900106e-06,
      "loss": 1.0796,
      "step": 23813
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.5142335891723633,
      "learning_rate": 5.4039548925023165e-06,
      "loss": 1.1575,
      "step": 23814
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.284184217453003,
      "learning_rate": 5.403315976104527e-06,
      "loss": 0.7561,
      "step": 23815
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.117563247680664,
      "learning_rate": 5.402677059706738e-06,
      "loss": 0.9608,
      "step": 23816
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4365650415420532,
      "learning_rate": 5.402038143308949e-06,
      "loss": 0.7588,
      "step": 23817
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.132544994354248,
      "learning_rate": 5.401399226911159e-06,
      "loss": 1.1228,
      "step": 23818
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.9111177921295166,
      "learning_rate": 5.40076031051337e-06,
      "loss": 0.9212,
      "step": 23819
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.828879952430725,
      "learning_rate": 5.400121394115581e-06,
      "loss": 1.1175,
      "step": 23820
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0322158336639404,
      "learning_rate": 5.399482477717791e-06,
      "loss": 0.9243,
      "step": 23821
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.9974255561828613,
      "learning_rate": 5.398843561320002e-06,
      "loss": 1.5059,
      "step": 23822
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8652026653289795,
      "learning_rate": 5.398204644922213e-06,
      "loss": 1.2119,
      "step": 23823
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.9978418350219727,
      "learning_rate": 5.3975657285244235e-06,
      "loss": 0.9642,
      "step": 23824
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1163103580474854,
      "learning_rate": 5.396926812126634e-06,
      "loss": 0.9701,
      "step": 23825
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.200362682342529,
      "learning_rate": 5.396287895728845e-06,
      "loss": 1.2766,
      "step": 23826
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.58176326751709,
      "learning_rate": 5.395648979331055e-06,
      "loss": 0.9021,
      "step": 23827
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.15986967086792,
      "learning_rate": 5.3950100629332654e-06,
      "loss": 0.91,
      "step": 23828
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.854981780052185,
      "learning_rate": 5.394371146535476e-06,
      "loss": 0.81,
      "step": 23829
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6663565635681152,
      "learning_rate": 5.393732230137687e-06,
      "loss": 0.8146,
      "step": 23830
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.15477991104126,
      "learning_rate": 5.3930933137398975e-06,
      "loss": 1.2027,
      "step": 23831
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0526082515716553,
      "learning_rate": 5.392454397342108e-06,
      "loss": 0.7513,
      "step": 23832
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.266021966934204,
      "learning_rate": 5.391815480944319e-06,
      "loss": 0.8896,
      "step": 23833
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9986363649368286,
      "learning_rate": 5.39117656454653e-06,
      "loss": 0.78,
      "step": 23834
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7085613012313843,
      "learning_rate": 5.39053764814874e-06,
      "loss": 0.7025,
      "step": 23835
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.025143623352051,
      "learning_rate": 5.389898731750951e-06,
      "loss": 0.8627,
      "step": 23836
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.748921513557434,
      "learning_rate": 5.389259815353162e-06,
      "loss": 0.8474,
      "step": 23837
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7355965375900269,
      "learning_rate": 5.3886208989553724e-06,
      "loss": 0.7447,
      "step": 23838
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5822263956069946,
      "learning_rate": 5.387981982557583e-06,
      "loss": 0.7719,
      "step": 23839
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.2753987312316895,
      "learning_rate": 5.387343066159794e-06,
      "loss": 1.051,
      "step": 23840
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4576561450958252,
      "learning_rate": 5.3867041497620045e-06,
      "loss": 0.3847,
      "step": 23841
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6494559049606323,
      "learning_rate": 5.386065233364215e-06,
      "loss": 0.8545,
      "step": 23842
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9960747957229614,
      "learning_rate": 5.385426316966426e-06,
      "loss": 0.8592,
      "step": 23843
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9357409477233887,
      "learning_rate": 5.384787400568637e-06,
      "loss": 0.923,
      "step": 23844
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6609989404678345,
      "learning_rate": 5.3841484841708465e-06,
      "loss": 0.7764,
      "step": 23845
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7070529460906982,
      "learning_rate": 5.383509567773057e-06,
      "loss": 0.8337,
      "step": 23846
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.712568998336792,
      "learning_rate": 5.382870651375268e-06,
      "loss": 0.613,
      "step": 23847
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0294137001037598,
      "learning_rate": 5.382231734977479e-06,
      "loss": 0.8617,
      "step": 23848
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.514327049255371,
      "learning_rate": 5.381592818579689e-06,
      "loss": 0.9045,
      "step": 23849
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4548207521438599,
      "learning_rate": 5.3809539021819e-06,
      "loss": 0.7426,
      "step": 23850
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.8598520755767822,
      "learning_rate": 5.380314985784111e-06,
      "loss": 0.9309,
      "step": 23851
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.322868824005127,
      "learning_rate": 5.379676069386321e-06,
      "loss": 0.847,
      "step": 23852
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6473413705825806,
      "learning_rate": 5.379037152988532e-06,
      "loss": 0.7989,
      "step": 23853
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.259204626083374,
      "learning_rate": 5.378398236590743e-06,
      "loss": 1.1521,
      "step": 23854
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.105116128921509,
      "learning_rate": 5.3777593201929535e-06,
      "loss": 0.7272,
      "step": 23855
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3667609691619873,
      "learning_rate": 5.377120403795164e-06,
      "loss": 0.5452,
      "step": 23856
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.681107759475708,
      "learning_rate": 5.376481487397375e-06,
      "loss": 0.6177,
      "step": 23857
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9067978858947754,
      "learning_rate": 5.375842570999586e-06,
      "loss": 0.919,
      "step": 23858
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2813642024993896,
      "learning_rate": 5.375203654601796e-06,
      "loss": 0.4101,
      "step": 23859
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6215457916259766,
      "learning_rate": 5.374564738204007e-06,
      "loss": 0.9531,
      "step": 23860
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.650179386138916,
      "learning_rate": 5.373925821806218e-06,
      "loss": 0.7305,
      "step": 23861
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.848494529724121,
      "learning_rate": 5.373286905408428e-06,
      "loss": 0.8245,
      "step": 23862
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.266902208328247,
      "learning_rate": 5.372647989010638e-06,
      "loss": 0.9817,
      "step": 23863
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.403911590576172,
      "learning_rate": 5.372009072612849e-06,
      "loss": 0.8952,
      "step": 23864
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.678771734237671,
      "learning_rate": 5.37137015621506e-06,
      "loss": 0.8681,
      "step": 23865
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.731051206588745,
      "learning_rate": 5.37073123981727e-06,
      "loss": 0.9609,
      "step": 23866
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7371264696121216,
      "learning_rate": 5.370092323419481e-06,
      "loss": 0.8121,
      "step": 23867
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.622483730316162,
      "learning_rate": 5.369453407021692e-06,
      "loss": 0.8797,
      "step": 23868
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.3554422855377197,
      "learning_rate": 5.3688144906239024e-06,
      "loss": 0.8692,
      "step": 23869
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.762192964553833,
      "learning_rate": 5.368175574226113e-06,
      "loss": 0.8481,
      "step": 23870
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7631083726882935,
      "learning_rate": 5.367536657828324e-06,
      "loss": 0.8332,
      "step": 23871
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.489048719406128,
      "learning_rate": 5.3668977414305345e-06,
      "loss": 0.6911,
      "step": 23872
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4286012649536133,
      "learning_rate": 5.366258825032745e-06,
      "loss": 0.6579,
      "step": 23873
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.2688000202178955,
      "learning_rate": 5.365619908634956e-06,
      "loss": 0.9383,
      "step": 23874
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1498806476593018,
      "learning_rate": 5.364980992237167e-06,
      "loss": 0.9631,
      "step": 23875
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.198537588119507,
      "learning_rate": 5.364342075839377e-06,
      "loss": 0.9839,
      "step": 23876
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9905328750610352,
      "learning_rate": 5.363703159441588e-06,
      "loss": 0.8118,
      "step": 23877
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.44057035446167,
      "learning_rate": 5.363064243043799e-06,
      "loss": 0.3595,
      "step": 23878
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.074359893798828,
      "learning_rate": 5.3624253266460094e-06,
      "loss": 0.8993,
      "step": 23879
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8070236444473267,
      "learning_rate": 5.36178641024822e-06,
      "loss": 0.8538,
      "step": 23880
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.988738775253296,
      "learning_rate": 5.361147493850431e-06,
      "loss": 0.4548,
      "step": 23881
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4601101875305176,
      "learning_rate": 5.360508577452641e-06,
      "loss": 0.8382,
      "step": 23882
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.531778335571289,
      "learning_rate": 5.359869661054851e-06,
      "loss": 1.0421,
      "step": 23883
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5549081563949585,
      "learning_rate": 5.359230744657062e-06,
      "loss": 0.9264,
      "step": 23884
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8150347471237183,
      "learning_rate": 5.358591828259273e-06,
      "loss": 0.972,
      "step": 23885
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6901354789733887,
      "learning_rate": 5.3579529118614835e-06,
      "loss": 0.8068,
      "step": 23886
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7014830112457275,
      "learning_rate": 5.357313995463694e-06,
      "loss": 0.9272,
      "step": 23887
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8236825466156006,
      "learning_rate": 5.356675079065905e-06,
      "loss": 0.8336,
      "step": 23888
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0657198429107666,
      "learning_rate": 5.356036162668116e-06,
      "loss": 0.7673,
      "step": 23889
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.6556694507598877,
      "learning_rate": 5.355397246270326e-06,
      "loss": 0.8812,
      "step": 23890
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.856532335281372,
      "learning_rate": 5.354758329872537e-06,
      "loss": 0.9088,
      "step": 23891
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1089508533477783,
      "learning_rate": 5.354119413474748e-06,
      "loss": 0.8934,
      "step": 23892
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.8422799110412598,
      "learning_rate": 5.353480497076958e-06,
      "loss": 1.0884,
      "step": 23893
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.1515607833862305,
      "learning_rate": 5.352841580679169e-06,
      "loss": 0.8028,
      "step": 23894
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.751287579536438,
      "learning_rate": 5.35220266428138e-06,
      "loss": 0.7725,
      "step": 23895
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4232351779937744,
      "learning_rate": 5.3515637478835905e-06,
      "loss": 1.0824,
      "step": 23896
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4877076148986816,
      "learning_rate": 5.350924831485801e-06,
      "loss": 0.558,
      "step": 23897
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5731799602508545,
      "learning_rate": 5.350285915088012e-06,
      "loss": 0.7773,
      "step": 23898
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5101457834243774,
      "learning_rate": 5.349646998690223e-06,
      "loss": 0.8503,
      "step": 23899
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8343125581741333,
      "learning_rate": 5.3490080822924324e-06,
      "loss": 0.7588,
      "step": 23900
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.033064126968384,
      "learning_rate": 5.348369165894643e-06,
      "loss": 0.9237,
      "step": 23901
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9699420928955078,
      "learning_rate": 5.347730249496854e-06,
      "loss": 1.2766,
      "step": 23902
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6433876752853394,
      "learning_rate": 5.3470913330990645e-06,
      "loss": 0.7791,
      "step": 23903
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8084323406219482,
      "learning_rate": 5.346452416701275e-06,
      "loss": 0.914,
      "step": 23904
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.3963541984558105,
      "learning_rate": 5.345813500303486e-06,
      "loss": 0.9401,
      "step": 23905
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.88239324092865,
      "learning_rate": 5.345174583905697e-06,
      "loss": 0.9463,
      "step": 23906
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.336958408355713,
      "learning_rate": 5.344535667507907e-06,
      "loss": 0.7434,
      "step": 23907
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.807032346725464,
      "learning_rate": 5.343896751110118e-06,
      "loss": 0.5462,
      "step": 23908
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6630797386169434,
      "learning_rate": 5.343257834712329e-06,
      "loss": 0.8588,
      "step": 23909
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.242523431777954,
      "learning_rate": 5.3426189183145394e-06,
      "loss": 0.9889,
      "step": 23910
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.7887043952941895,
      "learning_rate": 5.34198000191675e-06,
      "loss": 1.0279,
      "step": 23911
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7104207277297974,
      "learning_rate": 5.341341085518961e-06,
      "loss": 0.8393,
      "step": 23912
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9398729801177979,
      "learning_rate": 5.3407021691211715e-06,
      "loss": 0.7437,
      "step": 23913
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5424033403396606,
      "learning_rate": 5.340063252723382e-06,
      "loss": 0.8393,
      "step": 23914
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6599053144454956,
      "learning_rate": 5.339424336325593e-06,
      "loss": 0.8598,
      "step": 23915
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8202126026153564,
      "learning_rate": 5.338785419927804e-06,
      "loss": 0.929,
      "step": 23916
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.2335336208343506,
      "learning_rate": 5.338146503530014e-06,
      "loss": 1.0075,
      "step": 23917
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8027724027633667,
      "learning_rate": 5.337507587132224e-06,
      "loss": 0.9201,
      "step": 23918
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8374700546264648,
      "learning_rate": 5.336868670734435e-06,
      "loss": 0.8177,
      "step": 23919
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8512423038482666,
      "learning_rate": 5.336229754336646e-06,
      "loss": 0.7136,
      "step": 23920
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.32521390914917,
      "learning_rate": 5.335590837938856e-06,
      "loss": 1.2781,
      "step": 23921
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.539383888244629,
      "learning_rate": 5.334951921541067e-06,
      "loss": 0.8275,
      "step": 23922
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.6380200386047363,
      "learning_rate": 5.334313005143278e-06,
      "loss": 1.3242,
      "step": 23923
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.513288736343384,
      "learning_rate": 5.333674088745488e-06,
      "loss": 1.0898,
      "step": 23924
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.056042432785034,
      "learning_rate": 5.333035172347699e-06,
      "loss": 0.7993,
      "step": 23925
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1668660640716553,
      "learning_rate": 5.33239625594991e-06,
      "loss": 0.9638,
      "step": 23926
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3098185062408447,
      "learning_rate": 5.3317573395521205e-06,
      "loss": 0.7121,
      "step": 23927
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.2296574115753174,
      "learning_rate": 5.331118423154331e-06,
      "loss": 0.8167,
      "step": 23928
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3963799476623535,
      "learning_rate": 5.330479506756542e-06,
      "loss": 0.7291,
      "step": 23929
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6550520658493042,
      "learning_rate": 5.329840590358753e-06,
      "loss": 0.8162,
      "step": 23930
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.626774311065674,
      "learning_rate": 5.329201673960963e-06,
      "loss": 0.8255,
      "step": 23931
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1454241275787354,
      "learning_rate": 5.328562757563174e-06,
      "loss": 0.9789,
      "step": 23932
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.2755355834960938,
      "learning_rate": 5.327923841165385e-06,
      "loss": 0.5874,
      "step": 23933
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8627907037734985,
      "learning_rate": 5.327284924767595e-06,
      "loss": 0.9639,
      "step": 23934
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.296391248703003,
      "learning_rate": 5.326646008369806e-06,
      "loss": 1.1204,
      "step": 23935
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5330866575241089,
      "learning_rate": 5.326007091972016e-06,
      "loss": 0.8656,
      "step": 23936
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.971801996231079,
      "learning_rate": 5.325368175574227e-06,
      "loss": 0.8792,
      "step": 23937
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.274639844894409,
      "learning_rate": 5.324729259176437e-06,
      "loss": 1.0175,
      "step": 23938
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.159446954727173,
      "learning_rate": 5.324090342778648e-06,
      "loss": 0.888,
      "step": 23939
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.608638048171997,
      "learning_rate": 5.323451426380859e-06,
      "loss": 0.491,
      "step": 23940
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6582669019699097,
      "learning_rate": 5.3228125099830695e-06,
      "loss": 0.8172,
      "step": 23941
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.9279587268829346,
      "learning_rate": 5.32217359358528e-06,
      "loss": 0.8627,
      "step": 23942
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7552552223205566,
      "learning_rate": 5.321534677187491e-06,
      "loss": 0.8649,
      "step": 23943
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7722443342208862,
      "learning_rate": 5.3208957607897016e-06,
      "loss": 0.743,
      "step": 23944
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8813745975494385,
      "learning_rate": 5.320256844391912e-06,
      "loss": 0.8857,
      "step": 23945
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.802996039390564,
      "learning_rate": 5.319617927994123e-06,
      "loss": 0.8796,
      "step": 23946
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.1173832416534424,
      "learning_rate": 5.318979011596334e-06,
      "loss": 0.8309,
      "step": 23947
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5229859352111816,
      "learning_rate": 5.318340095198544e-06,
      "loss": 0.8431,
      "step": 23948
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4645968675613403,
      "learning_rate": 5.317701178800755e-06,
      "loss": 0.822,
      "step": 23949
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.8081376552581787,
      "learning_rate": 5.317062262402966e-06,
      "loss": 0.6606,
      "step": 23950
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0849432945251465,
      "learning_rate": 5.3164233460051765e-06,
      "loss": 0.8978,
      "step": 23951
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6556888818740845,
      "learning_rate": 5.315784429607387e-06,
      "loss": 0.8906,
      "step": 23952
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.5889110565185547,
      "learning_rate": 5.315145513209598e-06,
      "loss": 0.793,
      "step": 23953
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9923229217529297,
      "learning_rate": 5.3145065968118086e-06,
      "loss": 1.1795,
      "step": 23954
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5329055786132812,
      "learning_rate": 5.313867680414018e-06,
      "loss": 0.8614,
      "step": 23955
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.256498336791992,
      "learning_rate": 5.313228764016229e-06,
      "loss": 0.6621,
      "step": 23956
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9392471313476562,
      "learning_rate": 5.31258984761844e-06,
      "loss": 0.8568,
      "step": 23957
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9137004613876343,
      "learning_rate": 5.3119509312206505e-06,
      "loss": 0.9443,
      "step": 23958
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.9483957290649414,
      "learning_rate": 5.311312014822861e-06,
      "loss": 0.5828,
      "step": 23959
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.042440414428711,
      "learning_rate": 5.310673098425072e-06,
      "loss": 0.9152,
      "step": 23960
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.161303520202637,
      "learning_rate": 5.310034182027283e-06,
      "loss": 1.4545,
      "step": 23961
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4909145832061768,
      "learning_rate": 5.309395265629493e-06,
      "loss": 0.7658,
      "step": 23962
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9132559299468994,
      "learning_rate": 5.308756349231704e-06,
      "loss": 0.6757,
      "step": 23963
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9833159446716309,
      "learning_rate": 5.308117432833914e-06,
      "loss": 0.8845,
      "step": 23964
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7164298295974731,
      "learning_rate": 5.3074785164361246e-06,
      "loss": 0.8981,
      "step": 23965
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4849514961242676,
      "learning_rate": 5.306839600038335e-06,
      "loss": 0.9631,
      "step": 23966
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9044184684753418,
      "learning_rate": 5.306200683640545e-06,
      "loss": 0.823,
      "step": 23967
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.279798984527588,
      "learning_rate": 5.305561767242756e-06,
      "loss": 0.9053,
      "step": 23968
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.887195110321045,
      "learning_rate": 5.3049228508449665e-06,
      "loss": 1.0043,
      "step": 23969
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9167338609695435,
      "learning_rate": 5.304283934447177e-06,
      "loss": 0.9443,
      "step": 23970
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5279051065444946,
      "learning_rate": 5.303645018049388e-06,
      "loss": 0.4034,
      "step": 23971
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5651988983154297,
      "learning_rate": 5.303006101651599e-06,
      "loss": 0.8793,
      "step": 23972
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8013888597488403,
      "learning_rate": 5.302367185253809e-06,
      "loss": 0.8113,
      "step": 23973
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.889939785003662,
      "learning_rate": 5.30172826885602e-06,
      "loss": 0.8088,
      "step": 23974
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5078810453414917,
      "learning_rate": 5.301089352458231e-06,
      "loss": 0.4119,
      "step": 23975
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3947741985321045,
      "learning_rate": 5.300450436060441e-06,
      "loss": 0.8898,
      "step": 23976
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.348109245300293,
      "learning_rate": 5.299811519662652e-06,
      "loss": 0.9,
      "step": 23977
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0971810817718506,
      "learning_rate": 5.299172603264863e-06,
      "loss": 0.8908,
      "step": 23978
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8820321559906006,
      "learning_rate": 5.2985336868670735e-06,
      "loss": 0.9024,
      "step": 23979
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5778356790542603,
      "learning_rate": 5.297894770469284e-06,
      "loss": 0.6382,
      "step": 23980
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.716693162918091,
      "learning_rate": 5.297255854071495e-06,
      "loss": 1.086,
      "step": 23981
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5764573812484741,
      "learning_rate": 5.296616937673706e-06,
      "loss": 0.7747,
      "step": 23982
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.320296287536621,
      "learning_rate": 5.295978021275916e-06,
      "loss": 0.8684,
      "step": 23983
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5384340286254883,
      "learning_rate": 5.295339104878127e-06,
      "loss": 0.8323,
      "step": 23984
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7104787826538086,
      "learning_rate": 5.294700188480338e-06,
      "loss": 0.8148,
      "step": 23985
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7537177801132202,
      "learning_rate": 5.2940612720825476e-06,
      "loss": 0.8326,
      "step": 23986
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0407040119171143,
      "learning_rate": 5.293422355684758e-06,
      "loss": 0.7912,
      "step": 23987
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.857156991958618,
      "learning_rate": 5.292783439286969e-06,
      "loss": 1.0301,
      "step": 23988
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.045774221420288,
      "learning_rate": 5.29214452288918e-06,
      "loss": 0.7224,
      "step": 23989
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4821383953094482,
      "learning_rate": 5.29150560649139e-06,
      "loss": 0.7556,
      "step": 23990
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7174158096313477,
      "learning_rate": 5.290866690093601e-06,
      "loss": 0.9251,
      "step": 23991
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.798433542251587,
      "learning_rate": 5.290227773695812e-06,
      "loss": 0.8104,
      "step": 23992
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.944714307785034,
      "learning_rate": 5.2895888572980225e-06,
      "loss": 0.7303,
      "step": 23993
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.5794947147369385,
      "learning_rate": 5.288949940900233e-06,
      "loss": 0.8852,
      "step": 23994
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.784592866897583,
      "learning_rate": 5.288311024502444e-06,
      "loss": 0.8141,
      "step": 23995
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0485846996307373,
      "learning_rate": 5.2876721081046546e-06,
      "loss": 1.3139,
      "step": 23996
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6254311800003052,
      "learning_rate": 5.287033191706865e-06,
      "loss": 0.8267,
      "step": 23997
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9984630346298218,
      "learning_rate": 5.286394275309076e-06,
      "loss": 0.9007,
      "step": 23998
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9303264617919922,
      "learning_rate": 5.285755358911287e-06,
      "loss": 1.0132,
      "step": 23999
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9737454652786255,
      "learning_rate": 5.285116442513497e-06,
      "loss": 1.1874,
      "step": 24000
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.03680419921875,
      "learning_rate": 5.284477526115708e-06,
      "loss": 1.3338,
      "step": 24001
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9040019512176514,
      "learning_rate": 5.283838609717919e-06,
      "loss": 0.9905,
      "step": 24002
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.5960886478424072,
      "learning_rate": 5.2831996933201295e-06,
      "loss": 0.8698,
      "step": 24003
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.438098669052124,
      "learning_rate": 5.282560776922339e-06,
      "loss": 0.7263,
      "step": 24004
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9376386404037476,
      "learning_rate": 5.28192186052455e-06,
      "loss": 1.0705,
      "step": 24005
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3932468891143799,
      "learning_rate": 5.281282944126761e-06,
      "loss": 0.3877,
      "step": 24006
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5920895338058472,
      "learning_rate": 5.280644027728971e-06,
      "loss": 0.8398,
      "step": 24007
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9869014024734497,
      "learning_rate": 5.280005111331182e-06,
      "loss": 0.8755,
      "step": 24008
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8908812999725342,
      "learning_rate": 5.279366194933393e-06,
      "loss": 0.9363,
      "step": 24009
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7176141738891602,
      "learning_rate": 5.2787272785356035e-06,
      "loss": 0.7952,
      "step": 24010
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0848755836486816,
      "learning_rate": 5.278088362137814e-06,
      "loss": 0.7889,
      "step": 24011
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.0044965744018555,
      "learning_rate": 5.277449445740025e-06,
      "loss": 1.0806,
      "step": 24012
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.966005802154541,
      "learning_rate": 5.276810529342236e-06,
      "loss": 1.1034,
      "step": 24013
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6540154218673706,
      "learning_rate": 5.276171612944446e-06,
      "loss": 0.8339,
      "step": 24014
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9164555072784424,
      "learning_rate": 5.275532696546657e-06,
      "loss": 0.8796,
      "step": 24015
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7559808492660522,
      "learning_rate": 5.274893780148868e-06,
      "loss": 0.8054,
      "step": 24016
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.053077697753906,
      "learning_rate": 5.274254863751078e-06,
      "loss": 1.1165,
      "step": 24017
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0363059043884277,
      "learning_rate": 5.273615947353289e-06,
      "loss": 0.8287,
      "step": 24018
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7033047676086426,
      "learning_rate": 5.2729770309555e-06,
      "loss": 0.8086,
      "step": 24019
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.161062002182007,
      "learning_rate": 5.2723381145577105e-06,
      "loss": 0.967,
      "step": 24020
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.546623706817627,
      "learning_rate": 5.271699198159921e-06,
      "loss": 1.2974,
      "step": 24021
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3794047832489014,
      "learning_rate": 5.271060281762131e-06,
      "loss": 0.9484,
      "step": 24022
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4420294761657715,
      "learning_rate": 5.270421365364342e-06,
      "loss": 0.4464,
      "step": 24023
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8298521041870117,
      "learning_rate": 5.2697824489665525e-06,
      "loss": 0.8314,
      "step": 24024
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.405190944671631,
      "learning_rate": 5.269143532568763e-06,
      "loss": 0.7398,
      "step": 24025
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9554893970489502,
      "learning_rate": 5.268504616170974e-06,
      "loss": 0.9445,
      "step": 24026
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9453901052474976,
      "learning_rate": 5.2678656997731846e-06,
      "loss": 0.7535,
      "step": 24027
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.982288122177124,
      "learning_rate": 5.267226783375395e-06,
      "loss": 1.1728,
      "step": 24028
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.397322416305542,
      "learning_rate": 5.266587866977606e-06,
      "loss": 0.9498,
      "step": 24029
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9265469312667847,
      "learning_rate": 5.265948950579817e-06,
      "loss": 0.9513,
      "step": 24030
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.46872615814209,
      "learning_rate": 5.265310034182027e-06,
      "loss": 0.9406,
      "step": 24031
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7954111099243164,
      "learning_rate": 5.264671117784238e-06,
      "loss": 0.8196,
      "step": 24032
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.911719799041748,
      "learning_rate": 5.264032201386449e-06,
      "loss": 0.8532,
      "step": 24033
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.546479344367981,
      "learning_rate": 5.2633932849886595e-06,
      "loss": 0.748,
      "step": 24034
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5618654489517212,
      "learning_rate": 5.26275436859087e-06,
      "loss": 0.8765,
      "step": 24035
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.191894292831421,
      "learning_rate": 5.262115452193081e-06,
      "loss": 0.6496,
      "step": 24036
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4325671195983887,
      "learning_rate": 5.2614765357952916e-06,
      "loss": 0.397,
      "step": 24037
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.262498378753662,
      "learning_rate": 5.260837619397502e-06,
      "loss": 0.9685,
      "step": 24038
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5978680849075317,
      "learning_rate": 5.260198702999713e-06,
      "loss": 1.0335,
      "step": 24039
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7219135761260986,
      "learning_rate": 5.259559786601923e-06,
      "loss": 0.7621,
      "step": 24040
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.837604522705078,
      "learning_rate": 5.2589208702041335e-06,
      "loss": 0.79,
      "step": 24041
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.401700019836426,
      "learning_rate": 5.258281953806344e-06,
      "loss": 1.5713,
      "step": 24042
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6898452043533325,
      "learning_rate": 5.257643037408555e-06,
      "loss": 0.929,
      "step": 24043
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9955806732177734,
      "learning_rate": 5.257004121010766e-06,
      "loss": 0.6044,
      "step": 24044
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.427004098892212,
      "learning_rate": 5.256365204612976e-06,
      "loss": 0.888,
      "step": 24045
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.253478527069092,
      "learning_rate": 5.255726288215187e-06,
      "loss": 0.891,
      "step": 24046
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.211848258972168,
      "learning_rate": 5.255087371817398e-06,
      "loss": 1.2883,
      "step": 24047
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9189270734786987,
      "learning_rate": 5.254448455419608e-06,
      "loss": 0.8626,
      "step": 24048
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.119645118713379,
      "learning_rate": 5.253809539021819e-06,
      "loss": 0.9225,
      "step": 24049
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3680756092071533,
      "learning_rate": 5.25317062262403e-06,
      "loss": 0.7712,
      "step": 24050
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.004591226577759,
      "learning_rate": 5.2525317062262405e-06,
      "loss": 0.791,
      "step": 24051
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2859208583831787,
      "learning_rate": 5.251892789828451e-06,
      "loss": 1.0852,
      "step": 24052
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6795316934585571,
      "learning_rate": 5.251253873430662e-06,
      "loss": 0.8914,
      "step": 24053
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.606462001800537,
      "learning_rate": 5.250614957032873e-06,
      "loss": 1.0944,
      "step": 24054
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1702709197998047,
      "learning_rate": 5.249976040635083e-06,
      "loss": 0.849,
      "step": 24055
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6929205656051636,
      "learning_rate": 5.249337124237294e-06,
      "loss": 0.8245,
      "step": 24056
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7121766805648804,
      "learning_rate": 5.248698207839505e-06,
      "loss": 0.8551,
      "step": 24057
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7723280191421509,
      "learning_rate": 5.248059291441715e-06,
      "loss": 0.9197,
      "step": 24058
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.166112422943115,
      "learning_rate": 5.247420375043925e-06,
      "loss": 1.2898,
      "step": 24059
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.596514344215393,
      "learning_rate": 5.246781458646136e-06,
      "loss": 0.8652,
      "step": 24060
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7544441223144531,
      "learning_rate": 5.246142542248347e-06,
      "loss": 0.792,
      "step": 24061
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.832576870918274,
      "learning_rate": 5.245503625850557e-06,
      "loss": 0.8644,
      "step": 24062
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6956071853637695,
      "learning_rate": 5.244864709452768e-06,
      "loss": 0.7048,
      "step": 24063
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4722963571548462,
      "learning_rate": 5.244225793054979e-06,
      "loss": 0.7969,
      "step": 24064
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9404486417770386,
      "learning_rate": 5.2435868766571895e-06,
      "loss": 0.8487,
      "step": 24065
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7295490503311157,
      "learning_rate": 5.2429479602594e-06,
      "loss": 0.717,
      "step": 24066
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2903637886047363,
      "learning_rate": 5.242309043861611e-06,
      "loss": 0.4532,
      "step": 24067
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5964534282684326,
      "learning_rate": 5.2416701274638216e-06,
      "loss": 0.7235,
      "step": 24068
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.130154609680176,
      "learning_rate": 5.241031211066032e-06,
      "loss": 1.059,
      "step": 24069
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6951547861099243,
      "learning_rate": 5.240392294668243e-06,
      "loss": 0.7804,
      "step": 24070
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8440091609954834,
      "learning_rate": 5.239753378270454e-06,
      "loss": 0.84,
      "step": 24071
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.4999709129333496,
      "learning_rate": 5.239114461872664e-06,
      "loss": 1.049,
      "step": 24072
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6436511278152466,
      "learning_rate": 5.238475545474875e-06,
      "loss": 0.9033,
      "step": 24073
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.168044090270996,
      "learning_rate": 5.237836629077086e-06,
      "loss": 0.4638,
      "step": 24074
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4703497886657715,
      "learning_rate": 5.2371977126792965e-06,
      "loss": 1.0457,
      "step": 24075
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.635222315788269,
      "learning_rate": 5.236558796281507e-06,
      "loss": 0.831,
      "step": 24076
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8952091932296753,
      "learning_rate": 5.235919879883717e-06,
      "loss": 0.9421,
      "step": 24077
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4785033464431763,
      "learning_rate": 5.235280963485928e-06,
      "loss": 0.7738,
      "step": 24078
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.50472092628479,
      "learning_rate": 5.2346420470881384e-06,
      "loss": 0.7935,
      "step": 24079
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8323795795440674,
      "learning_rate": 5.234003130690349e-06,
      "loss": 0.8281,
      "step": 24080
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.927289366722107,
      "learning_rate": 5.23336421429256e-06,
      "loss": 1.1722,
      "step": 24081
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5237133502960205,
      "learning_rate": 5.2327252978947705e-06,
      "loss": 0.8438,
      "step": 24082
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7575794458389282,
      "learning_rate": 5.232086381496981e-06,
      "loss": 0.9283,
      "step": 24083
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.738726258277893,
      "learning_rate": 5.231447465099192e-06,
      "loss": 0.8971,
      "step": 24084
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.559921145439148,
      "learning_rate": 5.230808548701403e-06,
      "loss": 0.9091,
      "step": 24085
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.8624045848846436,
      "learning_rate": 5.230169632303613e-06,
      "loss": 0.7689,
      "step": 24086
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8544143438339233,
      "learning_rate": 5.229530715905824e-06,
      "loss": 0.8585,
      "step": 24087
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8972002267837524,
      "learning_rate": 5.228891799508035e-06,
      "loss": 0.9066,
      "step": 24088
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.512320637702942,
      "learning_rate": 5.2282528831102454e-06,
      "loss": 0.8312,
      "step": 24089
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8414480686187744,
      "learning_rate": 5.227613966712456e-06,
      "loss": 0.8624,
      "step": 24090
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.688788414001465,
      "learning_rate": 5.226975050314667e-06,
      "loss": 1.4369,
      "step": 24091
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.435746192932129,
      "learning_rate": 5.2263361339168775e-06,
      "loss": 0.7426,
      "step": 24092
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.873809814453125,
      "learning_rate": 5.225697217519088e-06,
      "loss": 0.982,
      "step": 24093
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.152153491973877,
      "learning_rate": 5.225058301121299e-06,
      "loss": 0.7696,
      "step": 24094
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5453580617904663,
      "learning_rate": 5.224419384723509e-06,
      "loss": 0.7642,
      "step": 24095
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.634448528289795,
      "learning_rate": 5.2237804683257195e-06,
      "loss": 1.0791,
      "step": 24096
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.658272743225098,
      "learning_rate": 5.22314155192793e-06,
      "loss": 1.5214,
      "step": 24097
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8233193159103394,
      "learning_rate": 5.222502635530141e-06,
      "loss": 0.8161,
      "step": 24098
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2079887390136719,
      "learning_rate": 5.221863719132352e-06,
      "loss": 0.4248,
      "step": 24099
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9943052530288696,
      "learning_rate": 5.221224802734562e-06,
      "loss": 0.8175,
      "step": 24100
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.149305820465088,
      "learning_rate": 5.220585886336773e-06,
      "loss": 0.826,
      "step": 24101
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.080012321472168,
      "learning_rate": 5.219946969938984e-06,
      "loss": 0.8572,
      "step": 24102
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.059333324432373,
      "learning_rate": 5.219308053541194e-06,
      "loss": 1.1185,
      "step": 24103
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9024578332901,
      "learning_rate": 5.218669137143405e-06,
      "loss": 0.8298,
      "step": 24104
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5057421922683716,
      "learning_rate": 5.218030220745616e-06,
      "loss": 0.8059,
      "step": 24105
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4366238117218018,
      "learning_rate": 5.2173913043478265e-06,
      "loss": 0.8805,
      "step": 24106
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.335419178009033,
      "learning_rate": 5.216752387950037e-06,
      "loss": 0.9714,
      "step": 24107
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3089611530303955,
      "learning_rate": 5.216113471552248e-06,
      "loss": 1.0536,
      "step": 24108
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.153825521469116,
      "learning_rate": 5.215474555154459e-06,
      "loss": 1.0688,
      "step": 24109
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.0954155921936035,
      "learning_rate": 5.214835638756669e-06,
      "loss": 0.9429,
      "step": 24110
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.687283754348755,
      "learning_rate": 5.21419672235888e-06,
      "loss": 0.6058,
      "step": 24111
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3827141523361206,
      "learning_rate": 5.213557805961091e-06,
      "loss": 0.7923,
      "step": 24112
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9288008213043213,
      "learning_rate": 5.2129188895633005e-06,
      "loss": 0.8857,
      "step": 24113
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.062007188796997,
      "learning_rate": 5.212279973165511e-06,
      "loss": 0.8929,
      "step": 24114
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.844667673110962,
      "learning_rate": 5.211641056767722e-06,
      "loss": 0.7971,
      "step": 24115
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.656808853149414,
      "learning_rate": 5.211002140369933e-06,
      "loss": 0.8896,
      "step": 24116
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.923363208770752,
      "learning_rate": 5.210363223972143e-06,
      "loss": 0.8087,
      "step": 24117
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.8611230850219727,
      "learning_rate": 5.209724307574354e-06,
      "loss": 0.9549,
      "step": 24118
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6987868547439575,
      "learning_rate": 5.209085391176565e-06,
      "loss": 0.7771,
      "step": 24119
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3384690284729004,
      "learning_rate": 5.2084464747787754e-06,
      "loss": 0.7524,
      "step": 24120
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8352762460708618,
      "learning_rate": 5.207807558380986e-06,
      "loss": 0.8257,
      "step": 24121
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9545724391937256,
      "learning_rate": 5.207168641983197e-06,
      "loss": 0.8171,
      "step": 24122
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.897512435913086,
      "learning_rate": 5.2065297255854075e-06,
      "loss": 0.8595,
      "step": 24123
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.3649463653564453,
      "learning_rate": 5.205890809187618e-06,
      "loss": 0.8492,
      "step": 24124
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9607120752334595,
      "learning_rate": 5.205251892789829e-06,
      "loss": 0.8054,
      "step": 24125
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.729448676109314,
      "learning_rate": 5.20461297639204e-06,
      "loss": 0.8383,
      "step": 24126
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.5058481693267822,
      "learning_rate": 5.20397405999425e-06,
      "loss": 1.0095,
      "step": 24127
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6591264009475708,
      "learning_rate": 5.203335143596461e-06,
      "loss": 0.8583,
      "step": 24128
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0179760456085205,
      "learning_rate": 5.202696227198672e-06,
      "loss": 0.8382,
      "step": 24129
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.7345030307769775,
      "learning_rate": 5.2020573108008824e-06,
      "loss": 0.9496,
      "step": 24130
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6802014112472534,
      "learning_rate": 5.201418394403093e-06,
      "loss": 0.936,
      "step": 24131
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.6745357513427734,
      "learning_rate": 5.200779478005303e-06,
      "loss": 0.9826,
      "step": 24132
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.496345281600952,
      "learning_rate": 5.200140561607514e-06,
      "loss": 1.0847,
      "step": 24133
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9762204885482788,
      "learning_rate": 5.199501645209724e-06,
      "loss": 0.9058,
      "step": 24134
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.042398452758789,
      "learning_rate": 5.198862728811935e-06,
      "loss": 0.997,
      "step": 24135
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6555838584899902,
      "learning_rate": 5.198223812414146e-06,
      "loss": 0.8777,
      "step": 24136
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.396691799163818,
      "learning_rate": 5.1975848960163565e-06,
      "loss": 1.1024,
      "step": 24137
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5526856184005737,
      "learning_rate": 5.196945979618567e-06,
      "loss": 0.8475,
      "step": 24138
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0456972122192383,
      "learning_rate": 5.196307063220778e-06,
      "loss": 1.1865,
      "step": 24139
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.670161485671997,
      "learning_rate": 5.195668146822989e-06,
      "loss": 0.7878,
      "step": 24140
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.152848243713379,
      "learning_rate": 5.195029230425199e-06,
      "loss": 0.7091,
      "step": 24141
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2642505168914795,
      "learning_rate": 5.19439031402741e-06,
      "loss": 0.7236,
      "step": 24142
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6367082595825195,
      "learning_rate": 5.193751397629621e-06,
      "loss": 0.8459,
      "step": 24143
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.072880983352661,
      "learning_rate": 5.193112481231831e-06,
      "loss": 0.9066,
      "step": 24144
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7289848327636719,
      "learning_rate": 5.192473564834042e-06,
      "loss": 0.8604,
      "step": 24145
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.641833782196045,
      "learning_rate": 5.191834648436253e-06,
      "loss": 0.6578,
      "step": 24146
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0913689136505127,
      "learning_rate": 5.1911957320384635e-06,
      "loss": 0.9513,
      "step": 24147
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.369323253631592,
      "learning_rate": 5.190556815640674e-06,
      "loss": 0.7752,
      "step": 24148
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0124950408935547,
      "learning_rate": 5.189917899242885e-06,
      "loss": 1.1794,
      "step": 24149
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.0429186820983887,
      "learning_rate": 5.189278982845095e-06,
      "loss": 1.2824,
      "step": 24150
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.0185916423797607,
      "learning_rate": 5.1886400664473054e-06,
      "loss": 0.898,
      "step": 24151
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9776115417480469,
      "learning_rate": 5.188001150049516e-06,
      "loss": 0.9486,
      "step": 24152
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.053490161895752,
      "learning_rate": 5.187362233651727e-06,
      "loss": 0.9606,
      "step": 24153
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9348022937774658,
      "learning_rate": 5.1867233172539375e-06,
      "loss": 1.0773,
      "step": 24154
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.811143159866333,
      "learning_rate": 5.186084400856148e-06,
      "loss": 0.9355,
      "step": 24155
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.713347315788269,
      "learning_rate": 5.185445484458359e-06,
      "loss": 0.8262,
      "step": 24156
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.692742109298706,
      "learning_rate": 5.18480656806057e-06,
      "loss": 0.8878,
      "step": 24157
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6184210777282715,
      "learning_rate": 5.18416765166278e-06,
      "loss": 0.8691,
      "step": 24158
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.6726906299591064,
      "learning_rate": 5.183528735264991e-06,
      "loss": 0.7588,
      "step": 24159
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.6154606342315674,
      "learning_rate": 5.182889818867202e-06,
      "loss": 0.7699,
      "step": 24160
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8935352563858032,
      "learning_rate": 5.1822509024694124e-06,
      "loss": 0.8267,
      "step": 24161
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.837319016456604,
      "learning_rate": 5.181611986071623e-06,
      "loss": 0.7784,
      "step": 24162
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2271835803985596,
      "learning_rate": 5.180973069673834e-06,
      "loss": 0.8487,
      "step": 24163
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.4893879890441895,
      "learning_rate": 5.1803341532760445e-06,
      "loss": 1.0479,
      "step": 24164
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4397433996200562,
      "learning_rate": 5.179695236878255e-06,
      "loss": 0.8251,
      "step": 24165
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7162894010543823,
      "learning_rate": 5.179056320480466e-06,
      "loss": 0.875,
      "step": 24166
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.797004222869873,
      "learning_rate": 5.178417404082677e-06,
      "loss": 1.0468,
      "step": 24167
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.781108021736145,
      "learning_rate": 5.1777784876848865e-06,
      "loss": 0.8544,
      "step": 24168
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3864562511444092,
      "learning_rate": 5.177139571287097e-06,
      "loss": 0.5239,
      "step": 24169
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8228942155838013,
      "learning_rate": 5.176500654889308e-06,
      "loss": 1.2428,
      "step": 24170
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.329223155975342,
      "learning_rate": 5.175861738491519e-06,
      "loss": 1.039,
      "step": 24171
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.765256404876709,
      "learning_rate": 5.175222822093729e-06,
      "loss": 0.833,
      "step": 24172
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.009161949157715,
      "learning_rate": 5.17458390569594e-06,
      "loss": 0.8528,
      "step": 24173
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.607771396636963,
      "learning_rate": 5.173944989298151e-06,
      "loss": 0.9073,
      "step": 24174
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.820040702819824,
      "learning_rate": 5.173306072900361e-06,
      "loss": 0.4535,
      "step": 24175
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5635712146759033,
      "learning_rate": 5.172667156502572e-06,
      "loss": 1.1669,
      "step": 24176
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4661179780960083,
      "learning_rate": 5.172028240104783e-06,
      "loss": 1.0504,
      "step": 24177
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.27717924118042,
      "learning_rate": 5.1713893237069935e-06,
      "loss": 0.8667,
      "step": 24178
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.135514974594116,
      "learning_rate": 5.170750407309204e-06,
      "loss": 1.1019,
      "step": 24179
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.6708827018737793,
      "learning_rate": 5.170111490911415e-06,
      "loss": 0.9488,
      "step": 24180
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.716331958770752,
      "learning_rate": 5.169472574513626e-06,
      "loss": 1.26,
      "step": 24181
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7482064962387085,
      "learning_rate": 5.168833658115836e-06,
      "loss": 0.9417,
      "step": 24182
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.210765838623047,
      "learning_rate": 5.168194741718047e-06,
      "loss": 1.1757,
      "step": 24183
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.884917974472046,
      "learning_rate": 5.167555825320258e-06,
      "loss": 0.8178,
      "step": 24184
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0478415489196777,
      "learning_rate": 5.166916908922468e-06,
      "loss": 1.007,
      "step": 24185
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.43599271774292,
      "learning_rate": 5.166277992524678e-06,
      "loss": 0.8871,
      "step": 24186
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6393896341323853,
      "learning_rate": 5.165639076126889e-06,
      "loss": 0.9282,
      "step": 24187
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.1338508129119873,
      "learning_rate": 5.1650001597291e-06,
      "loss": 1.1177,
      "step": 24188
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.498551845550537,
      "learning_rate": 5.16436124333131e-06,
      "loss": 1.1709,
      "step": 24189
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9572551250457764,
      "learning_rate": 5.163722326933521e-06,
      "loss": 0.8869,
      "step": 24190
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1904683113098145,
      "learning_rate": 5.163083410535732e-06,
      "loss": 0.8485,
      "step": 24191
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.365934133529663,
      "learning_rate": 5.1624444941379424e-06,
      "loss": 0.9434,
      "step": 24192
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.9701082706451416,
      "learning_rate": 5.161805577740153e-06,
      "loss": 1.2177,
      "step": 24193
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9068642854690552,
      "learning_rate": 5.161166661342364e-06,
      "loss": 0.8797,
      "step": 24194
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.5320944786071777,
      "learning_rate": 5.1605277449445745e-06,
      "loss": 0.6584,
      "step": 24195
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.775727391242981,
      "learning_rate": 5.159888828546785e-06,
      "loss": 0.8137,
      "step": 24196
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.1071455478668213,
      "learning_rate": 5.159249912148996e-06,
      "loss": 0.8164,
      "step": 24197
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5176899433135986,
      "learning_rate": 5.158610995751207e-06,
      "loss": 0.3712,
      "step": 24198
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0864999294281006,
      "learning_rate": 5.157972079353417e-06,
      "loss": 0.8532,
      "step": 24199
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5497009754180908,
      "learning_rate": 5.157333162955628e-06,
      "loss": 0.8003,
      "step": 24200
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1007351875305176,
      "learning_rate": 5.156694246557839e-06,
      "loss": 1.178,
      "step": 24201
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8645000457763672,
      "learning_rate": 5.1560553301600494e-06,
      "loss": 0.7879,
      "step": 24202
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0161311626434326,
      "learning_rate": 5.15541641376226e-06,
      "loss": 0.8634,
      "step": 24203
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8208787441253662,
      "learning_rate": 5.154777497364471e-06,
      "loss": 0.7974,
      "step": 24204
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9223498106002808,
      "learning_rate": 5.154138580966681e-06,
      "loss": 0.8233,
      "step": 24205
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6730488538742065,
      "learning_rate": 5.153499664568891e-06,
      "loss": 0.8372,
      "step": 24206
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.132556438446045,
      "learning_rate": 5.152860748171102e-06,
      "loss": 0.7969,
      "step": 24207
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.853584885597229,
      "learning_rate": 5.152221831773313e-06,
      "loss": 0.9628,
      "step": 24208
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.53645658493042,
      "learning_rate": 5.1515829153755235e-06,
      "loss": 1.2412,
      "step": 24209
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7853891849517822,
      "learning_rate": 5.150943998977734e-06,
      "loss": 1.0721,
      "step": 24210
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9615395069122314,
      "learning_rate": 5.150305082579945e-06,
      "loss": 0.8107,
      "step": 24211
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.602639198303223,
      "learning_rate": 5.149666166182156e-06,
      "loss": 1.5317,
      "step": 24212
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.5516462326049805,
      "learning_rate": 5.149027249784366e-06,
      "loss": 1.2006,
      "step": 24213
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.021170616149902,
      "learning_rate": 5.148388333386577e-06,
      "loss": 1.0145,
      "step": 24214
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.53113055229187,
      "learning_rate": 5.147749416988788e-06,
      "loss": 0.9511,
      "step": 24215
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.502652168273926,
      "learning_rate": 5.147110500590998e-06,
      "loss": 1.0233,
      "step": 24216
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7790369987487793,
      "learning_rate": 5.146471584193209e-06,
      "loss": 0.7858,
      "step": 24217
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.226020336151123,
      "learning_rate": 5.14583266779542e-06,
      "loss": 0.8128,
      "step": 24218
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.435645580291748,
      "learning_rate": 5.1451937513976305e-06,
      "loss": 1.7534,
      "step": 24219
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8750022649765015,
      "learning_rate": 5.144554834999841e-06,
      "loss": 0.5296,
      "step": 24220
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9734710454940796,
      "learning_rate": 5.143915918602052e-06,
      "loss": 0.8617,
      "step": 24221
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.462475538253784,
      "learning_rate": 5.143277002204263e-06,
      "loss": 0.7219,
      "step": 24222
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.601465106010437,
      "learning_rate": 5.1426380858064724e-06,
      "loss": 0.8944,
      "step": 24223
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8629534244537354,
      "learning_rate": 5.141999169408683e-06,
      "loss": 0.9397,
      "step": 24224
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0757856369018555,
      "learning_rate": 5.141360253010894e-06,
      "loss": 1.2675,
      "step": 24225
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.175992965698242,
      "learning_rate": 5.1407213366131045e-06,
      "loss": 1.2494,
      "step": 24226
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8105677366256714,
      "learning_rate": 5.140082420215315e-06,
      "loss": 0.8647,
      "step": 24227
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8134266138076782,
      "learning_rate": 5.139443503817526e-06,
      "loss": 0.8523,
      "step": 24228
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.808736562728882,
      "learning_rate": 5.138804587419737e-06,
      "loss": 0.8545,
      "step": 24229
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.110764980316162,
      "learning_rate": 5.138165671021947e-06,
      "loss": 1.611,
      "step": 24230
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.342726945877075,
      "learning_rate": 5.137526754624158e-06,
      "loss": 0.868,
      "step": 24231
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.032254457473755,
      "learning_rate": 5.136887838226369e-06,
      "loss": 0.8876,
      "step": 24232
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8262062072753906,
      "learning_rate": 5.1362489218285794e-06,
      "loss": 0.8896,
      "step": 24233
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8510627746582031,
      "learning_rate": 5.13561000543079e-06,
      "loss": 0.8082,
      "step": 24234
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.602696180343628,
      "learning_rate": 5.134971089033001e-06,
      "loss": 0.9105,
      "step": 24235
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6957930326461792,
      "learning_rate": 5.1343321726352115e-06,
      "loss": 0.8053,
      "step": 24236
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.253375768661499,
      "learning_rate": 5.133693256237422e-06,
      "loss": 0.5647,
      "step": 24237
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.57980215549469,
      "learning_rate": 5.133054339839633e-06,
      "loss": 0.7116,
      "step": 24238
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5153695344924927,
      "learning_rate": 5.132415423441844e-06,
      "loss": 0.8249,
      "step": 24239
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7798290252685547,
      "learning_rate": 5.131776507044054e-06,
      "loss": 0.8241,
      "step": 24240
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2563973665237427,
      "learning_rate": 5.131137590646264e-06,
      "loss": 0.4207,
      "step": 24241
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9543843269348145,
      "learning_rate": 5.130498674248475e-06,
      "loss": 0.7139,
      "step": 24242
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4394160509109497,
      "learning_rate": 5.129859757850686e-06,
      "loss": 0.3929,
      "step": 24243
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.7936019897460938,
      "learning_rate": 5.129220841452896e-06,
      "loss": 1.0719,
      "step": 24244
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7691634893417358,
      "learning_rate": 5.128581925055107e-06,
      "loss": 0.8846,
      "step": 24245
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.374293804168701,
      "learning_rate": 5.127943008657318e-06,
      "loss": 1.0474,
      "step": 24246
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4091663360595703,
      "learning_rate": 5.127304092259528e-06,
      "loss": 0.6913,
      "step": 24247
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.067387580871582,
      "learning_rate": 5.126665175861739e-06,
      "loss": 1.026,
      "step": 24248
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4759321212768555,
      "learning_rate": 5.12602625946395e-06,
      "loss": 0.7896,
      "step": 24249
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.2305526733398438,
      "learning_rate": 5.1253873430661605e-06,
      "loss": 0.933,
      "step": 24250
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0488553047180176,
      "learning_rate": 5.124748426668371e-06,
      "loss": 0.7829,
      "step": 24251
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7619863748550415,
      "learning_rate": 5.124109510270582e-06,
      "loss": 0.89,
      "step": 24252
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1399619579315186,
      "learning_rate": 5.123470593872793e-06,
      "loss": 1.0456,
      "step": 24253
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.6479785442352295,
      "learning_rate": 5.122831677475003e-06,
      "loss": 1.0009,
      "step": 24254
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.43821382522583,
      "learning_rate": 5.122192761077214e-06,
      "loss": 0.8771,
      "step": 24255
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9217333793640137,
      "learning_rate": 5.121553844679425e-06,
      "loss": 0.9024,
      "step": 24256
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.249233245849609,
      "learning_rate": 5.120914928281635e-06,
      "loss": 1.1432,
      "step": 24257
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.3628997802734375,
      "learning_rate": 5.120276011883846e-06,
      "loss": 1.0642,
      "step": 24258
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1021616458892822,
      "learning_rate": 5.119637095486056e-06,
      "loss": 0.852,
      "step": 24259
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.057427406311035,
      "learning_rate": 5.118998179088267e-06,
      "loss": 1.1453,
      "step": 24260
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2733986377716064,
      "learning_rate": 5.118359262690477e-06,
      "loss": 0.7496,
      "step": 24261
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.0476861000061035,
      "learning_rate": 5.117720346292688e-06,
      "loss": 1.184,
      "step": 24262
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.038027286529541,
      "learning_rate": 5.117081429894899e-06,
      "loss": 0.6836,
      "step": 24263
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.166657447814941,
      "learning_rate": 5.1164425134971094e-06,
      "loss": 0.9472,
      "step": 24264
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.204280138015747,
      "learning_rate": 5.11580359709932e-06,
      "loss": 0.677,
      "step": 24265
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9252898693084717,
      "learning_rate": 5.115164680701531e-06,
      "loss": 0.7539,
      "step": 24266
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7925686836242676,
      "learning_rate": 5.1145257643037415e-06,
      "loss": 0.8633,
      "step": 24267
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9568434953689575,
      "learning_rate": 5.113886847905952e-06,
      "loss": 0.848,
      "step": 24268
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7836934328079224,
      "learning_rate": 5.113247931508163e-06,
      "loss": 0.8082,
      "step": 24269
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.011035919189453,
      "learning_rate": 5.112609015110374e-06,
      "loss": 0.7324,
      "step": 24270
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.733834743499756,
      "learning_rate": 5.111970098712584e-06,
      "loss": 1.0944,
      "step": 24271
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5671343803405762,
      "learning_rate": 5.111331182314795e-06,
      "loss": 0.7934,
      "step": 24272
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5927259922027588,
      "learning_rate": 5.110692265917006e-06,
      "loss": 0.8737,
      "step": 24273
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8587584495544434,
      "learning_rate": 5.1100533495192164e-06,
      "loss": 0.7951,
      "step": 24274
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8078588247299194,
      "learning_rate": 5.109414433121427e-06,
      "loss": 1.1621,
      "step": 24275
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4218387603759766,
      "learning_rate": 5.108775516723638e-06,
      "loss": 0.7952,
      "step": 24276
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2755521535873413,
      "learning_rate": 5.1081366003258485e-06,
      "loss": 0.475,
      "step": 24277
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.554349899291992,
      "learning_rate": 5.107497683928058e-06,
      "loss": 0.8294,
      "step": 24278
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1235573291778564,
      "learning_rate": 5.106858767530269e-06,
      "loss": 0.8053,
      "step": 24279
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2899796962738037,
      "learning_rate": 5.10621985113248e-06,
      "loss": 0.6814,
      "step": 24280
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9461603164672852,
      "learning_rate": 5.1055809347346905e-06,
      "loss": 0.8729,
      "step": 24281
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4639806747436523,
      "learning_rate": 5.104942018336901e-06,
      "loss": 0.8972,
      "step": 24282
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3602041006088257,
      "learning_rate": 5.104303101939112e-06,
      "loss": 0.3956,
      "step": 24283
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6152316331863403,
      "learning_rate": 5.103664185541323e-06,
      "loss": 0.7915,
      "step": 24284
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.758777141571045,
      "learning_rate": 5.103025269143533e-06,
      "loss": 0.9473,
      "step": 24285
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3149445056915283,
      "learning_rate": 5.102386352745744e-06,
      "loss": 0.3114,
      "step": 24286
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8299312591552734,
      "learning_rate": 5.101747436347955e-06,
      "loss": 0.9366,
      "step": 24287
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.92384934425354,
      "learning_rate": 5.101108519950165e-06,
      "loss": 0.8042,
      "step": 24288
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.29409122467041,
      "learning_rate": 5.100469603552376e-06,
      "loss": 0.8773,
      "step": 24289
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3399672508239746,
      "learning_rate": 5.099830687154587e-06,
      "loss": 0.9925,
      "step": 24290
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.642425537109375,
      "learning_rate": 5.0991917707567975e-06,
      "loss": 0.9459,
      "step": 24291
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3376554250717163,
      "learning_rate": 5.098552854359008e-06,
      "loss": 0.3285,
      "step": 24292
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.610757827758789,
      "learning_rate": 5.097913937961219e-06,
      "loss": 0.7814,
      "step": 24293
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0975711345672607,
      "learning_rate": 5.09727502156343e-06,
      "loss": 0.9984,
      "step": 24294
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.702258586883545,
      "learning_rate": 5.09663610516564e-06,
      "loss": 0.8357,
      "step": 24295
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7876989841461182,
      "learning_rate": 5.09599718876785e-06,
      "loss": 0.8238,
      "step": 24296
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7117999792099,
      "learning_rate": 5.095358272370061e-06,
      "loss": 0.8197,
      "step": 24297
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8782892227172852,
      "learning_rate": 5.0947193559722716e-06,
      "loss": 0.8468,
      "step": 24298
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7946902513504028,
      "learning_rate": 5.094080439574482e-06,
      "loss": 0.7885,
      "step": 24299
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.701033115386963,
      "learning_rate": 5.093441523176693e-06,
      "loss": 0.9974,
      "step": 24300
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4361860752105713,
      "learning_rate": 5.092802606778904e-06,
      "loss": 0.911,
      "step": 24301
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.559464454650879,
      "learning_rate": 5.092163690381114e-06,
      "loss": 0.7072,
      "step": 24302
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1691625118255615,
      "learning_rate": 5.091524773983325e-06,
      "loss": 0.9887,
      "step": 24303
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.5233025550842285,
      "learning_rate": 5.090885857585536e-06,
      "loss": 1.0403,
      "step": 24304
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6096243858337402,
      "learning_rate": 5.0902469411877465e-06,
      "loss": 0.8629,
      "step": 24305
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.911543369293213,
      "learning_rate": 5.089608024789957e-06,
      "loss": 1.0018,
      "step": 24306
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1425604820251465,
      "learning_rate": 5.088969108392168e-06,
      "loss": 0.9325,
      "step": 24307
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.636838674545288,
      "learning_rate": 5.0883301919943786e-06,
      "loss": 0.7574,
      "step": 24308
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.013378381729126,
      "learning_rate": 5.087691275596589e-06,
      "loss": 0.9157,
      "step": 24309
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6349376440048218,
      "learning_rate": 5.0870523591988e-06,
      "loss": 0.8899,
      "step": 24310
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5304512977600098,
      "learning_rate": 5.086413442801011e-06,
      "loss": 0.7701,
      "step": 24311
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.843270301818848,
      "learning_rate": 5.085774526403221e-06,
      "loss": 1.236,
      "step": 24312
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.4758124351501465,
      "learning_rate": 5.085135610005432e-06,
      "loss": 1.0638,
      "step": 24313
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9295977354049683,
      "learning_rate": 5.084496693607642e-06,
      "loss": 0.8427,
      "step": 24314
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.676230788230896,
      "learning_rate": 5.083857777209853e-06,
      "loss": 0.8019,
      "step": 24315
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.118760108947754,
      "learning_rate": 5.083218860812063e-06,
      "loss": 0.8166,
      "step": 24316
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.635786771774292,
      "learning_rate": 5.082579944414274e-06,
      "loss": 0.3861,
      "step": 24317
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.563590168952942,
      "learning_rate": 5.081941028016485e-06,
      "loss": 0.7877,
      "step": 24318
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3113195896148682,
      "learning_rate": 5.081302111618695e-06,
      "loss": 0.8052,
      "step": 24319
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6278421878814697,
      "learning_rate": 5.080663195220906e-06,
      "loss": 0.7731,
      "step": 24320
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.391853332519531,
      "learning_rate": 5.080024278823117e-06,
      "loss": 0.978,
      "step": 24321
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7654075622558594,
      "learning_rate": 5.0793853624253275e-06,
      "loss": 0.8235,
      "step": 24322
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0643320083618164,
      "learning_rate": 5.078746446027538e-06,
      "loss": 0.8074,
      "step": 24323
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.1910316944122314,
      "learning_rate": 5.078107529629749e-06,
      "loss": 0.8731,
      "step": 24324
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7465723752975464,
      "learning_rate": 5.07746861323196e-06,
      "loss": 0.8898,
      "step": 24325
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1979153156280518,
      "learning_rate": 5.07682969683417e-06,
      "loss": 0.9148,
      "step": 24326
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5830988883972168,
      "learning_rate": 5.076190780436381e-06,
      "loss": 0.7523,
      "step": 24327
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.1579596996307373,
      "learning_rate": 5.075551864038592e-06,
      "loss": 0.803,
      "step": 24328
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2517473697662354,
      "learning_rate": 5.074912947640802e-06,
      "loss": 0.9433,
      "step": 24329
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1089842319488525,
      "learning_rate": 5.074274031243013e-06,
      "loss": 0.8077,
      "step": 24330
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.918359398841858,
      "learning_rate": 5.073635114845224e-06,
      "loss": 0.7131,
      "step": 24331
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.0146877765655518,
      "learning_rate": 5.072996198447434e-06,
      "loss": 0.6526,
      "step": 24332
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6672848463058472,
      "learning_rate": 5.072357282049644e-06,
      "loss": 0.8731,
      "step": 24333
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3229676485061646,
      "learning_rate": 5.071718365651855e-06,
      "loss": 0.7922,
      "step": 24334
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7173330783843994,
      "learning_rate": 5.071079449254066e-06,
      "loss": 0.9709,
      "step": 24335
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.168243885040283,
      "learning_rate": 5.0704405328562765e-06,
      "loss": 0.8827,
      "step": 24336
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8129503726959229,
      "learning_rate": 5.069801616458487e-06,
      "loss": 0.9146,
      "step": 24337
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6075104475021362,
      "learning_rate": 5.069162700060698e-06,
      "loss": 0.801,
      "step": 24338
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0146820545196533,
      "learning_rate": 5.0685237836629086e-06,
      "loss": 0.779,
      "step": 24339
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7706090211868286,
      "learning_rate": 5.067884867265119e-06,
      "loss": 0.7861,
      "step": 24340
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.530276894569397,
      "learning_rate": 5.06724595086733e-06,
      "loss": 0.7708,
      "step": 24341
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.157419681549072,
      "learning_rate": 5.066607034469541e-06,
      "loss": 1.5902,
      "step": 24342
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.173149585723877,
      "learning_rate": 5.065968118071751e-06,
      "loss": 0.9205,
      "step": 24343
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.666404128074646,
      "learning_rate": 5.065329201673962e-06,
      "loss": 0.8392,
      "step": 24344
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1817989349365234,
      "learning_rate": 5.064690285276173e-06,
      "loss": 1.0231,
      "step": 24345
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6563283205032349,
      "learning_rate": 5.0640513688783835e-06,
      "loss": 0.8558,
      "step": 24346
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.5335659980773926,
      "learning_rate": 5.063412452480594e-06,
      "loss": 0.9713,
      "step": 24347
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.4893908500671387,
      "learning_rate": 5.062773536082805e-06,
      "loss": 1.0248,
      "step": 24348
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7277483940124512,
      "learning_rate": 5.0621346196850156e-06,
      "loss": 0.4517,
      "step": 24349
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.2808265686035156,
      "learning_rate": 5.0614957032872246e-06,
      "loss": 0.9046,
      "step": 24350
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3851945400238037,
      "learning_rate": 5.060856786889435e-06,
      "loss": 0.4033,
      "step": 24351
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5115021467208862,
      "learning_rate": 5.060217870491646e-06,
      "loss": 0.8278,
      "step": 24352
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7327091693878174,
      "learning_rate": 5.059578954093857e-06,
      "loss": 0.8895,
      "step": 24353
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9025015830993652,
      "learning_rate": 5.058940037696067e-06,
      "loss": 0.8083,
      "step": 24354
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1250197887420654,
      "learning_rate": 5.058301121298278e-06,
      "loss": 0.9003,
      "step": 24355
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.3779919147491455,
      "learning_rate": 5.057662204900489e-06,
      "loss": 0.6845,
      "step": 24356
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5410313606262207,
      "learning_rate": 5.0570232885026995e-06,
      "loss": 0.7034,
      "step": 24357
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4925761222839355,
      "learning_rate": 5.05638437210491e-06,
      "loss": 0.8346,
      "step": 24358
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.980153203010559,
      "learning_rate": 5.055745455707121e-06,
      "loss": 0.8654,
      "step": 24359
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2871969938278198,
      "learning_rate": 5.0551065393093316e-06,
      "loss": 0.3757,
      "step": 24360
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7589819431304932,
      "learning_rate": 5.054467622911542e-06,
      "loss": 0.8838,
      "step": 24361
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.413881540298462,
      "learning_rate": 5.053828706513753e-06,
      "loss": 0.8394,
      "step": 24362
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4019075632095337,
      "learning_rate": 5.053189790115963e-06,
      "loss": 0.3822,
      "step": 24363
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.475473880767822,
      "learning_rate": 5.0525508737181735e-06,
      "loss": 1.1141,
      "step": 24364
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9699132442474365,
      "learning_rate": 5.051911957320384e-06,
      "loss": 0.9701,
      "step": 24365
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.256837844848633,
      "learning_rate": 5.051273040922595e-06,
      "loss": 0.9282,
      "step": 24366
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5197890996932983,
      "learning_rate": 5.050634124524806e-06,
      "loss": 0.7449,
      "step": 24367
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.752063274383545,
      "learning_rate": 5.049995208127016e-06,
      "loss": 0.9851,
      "step": 24368
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8957183361053467,
      "learning_rate": 5.049356291729227e-06,
      "loss": 0.8237,
      "step": 24369
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.117130756378174,
      "learning_rate": 5.048717375331438e-06,
      "loss": 0.9712,
      "step": 24370
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6584275960922241,
      "learning_rate": 5.048078458933648e-06,
      "loss": 0.9518,
      "step": 24371
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9337347745895386,
      "learning_rate": 5.047439542535859e-06,
      "loss": 0.7944,
      "step": 24372
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8615752458572388,
      "learning_rate": 5.04680062613807e-06,
      "loss": 0.7715,
      "step": 24373
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.5652883052825928,
      "learning_rate": 5.0461617097402805e-06,
      "loss": 0.8828,
      "step": 24374
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.552300214767456,
      "learning_rate": 5.045522793342491e-06,
      "loss": 0.8596,
      "step": 24375
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.888695001602173,
      "learning_rate": 5.044883876944702e-06,
      "loss": 0.6941,
      "step": 24376
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7379474639892578,
      "learning_rate": 5.044244960546913e-06,
      "loss": 0.8022,
      "step": 24377
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.738413095474243,
      "learning_rate": 5.043606044149123e-06,
      "loss": 0.9541,
      "step": 24378
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8527852296829224,
      "learning_rate": 5.042967127751334e-06,
      "loss": 0.706,
      "step": 24379
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.713744044303894,
      "learning_rate": 5.042328211353545e-06,
      "loss": 0.8369,
      "step": 24380
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.5640459060668945,
      "learning_rate": 5.0416892949557546e-06,
      "loss": 1.0674,
      "step": 24381
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7088932991027832,
      "learning_rate": 5.041050378557965e-06,
      "loss": 0.8169,
      "step": 24382
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1680781841278076,
      "learning_rate": 5.040411462160176e-06,
      "loss": 0.7043,
      "step": 24383
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7386902570724487,
      "learning_rate": 5.039772545762387e-06,
      "loss": 0.9042,
      "step": 24384
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8631536960601807,
      "learning_rate": 5.039133629364597e-06,
      "loss": 0.8712,
      "step": 24385
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.719863176345825,
      "learning_rate": 5.038494712966808e-06,
      "loss": 0.9865,
      "step": 24386
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5515556335449219,
      "learning_rate": 5.037855796569019e-06,
      "loss": 0.7919,
      "step": 24387
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7692410945892334,
      "learning_rate": 5.0372168801712295e-06,
      "loss": 0.855,
      "step": 24388
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6745530366897583,
      "learning_rate": 5.03657796377344e-06,
      "loss": 0.8959,
      "step": 24389
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3695106506347656,
      "learning_rate": 5.035939047375651e-06,
      "loss": 0.3733,
      "step": 24390
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3632256984710693,
      "learning_rate": 5.0353001309778616e-06,
      "loss": 0.7997,
      "step": 24391
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0001721382141113,
      "learning_rate": 5.034661214580072e-06,
      "loss": 0.8226,
      "step": 24392
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.054593324661255,
      "learning_rate": 5.034022298182283e-06,
      "loss": 0.9916,
      "step": 24393
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0263595581054688,
      "learning_rate": 5.033383381784494e-06,
      "loss": 0.8886,
      "step": 24394
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.736900806427002,
      "learning_rate": 5.032744465386704e-06,
      "loss": 0.8585,
      "step": 24395
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4878435134887695,
      "learning_rate": 5.032105548988915e-06,
      "loss": 0.4333,
      "step": 24396
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7127869129180908,
      "learning_rate": 5.031466632591126e-06,
      "loss": 0.8837,
      "step": 24397
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7150040864944458,
      "learning_rate": 5.0308277161933365e-06,
      "loss": 0.9353,
      "step": 24398
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.046322822570801,
      "learning_rate": 5.030188799795547e-06,
      "loss": 0.9289,
      "step": 24399
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9745535850524902,
      "learning_rate": 5.029549883397757e-06,
      "loss": 1.101,
      "step": 24400
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3874760866165161,
      "learning_rate": 5.028910966999968e-06,
      "loss": 0.8258,
      "step": 24401
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.5255887508392334,
      "learning_rate": 5.028272050602178e-06,
      "loss": 0.8844,
      "step": 24402
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.226912498474121,
      "learning_rate": 5.027633134204389e-06,
      "loss": 0.8261,
      "step": 24403
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9412341117858887,
      "learning_rate": 5.0269942178066e-06,
      "loss": 0.858,
      "step": 24404
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.3977177143096924,
      "learning_rate": 5.0263553014088105e-06,
      "loss": 0.732,
      "step": 24405
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.449688196182251,
      "learning_rate": 5.025716385011021e-06,
      "loss": 0.6139,
      "step": 24406
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7824854850769043,
      "learning_rate": 5.025077468613232e-06,
      "loss": 0.8519,
      "step": 24407
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8015012741088867,
      "learning_rate": 5.024438552215443e-06,
      "loss": 0.7767,
      "step": 24408
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.18562388420105,
      "learning_rate": 5.023799635817653e-06,
      "loss": 0.7949,
      "step": 24409
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.9322826862335205,
      "learning_rate": 5.023160719419864e-06,
      "loss": 1.1245,
      "step": 24410
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4307823181152344,
      "learning_rate": 5.022521803022075e-06,
      "loss": 0.3918,
      "step": 24411
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5579630136489868,
      "learning_rate": 5.021882886624285e-06,
      "loss": 0.7612,
      "step": 24412
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.256221055984497,
      "learning_rate": 5.021243970226496e-06,
      "loss": 0.8187,
      "step": 24413
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.007591724395752,
      "learning_rate": 5.020605053828707e-06,
      "loss": 0.8804,
      "step": 24414
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6890010833740234,
      "learning_rate": 5.0199661374309175e-06,
      "loss": 0.8485,
      "step": 24415
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0241172313690186,
      "learning_rate": 5.019327221033128e-06,
      "loss": 0.7495,
      "step": 24416
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7965435981750488,
      "learning_rate": 5.018688304635339e-06,
      "loss": 0.9304,
      "step": 24417
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.956191301345825,
      "learning_rate": 5.018049388237549e-06,
      "loss": 0.8892,
      "step": 24418
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.48815393447876,
      "learning_rate": 5.0174104718397595e-06,
      "loss": 0.9133,
      "step": 24419
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3874146938323975,
      "learning_rate": 5.01677155544197e-06,
      "loss": 0.4113,
      "step": 24420
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.744507908821106,
      "learning_rate": 5.016132639044181e-06,
      "loss": 0.8534,
      "step": 24421
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6681840419769287,
      "learning_rate": 5.0154937226463916e-06,
      "loss": 0.8122,
      "step": 24422
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8996174335479736,
      "learning_rate": 5.014854806248602e-06,
      "loss": 0.6868,
      "step": 24423
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8916194438934326,
      "learning_rate": 5.014215889850813e-06,
      "loss": 0.9697,
      "step": 24424
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5365177392959595,
      "learning_rate": 5.013576973453024e-06,
      "loss": 0.7388,
      "step": 24425
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.907828688621521,
      "learning_rate": 5.012938057055234e-06,
      "loss": 0.884,
      "step": 24426
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7588231563568115,
      "learning_rate": 5.012299140657445e-06,
      "loss": 0.8413,
      "step": 24427
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.988053798675537,
      "learning_rate": 5.011660224259656e-06,
      "loss": 1.408,
      "step": 24428
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.139329433441162,
      "learning_rate": 5.0110213078618665e-06,
      "loss": 0.9641,
      "step": 24429
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5502346754074097,
      "learning_rate": 5.010382391464077e-06,
      "loss": 0.809,
      "step": 24430
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6208913326263428,
      "learning_rate": 5.009743475066288e-06,
      "loss": 0.8536,
      "step": 24431
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.26420521736145,
      "learning_rate": 5.0091045586684986e-06,
      "loss": 1.1009,
      "step": 24432
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8942855596542358,
      "learning_rate": 5.008465642270709e-06,
      "loss": 0.8133,
      "step": 24433
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.8217697143554688,
      "learning_rate": 5.00782672587292e-06,
      "loss": 1.2057,
      "step": 24434
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2125625610351562,
      "learning_rate": 5.007187809475131e-06,
      "loss": 0.9575,
      "step": 24435
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.99431574344635,
      "learning_rate": 5.0065488930773405e-06,
      "loss": 0.9605,
      "step": 24436
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7038514614105225,
      "learning_rate": 5.005909976679551e-06,
      "loss": 0.7866,
      "step": 24437
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8509117364883423,
      "learning_rate": 5.005271060281762e-06,
      "loss": 0.7994,
      "step": 24438
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7337056398391724,
      "learning_rate": 5.004632143883973e-06,
      "loss": 1.0324,
      "step": 24439
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.567366361618042,
      "learning_rate": 5.003993227486183e-06,
      "loss": 0.903,
      "step": 24440
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1860251426696777,
      "learning_rate": 5.003354311088394e-06,
      "loss": 0.8572,
      "step": 24441
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8711352348327637,
      "learning_rate": 5.002715394690605e-06,
      "loss": 0.9272,
      "step": 24442
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9335601329803467,
      "learning_rate": 5.002076478292815e-06,
      "loss": 1.2002,
      "step": 24443
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1133358478546143,
      "learning_rate": 5.001437561895026e-06,
      "loss": 0.8486,
      "step": 24444
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.6751327514648438,
      "learning_rate": 5.000798645497237e-06,
      "loss": 0.8292,
      "step": 24445
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6448814868927002,
      "learning_rate": 5.0001597290994475e-06,
      "loss": 0.9228,
      "step": 24446
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.155155658721924,
      "learning_rate": 4.999520812701658e-06,
      "loss": 0.875,
      "step": 24447
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9751925468444824,
      "learning_rate": 4.998881896303869e-06,
      "loss": 0.7232,
      "step": 24448
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.500009059906006,
      "learning_rate": 4.99824297990608e-06,
      "loss": 0.9359,
      "step": 24449
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.341431140899658,
      "learning_rate": 4.99760406350829e-06,
      "loss": 0.8812,
      "step": 24450
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6194398403167725,
      "learning_rate": 4.996965147110501e-06,
      "loss": 0.7852,
      "step": 24451
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9052730798721313,
      "learning_rate": 4.996326230712712e-06,
      "loss": 0.8833,
      "step": 24452
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8206939697265625,
      "learning_rate": 4.995687314314922e-06,
      "loss": 0.8199,
      "step": 24453
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3598432540893555,
      "learning_rate": 4.995048397917132e-06,
      "loss": 0.7988,
      "step": 24454
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9069174528121948,
      "learning_rate": 4.994409481519343e-06,
      "loss": 0.7983,
      "step": 24455
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8797142505645752,
      "learning_rate": 4.993770565121554e-06,
      "loss": 0.8407,
      "step": 24456
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7133443355560303,
      "learning_rate": 4.993131648723764e-06,
      "loss": 0.7717,
      "step": 24457
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.5484516620635986,
      "learning_rate": 4.992492732325975e-06,
      "loss": 1.0968,
      "step": 24458
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.678816795349121,
      "learning_rate": 4.991853815928186e-06,
      "loss": 0.9322,
      "step": 24459
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.068711996078491,
      "learning_rate": 4.9912148995303965e-06,
      "loss": 0.8237,
      "step": 24460
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.712145447731018,
      "learning_rate": 4.990575983132607e-06,
      "loss": 0.7739,
      "step": 24461
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1108829975128174,
      "learning_rate": 4.989937066734818e-06,
      "loss": 0.8542,
      "step": 24462
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9572277069091797,
      "learning_rate": 4.9892981503370286e-06,
      "loss": 0.8263,
      "step": 24463
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9462382793426514,
      "learning_rate": 4.988659233939239e-06,
      "loss": 0.8427,
      "step": 24464
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.542165994644165,
      "learning_rate": 4.98802031754145e-06,
      "loss": 0.7904,
      "step": 24465
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6663089990615845,
      "learning_rate": 4.987381401143661e-06,
      "loss": 0.8812,
      "step": 24466
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.096533298492432,
      "learning_rate": 4.986742484745871e-06,
      "loss": 1.5611,
      "step": 24467
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.151235103607178,
      "learning_rate": 4.986103568348082e-06,
      "loss": 1.1897,
      "step": 24468
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8234405517578125,
      "learning_rate": 4.985464651950293e-06,
      "loss": 0.7386,
      "step": 24469
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4310872554779053,
      "learning_rate": 4.9848257355525035e-06,
      "loss": 0.4113,
      "step": 24470
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9023298025131226,
      "learning_rate": 4.984186819154714e-06,
      "loss": 0.8147,
      "step": 24471
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.538461446762085,
      "learning_rate": 4.983547902756925e-06,
      "loss": 0.7213,
      "step": 24472
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8064773082733154,
      "learning_rate": 4.982908986359135e-06,
      "loss": 0.8962,
      "step": 24473
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7731225490570068,
      "learning_rate": 4.9822700699613454e-06,
      "loss": 0.8096,
      "step": 24474
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7403125762939453,
      "learning_rate": 4.981631153563556e-06,
      "loss": 0.4842,
      "step": 24475
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8844282627105713,
      "learning_rate": 4.980992237165767e-06,
      "loss": 1.0192,
      "step": 24476
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6299328804016113,
      "learning_rate": 4.9803533207679775e-06,
      "loss": 0.7418,
      "step": 24477
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.097900867462158,
      "learning_rate": 4.979714404370188e-06,
      "loss": 1.0697,
      "step": 24478
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6780210733413696,
      "learning_rate": 4.979075487972399e-06,
      "loss": 0.9127,
      "step": 24479
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.110275983810425,
      "learning_rate": 4.97843657157461e-06,
      "loss": 0.8809,
      "step": 24480
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4479973316192627,
      "learning_rate": 4.97779765517682e-06,
      "loss": 0.7867,
      "step": 24481
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7662556171417236,
      "learning_rate": 4.977158738779031e-06,
      "loss": 0.9445,
      "step": 24482
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.418888568878174,
      "learning_rate": 4.976519822381242e-06,
      "loss": 0.8686,
      "step": 24483
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.2704570293426514,
      "learning_rate": 4.9758809059834524e-06,
      "loss": 1.0136,
      "step": 24484
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1656296253204346,
      "learning_rate": 4.975241989585663e-06,
      "loss": 0.8633,
      "step": 24485
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4539047479629517,
      "learning_rate": 4.974603073187874e-06,
      "loss": 0.8553,
      "step": 24486
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1083197593688965,
      "learning_rate": 4.9739641567900845e-06,
      "loss": 1.1828,
      "step": 24487
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.58004903793335,
      "learning_rate": 4.973325240392295e-06,
      "loss": 1.0717,
      "step": 24488
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5398579835891724,
      "learning_rate": 4.972686323994506e-06,
      "loss": 0.3653,
      "step": 24489
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1155319213867188,
      "learning_rate": 4.972047407596717e-06,
      "loss": 0.9195,
      "step": 24490
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.238933563232422,
      "learning_rate": 4.9714084911989265e-06,
      "loss": 0.8724,
      "step": 24491
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6530952453613281,
      "learning_rate": 4.970769574801137e-06,
      "loss": 0.86,
      "step": 24492
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3010114431381226,
      "learning_rate": 4.970130658403348e-06,
      "loss": 0.3577,
      "step": 24493
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.081721544265747,
      "learning_rate": 4.969491742005559e-06,
      "loss": 0.8719,
      "step": 24494
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4672002792358398,
      "learning_rate": 4.968852825607769e-06,
      "loss": 0.4619,
      "step": 24495
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.067484378814697,
      "learning_rate": 4.96821390920998e-06,
      "loss": 1.1476,
      "step": 24496
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.168124198913574,
      "learning_rate": 4.967574992812191e-06,
      "loss": 1.1859,
      "step": 24497
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.353004217147827,
      "learning_rate": 4.966936076414401e-06,
      "loss": 1.2837,
      "step": 24498
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7577661275863647,
      "learning_rate": 4.966297160016612e-06,
      "loss": 0.8788,
      "step": 24499
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9488939046859741,
      "learning_rate": 4.965658243618823e-06,
      "loss": 1.1756,
      "step": 24500
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5741757154464722,
      "learning_rate": 4.9650193272210335e-06,
      "loss": 1.1474,
      "step": 24501
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.181145429611206,
      "learning_rate": 4.964380410823244e-06,
      "loss": 0.8974,
      "step": 24502
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7147027254104614,
      "learning_rate": 4.963741494425455e-06,
      "loss": 0.863,
      "step": 24503
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6047104597091675,
      "learning_rate": 4.963102578027666e-06,
      "loss": 0.8019,
      "step": 24504
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.7481770515441895,
      "learning_rate": 4.962463661629876e-06,
      "loss": 1.3061,
      "step": 24505
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.108351469039917,
      "learning_rate": 4.961824745232087e-06,
      "loss": 0.9627,
      "step": 24506
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.852325439453125,
      "learning_rate": 4.961185828834298e-06,
      "loss": 0.7887,
      "step": 24507
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.6622393131256104,
      "learning_rate": 4.960546912436508e-06,
      "loss": 0.8258,
      "step": 24508
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.873349905014038,
      "learning_rate": 4.959907996038718e-06,
      "loss": 0.8048,
      "step": 24509
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.477967381477356,
      "learning_rate": 4.959269079640929e-06,
      "loss": 0.8381,
      "step": 24510
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4658900499343872,
      "learning_rate": 4.95863016324314e-06,
      "loss": 0.3444,
      "step": 24511
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8858036994934082,
      "learning_rate": 4.95799124684535e-06,
      "loss": 1.1053,
      "step": 24512
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.870521068572998,
      "learning_rate": 4.957352330447561e-06,
      "loss": 0.9031,
      "step": 24513
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8595759868621826,
      "learning_rate": 4.956713414049772e-06,
      "loss": 0.7859,
      "step": 24514
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2386741638183594,
      "learning_rate": 4.9560744976519824e-06,
      "loss": 0.8231,
      "step": 24515
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6793354749679565,
      "learning_rate": 4.955435581254193e-06,
      "loss": 0.8982,
      "step": 24516
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.311387538909912,
      "learning_rate": 4.954796664856404e-06,
      "loss": 1.7594,
      "step": 24517
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7560398578643799,
      "learning_rate": 4.9541577484586145e-06,
      "loss": 0.8223,
      "step": 24518
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9759154319763184,
      "learning_rate": 4.953518832060825e-06,
      "loss": 0.9086,
      "step": 24519
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.556382656097412,
      "learning_rate": 4.952879915663036e-06,
      "loss": 0.764,
      "step": 24520
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1118621826171875,
      "learning_rate": 4.952240999265247e-06,
      "loss": 0.8228,
      "step": 24521
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7386767864227295,
      "learning_rate": 4.951602082867457e-06,
      "loss": 0.8669,
      "step": 24522
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6466423273086548,
      "learning_rate": 4.950963166469668e-06,
      "loss": 0.3848,
      "step": 24523
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.8490660190582275,
      "learning_rate": 4.950324250071879e-06,
      "loss": 1.213,
      "step": 24524
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8143534660339355,
      "learning_rate": 4.9496853336740894e-06,
      "loss": 0.8707,
      "step": 24525
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6428990364074707,
      "learning_rate": 4.9490464172763e-06,
      "loss": 0.7197,
      "step": 24526
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6074129343032837,
      "learning_rate": 4.94840750087851e-06,
      "loss": 0.8743,
      "step": 24527
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.763760089874268,
      "learning_rate": 4.947768584480721e-06,
      "loss": 1.8703,
      "step": 24528
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1865644454956055,
      "learning_rate": 4.947129668082931e-06,
      "loss": 0.7355,
      "step": 24529
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8414665460586548,
      "learning_rate": 4.946490751685142e-06,
      "loss": 0.7383,
      "step": 24530
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8107950687408447,
      "learning_rate": 4.945851835287353e-06,
      "loss": 0.8084,
      "step": 24531
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.573121547698975,
      "learning_rate": 4.9452129188895635e-06,
      "loss": 1.2309,
      "step": 24532
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3691883087158203,
      "learning_rate": 4.944574002491774e-06,
      "loss": 0.8948,
      "step": 24533
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.417069911956787,
      "learning_rate": 4.943935086093985e-06,
      "loss": 0.5679,
      "step": 24534
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9976271390914917,
      "learning_rate": 4.943296169696196e-06,
      "loss": 0.9615,
      "step": 24535
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8495081663131714,
      "learning_rate": 4.942657253298406e-06,
      "loss": 0.9225,
      "step": 24536
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.82905113697052,
      "learning_rate": 4.942018336900617e-06,
      "loss": 0.7503,
      "step": 24537
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.079145669937134,
      "learning_rate": 4.941379420502828e-06,
      "loss": 0.8014,
      "step": 24538
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7374078035354614,
      "learning_rate": 4.940740504105038e-06,
      "loss": 0.9964,
      "step": 24539
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8346723318099976,
      "learning_rate": 4.940101587707249e-06,
      "loss": 0.935,
      "step": 24540
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8445067405700684,
      "learning_rate": 4.93946267130946e-06,
      "loss": 0.7858,
      "step": 24541
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.1847546100616455,
      "learning_rate": 4.9388237549116705e-06,
      "loss": 0.7802,
      "step": 24542
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.0610809326171875,
      "learning_rate": 4.938184838513881e-06,
      "loss": 1.0852,
      "step": 24543
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.0020580291748047,
      "learning_rate": 4.937545922116092e-06,
      "loss": 1.0839,
      "step": 24544
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.785258412361145,
      "learning_rate": 4.936907005718303e-06,
      "loss": 0.6093,
      "step": 24545
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.618129014968872,
      "learning_rate": 4.9362680893205124e-06,
      "loss": 0.8252,
      "step": 24546
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.123593330383301,
      "learning_rate": 4.935629172922723e-06,
      "loss": 0.6726,
      "step": 24547
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.17846417427063,
      "learning_rate": 4.934990256524934e-06,
      "loss": 0.8155,
      "step": 24548
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4633314609527588,
      "learning_rate": 4.9343513401271445e-06,
      "loss": 0.8125,
      "step": 24549
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3495376110076904,
      "learning_rate": 4.933712423729355e-06,
      "loss": 0.4156,
      "step": 24550
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.614789605140686,
      "learning_rate": 4.933073507331566e-06,
      "loss": 0.8172,
      "step": 24551
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7171399593353271,
      "learning_rate": 4.932434590933777e-06,
      "loss": 0.9689,
      "step": 24552
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8688132762908936,
      "learning_rate": 4.931795674535987e-06,
      "loss": 0.9054,
      "step": 24553
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.383082151412964,
      "learning_rate": 4.931156758138198e-06,
      "loss": 0.8458,
      "step": 24554
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.10394549369812,
      "learning_rate": 4.930517841740409e-06,
      "loss": 0.8379,
      "step": 24555
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.093613624572754,
      "learning_rate": 4.9298789253426194e-06,
      "loss": 0.7451,
      "step": 24556
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6210932731628418,
      "learning_rate": 4.92924000894483e-06,
      "loss": 0.854,
      "step": 24557
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1981074810028076,
      "learning_rate": 4.928601092547041e-06,
      "loss": 0.4384,
      "step": 24558
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.12558913230896,
      "learning_rate": 4.9279621761492515e-06,
      "loss": 1.0706,
      "step": 24559
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9305908679962158,
      "learning_rate": 4.927323259751462e-06,
      "loss": 0.8579,
      "step": 24560
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.67763614654541,
      "learning_rate": 4.926684343353673e-06,
      "loss": 0.6102,
      "step": 24561
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.6513800621032715,
      "learning_rate": 4.926045426955884e-06,
      "loss": 0.8239,
      "step": 24562
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9957219362258911,
      "learning_rate": 4.925406510558094e-06,
      "loss": 0.7483,
      "step": 24563
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.748655080795288,
      "learning_rate": 4.924767594160304e-06,
      "loss": 0.8173,
      "step": 24564
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.111088752746582,
      "learning_rate": 4.924128677762515e-06,
      "loss": 1.5558,
      "step": 24565
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.26322078704834,
      "learning_rate": 4.923489761364726e-06,
      "loss": 0.8115,
      "step": 24566
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1437795162200928,
      "learning_rate": 4.922850844966936e-06,
      "loss": 0.8602,
      "step": 24567
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.19551944732666,
      "learning_rate": 4.922211928569147e-06,
      "loss": 1.0818,
      "step": 24568
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.25679612159729,
      "learning_rate": 4.921573012171358e-06,
      "loss": 0.7692,
      "step": 24569
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1096205711364746,
      "learning_rate": 4.920934095773568e-06,
      "loss": 0.9513,
      "step": 24570
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6267105340957642,
      "learning_rate": 4.920295179375779e-06,
      "loss": 0.8764,
      "step": 24571
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0414342880249023,
      "learning_rate": 4.91965626297799e-06,
      "loss": 0.8304,
      "step": 24572
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.067373275756836,
      "learning_rate": 4.9190173465802005e-06,
      "loss": 0.7783,
      "step": 24573
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2434310913085938,
      "learning_rate": 4.918378430182411e-06,
      "loss": 0.8466,
      "step": 24574
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5799678564071655,
      "learning_rate": 4.917739513784622e-06,
      "loss": 0.8316,
      "step": 24575
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.502600073814392,
      "learning_rate": 4.917100597386833e-06,
      "loss": 0.8645,
      "step": 24576
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5247411727905273,
      "learning_rate": 4.916461680989043e-06,
      "loss": 0.4369,
      "step": 24577
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8847930431365967,
      "learning_rate": 4.915822764591254e-06,
      "loss": 0.8245,
      "step": 24578
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.76877760887146,
      "learning_rate": 4.915183848193465e-06,
      "loss": 0.907,
      "step": 24579
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.042226791381836,
      "learning_rate": 4.914544931795675e-06,
      "loss": 1.1002,
      "step": 24580
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7793283462524414,
      "learning_rate": 4.913906015397886e-06,
      "loss": 0.8587,
      "step": 24581
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9131499528884888,
      "learning_rate": 4.913267099000096e-06,
      "loss": 0.8579,
      "step": 24582
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.4134883880615234,
      "learning_rate": 4.912628182602307e-06,
      "loss": 0.9528,
      "step": 24583
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.729004144668579,
      "learning_rate": 4.911989266204517e-06,
      "loss": 0.7563,
      "step": 24584
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7060041427612305,
      "learning_rate": 4.911350349806728e-06,
      "loss": 1.0118,
      "step": 24585
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2764266729354858,
      "learning_rate": 4.910711433408939e-06,
      "loss": 0.4024,
      "step": 24586
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6822175979614258,
      "learning_rate": 4.9100725170111494e-06,
      "loss": 0.8396,
      "step": 24587
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.107520580291748,
      "learning_rate": 4.90943360061336e-06,
      "loss": 1.0478,
      "step": 24588
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3968467712402344,
      "learning_rate": 4.908794684215571e-06,
      "loss": 0.8369,
      "step": 24589
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5685564279556274,
      "learning_rate": 4.9081557678177815e-06,
      "loss": 0.7516,
      "step": 24590
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.4478256702423096,
      "learning_rate": 4.907516851419992e-06,
      "loss": 1.5083,
      "step": 24591
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8478139638900757,
      "learning_rate": 4.906877935022203e-06,
      "loss": 0.8115,
      "step": 24592
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2863452434539795,
      "learning_rate": 4.906239018624414e-06,
      "loss": 0.3477,
      "step": 24593
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.543694257736206,
      "learning_rate": 4.905600102226624e-06,
      "loss": 1.1311,
      "step": 24594
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0197744369506836,
      "learning_rate": 4.904961185828835e-06,
      "loss": 0.9029,
      "step": 24595
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6376551389694214,
      "learning_rate": 4.904322269431046e-06,
      "loss": 1.0868,
      "step": 24596
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.866287350654602,
      "learning_rate": 4.9036833530332564e-06,
      "loss": 0.849,
      "step": 24597
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5052770376205444,
      "learning_rate": 4.903044436635467e-06,
      "loss": 0.854,
      "step": 24598
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7076802253723145,
      "learning_rate": 4.902405520237678e-06,
      "loss": 0.9099,
      "step": 24599
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3509870767593384,
      "learning_rate": 4.901766603839888e-06,
      "loss": 0.3532,
      "step": 24600
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5779732465744019,
      "learning_rate": 4.901127687442098e-06,
      "loss": 0.8371,
      "step": 24601
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5704734325408936,
      "learning_rate": 4.900488771044309e-06,
      "loss": 0.7975,
      "step": 24602
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.2062554359436035,
      "learning_rate": 4.89984985464652e-06,
      "loss": 0.7641,
      "step": 24603
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.737165451049805,
      "learning_rate": 4.8992109382487305e-06,
      "loss": 1.5739,
      "step": 24604
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.284672737121582,
      "learning_rate": 4.898572021850941e-06,
      "loss": 1.2124,
      "step": 24605
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0754177570343018,
      "learning_rate": 4.897933105453152e-06,
      "loss": 1.1981,
      "step": 24606
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.3460521697998047,
      "learning_rate": 4.897294189055363e-06,
      "loss": 0.9652,
      "step": 24607
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.085355758666992,
      "learning_rate": 4.896655272657573e-06,
      "loss": 0.9813,
      "step": 24608
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0650594234466553,
      "learning_rate": 4.896016356259784e-06,
      "loss": 0.8927,
      "step": 24609
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6349753141403198,
      "learning_rate": 4.895377439861995e-06,
      "loss": 0.8061,
      "step": 24610
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8850165605545044,
      "learning_rate": 4.894738523464205e-06,
      "loss": 0.962,
      "step": 24611
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7374359369277954,
      "learning_rate": 4.894099607066416e-06,
      "loss": 0.8359,
      "step": 24612
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.4484689235687256,
      "learning_rate": 4.893460690668627e-06,
      "loss": 0.8575,
      "step": 24613
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.959650754928589,
      "learning_rate": 4.8928217742708375e-06,
      "loss": 0.9235,
      "step": 24614
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.475985288619995,
      "learning_rate": 4.892182857873048e-06,
      "loss": 0.9683,
      "step": 24615
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.413750410079956,
      "learning_rate": 4.891543941475259e-06,
      "loss": 1.1644,
      "step": 24616
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.599005937576294,
      "learning_rate": 4.89090502507747e-06,
      "loss": 0.8625,
      "step": 24617
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.6841089725494385,
      "learning_rate": 4.89026610867968e-06,
      "loss": 0.8758,
      "step": 24618
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.6336753368377686,
      "learning_rate": 4.88962719228189e-06,
      "loss": 0.9583,
      "step": 24619
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5935500860214233,
      "learning_rate": 4.888988275884101e-06,
      "loss": 1.1457,
      "step": 24620
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.94198739528656,
      "learning_rate": 4.8883493594863115e-06,
      "loss": 1.0427,
      "step": 24621
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.4973294734954834,
      "learning_rate": 4.887710443088522e-06,
      "loss": 0.9158,
      "step": 24622
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4100054502487183,
      "learning_rate": 4.887071526690733e-06,
      "loss": 0.4125,
      "step": 24623
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1917896270751953,
      "learning_rate": 4.886432610292944e-06,
      "loss": 0.9657,
      "step": 24624
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.350341796875,
      "learning_rate": 4.885793693895154e-06,
      "loss": 1.4178,
      "step": 24625
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.812165379524231,
      "learning_rate": 4.885154777497365e-06,
      "loss": 0.7844,
      "step": 24626
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9706958532333374,
      "learning_rate": 4.884515861099576e-06,
      "loss": 0.9212,
      "step": 24627
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.667806148529053,
      "learning_rate": 4.8838769447017864e-06,
      "loss": 1.9708,
      "step": 24628
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.2565174102783203,
      "learning_rate": 4.883238028303997e-06,
      "loss": 1.1872,
      "step": 24629
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4137006998062134,
      "learning_rate": 4.882599111906208e-06,
      "loss": 0.8667,
      "step": 24630
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7487823963165283,
      "learning_rate": 4.8819601955084185e-06,
      "loss": 0.8944,
      "step": 24631
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3553638458251953,
      "learning_rate": 4.881321279110629e-06,
      "loss": 0.3904,
      "step": 24632
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7693610191345215,
      "learning_rate": 4.88068236271284e-06,
      "loss": 0.8612,
      "step": 24633
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.171550750732422,
      "learning_rate": 4.880043446315051e-06,
      "loss": 0.9363,
      "step": 24634
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3083980083465576,
      "learning_rate": 4.879404529917261e-06,
      "loss": 0.9045,
      "step": 24635
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5395045280456543,
      "learning_rate": 4.878765613519472e-06,
      "loss": 0.3805,
      "step": 24636
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.6761269569396973,
      "learning_rate": 4.878126697121682e-06,
      "loss": 0.841,
      "step": 24637
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7010421752929688,
      "learning_rate": 4.877487780723893e-06,
      "loss": 0.763,
      "step": 24638
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8336262702941895,
      "learning_rate": 4.876848864326103e-06,
      "loss": 0.8989,
      "step": 24639
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6326348781585693,
      "learning_rate": 4.876209947928314e-06,
      "loss": 0.731,
      "step": 24640
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0576465129852295,
      "learning_rate": 4.875571031530525e-06,
      "loss": 0.816,
      "step": 24641
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3029122352600098,
      "learning_rate": 4.874932115132735e-06,
      "loss": 0.7721,
      "step": 24642
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7482023239135742,
      "learning_rate": 4.874293198734945e-06,
      "loss": 0.9029,
      "step": 24643
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8555630445480347,
      "learning_rate": 4.873654282337156e-06,
      "loss": 0.8033,
      "step": 24644
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1278574466705322,
      "learning_rate": 4.873015365939367e-06,
      "loss": 0.7774,
      "step": 24645
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7428512573242188,
      "learning_rate": 4.872376449541577e-06,
      "loss": 1.0924,
      "step": 24646
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2969543933868408,
      "learning_rate": 4.871737533143788e-06,
      "loss": 0.7947,
      "step": 24647
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.3090851306915283,
      "learning_rate": 4.871098616745999e-06,
      "loss": 0.666,
      "step": 24648
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7217475175857544,
      "learning_rate": 4.8704597003482094e-06,
      "loss": 0.7784,
      "step": 24649
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.756956100463867,
      "learning_rate": 4.86982078395042e-06,
      "loss": 0.6906,
      "step": 24650
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.9038617610931396,
      "learning_rate": 4.869181867552631e-06,
      "loss": 0.837,
      "step": 24651
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.560246229171753,
      "learning_rate": 4.8685429511548415e-06,
      "loss": 0.8028,
      "step": 24652
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.962170958518982,
      "learning_rate": 4.867904034757052e-06,
      "loss": 0.8863,
      "step": 24653
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.541563034057617,
      "learning_rate": 4.867265118359263e-06,
      "loss": 0.8712,
      "step": 24654
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.451640248298645,
      "learning_rate": 4.866626201961474e-06,
      "loss": 0.7662,
      "step": 24655
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.3646748065948486,
      "learning_rate": 4.865987285563684e-06,
      "loss": 0.9524,
      "step": 24656
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5998116731643677,
      "learning_rate": 4.865348369165895e-06,
      "loss": 0.7489,
      "step": 24657
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8074021339416504,
      "learning_rate": 4.864709452768106e-06,
      "loss": 0.8519,
      "step": 24658
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0132267475128174,
      "learning_rate": 4.8640705363703164e-06,
      "loss": 0.8151,
      "step": 24659
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.765185832977295,
      "learning_rate": 4.863431619972527e-06,
      "loss": 1.058,
      "step": 24660
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.091719150543213,
      "learning_rate": 4.862792703574738e-06,
      "loss": 0.8552,
      "step": 24661
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8056172132492065,
      "learning_rate": 4.862153787176948e-06,
      "loss": 0.8663,
      "step": 24662
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.9677670001983643,
      "learning_rate": 4.861514870779158e-06,
      "loss": 0.8586,
      "step": 24663
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8360874652862549,
      "learning_rate": 4.860875954381369e-06,
      "loss": 0.8572,
      "step": 24664
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.0974161624908447,
      "learning_rate": 4.86023703798358e-06,
      "loss": 1.067,
      "step": 24665
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.837161660194397,
      "learning_rate": 4.8595981215857905e-06,
      "loss": 1.1553,
      "step": 24666
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.05397629737854,
      "learning_rate": 4.858959205188001e-06,
      "loss": 0.7879,
      "step": 24667
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9211496114730835,
      "learning_rate": 4.858320288790212e-06,
      "loss": 0.84,
      "step": 24668
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.557651996612549,
      "learning_rate": 4.857681372392423e-06,
      "loss": 0.6489,
      "step": 24669
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.375214099884033,
      "learning_rate": 4.857042455994633e-06,
      "loss": 0.8568,
      "step": 24670
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5256458520889282,
      "learning_rate": 4.856403539596844e-06,
      "loss": 0.7564,
      "step": 24671
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.08715558052063,
      "learning_rate": 4.855764623199055e-06,
      "loss": 0.8718,
      "step": 24672
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.849709391593933,
      "learning_rate": 4.855125706801265e-06,
      "loss": 0.9339,
      "step": 24673
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.1884779930114746,
      "learning_rate": 4.854486790403476e-06,
      "loss": 0.5889,
      "step": 24674
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7982686758041382,
      "learning_rate": 4.853847874005687e-06,
      "loss": 1.0052,
      "step": 24675
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8411242961883545,
      "learning_rate": 4.8532089576078975e-06,
      "loss": 0.8056,
      "step": 24676
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.416628360748291,
      "learning_rate": 4.852570041210108e-06,
      "loss": 0.8653,
      "step": 24677
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9351253509521484,
      "learning_rate": 4.851931124812319e-06,
      "loss": 0.9215,
      "step": 24678
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7427347898483276,
      "learning_rate": 4.85129220841453e-06,
      "loss": 0.8518,
      "step": 24679
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5263233184814453,
      "learning_rate": 4.8506532920167395e-06,
      "loss": 0.8082,
      "step": 24680
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7958678007125854,
      "learning_rate": 4.85001437561895e-06,
      "loss": 0.7201,
      "step": 24681
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7079774141311646,
      "learning_rate": 4.849375459221161e-06,
      "loss": 0.926,
      "step": 24682
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7237768173217773,
      "learning_rate": 4.8487365428233716e-06,
      "loss": 0.8983,
      "step": 24683
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.787710428237915,
      "learning_rate": 4.848097626425582e-06,
      "loss": 0.9107,
      "step": 24684
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1931517124176025,
      "learning_rate": 4.847458710027793e-06,
      "loss": 0.7662,
      "step": 24685
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.2314977645874023,
      "learning_rate": 4.846819793630004e-06,
      "loss": 0.9376,
      "step": 24686
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7013750076293945,
      "learning_rate": 4.846180877232214e-06,
      "loss": 0.8359,
      "step": 24687
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1108415126800537,
      "learning_rate": 4.845541960834425e-06,
      "loss": 0.7917,
      "step": 24688
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0973691940307617,
      "learning_rate": 4.844903044436636e-06,
      "loss": 0.706,
      "step": 24689
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3771767616271973,
      "learning_rate": 4.8442641280388465e-06,
      "loss": 0.3487,
      "step": 24690
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7165011167526245,
      "learning_rate": 4.843625211641057e-06,
      "loss": 0.88,
      "step": 24691
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8456332683563232,
      "learning_rate": 4.842986295243268e-06,
      "loss": 0.9187,
      "step": 24692
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9558234214782715,
      "learning_rate": 4.8423473788454786e-06,
      "loss": 0.9397,
      "step": 24693
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7806257009506226,
      "learning_rate": 4.841708462447689e-06,
      "loss": 0.8016,
      "step": 24694
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6289230585098267,
      "learning_rate": 4.8410695460499e-06,
      "loss": 0.7594,
      "step": 24695
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4344407320022583,
      "learning_rate": 4.840430629652111e-06,
      "loss": 0.7887,
      "step": 24696
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.6304028034210205,
      "learning_rate": 4.839791713254321e-06,
      "loss": 0.6764,
      "step": 24697
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.598548412322998,
      "learning_rate": 4.839152796856531e-06,
      "loss": 0.6172,
      "step": 24698
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5047110319137573,
      "learning_rate": 4.838513880458742e-06,
      "loss": 0.8313,
      "step": 24699
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.035776138305664,
      "learning_rate": 4.837874964060953e-06,
      "loss": 1.0593,
      "step": 24700
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.594855785369873,
      "learning_rate": 4.837236047663163e-06,
      "loss": 0.642,
      "step": 24701
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4591113328933716,
      "learning_rate": 4.836597131265374e-06,
      "loss": 0.3026,
      "step": 24702
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9228435754776,
      "learning_rate": 4.835958214867585e-06,
      "loss": 0.8492,
      "step": 24703
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1419777870178223,
      "learning_rate": 4.835319298469795e-06,
      "loss": 0.9035,
      "step": 24704
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.156876564025879,
      "learning_rate": 4.834680382072006e-06,
      "loss": 1.5172,
      "step": 24705
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.361222982406616,
      "learning_rate": 4.834041465674217e-06,
      "loss": 0.5921,
      "step": 24706
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.1722819805145264,
      "learning_rate": 4.8334025492764275e-06,
      "loss": 0.6867,
      "step": 24707
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.3194077014923096,
      "learning_rate": 4.832763632878638e-06,
      "loss": 0.8847,
      "step": 24708
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.643598794937134,
      "learning_rate": 4.832124716480849e-06,
      "loss": 0.6022,
      "step": 24709
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.54108464717865,
      "learning_rate": 4.83148580008306e-06,
      "loss": 0.8142,
      "step": 24710
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7543729543685913,
      "learning_rate": 4.83084688368527e-06,
      "loss": 0.8429,
      "step": 24711
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8262219429016113,
      "learning_rate": 4.830207967287481e-06,
      "loss": 0.7829,
      "step": 24712
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.283320188522339,
      "learning_rate": 4.829569050889692e-06,
      "loss": 1.1273,
      "step": 24713
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5816959142684937,
      "learning_rate": 4.828930134491902e-06,
      "loss": 0.8447,
      "step": 24714
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0157830715179443,
      "learning_rate": 4.828291218094113e-06,
      "loss": 0.9371,
      "step": 24715
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0514075756073,
      "learning_rate": 4.827652301696323e-06,
      "loss": 0.8661,
      "step": 24716
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.7756662368774414,
      "learning_rate": 4.827013385298534e-06,
      "loss": 0.7348,
      "step": 24717
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0164201259613037,
      "learning_rate": 4.826374468900744e-06,
      "loss": 0.9153,
      "step": 24718
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6300898790359497,
      "learning_rate": 4.825735552502955e-06,
      "loss": 0.9355,
      "step": 24719
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7786842584609985,
      "learning_rate": 4.825096636105166e-06,
      "loss": 0.6407,
      "step": 24720
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5774990320205688,
      "learning_rate": 4.8244577197073765e-06,
      "loss": 0.4155,
      "step": 24721
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.843628406524658,
      "learning_rate": 4.823818803309587e-06,
      "loss": 1.3521,
      "step": 24722
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8484642505645752,
      "learning_rate": 4.823179886911798e-06,
      "loss": 0.8404,
      "step": 24723
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.755025029182434,
      "learning_rate": 4.8225409705140086e-06,
      "loss": 0.7946,
      "step": 24724
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.369838237762451,
      "learning_rate": 4.821902054116219e-06,
      "loss": 0.8125,
      "step": 24725
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.819555401802063,
      "learning_rate": 4.82126313771843e-06,
      "loss": 0.872,
      "step": 24726
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5670474767684937,
      "learning_rate": 4.820624221320641e-06,
      "loss": 0.751,
      "step": 24727
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4336405992507935,
      "learning_rate": 4.819985304922851e-06,
      "loss": 0.7022,
      "step": 24728
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6955803632736206,
      "learning_rate": 4.819346388525062e-06,
      "loss": 0.829,
      "step": 24729
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.52338171005249,
      "learning_rate": 4.818707472127273e-06,
      "loss": 1.4303,
      "step": 24730
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0883116722106934,
      "learning_rate": 4.8180685557294835e-06,
      "loss": 0.8615,
      "step": 24731
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.73502254486084,
      "learning_rate": 4.817429639331694e-06,
      "loss": 1.2737,
      "step": 24732
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7016671895980835,
      "learning_rate": 4.816790722933905e-06,
      "loss": 0.8425,
      "step": 24733
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2821097373962402,
      "learning_rate": 4.8161518065361156e-06,
      "loss": 0.9285,
      "step": 24734
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0317912101745605,
      "learning_rate": 4.815512890138325e-06,
      "loss": 0.8458,
      "step": 24735
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6569229364395142,
      "learning_rate": 4.814873973740536e-06,
      "loss": 0.7824,
      "step": 24736
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.802468776702881,
      "learning_rate": 4.814235057342747e-06,
      "loss": 1.1485,
      "step": 24737
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9433517456054688,
      "learning_rate": 4.8135961409449575e-06,
      "loss": 1.1488,
      "step": 24738
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9261716604232788,
      "learning_rate": 4.812957224547168e-06,
      "loss": 0.9326,
      "step": 24739
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.123009443283081,
      "learning_rate": 4.812318308149379e-06,
      "loss": 0.9594,
      "step": 24740
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7771363258361816,
      "learning_rate": 4.81167939175159e-06,
      "loss": 0.7552,
      "step": 24741
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0008299350738525,
      "learning_rate": 4.8110404753538e-06,
      "loss": 0.5971,
      "step": 24742
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4390838146209717,
      "learning_rate": 4.810401558956011e-06,
      "loss": 0.7913,
      "step": 24743
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6368279457092285,
      "learning_rate": 4.809762642558222e-06,
      "loss": 0.4215,
      "step": 24744
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6140861511230469,
      "learning_rate": 4.809123726160432e-06,
      "loss": 0.3679,
      "step": 24745
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.532821178436279,
      "learning_rate": 4.808484809762643e-06,
      "loss": 0.878,
      "step": 24746
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9560985565185547,
      "learning_rate": 4.807845893364854e-06,
      "loss": 1.0959,
      "step": 24747
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3199347257614136,
      "learning_rate": 4.8072069769670645e-06,
      "loss": 0.4194,
      "step": 24748
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5353624820709229,
      "learning_rate": 4.806568060569275e-06,
      "loss": 0.7304,
      "step": 24749
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.483316421508789,
      "learning_rate": 4.805929144171486e-06,
      "loss": 0.9008,
      "step": 24750
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8361459970474243,
      "learning_rate": 4.805290227773697e-06,
      "loss": 0.884,
      "step": 24751
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8453470468521118,
      "learning_rate": 4.804651311375907e-06,
      "loss": 0.9058,
      "step": 24752
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5500595569610596,
      "learning_rate": 4.804012394978117e-06,
      "loss": 0.7235,
      "step": 24753
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.940979480743408,
      "learning_rate": 4.803373478580328e-06,
      "loss": 1.3416,
      "step": 24754
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7956702709197998,
      "learning_rate": 4.8027345621825386e-06,
      "loss": 0.9143,
      "step": 24755
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.494518518447876,
      "learning_rate": 4.802095645784749e-06,
      "loss": 0.8682,
      "step": 24756
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2701613903045654,
      "learning_rate": 4.80145672938696e-06,
      "loss": 0.663,
      "step": 24757
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.248602867126465,
      "learning_rate": 4.800817812989171e-06,
      "loss": 0.7913,
      "step": 24758
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6697582006454468,
      "learning_rate": 4.800178896591381e-06,
      "loss": 0.7247,
      "step": 24759
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.395613193511963,
      "learning_rate": 4.799539980193592e-06,
      "loss": 0.7421,
      "step": 24760
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.209558486938477,
      "learning_rate": 4.798901063795803e-06,
      "loss": 1.438,
      "step": 24761
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6675834655761719,
      "learning_rate": 4.7982621473980135e-06,
      "loss": 0.8391,
      "step": 24762
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.914316177368164,
      "learning_rate": 4.797623231000224e-06,
      "loss": 1.0303,
      "step": 24763
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.357496738433838,
      "learning_rate": 4.796984314602435e-06,
      "loss": 1.0911,
      "step": 24764
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9409362077713013,
      "learning_rate": 4.7963453982046456e-06,
      "loss": 0.9144,
      "step": 24765
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6324056386947632,
      "learning_rate": 4.795706481806856e-06,
      "loss": 0.8435,
      "step": 24766
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9774916172027588,
      "learning_rate": 4.795067565409067e-06,
      "loss": 0.8108,
      "step": 24767
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3249183893203735,
      "learning_rate": 4.794428649011278e-06,
      "loss": 0.3574,
      "step": 24768
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3338390588760376,
      "learning_rate": 4.793789732613488e-06,
      "loss": 0.7233,
      "step": 24769
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6253324747085571,
      "learning_rate": 4.793150816215699e-06,
      "loss": 0.8424,
      "step": 24770
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.059410572052002,
      "learning_rate": 4.792511899817909e-06,
      "loss": 0.9981,
      "step": 24771
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.9496049880981445,
      "learning_rate": 4.79187298342012e-06,
      "loss": 0.8247,
      "step": 24772
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.73073148727417,
      "learning_rate": 4.79123406702233e-06,
      "loss": 0.9016,
      "step": 24773
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7672878503799438,
      "learning_rate": 4.790595150624541e-06,
      "loss": 0.8497,
      "step": 24774
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.686995267868042,
      "learning_rate": 4.789956234226752e-06,
      "loss": 0.81,
      "step": 24775
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6783416271209717,
      "learning_rate": 4.789317317828962e-06,
      "loss": 0.8725,
      "step": 24776
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9387074708938599,
      "learning_rate": 4.788678401431173e-06,
      "loss": 0.7962,
      "step": 24777
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.981970191001892,
      "learning_rate": 4.788039485033384e-06,
      "loss": 0.9483,
      "step": 24778
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9130795001983643,
      "learning_rate": 4.7874005686355945e-06,
      "loss": 0.8601,
      "step": 24779
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9475246667861938,
      "learning_rate": 4.786761652237805e-06,
      "loss": 1.0516,
      "step": 24780
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.747020959854126,
      "learning_rate": 4.786122735840016e-06,
      "loss": 0.6115,
      "step": 24781
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5646741390228271,
      "learning_rate": 4.785483819442227e-06,
      "loss": 0.8279,
      "step": 24782
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.557165145874023,
      "learning_rate": 4.784844903044437e-06,
      "loss": 1.1132,
      "step": 24783
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9844813346862793,
      "learning_rate": 4.784205986646648e-06,
      "loss": 0.8778,
      "step": 24784
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.6011228561401367,
      "learning_rate": 4.783567070248859e-06,
      "loss": 0.7582,
      "step": 24785
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.437954902648926,
      "learning_rate": 4.782928153851069e-06,
      "loss": 1.2587,
      "step": 24786
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8635720014572144,
      "learning_rate": 4.78228923745328e-06,
      "loss": 0.9242,
      "step": 24787
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.618590831756592,
      "learning_rate": 4.781650321055491e-06,
      "loss": 0.8086,
      "step": 24788
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.095963954925537,
      "learning_rate": 4.781011404657701e-06,
      "loss": 1.1221,
      "step": 24789
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.650537133216858,
      "learning_rate": 4.780372488259911e-06,
      "loss": 0.8509,
      "step": 24790
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6375815868377686,
      "learning_rate": 4.779733571862122e-06,
      "loss": 0.9134,
      "step": 24791
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.62369966506958,
      "learning_rate": 4.779094655464333e-06,
      "loss": 0.7433,
      "step": 24792
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2422993183135986,
      "learning_rate": 4.7784557390665435e-06,
      "loss": 0.7948,
      "step": 24793
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.574903130531311,
      "learning_rate": 4.777816822668754e-06,
      "loss": 0.8895,
      "step": 24794
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.269430637359619,
      "learning_rate": 4.777177906270965e-06,
      "loss": 0.8637,
      "step": 24795
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5757461786270142,
      "learning_rate": 4.7765389898731756e-06,
      "loss": 0.6955,
      "step": 24796
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1989433765411377,
      "learning_rate": 4.775900073475386e-06,
      "loss": 0.8784,
      "step": 24797
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.216240882873535,
      "learning_rate": 4.775261157077597e-06,
      "loss": 0.9552,
      "step": 24798
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.5263543128967285,
      "learning_rate": 4.774622240679808e-06,
      "loss": 0.7123,
      "step": 24799
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.6081383228302,
      "learning_rate": 4.773983324282018e-06,
      "loss": 0.812,
      "step": 24800
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5238100290298462,
      "learning_rate": 4.773344407884229e-06,
      "loss": 0.9643,
      "step": 24801
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.198680877685547,
      "learning_rate": 4.77270549148644e-06,
      "loss": 1.0011,
      "step": 24802
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4805723428726196,
      "learning_rate": 4.7720665750886505e-06,
      "loss": 0.7895,
      "step": 24803
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7584563493728638,
      "learning_rate": 4.771427658690861e-06,
      "loss": 0.861,
      "step": 24804
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.609027862548828,
      "learning_rate": 4.770788742293072e-06,
      "loss": 1.0398,
      "step": 24805
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.386789321899414,
      "learning_rate": 4.7701498258952826e-06,
      "loss": 0.9697,
      "step": 24806
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0051980018615723,
      "learning_rate": 4.769510909497493e-06,
      "loss": 0.9407,
      "step": 24807
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.335434675216675,
      "learning_rate": 4.768871993099703e-06,
      "loss": 0.9582,
      "step": 24808
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9363415241241455,
      "learning_rate": 4.768233076701914e-06,
      "loss": 0.8798,
      "step": 24809
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6786072254180908,
      "learning_rate": 4.7675941603041245e-06,
      "loss": 0.8427,
      "step": 24810
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2711021900177002,
      "learning_rate": 4.766955243906335e-06,
      "loss": 0.3376,
      "step": 24811
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0438289642333984,
      "learning_rate": 4.766316327508546e-06,
      "loss": 0.8638,
      "step": 24812
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.174854278564453,
      "learning_rate": 4.765677411110757e-06,
      "loss": 0.7867,
      "step": 24813
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9567896127700806,
      "learning_rate": 4.765038494712967e-06,
      "loss": 0.8075,
      "step": 24814
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6875975131988525,
      "learning_rate": 4.764399578315178e-06,
      "loss": 0.6419,
      "step": 24815
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7892602682113647,
      "learning_rate": 4.763760661917389e-06,
      "loss": 0.879,
      "step": 24816
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5588791370391846,
      "learning_rate": 4.763121745519599e-06,
      "loss": 0.7393,
      "step": 24817
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9533747434616089,
      "learning_rate": 4.76248282912181e-06,
      "loss": 1.1531,
      "step": 24818
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3326122760772705,
      "learning_rate": 4.761843912724021e-06,
      "loss": 0.8688,
      "step": 24819
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5120726823806763,
      "learning_rate": 4.7612049963262315e-06,
      "loss": 0.9034,
      "step": 24820
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3104196786880493,
      "learning_rate": 4.760566079928442e-06,
      "loss": 0.3724,
      "step": 24821
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8805721998214722,
      "learning_rate": 4.759927163530653e-06,
      "loss": 0.8734,
      "step": 24822
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.4690301418304443,
      "learning_rate": 4.759288247132864e-06,
      "loss": 0.9487,
      "step": 24823
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8827723264694214,
      "learning_rate": 4.758649330735074e-06,
      "loss": 0.874,
      "step": 24824
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.615615725517273,
      "learning_rate": 4.758010414337285e-06,
      "loss": 0.4257,
      "step": 24825
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.200198650360107,
      "learning_rate": 4.757371497939495e-06,
      "loss": 0.9588,
      "step": 24826
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6436433792114258,
      "learning_rate": 4.7567325815417056e-06,
      "loss": 0.8173,
      "step": 24827
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1066651344299316,
      "learning_rate": 4.756093665143916e-06,
      "loss": 0.8704,
      "step": 24828
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.296809673309326,
      "learning_rate": 4.755454748746127e-06,
      "loss": 1.1747,
      "step": 24829
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7273539304733276,
      "learning_rate": 4.754815832348338e-06,
      "loss": 0.7467,
      "step": 24830
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.990724802017212,
      "learning_rate": 4.754176915950548e-06,
      "loss": 0.5126,
      "step": 24831
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.968138337135315,
      "learning_rate": 4.753537999552759e-06,
      "loss": 0.9686,
      "step": 24832
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.097002983093262,
      "learning_rate": 4.752899083154969e-06,
      "loss": 1.3802,
      "step": 24833
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1043248176574707,
      "learning_rate": 4.75226016675718e-06,
      "loss": 1.0135,
      "step": 24834
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3770956993103027,
      "learning_rate": 4.75162125035939e-06,
      "loss": 0.338,
      "step": 24835
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5582998991012573,
      "learning_rate": 4.750982333961601e-06,
      "loss": 0.7781,
      "step": 24836
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.7462244033813477,
      "learning_rate": 4.750343417563812e-06,
      "loss": 0.824,
      "step": 24837
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1613619327545166,
      "learning_rate": 4.7497045011660224e-06,
      "loss": 0.8341,
      "step": 24838
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6850030422210693,
      "learning_rate": 4.749065584768233e-06,
      "loss": 1.1074,
      "step": 24839
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9203120470046997,
      "learning_rate": 4.748426668370444e-06,
      "loss": 0.9526,
      "step": 24840
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.084991216659546,
      "learning_rate": 4.7477877519726545e-06,
      "loss": 0.6927,
      "step": 24841
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4237275123596191,
      "learning_rate": 4.747148835574865e-06,
      "loss": 0.8247,
      "step": 24842
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.455108642578125,
      "learning_rate": 4.746509919177076e-06,
      "loss": 0.8064,
      "step": 24843
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.767501950263977,
      "learning_rate": 4.745871002779287e-06,
      "loss": 0.8181,
      "step": 24844
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8168935775756836,
      "learning_rate": 4.745232086381497e-06,
      "loss": 1.2093,
      "step": 24845
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.009758234024048,
      "learning_rate": 4.744593169983708e-06,
      "loss": 0.7468,
      "step": 24846
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.139359474182129,
      "learning_rate": 4.743954253585919e-06,
      "loss": 1.41,
      "step": 24847
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.001176357269287,
      "learning_rate": 4.7433153371881294e-06,
      "loss": 0.8169,
      "step": 24848
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9272359609603882,
      "learning_rate": 4.74267642079034e-06,
      "loss": 0.9363,
      "step": 24849
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.674202919006348,
      "learning_rate": 4.74203750439255e-06,
      "loss": 1.8011,
      "step": 24850
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.2526979446411133,
      "learning_rate": 4.741398587994761e-06,
      "loss": 1.0271,
      "step": 24851
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.142301559448242,
      "learning_rate": 4.740759671596971e-06,
      "loss": 0.6575,
      "step": 24852
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8233450651168823,
      "learning_rate": 4.740120755199182e-06,
      "loss": 0.7371,
      "step": 24853
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.649996757507324,
      "learning_rate": 4.739481838801393e-06,
      "loss": 1.1458,
      "step": 24854
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.210946798324585,
      "learning_rate": 4.7388429224036035e-06,
      "loss": 1.0294,
      "step": 24855
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0154387950897217,
      "learning_rate": 4.738204006005814e-06,
      "loss": 0.9291,
      "step": 24856
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0659115314483643,
      "learning_rate": 4.737565089608025e-06,
      "loss": 0.7809,
      "step": 24857
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6492494344711304,
      "learning_rate": 4.736926173210236e-06,
      "loss": 0.8089,
      "step": 24858
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7019387483596802,
      "learning_rate": 4.736287256812446e-06,
      "loss": 0.941,
      "step": 24859
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.063082695007324,
      "learning_rate": 4.735648340414657e-06,
      "loss": 0.8657,
      "step": 24860
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.842340111732483,
      "learning_rate": 4.735009424016868e-06,
      "loss": 0.8694,
      "step": 24861
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.6861188411712646,
      "learning_rate": 4.734370507619078e-06,
      "loss": 0.8054,
      "step": 24862
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6394455432891846,
      "learning_rate": 4.733731591221289e-06,
      "loss": 0.8152,
      "step": 24863
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9167728424072266,
      "learning_rate": 4.7330926748235e-06,
      "loss": 0.8197,
      "step": 24864
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6504743099212646,
      "learning_rate": 4.7324537584257105e-06,
      "loss": 0.8277,
      "step": 24865
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3891702890396118,
      "learning_rate": 4.731814842027921e-06,
      "loss": 0.7897,
      "step": 24866
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.108720302581787,
      "learning_rate": 4.731175925630132e-06,
      "loss": 1.2065,
      "step": 24867
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8205147981643677,
      "learning_rate": 4.7305370092323426e-06,
      "loss": 0.875,
      "step": 24868
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9701741933822632,
      "learning_rate": 4.7298980928345524e-06,
      "loss": 0.8475,
      "step": 24869
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5900086164474487,
      "learning_rate": 4.729259176436763e-06,
      "loss": 0.7952,
      "step": 24870
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.654153347015381,
      "learning_rate": 4.728620260038974e-06,
      "loss": 0.9776,
      "step": 24871
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9575042724609375,
      "learning_rate": 4.7279813436411845e-06,
      "loss": 0.8946,
      "step": 24872
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.360300540924072,
      "learning_rate": 4.727342427243395e-06,
      "loss": 1.3696,
      "step": 24873
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.677902936935425,
      "learning_rate": 4.726703510845606e-06,
      "loss": 0.6511,
      "step": 24874
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.383495569229126,
      "learning_rate": 4.726064594447817e-06,
      "loss": 0.9384,
      "step": 24875
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.677203893661499,
      "learning_rate": 4.725425678050027e-06,
      "loss": 0.9651,
      "step": 24876
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.062180519104004,
      "learning_rate": 4.724786761652238e-06,
      "loss": 1.0214,
      "step": 24877
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.913430690765381,
      "learning_rate": 4.724147845254449e-06,
      "loss": 1.147,
      "step": 24878
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.5510189533233643,
      "learning_rate": 4.7235089288566594e-06,
      "loss": 0.7319,
      "step": 24879
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.945568799972534,
      "learning_rate": 4.72287001245887e-06,
      "loss": 1.1692,
      "step": 24880
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5404618978500366,
      "learning_rate": 4.722231096061081e-06,
      "loss": 0.3875,
      "step": 24881
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9907188415527344,
      "learning_rate": 4.7215921796632915e-06,
      "loss": 0.9029,
      "step": 24882
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.415555000305176,
      "learning_rate": 4.720953263265502e-06,
      "loss": 1.1348,
      "step": 24883
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.3549644947052,
      "learning_rate": 4.720314346867713e-06,
      "loss": 0.9103,
      "step": 24884
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.827900767326355,
      "learning_rate": 4.719675430469924e-06,
      "loss": 0.7904,
      "step": 24885
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.775775909423828,
      "learning_rate": 4.719036514072134e-06,
      "loss": 0.8082,
      "step": 24886
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.8715813159942627,
      "learning_rate": 4.718397597674344e-06,
      "loss": 0.8461,
      "step": 24887
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8413809537887573,
      "learning_rate": 4.717758681276555e-06,
      "loss": 0.8151,
      "step": 24888
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7034188508987427,
      "learning_rate": 4.717119764878766e-06,
      "loss": 0.8072,
      "step": 24889
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8368730545043945,
      "learning_rate": 4.716480848480976e-06,
      "loss": 0.8518,
      "step": 24890
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1093671321868896,
      "learning_rate": 4.715841932083187e-06,
      "loss": 0.7799,
      "step": 24891
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.4959309101104736,
      "learning_rate": 4.715203015685398e-06,
      "loss": 0.7791,
      "step": 24892
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6916980743408203,
      "learning_rate": 4.714564099287608e-06,
      "loss": 0.7659,
      "step": 24893
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9388530254364014,
      "learning_rate": 4.713925182889819e-06,
      "loss": 0.8839,
      "step": 24894
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8139852285385132,
      "learning_rate": 4.71328626649203e-06,
      "loss": 1.3283,
      "step": 24895
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6078133583068848,
      "learning_rate": 4.7126473500942405e-06,
      "loss": 0.7778,
      "step": 24896
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0461537837982178,
      "learning_rate": 4.712008433696451e-06,
      "loss": 0.9709,
      "step": 24897
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.430537223815918,
      "learning_rate": 4.711369517298662e-06,
      "loss": 0.7732,
      "step": 24898
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.488572835922241,
      "learning_rate": 4.710730600900873e-06,
      "loss": 1.1955,
      "step": 24899
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.515545606613159,
      "learning_rate": 4.710091684503083e-06,
      "loss": 0.9723,
      "step": 24900
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6246345043182373,
      "learning_rate": 4.709452768105294e-06,
      "loss": 0.7563,
      "step": 24901
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6753331422805786,
      "learning_rate": 4.708813851707505e-06,
      "loss": 0.7781,
      "step": 24902
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.786475658416748,
      "learning_rate": 4.708174935309715e-06,
      "loss": 0.7677,
      "step": 24903
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1168224811553955,
      "learning_rate": 4.707536018911926e-06,
      "loss": 1.0099,
      "step": 24904
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.093581438064575,
      "learning_rate": 4.706897102514136e-06,
      "loss": 0.9401,
      "step": 24905
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.08789324760437,
      "learning_rate": 4.706258186116347e-06,
      "loss": 0.8964,
      "step": 24906
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7031629085540771,
      "learning_rate": 4.705619269718557e-06,
      "loss": 0.8435,
      "step": 24907
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9459447860717773,
      "learning_rate": 4.704980353320768e-06,
      "loss": 0.8312,
      "step": 24908
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3922066688537598,
      "learning_rate": 4.704341436922979e-06,
      "loss": 1.0329,
      "step": 24909
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1785988807678223,
      "learning_rate": 4.7037025205251894e-06,
      "loss": 0.7982,
      "step": 24910
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8590893745422363,
      "learning_rate": 4.7030636041274e-06,
      "loss": 0.8607,
      "step": 24911
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.867663621902466,
      "learning_rate": 4.702424687729611e-06,
      "loss": 0.9799,
      "step": 24912
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.5604188442230225,
      "learning_rate": 4.7017857713318215e-06,
      "loss": 0.8381,
      "step": 24913
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.7213187217712402,
      "learning_rate": 4.701146854934032e-06,
      "loss": 1.537,
      "step": 24914
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4681519269943237,
      "learning_rate": 4.700507938536243e-06,
      "loss": 0.4473,
      "step": 24915
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.018187999725342,
      "learning_rate": 4.699869022138454e-06,
      "loss": 0.9579,
      "step": 24916
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6377567052841187,
      "learning_rate": 4.699230105740664e-06,
      "loss": 0.8289,
      "step": 24917
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6819548606872559,
      "learning_rate": 4.698591189342875e-06,
      "loss": 0.8114,
      "step": 24918
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.667114019393921,
      "learning_rate": 4.697952272945086e-06,
      "loss": 0.8654,
      "step": 24919
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.171124219894409,
      "learning_rate": 4.6973133565472964e-06,
      "loss": 0.7564,
      "step": 24920
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6934940814971924,
      "learning_rate": 4.696674440149507e-06,
      "loss": 0.8965,
      "step": 24921
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.523782253265381,
      "learning_rate": 4.696035523751718e-06,
      "loss": 1.1551,
      "step": 24922
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.567347526550293,
      "learning_rate": 4.695396607353928e-06,
      "loss": 0.8898,
      "step": 24923
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7016937732696533,
      "learning_rate": 4.694757690956138e-06,
      "loss": 0.8685,
      "step": 24924
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6491347551345825,
      "learning_rate": 4.694118774558349e-06,
      "loss": 1.1872,
      "step": 24925
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7427653074264526,
      "learning_rate": 4.69347985816056e-06,
      "loss": 0.8377,
      "step": 24926
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.23077392578125,
      "learning_rate": 4.6928409417627705e-06,
      "loss": 0.757,
      "step": 24927
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6824281215667725,
      "learning_rate": 4.692202025364981e-06,
      "loss": 0.8306,
      "step": 24928
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.51759934425354,
      "learning_rate": 4.691563108967192e-06,
      "loss": 1.0645,
      "step": 24929
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.788267970085144,
      "learning_rate": 4.690924192569403e-06,
      "loss": 0.8434,
      "step": 24930
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0191280841827393,
      "learning_rate": 4.690285276171613e-06,
      "loss": 1.1902,
      "step": 24931
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6576734781265259,
      "learning_rate": 4.689646359773824e-06,
      "loss": 0.8029,
      "step": 24932
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9169968366622925,
      "learning_rate": 4.689007443376035e-06,
      "loss": 0.7703,
      "step": 24933
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6703588962554932,
      "learning_rate": 4.688368526978245e-06,
      "loss": 0.7616,
      "step": 24934
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.14070463180542,
      "learning_rate": 4.687729610580456e-06,
      "loss": 1.1796,
      "step": 24935
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.726346492767334,
      "learning_rate": 4.687090694182667e-06,
      "loss": 1.4444,
      "step": 24936
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3803008794784546,
      "learning_rate": 4.6864517777848775e-06,
      "loss": 0.8047,
      "step": 24937
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9650367498397827,
      "learning_rate": 4.685812861387088e-06,
      "loss": 0.8141,
      "step": 24938
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6612553596496582,
      "learning_rate": 4.685173944989299e-06,
      "loss": 0.722,
      "step": 24939
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8639171123504639,
      "learning_rate": 4.68453502859151e-06,
      "loss": 0.7154,
      "step": 24940
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9523080587387085,
      "learning_rate": 4.68389611219372e-06,
      "loss": 0.9248,
      "step": 24941
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7741999626159668,
      "learning_rate": 4.68325719579593e-06,
      "loss": 0.7955,
      "step": 24942
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9725672006607056,
      "learning_rate": 4.682618279398141e-06,
      "loss": 1.0178,
      "step": 24943
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1748270988464355,
      "learning_rate": 4.6819793630003515e-06,
      "loss": 0.5801,
      "step": 24944
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7244000434875488,
      "learning_rate": 4.681340446602562e-06,
      "loss": 0.8705,
      "step": 24945
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.261049509048462,
      "learning_rate": 4.680701530204773e-06,
      "loss": 0.696,
      "step": 24946
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.113044261932373,
      "learning_rate": 4.680062613806984e-06,
      "loss": 1.3935,
      "step": 24947
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6560602188110352,
      "learning_rate": 4.679423697409194e-06,
      "loss": 0.7795,
      "step": 24948
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.132042646408081,
      "learning_rate": 4.678784781011405e-06,
      "loss": 0.9341,
      "step": 24949
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7700297832489014,
      "learning_rate": 4.678145864613616e-06,
      "loss": 0.8072,
      "step": 24950
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.012872695922852,
      "learning_rate": 4.6775069482158264e-06,
      "loss": 1.232,
      "step": 24951
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8785099983215332,
      "learning_rate": 4.676868031818037e-06,
      "loss": 1.0214,
      "step": 24952
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.447807788848877,
      "learning_rate": 4.676229115420248e-06,
      "loss": 0.6974,
      "step": 24953
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7162261009216309,
      "learning_rate": 4.6755901990224585e-06,
      "loss": 0.8778,
      "step": 24954
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7812633514404297,
      "learning_rate": 4.674951282624669e-06,
      "loss": 0.7664,
      "step": 24955
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1975555419921875,
      "learning_rate": 4.67431236622688e-06,
      "loss": 0.7951,
      "step": 24956
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2474238872528076,
      "learning_rate": 4.673673449829091e-06,
      "loss": 0.8784,
      "step": 24957
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.216015815734863,
      "learning_rate": 4.673034533431301e-06,
      "loss": 1.1739,
      "step": 24958
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.9931418895721436,
      "learning_rate": 4.672395617033512e-06,
      "loss": 1.1467,
      "step": 24959
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8948148488998413,
      "learning_rate": 4.671756700635722e-06,
      "loss": 1.0396,
      "step": 24960
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.812291145324707,
      "learning_rate": 4.671117784237933e-06,
      "loss": 0.683,
      "step": 24961
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.7788023948669434,
      "learning_rate": 4.670478867840143e-06,
      "loss": 0.9914,
      "step": 24962
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.555595636367798,
      "learning_rate": 4.669839951442354e-06,
      "loss": 0.5556,
      "step": 24963
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.752193808555603,
      "learning_rate": 4.669201035044565e-06,
      "loss": 0.8853,
      "step": 24964
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5667002201080322,
      "learning_rate": 4.668562118646775e-06,
      "loss": 0.8097,
      "step": 24965
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9802557229995728,
      "learning_rate": 4.667923202248986e-06,
      "loss": 0.9016,
      "step": 24966
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7305891513824463,
      "learning_rate": 4.667284285851197e-06,
      "loss": 0.7797,
      "step": 24967
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.116044521331787,
      "learning_rate": 4.6666453694534075e-06,
      "loss": 0.9236,
      "step": 24968
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.713631272315979,
      "learning_rate": 4.666006453055618e-06,
      "loss": 0.8439,
      "step": 24969
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.3439347743988037,
      "learning_rate": 4.665367536657829e-06,
      "loss": 0.8466,
      "step": 24970
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7474009990692139,
      "learning_rate": 4.66472862026004e-06,
      "loss": 0.8563,
      "step": 24971
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.060476064682007,
      "learning_rate": 4.66408970386225e-06,
      "loss": 0.9944,
      "step": 24972
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6765570640563965,
      "learning_rate": 4.663450787464461e-06,
      "loss": 0.8672,
      "step": 24973
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.231950283050537,
      "learning_rate": 4.662811871066672e-06,
      "loss": 1.213,
      "step": 24974
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.384188175201416,
      "learning_rate": 4.662172954668882e-06,
      "loss": 1.2791,
      "step": 24975
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6699687242507935,
      "learning_rate": 4.661534038271093e-06,
      "loss": 0.4023,
      "step": 24976
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1174371242523193,
      "learning_rate": 4.660895121873304e-06,
      "loss": 0.8895,
      "step": 24977
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.082887649536133,
      "learning_rate": 4.660256205475514e-06,
      "loss": 0.8,
      "step": 24978
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.992819309234619,
      "learning_rate": 4.659617289077724e-06,
      "loss": 1.2725,
      "step": 24979
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1939077377319336,
      "learning_rate": 4.658978372679935e-06,
      "loss": 0.9219,
      "step": 24980
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6016592979431152,
      "learning_rate": 4.658339456282146e-06,
      "loss": 0.7863,
      "step": 24981
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.9883275032043457,
      "learning_rate": 4.6577005398843564e-06,
      "loss": 1.0646,
      "step": 24982
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6772541999816895,
      "learning_rate": 4.657061623486567e-06,
      "loss": 0.8724,
      "step": 24983
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7594224214553833,
      "learning_rate": 4.656422707088778e-06,
      "loss": 0.8381,
      "step": 24984
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.016563892364502,
      "learning_rate": 4.6557837906909885e-06,
      "loss": 0.7589,
      "step": 24985
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9782657623291016,
      "learning_rate": 4.655144874293199e-06,
      "loss": 0.8686,
      "step": 24986
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.179048776626587,
      "learning_rate": 4.65450595789541e-06,
      "loss": 0.7668,
      "step": 24987
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.8172731399536133,
      "learning_rate": 4.653867041497621e-06,
      "loss": 0.7789,
      "step": 24988
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6088021993637085,
      "learning_rate": 4.653228125099831e-06,
      "loss": 0.8996,
      "step": 24989
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9167922735214233,
      "learning_rate": 4.652589208702042e-06,
      "loss": 0.9632,
      "step": 24990
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5700150728225708,
      "learning_rate": 4.651950292304253e-06,
      "loss": 0.832,
      "step": 24991
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.348475456237793,
      "learning_rate": 4.6513113759064634e-06,
      "loss": 0.7415,
      "step": 24992
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.0358452796936035,
      "learning_rate": 4.650672459508674e-06,
      "loss": 1.7919,
      "step": 24993
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.1632280349731445,
      "learning_rate": 4.650033543110885e-06,
      "loss": 0.6488,
      "step": 24994
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.311563491821289,
      "learning_rate": 4.6493946267130955e-06,
      "loss": 1.3202,
      "step": 24995
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8490056991577148,
      "learning_rate": 4.648755710315305e-06,
      "loss": 0.948,
      "step": 24996
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8924705982208252,
      "learning_rate": 4.648116793917516e-06,
      "loss": 0.8277,
      "step": 24997
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4730758666992188,
      "learning_rate": 4.647477877519727e-06,
      "loss": 0.8184,
      "step": 24998
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1946866512298584,
      "learning_rate": 4.6468389611219375e-06,
      "loss": 0.8105,
      "step": 24999
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3214468955993652,
      "learning_rate": 4.646200044724148e-06,
      "loss": 1.1107,
      "step": 25000
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6192777156829834,
      "learning_rate": 4.645561128326359e-06,
      "loss": 0.8886,
      "step": 25001
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6015408039093018,
      "learning_rate": 4.64492221192857e-06,
      "loss": 0.7813,
      "step": 25002
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.060575485229492,
      "learning_rate": 4.64428329553078e-06,
      "loss": 1.2108,
      "step": 25003
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6352059841156006,
      "learning_rate": 4.643644379132991e-06,
      "loss": 0.8022,
      "step": 25004
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.81475830078125,
      "learning_rate": 4.643005462735202e-06,
      "loss": 0.8753,
      "step": 25005
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.0264556407928467,
      "learning_rate": 4.642366546337412e-06,
      "loss": 0.6273,
      "step": 25006
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6799309253692627,
      "learning_rate": 4.641727629939623e-06,
      "loss": 0.9626,
      "step": 25007
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0388503074645996,
      "learning_rate": 4.641088713541834e-06,
      "loss": 0.9414,
      "step": 25008
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8155996799468994,
      "learning_rate": 4.6404497971440445e-06,
      "loss": 0.9081,
      "step": 25009
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6834509372711182,
      "learning_rate": 4.639810880746255e-06,
      "loss": 0.9162,
      "step": 25010
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.6047797203063965,
      "learning_rate": 4.639171964348466e-06,
      "loss": 1.4118,
      "step": 25011
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9492590427398682,
      "learning_rate": 4.638533047950677e-06,
      "loss": 0.9141,
      "step": 25012
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0667240619659424,
      "learning_rate": 4.637894131552887e-06,
      "loss": 0.8108,
      "step": 25013
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0708506107330322,
      "learning_rate": 4.637255215155098e-06,
      "loss": 0.8898,
      "step": 25014
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.882864952087402,
      "learning_rate": 4.636616298757308e-06,
      "loss": 1.2068,
      "step": 25015
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.738267183303833,
      "learning_rate": 4.6359773823595185e-06,
      "loss": 0.8473,
      "step": 25016
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.821887493133545,
      "learning_rate": 4.635338465961729e-06,
      "loss": 0.7654,
      "step": 25017
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0535337924957275,
      "learning_rate": 4.63469954956394e-06,
      "loss": 1.2324,
      "step": 25018
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.069986581802368,
      "learning_rate": 4.634060633166151e-06,
      "loss": 0.8048,
      "step": 25019
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3549389839172363,
      "learning_rate": 4.633421716768361e-06,
      "loss": 0.8971,
      "step": 25020
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.2509896755218506,
      "learning_rate": 4.632782800370572e-06,
      "loss": 1.0244,
      "step": 25021
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.447840690612793,
      "learning_rate": 4.632143883972783e-06,
      "loss": 1.0066,
      "step": 25022
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.043628454208374,
      "learning_rate": 4.6315049675749934e-06,
      "loss": 0.9185,
      "step": 25023
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.33207368850708,
      "learning_rate": 4.630866051177204e-06,
      "loss": 1.2318,
      "step": 25024
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.0737671852111816,
      "learning_rate": 4.630227134779415e-06,
      "loss": 0.9085,
      "step": 25025
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.060051918029785,
      "learning_rate": 4.629588218381625e-06,
      "loss": 0.8438,
      "step": 25026
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.446652889251709,
      "learning_rate": 4.628949301983835e-06,
      "loss": 0.3815,
      "step": 25027
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8582639694213867,
      "learning_rate": 4.628310385586046e-06,
      "loss": 0.7396,
      "step": 25028
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.443284511566162,
      "learning_rate": 4.627671469188257e-06,
      "loss": 1.5512,
      "step": 25029
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6633446216583252,
      "learning_rate": 4.6270325527904675e-06,
      "loss": 0.4687,
      "step": 25030
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8504360914230347,
      "learning_rate": 4.626393636392678e-06,
      "loss": 0.8903,
      "step": 25031
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7324110269546509,
      "learning_rate": 4.625754719994889e-06,
      "loss": 0.8362,
      "step": 25032
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.195528507232666,
      "learning_rate": 4.6251158035971e-06,
      "loss": 0.746,
      "step": 25033
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.038841724395752,
      "learning_rate": 4.62447688719931e-06,
      "loss": 0.976,
      "step": 25034
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.01806378364563,
      "learning_rate": 4.623837970801521e-06,
      "loss": 0.9525,
      "step": 25035
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.9650280475616455,
      "learning_rate": 4.623199054403732e-06,
      "loss": 0.9548,
      "step": 25036
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5739057064056396,
      "learning_rate": 4.622560138005942e-06,
      "loss": 0.7231,
      "step": 25037
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.574796438217163,
      "learning_rate": 4.621921221608153e-06,
      "loss": 0.8952,
      "step": 25038
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2684555053710938,
      "learning_rate": 4.621282305210363e-06,
      "loss": 0.825,
      "step": 25039
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0272371768951416,
      "learning_rate": 4.620643388812574e-06,
      "loss": 0.9486,
      "step": 25040
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0239429473876953,
      "learning_rate": 4.620004472414784e-06,
      "loss": 0.9076,
      "step": 25041
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6255708932876587,
      "learning_rate": 4.619365556016995e-06,
      "loss": 0.8332,
      "step": 25042
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.594163179397583,
      "learning_rate": 4.618726639619206e-06,
      "loss": 1.1702,
      "step": 25043
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.907589316368103,
      "learning_rate": 4.6180877232214165e-06,
      "loss": 0.8807,
      "step": 25044
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3385651111602783,
      "learning_rate": 4.617448806823627e-06,
      "loss": 0.7029,
      "step": 25045
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.666822910308838,
      "learning_rate": 4.616809890425838e-06,
      "loss": 0.8059,
      "step": 25046
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.220309019088745,
      "learning_rate": 4.6161709740280486e-06,
      "loss": 0.8649,
      "step": 25047
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.471649646759033,
      "learning_rate": 4.615532057630259e-06,
      "loss": 1.3487,
      "step": 25048
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6686830520629883,
      "learning_rate": 4.61489314123247e-06,
      "loss": 0.7559,
      "step": 25049
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5208474397659302,
      "learning_rate": 4.614254224834681e-06,
      "loss": 0.8005,
      "step": 25050
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8097151517868042,
      "learning_rate": 4.613615308436891e-06,
      "loss": 0.927,
      "step": 25051
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7405449151992798,
      "learning_rate": 4.612976392039102e-06,
      "loss": 0.7776,
      "step": 25052
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.448741912841797,
      "learning_rate": 4.612337475641313e-06,
      "loss": 0.8918,
      "step": 25053
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.8099215030670166,
      "learning_rate": 4.6116985592435235e-06,
      "loss": 0.7841,
      "step": 25054
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.696071982383728,
      "learning_rate": 4.611059642845734e-06,
      "loss": 0.9156,
      "step": 25055
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7279894351959229,
      "learning_rate": 4.610420726447945e-06,
      "loss": 0.7786,
      "step": 25056
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2758166790008545,
      "learning_rate": 4.609781810050155e-06,
      "loss": 0.8441,
      "step": 25057
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.213688850402832,
      "learning_rate": 4.609142893652365e-06,
      "loss": 0.8447,
      "step": 25058
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.996721029281616,
      "learning_rate": 4.608503977254576e-06,
      "loss": 1.1254,
      "step": 25059
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.64304518699646,
      "learning_rate": 4.607865060856787e-06,
      "loss": 0.8202,
      "step": 25060
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7009021043777466,
      "learning_rate": 4.6072261444589975e-06,
      "loss": 0.8536,
      "step": 25061
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.803930401802063,
      "learning_rate": 4.606587228061208e-06,
      "loss": 0.8793,
      "step": 25062
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3193142414093018,
      "learning_rate": 4.605948311663419e-06,
      "loss": 0.5314,
      "step": 25063
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9329423904418945,
      "learning_rate": 4.60530939526563e-06,
      "loss": 0.8308,
      "step": 25064
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7987966537475586,
      "learning_rate": 4.60467047886784e-06,
      "loss": 0.8984,
      "step": 25065
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.839266061782837,
      "learning_rate": 4.604031562470051e-06,
      "loss": 0.892,
      "step": 25066
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1815452575683594,
      "learning_rate": 4.603392646072262e-06,
      "loss": 0.6834,
      "step": 25067
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.821628451347351,
      "learning_rate": 4.602753729674472e-06,
      "loss": 0.8825,
      "step": 25068
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.16324520111084,
      "learning_rate": 4.602114813276683e-06,
      "loss": 0.8947,
      "step": 25069
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.834812045097351,
      "learning_rate": 4.601475896878894e-06,
      "loss": 0.932,
      "step": 25070
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6863914728164673,
      "learning_rate": 4.6008369804811045e-06,
      "loss": 0.6867,
      "step": 25071
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.139659881591797,
      "learning_rate": 4.600198064083315e-06,
      "loss": 0.8532,
      "step": 25072
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6296347379684448,
      "learning_rate": 4.599559147685526e-06,
      "loss": 0.726,
      "step": 25073
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4206011295318604,
      "learning_rate": 4.598920231287737e-06,
      "loss": 0.8225,
      "step": 25074
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8038409948349,
      "learning_rate": 4.598281314889947e-06,
      "loss": 0.7538,
      "step": 25075
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.105813503265381,
      "learning_rate": 4.597642398492157e-06,
      "loss": 0.9707,
      "step": 25076
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3861368894577026,
      "learning_rate": 4.597003482094368e-06,
      "loss": 0.7915,
      "step": 25077
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.969416618347168,
      "learning_rate": 4.5963645656965786e-06,
      "loss": 0.9354,
      "step": 25078
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.149416446685791,
      "learning_rate": 4.595725649298789e-06,
      "loss": 0.8001,
      "step": 25079
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6442524194717407,
      "learning_rate": 4.595086732901e-06,
      "loss": 0.8529,
      "step": 25080
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6462819576263428,
      "learning_rate": 4.594447816503211e-06,
      "loss": 0.8618,
      "step": 25081
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4218734502792358,
      "learning_rate": 4.593808900105421e-06,
      "loss": 0.37,
      "step": 25082
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.786374568939209,
      "learning_rate": 4.593169983707632e-06,
      "loss": 0.7916,
      "step": 25083
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8749898672103882,
      "learning_rate": 4.592531067309843e-06,
      "loss": 0.6466,
      "step": 25084
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.68423330783844,
      "learning_rate": 4.5918921509120535e-06,
      "loss": 0.8222,
      "step": 25085
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6870015859603882,
      "learning_rate": 4.591253234514264e-06,
      "loss": 0.8706,
      "step": 25086
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.396322727203369,
      "learning_rate": 4.590614318116475e-06,
      "loss": 0.9115,
      "step": 25087
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7814873456954956,
      "learning_rate": 4.5899754017186856e-06,
      "loss": 0.8907,
      "step": 25088
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6830040216445923,
      "learning_rate": 4.589336485320896e-06,
      "loss": 0.7561,
      "step": 25089
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2852652072906494,
      "learning_rate": 4.588697568923107e-06,
      "loss": 0.8203,
      "step": 25090
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.013519287109375,
      "learning_rate": 4.588058652525318e-06,
      "loss": 0.9186,
      "step": 25091
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.056355953216553,
      "learning_rate": 4.587419736127528e-06,
      "loss": 1.2596,
      "step": 25092
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.78801429271698,
      "learning_rate": 4.586780819729739e-06,
      "loss": 1.075,
      "step": 25093
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7715208530426025,
      "learning_rate": 4.586141903331949e-06,
      "loss": 0.6605,
      "step": 25094
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7361407279968262,
      "learning_rate": 4.58550298693416e-06,
      "loss": 0.8852,
      "step": 25095
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1391308307647705,
      "learning_rate": 4.58486407053637e-06,
      "loss": 1.3096,
      "step": 25096
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6366914510726929,
      "learning_rate": 4.584225154138581e-06,
      "loss": 0.8803,
      "step": 25097
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3229598999023438,
      "learning_rate": 4.583586237740792e-06,
      "loss": 0.7293,
      "step": 25098
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.5585269927978516,
      "learning_rate": 4.582947321343002e-06,
      "loss": 1.3246,
      "step": 25099
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9918875694274902,
      "learning_rate": 4.582308404945213e-06,
      "loss": 0.8813,
      "step": 25100
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6211628913879395,
      "learning_rate": 4.581669488547424e-06,
      "loss": 0.7685,
      "step": 25101
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7889525890350342,
      "learning_rate": 4.5810305721496345e-06,
      "loss": 0.8133,
      "step": 25102
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7707164287567139,
      "learning_rate": 4.580391655751845e-06,
      "loss": 0.8844,
      "step": 25103
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.8033792972564697,
      "learning_rate": 4.579752739354056e-06,
      "loss": 0.695,
      "step": 25104
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0972838401794434,
      "learning_rate": 4.579113822956267e-06,
      "loss": 1.2716,
      "step": 25105
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1511623859405518,
      "learning_rate": 4.578474906558477e-06,
      "loss": 0.9419,
      "step": 25106
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2740318775177,
      "learning_rate": 4.577835990160688e-06,
      "loss": 1.1425,
      "step": 25107
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7050961256027222,
      "learning_rate": 4.577197073762899e-06,
      "loss": 0.8713,
      "step": 25108
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.7483341693878174,
      "learning_rate": 4.576558157365109e-06,
      "loss": 1.225,
      "step": 25109
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1569275856018066,
      "learning_rate": 4.57591924096732e-06,
      "loss": 0.8193,
      "step": 25110
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0803370475769043,
      "learning_rate": 4.575280324569531e-06,
      "loss": 1.0595,
      "step": 25111
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8086237907409668,
      "learning_rate": 4.574641408171741e-06,
      "loss": 0.8656,
      "step": 25112
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.037550210952759,
      "learning_rate": 4.574002491773951e-06,
      "loss": 0.8769,
      "step": 25113
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.9863877296447754,
      "learning_rate": 4.573363575376162e-06,
      "loss": 0.8737,
      "step": 25114
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.027115821838379,
      "learning_rate": 4.572724658978373e-06,
      "loss": 0.818,
      "step": 25115
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.625361680984497,
      "learning_rate": 4.5720857425805835e-06,
      "loss": 0.71,
      "step": 25116
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.669344425201416,
      "learning_rate": 4.571446826182794e-06,
      "loss": 0.9295,
      "step": 25117
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.58565092086792,
      "learning_rate": 4.570807909785005e-06,
      "loss": 1.1367,
      "step": 25118
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.870112180709839,
      "learning_rate": 4.5701689933872156e-06,
      "loss": 0.848,
      "step": 25119
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.0585825443267822,
      "learning_rate": 4.569530076989426e-06,
      "loss": 0.9783,
      "step": 25120
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5038888454437256,
      "learning_rate": 4.568891160591637e-06,
      "loss": 0.3355,
      "step": 25121
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6742936372756958,
      "learning_rate": 4.568252244193848e-06,
      "loss": 0.8009,
      "step": 25122
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.381571054458618,
      "learning_rate": 4.567613327796058e-06,
      "loss": 0.9164,
      "step": 25123
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.352052927017212,
      "learning_rate": 4.566974411398269e-06,
      "loss": 0.8452,
      "step": 25124
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1620943546295166,
      "learning_rate": 4.56633549500048e-06,
      "loss": 0.9693,
      "step": 25125
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.462417721748352,
      "learning_rate": 4.5656965786026905e-06,
      "loss": 0.7518,
      "step": 25126
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8909813165664673,
      "learning_rate": 4.565057662204901e-06,
      "loss": 0.916,
      "step": 25127
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.176854372024536,
      "learning_rate": 4.564418745807112e-06,
      "loss": 0.9331,
      "step": 25128
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9209824800491333,
      "learning_rate": 4.5637798294093226e-06,
      "loss": 0.8419,
      "step": 25129
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6366666555404663,
      "learning_rate": 4.563140913011532e-06,
      "loss": 0.8787,
      "step": 25130
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6434837579727173,
      "learning_rate": 4.562501996613743e-06,
      "loss": 0.4606,
      "step": 25131
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.044915199279785,
      "learning_rate": 4.561863080215954e-06,
      "loss": 1.3419,
      "step": 25132
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.80201256275177,
      "learning_rate": 4.5612241638181645e-06,
      "loss": 0.8226,
      "step": 25133
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1953370571136475,
      "learning_rate": 4.560585247420375e-06,
      "loss": 0.8684,
      "step": 25134
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.898398995399475,
      "learning_rate": 4.559946331022586e-06,
      "loss": 0.8548,
      "step": 25135
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4850738048553467,
      "learning_rate": 4.559307414624797e-06,
      "loss": 0.756,
      "step": 25136
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0402681827545166,
      "learning_rate": 4.558668498227007e-06,
      "loss": 0.884,
      "step": 25137
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0129947662353516,
      "learning_rate": 4.558029581829218e-06,
      "loss": 1.1374,
      "step": 25138
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7180017232894897,
      "learning_rate": 4.557390665431429e-06,
      "loss": 0.8968,
      "step": 25139
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.9732913970947266,
      "learning_rate": 4.556751749033639e-06,
      "loss": 1.1331,
      "step": 25140
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.127974271774292,
      "learning_rate": 4.55611283263585e-06,
      "loss": 0.9823,
      "step": 25141
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5971786975860596,
      "learning_rate": 4.555473916238061e-06,
      "loss": 0.7017,
      "step": 25142
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7928909063339233,
      "learning_rate": 4.5548349998402715e-06,
      "loss": 0.9291,
      "step": 25143
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8772531747817993,
      "learning_rate": 4.554196083442482e-06,
      "loss": 0.9348,
      "step": 25144
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5479665994644165,
      "learning_rate": 4.553557167044693e-06,
      "loss": 0.7976,
      "step": 25145
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.284657001495361,
      "learning_rate": 4.552918250646904e-06,
      "loss": 1.7076,
      "step": 25146
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0658140182495117,
      "learning_rate": 4.552279334249114e-06,
      "loss": 0.8735,
      "step": 25147
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8953628540039062,
      "learning_rate": 4.551640417851325e-06,
      "loss": 1.1472,
      "step": 25148
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9847458600997925,
      "learning_rate": 4.551001501453535e-06,
      "loss": 0.748,
      "step": 25149
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0933315753936768,
      "learning_rate": 4.5503625850557456e-06,
      "loss": 0.824,
      "step": 25150
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6217422485351562,
      "learning_rate": 4.549723668657956e-06,
      "loss": 0.7609,
      "step": 25151
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0312094688415527,
      "learning_rate": 4.549084752260167e-06,
      "loss": 0.8567,
      "step": 25152
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0931060314178467,
      "learning_rate": 4.548445835862378e-06,
      "loss": 0.6451,
      "step": 25153
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.607039213180542,
      "learning_rate": 4.547806919464588e-06,
      "loss": 0.7646,
      "step": 25154
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.839475393295288,
      "learning_rate": 4.547168003066799e-06,
      "loss": 0.7698,
      "step": 25155
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.439621925354004,
      "learning_rate": 4.54652908666901e-06,
      "loss": 0.3271,
      "step": 25156
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.481224775314331,
      "learning_rate": 4.5458901702712205e-06,
      "loss": 0.9057,
      "step": 25157
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8690433502197266,
      "learning_rate": 4.545251253873431e-06,
      "loss": 0.9328,
      "step": 25158
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3871986865997314,
      "learning_rate": 4.544612337475642e-06,
      "loss": 0.5749,
      "step": 25159
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1212158203125,
      "learning_rate": 4.5439734210778526e-06,
      "loss": 0.9592,
      "step": 25160
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2783138751983643,
      "learning_rate": 4.543334504680063e-06,
      "loss": 0.8387,
      "step": 25161
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.796749114990234,
      "learning_rate": 4.542695588282274e-06,
      "loss": 0.9162,
      "step": 25162
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2676571607589722,
      "learning_rate": 4.542056671884485e-06,
      "loss": 0.3761,
      "step": 25163
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.242065906524658,
      "learning_rate": 4.541417755486695e-06,
      "loss": 0.8491,
      "step": 25164
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.924791932106018,
      "learning_rate": 4.540778839088906e-06,
      "loss": 0.7593,
      "step": 25165
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6561145782470703,
      "learning_rate": 4.540139922691117e-06,
      "loss": 0.8123,
      "step": 25166
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6533926725387573,
      "learning_rate": 4.539501006293327e-06,
      "loss": 0.7511,
      "step": 25167
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.715227484703064,
      "learning_rate": 4.538862089895537e-06,
      "loss": 0.8368,
      "step": 25168
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0740418434143066,
      "learning_rate": 4.538223173497748e-06,
      "loss": 0.8893,
      "step": 25169
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.157271385192871,
      "learning_rate": 4.537584257099959e-06,
      "loss": 1.1945,
      "step": 25170
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.344188690185547,
      "learning_rate": 4.536945340702169e-06,
      "loss": 0.6553,
      "step": 25171
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7312777042388916,
      "learning_rate": 4.53630642430438e-06,
      "loss": 0.8242,
      "step": 25172
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5040245056152344,
      "learning_rate": 4.535667507906591e-06,
      "loss": 1.0166,
      "step": 25173
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8059693574905396,
      "learning_rate": 4.5350285915088015e-06,
      "loss": 0.824,
      "step": 25174
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7581206560134888,
      "learning_rate": 4.534389675111012e-06,
      "loss": 0.7873,
      "step": 25175
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.034416913986206,
      "learning_rate": 4.533750758713223e-06,
      "loss": 0.8296,
      "step": 25176
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2574191093444824,
      "learning_rate": 4.533111842315434e-06,
      "loss": 0.9091,
      "step": 25177
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1012840270996094,
      "learning_rate": 4.532472925917644e-06,
      "loss": 1.216,
      "step": 25178
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6535584926605225,
      "learning_rate": 4.531834009519855e-06,
      "loss": 0.8439,
      "step": 25179
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.379953145980835,
      "learning_rate": 4.531195093122066e-06,
      "loss": 0.5395,
      "step": 25180
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6610437631607056,
      "learning_rate": 4.530556176724276e-06,
      "loss": 0.8554,
      "step": 25181
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.136793613433838,
      "learning_rate": 4.529917260326487e-06,
      "loss": 0.8542,
      "step": 25182
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8778576850891113,
      "learning_rate": 4.529278343928698e-06,
      "loss": 0.8747,
      "step": 25183
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7122724056243896,
      "learning_rate": 4.5286394275309085e-06,
      "loss": 0.8298,
      "step": 25184
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.13513445854187,
      "learning_rate": 4.528000511133118e-06,
      "loss": 0.9493,
      "step": 25185
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.2303152084350586,
      "learning_rate": 4.527361594735329e-06,
      "loss": 0.6741,
      "step": 25186
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.283616065979004,
      "learning_rate": 4.52672267833754e-06,
      "loss": 0.9275,
      "step": 25187
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.906794786453247,
      "learning_rate": 4.5260837619397505e-06,
      "loss": 0.9336,
      "step": 25188
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2456910610198975,
      "learning_rate": 4.525444845541961e-06,
      "loss": 0.79,
      "step": 25189
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8323519229888916,
      "learning_rate": 4.524805929144172e-06,
      "loss": 1.1172,
      "step": 25190
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2825307846069336,
      "learning_rate": 4.5241670127463826e-06,
      "loss": 0.7375,
      "step": 25191
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9872907400131226,
      "learning_rate": 4.523528096348593e-06,
      "loss": 0.7379,
      "step": 25192
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.645790457725525,
      "learning_rate": 4.522889179950804e-06,
      "loss": 0.8689,
      "step": 25193
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7234998941421509,
      "learning_rate": 4.522250263553015e-06,
      "loss": 0.829,
      "step": 25194
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8817147016525269,
      "learning_rate": 4.521611347155225e-06,
      "loss": 0.9282,
      "step": 25195
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2215709686279297,
      "learning_rate": 4.520972430757436e-06,
      "loss": 0.8799,
      "step": 25196
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.471273183822632,
      "learning_rate": 4.520333514359647e-06,
      "loss": 1.0469,
      "step": 25197
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7274327278137207,
      "learning_rate": 4.5196945979618575e-06,
      "loss": 0.8068,
      "step": 25198
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5672916173934937,
      "learning_rate": 4.519055681564068e-06,
      "loss": 0.7559,
      "step": 25199
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9667940139770508,
      "learning_rate": 4.518416765166279e-06,
      "loss": 0.8444,
      "step": 25200
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.152808427810669,
      "learning_rate": 4.5177778487684896e-06,
      "loss": 0.7636,
      "step": 25201
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9226711988449097,
      "learning_rate": 4.5171389323707e-06,
      "loss": 0.763,
      "step": 25202
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.791715145111084,
      "learning_rate": 4.516500015972911e-06,
      "loss": 1.167,
      "step": 25203
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6117717027664185,
      "learning_rate": 4.515861099575121e-06,
      "loss": 0.8501,
      "step": 25204
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1170780658721924,
      "learning_rate": 4.5152221831773315e-06,
      "loss": 0.9994,
      "step": 25205
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7254949808120728,
      "learning_rate": 4.514583266779542e-06,
      "loss": 0.8159,
      "step": 25206
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.918139100074768,
      "learning_rate": 4.513944350381753e-06,
      "loss": 0.8703,
      "step": 25207
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.2096943855285645,
      "learning_rate": 4.513305433983964e-06,
      "loss": 1.1633,
      "step": 25208
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3866963386535645,
      "learning_rate": 4.512666517586174e-06,
      "loss": 0.9202,
      "step": 25209
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6472244262695312,
      "learning_rate": 4.512027601188385e-06,
      "loss": 0.6679,
      "step": 25210
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5282506942749023,
      "learning_rate": 4.511388684790596e-06,
      "loss": 0.9038,
      "step": 25211
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9502347707748413,
      "learning_rate": 4.510749768392806e-06,
      "loss": 0.7748,
      "step": 25212
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3912042379379272,
      "learning_rate": 4.510110851995017e-06,
      "loss": 0.8044,
      "step": 25213
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3559722900390625,
      "learning_rate": 4.509471935597228e-06,
      "loss": 0.8278,
      "step": 25214
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.6071979999542236,
      "learning_rate": 4.5088330191994385e-06,
      "loss": 1.0656,
      "step": 25215
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.4975807666778564,
      "learning_rate": 4.508194102801649e-06,
      "loss": 0.9073,
      "step": 25216
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.305248260498047,
      "learning_rate": 4.50755518640386e-06,
      "loss": 0.7507,
      "step": 25217
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.4259326457977295,
      "learning_rate": 4.506916270006071e-06,
      "loss": 1.2799,
      "step": 25218
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3832170963287354,
      "learning_rate": 4.506277353608281e-06,
      "loss": 0.3558,
      "step": 25219
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8936347961425781,
      "learning_rate": 4.505638437210491e-06,
      "loss": 0.888,
      "step": 25220
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6260097026824951,
      "learning_rate": 4.504999520812702e-06,
      "loss": 0.816,
      "step": 25221
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6107397079467773,
      "learning_rate": 4.5043606044149126e-06,
      "loss": 0.8198,
      "step": 25222
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.477670907974243,
      "learning_rate": 4.503721688017123e-06,
      "loss": 0.813,
      "step": 25223
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.087738037109375,
      "learning_rate": 4.503082771619334e-06,
      "loss": 1.4529,
      "step": 25224
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3351857662200928,
      "learning_rate": 4.502443855221545e-06,
      "loss": 0.8557,
      "step": 25225
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7759596109390259,
      "learning_rate": 4.501804938823755e-06,
      "loss": 0.7999,
      "step": 25226
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.668874502182007,
      "learning_rate": 4.501166022425966e-06,
      "loss": 1.0681,
      "step": 25227
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.294628143310547,
      "learning_rate": 4.500527106028176e-06,
      "loss": 0.706,
      "step": 25228
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7162058353424072,
      "learning_rate": 4.499888189630387e-06,
      "loss": 0.7552,
      "step": 25229
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0111634731292725,
      "learning_rate": 4.499249273232597e-06,
      "loss": 0.8251,
      "step": 25230
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5261553525924683,
      "learning_rate": 4.498610356834808e-06,
      "loss": 0.8042,
      "step": 25231
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.059041976928711,
      "learning_rate": 4.497971440437019e-06,
      "loss": 0.9345,
      "step": 25232
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.3749115467071533,
      "learning_rate": 4.4973325240392294e-06,
      "loss": 0.8647,
      "step": 25233
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7724158763885498,
      "learning_rate": 4.49669360764144e-06,
      "loss": 0.9157,
      "step": 25234
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.075559616088867,
      "learning_rate": 4.496054691243651e-06,
      "loss": 0.8226,
      "step": 25235
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6719011068344116,
      "learning_rate": 4.4954157748458615e-06,
      "loss": 0.9252,
      "step": 25236
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6474796533584595,
      "learning_rate": 4.494776858448072e-06,
      "loss": 0.8167,
      "step": 25237
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9666591882705688,
      "learning_rate": 4.494137942050283e-06,
      "loss": 0.8768,
      "step": 25238
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7317241430282593,
      "learning_rate": 4.493499025652494e-06,
      "loss": 0.8976,
      "step": 25239
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.750020146369934,
      "learning_rate": 4.492860109254704e-06,
      "loss": 0.8015,
      "step": 25240
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3162884712219238,
      "learning_rate": 4.492221192856915e-06,
      "loss": 0.3548,
      "step": 25241
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9663960933685303,
      "learning_rate": 4.491582276459126e-06,
      "loss": 0.7826,
      "step": 25242
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0397112369537354,
      "learning_rate": 4.4909433600613364e-06,
      "loss": 0.8939,
      "step": 25243
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.0086050033569336,
      "learning_rate": 4.490304443663547e-06,
      "loss": 0.7742,
      "step": 25244
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.445042610168457,
      "learning_rate": 4.489665527265758e-06,
      "loss": 0.8457,
      "step": 25245
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1497936248779297,
      "learning_rate": 4.489026610867968e-06,
      "loss": 0.9404,
      "step": 25246
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.471279501914978,
      "learning_rate": 4.488387694470178e-06,
      "loss": 0.7607,
      "step": 25247
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.3513126373291016,
      "learning_rate": 4.487748778072389e-06,
      "loss": 0.905,
      "step": 25248
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.762283444404602,
      "learning_rate": 4.4871098616746e-06,
      "loss": 0.9087,
      "step": 25249
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.082470655441284,
      "learning_rate": 4.4864709452768105e-06,
      "loss": 0.883,
      "step": 25250
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.984910011291504,
      "learning_rate": 4.485832028879021e-06,
      "loss": 0.8696,
      "step": 25251
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.141756296157837,
      "learning_rate": 4.485193112481232e-06,
      "loss": 0.7796,
      "step": 25252
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.714844226837158,
      "learning_rate": 4.484554196083443e-06,
      "loss": 1.1362,
      "step": 25253
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.269580364227295,
      "learning_rate": 4.483915279685653e-06,
      "loss": 0.8122,
      "step": 25254
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4649149179458618,
      "learning_rate": 4.483276363287864e-06,
      "loss": 0.3934,
      "step": 25255
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4664820432662964,
      "learning_rate": 4.482637446890075e-06,
      "loss": 1.0962,
      "step": 25256
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5310614109039307,
      "learning_rate": 4.481998530492285e-06,
      "loss": 0.817,
      "step": 25257
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5259605646133423,
      "learning_rate": 4.481359614094496e-06,
      "loss": 0.7639,
      "step": 25258
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7279819250106812,
      "learning_rate": 4.480720697696707e-06,
      "loss": 0.8785,
      "step": 25259
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.49573814868927,
      "learning_rate": 4.4800817812989175e-06,
      "loss": 0.3223,
      "step": 25260
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.013208389282227,
      "learning_rate": 4.479442864901128e-06,
      "loss": 1.3864,
      "step": 25261
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.0129027366638184,
      "learning_rate": 4.478803948503339e-06,
      "loss": 1.0327,
      "step": 25262
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.066026449203491,
      "learning_rate": 4.47816503210555e-06,
      "loss": 0.8409,
      "step": 25263
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3920921087265015,
      "learning_rate": 4.47752611570776e-06,
      "loss": 0.3899,
      "step": 25264
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.924529552459717,
      "learning_rate": 4.47688719930997e-06,
      "loss": 0.5964,
      "step": 25265
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6622438430786133,
      "learning_rate": 4.476248282912181e-06,
      "loss": 0.7889,
      "step": 25266
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6835471391677856,
      "learning_rate": 4.4756093665143915e-06,
      "loss": 0.4913,
      "step": 25267
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1043455600738525,
      "learning_rate": 4.474970450116602e-06,
      "loss": 0.8364,
      "step": 25268
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.172415018081665,
      "learning_rate": 4.474331533718813e-06,
      "loss": 0.9098,
      "step": 25269
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.448502540588379,
      "learning_rate": 4.473692617321024e-06,
      "loss": 1.1856,
      "step": 25270
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.065079689025879,
      "learning_rate": 4.473053700923234e-06,
      "loss": 1.1185,
      "step": 25271
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4150630235671997,
      "learning_rate": 4.472414784525445e-06,
      "loss": 0.4752,
      "step": 25272
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1689839363098145,
      "learning_rate": 4.471775868127656e-06,
      "loss": 0.9414,
      "step": 25273
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.816685676574707,
      "learning_rate": 4.4711369517298664e-06,
      "loss": 0.8599,
      "step": 25274
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1713075637817383,
      "learning_rate": 4.470498035332077e-06,
      "loss": 1.2109,
      "step": 25275
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.959414005279541,
      "learning_rate": 4.469859118934288e-06,
      "loss": 1.1072,
      "step": 25276
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.335134744644165,
      "learning_rate": 4.4692202025364985e-06,
      "loss": 0.6603,
      "step": 25277
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.4692764282226562,
      "learning_rate": 4.468581286138709e-06,
      "loss": 1.0347,
      "step": 25278
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.739736557006836,
      "learning_rate": 4.46794236974092e-06,
      "loss": 0.5646,
      "step": 25279
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.393663167953491,
      "learning_rate": 4.467303453343131e-06,
      "loss": 0.721,
      "step": 25280
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1812703609466553,
      "learning_rate": 4.466664536945341e-06,
      "loss": 0.9921,
      "step": 25281
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3912665843963623,
      "learning_rate": 4.466025620547552e-06,
      "loss": 0.4894,
      "step": 25282
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7878130674362183,
      "learning_rate": 4.465386704149762e-06,
      "loss": 0.8101,
      "step": 25283
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.547379732131958,
      "learning_rate": 4.464747787751973e-06,
      "loss": 0.7719,
      "step": 25284
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8145290613174438,
      "learning_rate": 4.464108871354183e-06,
      "loss": 0.8895,
      "step": 25285
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.260449171066284,
      "learning_rate": 4.463469954956394e-06,
      "loss": 0.8432,
      "step": 25286
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8086812496185303,
      "learning_rate": 4.462831038558605e-06,
      "loss": 0.6864,
      "step": 25287
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7424513101577759,
      "learning_rate": 4.462192122160815e-06,
      "loss": 1.1395,
      "step": 25288
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.804564952850342,
      "learning_rate": 4.461553205763026e-06,
      "loss": 1.8377,
      "step": 25289
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.8843374252319336,
      "learning_rate": 4.460914289365237e-06,
      "loss": 1.1074,
      "step": 25290
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1159191131591797,
      "learning_rate": 4.4602753729674475e-06,
      "loss": 0.8201,
      "step": 25291
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9413882493972778,
      "learning_rate": 4.459636456569658e-06,
      "loss": 0.8193,
      "step": 25292
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.350473403930664,
      "learning_rate": 4.458997540171869e-06,
      "loss": 1.0696,
      "step": 25293
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.9232113361358643,
      "learning_rate": 4.45835862377408e-06,
      "loss": 1.3436,
      "step": 25294
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.134657621383667,
      "learning_rate": 4.45771970737629e-06,
      "loss": 0.7796,
      "step": 25295
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8674371242523193,
      "learning_rate": 4.457080790978501e-06,
      "loss": 1.0753,
      "step": 25296
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7457704544067383,
      "learning_rate": 4.456441874580712e-06,
      "loss": 0.8302,
      "step": 25297
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7899134159088135,
      "learning_rate": 4.455802958182922e-06,
      "loss": 0.9274,
      "step": 25298
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9167561531066895,
      "learning_rate": 4.455164041785133e-06,
      "loss": 0.6559,
      "step": 25299
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5693923234939575,
      "learning_rate": 4.454525125387344e-06,
      "loss": 0.7285,
      "step": 25300
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.544908046722412,
      "learning_rate": 4.453886208989554e-06,
      "loss": 0.7678,
      "step": 25301
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6327910423278809,
      "learning_rate": 4.453247292591764e-06,
      "loss": 0.7985,
      "step": 25302
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.31709623336792,
      "learning_rate": 4.452608376193975e-06,
      "loss": 0.9689,
      "step": 25303
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.793133020401001,
      "learning_rate": 4.451969459796186e-06,
      "loss": 0.8754,
      "step": 25304
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5349464416503906,
      "learning_rate": 4.4513305433983964e-06,
      "loss": 0.3996,
      "step": 25305
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8463072776794434,
      "learning_rate": 4.450691627000607e-06,
      "loss": 0.9377,
      "step": 25306
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9952315092086792,
      "learning_rate": 4.450052710602818e-06,
      "loss": 0.994,
      "step": 25307
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.563647985458374,
      "learning_rate": 4.4494137942050285e-06,
      "loss": 1.1901,
      "step": 25308
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.422135353088379,
      "learning_rate": 4.448774877807239e-06,
      "loss": 0.3236,
      "step": 25309
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9344221353530884,
      "learning_rate": 4.44813596140945e-06,
      "loss": 0.7998,
      "step": 25310
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4506484270095825,
      "learning_rate": 4.447497045011661e-06,
      "loss": 0.3603,
      "step": 25311
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.250319719314575,
      "learning_rate": 4.446858128613871e-06,
      "loss": 0.8812,
      "step": 25312
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5760754346847534,
      "learning_rate": 4.446219212216082e-06,
      "loss": 0.8204,
      "step": 25313
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.497128963470459,
      "learning_rate": 4.445580295818293e-06,
      "loss": 0.6482,
      "step": 25314
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7338433265686035,
      "learning_rate": 4.4449413794205034e-06,
      "loss": 1.0868,
      "step": 25315
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.357819080352783,
      "learning_rate": 4.444302463022714e-06,
      "loss": 0.5148,
      "step": 25316
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.321159601211548,
      "learning_rate": 4.443663546624925e-06,
      "loss": 0.8463,
      "step": 25317
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.670400857925415,
      "learning_rate": 4.4430246302271355e-06,
      "loss": 0.8964,
      "step": 25318
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5997161865234375,
      "learning_rate": 4.442385713829345e-06,
      "loss": 0.7884,
      "step": 25319
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7097848653793335,
      "learning_rate": 4.441746797431556e-06,
      "loss": 0.889,
      "step": 25320
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7644084692001343,
      "learning_rate": 4.441107881033767e-06,
      "loss": 0.8684,
      "step": 25321
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.406421422958374,
      "learning_rate": 4.4404689646359775e-06,
      "loss": 0.7894,
      "step": 25322
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8796777725219727,
      "learning_rate": 4.439830048238188e-06,
      "loss": 0.8721,
      "step": 25323
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7744905948638916,
      "learning_rate": 4.439191131840399e-06,
      "loss": 0.8831,
      "step": 25324
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.042452573776245,
      "learning_rate": 4.43855221544261e-06,
      "loss": 0.8221,
      "step": 25325
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7101035118103027,
      "learning_rate": 4.43791329904482e-06,
      "loss": 0.8709,
      "step": 25326
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.645506739616394,
      "learning_rate": 4.437274382647031e-06,
      "loss": 0.8693,
      "step": 25327
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5731091499328613,
      "learning_rate": 4.436635466249242e-06,
      "loss": 0.8541,
      "step": 25328
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.351990222930908,
      "learning_rate": 4.435996549851452e-06,
      "loss": 0.8961,
      "step": 25329
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5857406854629517,
      "learning_rate": 4.435357633453663e-06,
      "loss": 0.9347,
      "step": 25330
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.4764790534973145,
      "learning_rate": 4.434718717055874e-06,
      "loss": 1.3958,
      "step": 25331
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.032948017120361,
      "learning_rate": 4.4340798006580845e-06,
      "loss": 1.1412,
      "step": 25332
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.790842056274414,
      "learning_rate": 4.433440884260295e-06,
      "loss": 0.9725,
      "step": 25333
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.5519752502441406,
      "learning_rate": 4.432801967862506e-06,
      "loss": 1.6097,
      "step": 25334
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.174427509307861,
      "learning_rate": 4.432163051464717e-06,
      "loss": 0.7907,
      "step": 25335
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9111137390136719,
      "learning_rate": 4.431524135066927e-06,
      "loss": 0.934,
      "step": 25336
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7280491590499878,
      "learning_rate": 4.430885218669138e-06,
      "loss": 0.7395,
      "step": 25337
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9067286252975464,
      "learning_rate": 4.430246302271348e-06,
      "loss": 0.8837,
      "step": 25338
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.218221426010132,
      "learning_rate": 4.4296073858735585e-06,
      "loss": 0.7447,
      "step": 25339
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6922966241836548,
      "learning_rate": 4.428968469475769e-06,
      "loss": 0.7225,
      "step": 25340
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.680914282798767,
      "learning_rate": 4.42832955307798e-06,
      "loss": 0.9022,
      "step": 25341
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.00296688079834,
      "learning_rate": 4.427690636680191e-06,
      "loss": 0.9928,
      "step": 25342
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.07692551612854,
      "learning_rate": 4.427051720282401e-06,
      "loss": 0.8937,
      "step": 25343
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.186615467071533,
      "learning_rate": 4.426412803884612e-06,
      "loss": 0.8596,
      "step": 25344
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9945112466812134,
      "learning_rate": 4.425773887486823e-06,
      "loss": 0.9699,
      "step": 25345
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.052894353866577,
      "learning_rate": 4.4251349710890334e-06,
      "loss": 0.9083,
      "step": 25346
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0195109844207764,
      "learning_rate": 4.424496054691244e-06,
      "loss": 0.8858,
      "step": 25347
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6933977603912354,
      "learning_rate": 4.423857138293455e-06,
      "loss": 0.8145,
      "step": 25348
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.998740792274475,
      "learning_rate": 4.4232182218956655e-06,
      "loss": 0.8942,
      "step": 25349
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8851109743118286,
      "learning_rate": 4.422579305497876e-06,
      "loss": 0.8565,
      "step": 25350
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0335512161254883,
      "learning_rate": 4.421940389100087e-06,
      "loss": 1.0081,
      "step": 25351
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8162857294082642,
      "learning_rate": 4.421301472702298e-06,
      "loss": 0.8768,
      "step": 25352
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7906835079193115,
      "learning_rate": 4.420662556304508e-06,
      "loss": 0.9509,
      "step": 25353
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.144196033477783,
      "learning_rate": 4.420023639906719e-06,
      "loss": 1.2045,
      "step": 25354
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3306175470352173,
      "learning_rate": 4.41938472350893e-06,
      "loss": 0.4197,
      "step": 25355
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7266322374343872,
      "learning_rate": 4.41874580711114e-06,
      "loss": 0.7084,
      "step": 25356
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.597468614578247,
      "learning_rate": 4.41810689071335e-06,
      "loss": 0.8164,
      "step": 25357
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7122069597244263,
      "learning_rate": 4.417467974315561e-06,
      "loss": 0.8825,
      "step": 25358
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.773883104324341,
      "learning_rate": 4.416829057917772e-06,
      "loss": 0.93,
      "step": 25359
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.849224090576172,
      "learning_rate": 4.416190141519982e-06,
      "loss": 0.9735,
      "step": 25360
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.6286027431488037,
      "learning_rate": 4.415551225122193e-06,
      "loss": 0.9782,
      "step": 25361
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4971693754196167,
      "learning_rate": 4.414912308724404e-06,
      "loss": 0.9043,
      "step": 25362
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.1093266010284424,
      "learning_rate": 4.4142733923266145e-06,
      "loss": 0.8829,
      "step": 25363
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9670754671096802,
      "learning_rate": 4.413634475928825e-06,
      "loss": 0.8585,
      "step": 25364
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.768218755722046,
      "learning_rate": 4.412995559531036e-06,
      "loss": 0.8376,
      "step": 25365
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5439051389694214,
      "learning_rate": 4.412356643133247e-06,
      "loss": 0.3537,
      "step": 25366
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.720759868621826,
      "learning_rate": 4.411717726735457e-06,
      "loss": 1.0685,
      "step": 25367
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.928237795829773,
      "learning_rate": 4.411078810337668e-06,
      "loss": 1.0645,
      "step": 25368
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.542799472808838,
      "learning_rate": 4.410439893939879e-06,
      "loss": 1.3065,
      "step": 25369
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7359119653701782,
      "learning_rate": 4.409800977542089e-06,
      "loss": 0.8388,
      "step": 25370
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.071135997772217,
      "learning_rate": 4.4091620611443e-06,
      "loss": 1.1457,
      "step": 25371
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.821150302886963,
      "learning_rate": 4.408523144746511e-06,
      "loss": 0.8381,
      "step": 25372
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1026523113250732,
      "learning_rate": 4.4078842283487215e-06,
      "loss": 0.8261,
      "step": 25373
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8881824016571045,
      "learning_rate": 4.407245311950931e-06,
      "loss": 0.8411,
      "step": 25374
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.789501428604126,
      "learning_rate": 4.406606395553142e-06,
      "loss": 0.8081,
      "step": 25375
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.153678894042969,
      "learning_rate": 4.405967479155353e-06,
      "loss": 1.2824,
      "step": 25376
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.953139066696167,
      "learning_rate": 4.4053285627575634e-06,
      "loss": 0.797,
      "step": 25377
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.8366482257843018,
      "learning_rate": 4.404689646359774e-06,
      "loss": 0.8674,
      "step": 25378
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8736300468444824,
      "learning_rate": 4.404050729961985e-06,
      "loss": 0.9658,
      "step": 25379
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5412958860397339,
      "learning_rate": 4.4034118135641955e-06,
      "loss": 0.7639,
      "step": 25380
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3166664838790894,
      "learning_rate": 4.402772897166406e-06,
      "loss": 0.7989,
      "step": 25381
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4479091167449951,
      "learning_rate": 4.402133980768617e-06,
      "loss": 0.7936,
      "step": 25382
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7334096431732178,
      "learning_rate": 4.401495064370828e-06,
      "loss": 0.7631,
      "step": 25383
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0456314086914062,
      "learning_rate": 4.400856147973038e-06,
      "loss": 0.8865,
      "step": 25384
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.200406551361084,
      "learning_rate": 4.400217231575249e-06,
      "loss": 0.9504,
      "step": 25385
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6642050743103027,
      "learning_rate": 4.39957831517746e-06,
      "loss": 0.8467,
      "step": 25386
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.722045421600342,
      "learning_rate": 4.3989393987796704e-06,
      "loss": 0.9087,
      "step": 25387
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6882648468017578,
      "learning_rate": 4.398300482381881e-06,
      "loss": 0.8708,
      "step": 25388
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.982286214828491,
      "learning_rate": 4.397661565984092e-06,
      "loss": 0.818,
      "step": 25389
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5551000833511353,
      "learning_rate": 4.3970226495863025e-06,
      "loss": 0.7631,
      "step": 25390
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5057060718536377,
      "learning_rate": 4.396383733188513e-06,
      "loss": 0.8072,
      "step": 25391
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8287959098815918,
      "learning_rate": 4.395744816790723e-06,
      "loss": 0.8975,
      "step": 25392
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7977327108383179,
      "learning_rate": 4.395105900392934e-06,
      "loss": 0.8795,
      "step": 25393
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.541764497756958,
      "learning_rate": 4.3944669839951445e-06,
      "loss": 0.8191,
      "step": 25394
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8442643880844116,
      "learning_rate": 4.393828067597355e-06,
      "loss": 0.8157,
      "step": 25395
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.16389799118042,
      "learning_rate": 4.393189151199566e-06,
      "loss": 0.9359,
      "step": 25396
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.504049062728882,
      "learning_rate": 4.392550234801777e-06,
      "loss": 0.6581,
      "step": 25397
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.179936170578003,
      "learning_rate": 4.391911318403987e-06,
      "loss": 0.7545,
      "step": 25398
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0905237197875977,
      "learning_rate": 4.391272402006198e-06,
      "loss": 0.8378,
      "step": 25399
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.411150574684143,
      "learning_rate": 4.390633485608409e-06,
      "loss": 0.8168,
      "step": 25400
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5634104013442993,
      "learning_rate": 4.389994569210619e-06,
      "loss": 0.8579,
      "step": 25401
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.2251200675964355,
      "learning_rate": 4.38935565281283e-06,
      "loss": 0.5674,
      "step": 25402
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6475131511688232,
      "learning_rate": 4.388716736415041e-06,
      "loss": 0.8673,
      "step": 25403
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0917084217071533,
      "learning_rate": 4.3880778200172515e-06,
      "loss": 0.7186,
      "step": 25404
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6347033977508545,
      "learning_rate": 4.387438903619462e-06,
      "loss": 0.3289,
      "step": 25405
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7398812770843506,
      "learning_rate": 4.386799987221673e-06,
      "loss": 0.8098,
      "step": 25406
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.125269412994385,
      "learning_rate": 4.386161070823884e-06,
      "loss": 1.1173,
      "step": 25407
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3350830078125,
      "learning_rate": 4.385522154426094e-06,
      "loss": 0.9585,
      "step": 25408
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.286165714263916,
      "learning_rate": 4.384883238028305e-06,
      "loss": 0.9227,
      "step": 25409
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7875629663467407,
      "learning_rate": 4.384244321630516e-06,
      "loss": 0.8718,
      "step": 25410
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.817582607269287,
      "learning_rate": 4.3836054052327255e-06,
      "loss": 1.1508,
      "step": 25411
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.409161567687988,
      "learning_rate": 4.382966488834936e-06,
      "loss": 0.883,
      "step": 25412
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.6368980407714844,
      "learning_rate": 4.382327572437147e-06,
      "loss": 0.9364,
      "step": 25413
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6469700336456299,
      "learning_rate": 4.381688656039358e-06,
      "loss": 0.8773,
      "step": 25414
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.530022144317627,
      "learning_rate": 4.381049739641568e-06,
      "loss": 0.9418,
      "step": 25415
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5054317712783813,
      "learning_rate": 4.380410823243779e-06,
      "loss": 0.8381,
      "step": 25416
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.5304665565490723,
      "learning_rate": 4.379771906845989e-06,
      "loss": 1.0069,
      "step": 25417
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.665269136428833,
      "learning_rate": 4.3791329904482e-06,
      "loss": 0.8035,
      "step": 25418
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9167613983154297,
      "learning_rate": 4.37849407405041e-06,
      "loss": 1.0442,
      "step": 25419
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0053529739379883,
      "learning_rate": 4.377855157652621e-06,
      "loss": 0.7765,
      "step": 25420
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.023536205291748,
      "learning_rate": 4.377216241254832e-06,
      "loss": 0.857,
      "step": 25421
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.697089433670044,
      "learning_rate": 4.376577324857042e-06,
      "loss": 0.8087,
      "step": 25422
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4302784204483032,
      "learning_rate": 4.375938408459253e-06,
      "loss": 0.3307,
      "step": 25423
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8551231622695923,
      "learning_rate": 4.375299492061464e-06,
      "loss": 0.8787,
      "step": 25424
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0412402153015137,
      "learning_rate": 4.3746605756636745e-06,
      "loss": 0.8874,
      "step": 25425
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5563526153564453,
      "learning_rate": 4.374021659265885e-06,
      "loss": 0.8122,
      "step": 25426
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.209297180175781,
      "learning_rate": 4.373382742868096e-06,
      "loss": 1.2356,
      "step": 25427
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3984326124191284,
      "learning_rate": 4.372743826470307e-06,
      "loss": 0.797,
      "step": 25428
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.304384469985962,
      "learning_rate": 4.372104910072517e-06,
      "loss": 0.4057,
      "step": 25429
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.7895121574401855,
      "learning_rate": 4.371465993674728e-06,
      "loss": 1.1081,
      "step": 25430
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.159209728240967,
      "learning_rate": 4.370827077276939e-06,
      "loss": 0.9646,
      "step": 25431
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9829005002975464,
      "learning_rate": 4.370188160879149e-06,
      "loss": 0.8965,
      "step": 25432
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7796809673309326,
      "learning_rate": 4.36954924448136e-06,
      "loss": 0.8899,
      "step": 25433
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.836892604827881,
      "learning_rate": 4.368910328083571e-06,
      "loss": 0.9371,
      "step": 25434
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.215437412261963,
      "learning_rate": 4.368271411685781e-06,
      "loss": 0.7577,
      "step": 25435
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.49826180934906,
      "learning_rate": 4.367632495287991e-06,
      "loss": 0.4203,
      "step": 25436
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5744874477386475,
      "learning_rate": 4.366993578890202e-06,
      "loss": 0.3766,
      "step": 25437
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.061227321624756,
      "learning_rate": 4.366354662492413e-06,
      "loss": 1.1829,
      "step": 25438
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5148439407348633,
      "learning_rate": 4.3657157460946235e-06,
      "loss": 0.8233,
      "step": 25439
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.949068069458008,
      "learning_rate": 4.365076829696834e-06,
      "loss": 1.4355,
      "step": 25440
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6785392761230469,
      "learning_rate": 4.364437913299045e-06,
      "loss": 0.8007,
      "step": 25441
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.1969058513641357,
      "learning_rate": 4.3637989969012556e-06,
      "loss": 1.0565,
      "step": 25442
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0065340995788574,
      "learning_rate": 4.363160080503466e-06,
      "loss": 0.925,
      "step": 25443
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6651248931884766,
      "learning_rate": 4.362521164105677e-06,
      "loss": 0.8175,
      "step": 25444
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4134048223495483,
      "learning_rate": 4.361882247707888e-06,
      "loss": 0.3675,
      "step": 25445
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0292913913726807,
      "learning_rate": 4.361243331310098e-06,
      "loss": 1.0411,
      "step": 25446
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9744683504104614,
      "learning_rate": 4.360604414912309e-06,
      "loss": 0.9224,
      "step": 25447
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6865524053573608,
      "learning_rate": 4.35996549851452e-06,
      "loss": 0.7794,
      "step": 25448
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8186448812484741,
      "learning_rate": 4.3593265821167305e-06,
      "loss": 0.9155,
      "step": 25449
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.45795202255249,
      "learning_rate": 4.358687665718941e-06,
      "loss": 1.655,
      "step": 25450
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7502516508102417,
      "learning_rate": 4.358048749321152e-06,
      "loss": 0.8285,
      "step": 25451
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.250047206878662,
      "learning_rate": 4.3574098329233626e-06,
      "loss": 0.7638,
      "step": 25452
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.041057825088501,
      "learning_rate": 4.356770916525572e-06,
      "loss": 0.9139,
      "step": 25453
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6136410236358643,
      "learning_rate": 4.356132000127783e-06,
      "loss": 0.8568,
      "step": 25454
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9908562898635864,
      "learning_rate": 4.355493083729994e-06,
      "loss": 0.8076,
      "step": 25455
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5705814361572266,
      "learning_rate": 4.3548541673322045e-06,
      "loss": 0.837,
      "step": 25456
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9218008518218994,
      "learning_rate": 4.354215250934415e-06,
      "loss": 0.807,
      "step": 25457
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0463454723358154,
      "learning_rate": 4.353576334536626e-06,
      "loss": 1.1002,
      "step": 25458
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.666529655456543,
      "learning_rate": 4.352937418138837e-06,
      "loss": 1.1208,
      "step": 25459
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.67985999584198,
      "learning_rate": 4.352298501741047e-06,
      "loss": 0.8521,
      "step": 25460
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8241275548934937,
      "learning_rate": 4.351659585343258e-06,
      "loss": 0.4238,
      "step": 25461
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9928747415542603,
      "learning_rate": 4.351020668945469e-06,
      "loss": 0.7946,
      "step": 25462
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.030956745147705,
      "learning_rate": 4.350381752547679e-06,
      "loss": 0.9571,
      "step": 25463
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.012639999389648,
      "learning_rate": 4.34974283614989e-06,
      "loss": 1.0682,
      "step": 25464
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.42003059387207,
      "learning_rate": 4.349103919752101e-06,
      "loss": 0.9783,
      "step": 25465
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.543239712715149,
      "learning_rate": 4.3484650033543115e-06,
      "loss": 0.8614,
      "step": 25466
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0873959064483643,
      "learning_rate": 4.347826086956522e-06,
      "loss": 1.1931,
      "step": 25467
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1719117164611816,
      "learning_rate": 4.347187170558733e-06,
      "loss": 0.8109,
      "step": 25468
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6845701932907104,
      "learning_rate": 4.346548254160944e-06,
      "loss": 0.9638,
      "step": 25469
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.385677695274353,
      "learning_rate": 4.345909337763154e-06,
      "loss": 0.363,
      "step": 25470
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.496363878250122,
      "learning_rate": 4.345270421365365e-06,
      "loss": 0.8275,
      "step": 25471
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7507156133651733,
      "learning_rate": 4.344631504967575e-06,
      "loss": 0.8869,
      "step": 25472
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7997995615005493,
      "learning_rate": 4.3439925885697856e-06,
      "loss": 0.8504,
      "step": 25473
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.684222221374512,
      "learning_rate": 4.343353672171996e-06,
      "loss": 1.2643,
      "step": 25474
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9517287015914917,
      "learning_rate": 4.342714755774207e-06,
      "loss": 0.8129,
      "step": 25475
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.343477725982666,
      "learning_rate": 4.342075839376418e-06,
      "loss": 0.7518,
      "step": 25476
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.781214714050293,
      "learning_rate": 4.341436922978628e-06,
      "loss": 1.0022,
      "step": 25477
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.5115365982055664,
      "learning_rate": 4.340798006580839e-06,
      "loss": 0.8583,
      "step": 25478
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.871198058128357,
      "learning_rate": 4.34015909018305e-06,
      "loss": 0.8343,
      "step": 25479
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.753556728363037,
      "learning_rate": 4.3395201737852605e-06,
      "loss": 0.8329,
      "step": 25480
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6636452674865723,
      "learning_rate": 4.338881257387471e-06,
      "loss": 0.8349,
      "step": 25481
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9594188928604126,
      "learning_rate": 4.338242340989682e-06,
      "loss": 0.8258,
      "step": 25482
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.888907432556152,
      "learning_rate": 4.3376034245918926e-06,
      "loss": 1.0436,
      "step": 25483
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7694364786148071,
      "learning_rate": 4.336964508194103e-06,
      "loss": 0.9241,
      "step": 25484
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.3507943153381348,
      "learning_rate": 4.336325591796314e-06,
      "loss": 0.6982,
      "step": 25485
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7779749631881714,
      "learning_rate": 4.335686675398525e-06,
      "loss": 0.8073,
      "step": 25486
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7995550632476807,
      "learning_rate": 4.335047759000735e-06,
      "loss": 1.0842,
      "step": 25487
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.4054722785949707,
      "learning_rate": 4.334408842602946e-06,
      "loss": 0.9348,
      "step": 25488
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.532253623008728,
      "learning_rate": 4.333769926205157e-06,
      "loss": 0.7612,
      "step": 25489
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.611212968826294,
      "learning_rate": 4.333131009807367e-06,
      "loss": 0.8265,
      "step": 25490
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.262056827545166,
      "learning_rate": 4.332492093409577e-06,
      "loss": 0.9311,
      "step": 25491
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.446902871131897,
      "learning_rate": 4.331853177011788e-06,
      "loss": 0.7827,
      "step": 25492
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6188108921051025,
      "learning_rate": 4.331214260613999e-06,
      "loss": 0.7905,
      "step": 25493
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6080483198165894,
      "learning_rate": 4.330575344216209e-06,
      "loss": 0.7546,
      "step": 25494
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9851012229919434,
      "learning_rate": 4.32993642781842e-06,
      "loss": 1.1309,
      "step": 25495
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.559981107711792,
      "learning_rate": 4.329297511420631e-06,
      "loss": 0.7003,
      "step": 25496
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.8993632793426514,
      "learning_rate": 4.3286585950228415e-06,
      "loss": 1.4538,
      "step": 25497
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8773672580718994,
      "learning_rate": 4.328019678625052e-06,
      "loss": 0.7624,
      "step": 25498
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.839348316192627,
      "learning_rate": 4.327380762227263e-06,
      "loss": 0.8475,
      "step": 25499
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.139233350753784,
      "learning_rate": 4.326741845829474e-06,
      "loss": 0.8003,
      "step": 25500
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.150474548339844,
      "learning_rate": 4.326102929431684e-06,
      "loss": 1.2907,
      "step": 25501
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.289719820022583,
      "learning_rate": 4.325464013033895e-06,
      "loss": 0.9753,
      "step": 25502
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6933990716934204,
      "learning_rate": 4.324825096636106e-06,
      "loss": 0.8218,
      "step": 25503
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.656134605407715,
      "learning_rate": 4.324186180238316e-06,
      "loss": 0.8053,
      "step": 25504
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0595548152923584,
      "learning_rate": 4.323547263840527e-06,
      "loss": 0.8719,
      "step": 25505
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0556254386901855,
      "learning_rate": 4.322908347442738e-06,
      "loss": 0.7074,
      "step": 25506
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9301109313964844,
      "learning_rate": 4.3222694310449485e-06,
      "loss": 0.8604,
      "step": 25507
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5423232316970825,
      "learning_rate": 4.321630514647158e-06,
      "loss": 0.7945,
      "step": 25508
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2355010509490967,
      "learning_rate": 4.320991598249369e-06,
      "loss": 1.0288,
      "step": 25509
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5634129047393799,
      "learning_rate": 4.32035268185158e-06,
      "loss": 0.8305,
      "step": 25510
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.9208500385284424,
      "learning_rate": 4.3197137654537905e-06,
      "loss": 1.393,
      "step": 25511
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7516305446624756,
      "learning_rate": 4.319074849056001e-06,
      "loss": 0.8656,
      "step": 25512
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0122573375701904,
      "learning_rate": 4.318435932658212e-06,
      "loss": 0.9968,
      "step": 25513
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9417150020599365,
      "learning_rate": 4.3177970162604226e-06,
      "loss": 1.0302,
      "step": 25514
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.967554092407227,
      "learning_rate": 4.317158099862633e-06,
      "loss": 1.3181,
      "step": 25515
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9975093603134155,
      "learning_rate": 4.316519183464844e-06,
      "loss": 0.827,
      "step": 25516
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.157227039337158,
      "learning_rate": 4.315880267067055e-06,
      "loss": 1.0998,
      "step": 25517
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9217307567596436,
      "learning_rate": 4.315241350669265e-06,
      "loss": 0.7705,
      "step": 25518
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7333712577819824,
      "learning_rate": 4.314602434271476e-06,
      "loss": 0.9143,
      "step": 25519
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.16147518157959,
      "learning_rate": 4.313963517873687e-06,
      "loss": 2.2032,
      "step": 25520
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7417912483215332,
      "learning_rate": 4.3133246014758975e-06,
      "loss": 1.1261,
      "step": 25521
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8799188137054443,
      "learning_rate": 4.312685685078108e-06,
      "loss": 0.8037,
      "step": 25522
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8248310089111328,
      "learning_rate": 4.312046768680319e-06,
      "loss": 0.8882,
      "step": 25523
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.653834581375122,
      "learning_rate": 4.3114078522825296e-06,
      "loss": 0.7922,
      "step": 25524
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7724888324737549,
      "learning_rate": 4.31076893588474e-06,
      "loss": 0.9313,
      "step": 25525
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.790886402130127,
      "learning_rate": 4.31013001948695e-06,
      "loss": 0.7739,
      "step": 25526
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0329110622406006,
      "learning_rate": 4.309491103089161e-06,
      "loss": 0.9243,
      "step": 25527
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.7029736042022705,
      "learning_rate": 4.3088521866913715e-06,
      "loss": 0.8612,
      "step": 25528
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.584770917892456,
      "learning_rate": 4.308213270293582e-06,
      "loss": 0.7541,
      "step": 25529
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.596869945526123,
      "learning_rate": 4.307574353895793e-06,
      "loss": 0.8794,
      "step": 25530
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.493908643722534,
      "learning_rate": 4.306935437498004e-06,
      "loss": 0.9066,
      "step": 25531
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.240290403366089,
      "learning_rate": 4.306296521100214e-06,
      "loss": 0.8125,
      "step": 25532
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5115553140640259,
      "learning_rate": 4.305657604702425e-06,
      "loss": 0.791,
      "step": 25533
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.7135698795318604,
      "learning_rate": 4.305018688304636e-06,
      "loss": 0.5851,
      "step": 25534
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7669540643692017,
      "learning_rate": 4.304379771906846e-06,
      "loss": 1.0665,
      "step": 25535
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9183611869812012,
      "learning_rate": 4.303740855509057e-06,
      "loss": 1.0409,
      "step": 25536
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.2206594944000244,
      "learning_rate": 4.303101939111268e-06,
      "loss": 0.9957,
      "step": 25537
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.041598320007324,
      "learning_rate": 4.3024630227134785e-06,
      "loss": 1.5729,
      "step": 25538
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3863635063171387,
      "learning_rate": 4.301824106315689e-06,
      "loss": 0.7742,
      "step": 25539
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5699214935302734,
      "learning_rate": 4.3011851899179e-06,
      "loss": 0.7982,
      "step": 25540
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.1830999851226807,
      "learning_rate": 4.300546273520111e-06,
      "loss": 1.0862,
      "step": 25541
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6857565641403198,
      "learning_rate": 4.299907357122321e-06,
      "loss": 0.7828,
      "step": 25542
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1114885807037354,
      "learning_rate": 4.299268440724532e-06,
      "loss": 1.3549,
      "step": 25543
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6774530410766602,
      "learning_rate": 4.298629524326743e-06,
      "loss": 0.8715,
      "step": 25544
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.563277244567871,
      "learning_rate": 4.2979906079289526e-06,
      "loss": 1.5403,
      "step": 25545
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.856797695159912,
      "learning_rate": 4.297351691531163e-06,
      "loss": 0.618,
      "step": 25546
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.25390625,
      "learning_rate": 4.296712775133374e-06,
      "loss": 1.0323,
      "step": 25547
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.666647434234619,
      "learning_rate": 4.296073858735585e-06,
      "loss": 1.1075,
      "step": 25548
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8127939701080322,
      "learning_rate": 4.295434942337795e-06,
      "loss": 0.8918,
      "step": 25549
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.5610029697418213,
      "learning_rate": 4.294796025940006e-06,
      "loss": 0.7653,
      "step": 25550
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9156689643859863,
      "learning_rate": 4.294157109542217e-06,
      "loss": 0.5024,
      "step": 25551
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.683140993118286,
      "learning_rate": 4.2935181931444275e-06,
      "loss": 0.9027,
      "step": 25552
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.328232765197754,
      "learning_rate": 4.292879276746638e-06,
      "loss": 1.7281,
      "step": 25553
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2519657611846924,
      "learning_rate": 4.292240360348849e-06,
      "loss": 0.9637,
      "step": 25554
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7076250314712524,
      "learning_rate": 4.2916014439510596e-06,
      "loss": 1.0337,
      "step": 25555
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.911971092224121,
      "learning_rate": 4.29096252755327e-06,
      "loss": 0.5581,
      "step": 25556
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.527362108230591,
      "learning_rate": 4.290323611155481e-06,
      "loss": 0.7761,
      "step": 25557
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9767087697982788,
      "learning_rate": 4.289684694757692e-06,
      "loss": 0.8782,
      "step": 25558
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9585777521133423,
      "learning_rate": 4.289045778359902e-06,
      "loss": 0.7689,
      "step": 25559
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7828915119171143,
      "learning_rate": 4.288406861962113e-06,
      "loss": 0.8956,
      "step": 25560
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.5878872871398926,
      "learning_rate": 4.287767945564324e-06,
      "loss": 1.0781,
      "step": 25561
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9226362705230713,
      "learning_rate": 4.2871290291665345e-06,
      "loss": 0.9408,
      "step": 25562
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8283822536468506,
      "learning_rate": 4.286490112768744e-06,
      "loss": 0.8651,
      "step": 25563
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7377084493637085,
      "learning_rate": 4.285851196370955e-06,
      "loss": 0.8246,
      "step": 25564
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.40196418762207,
      "learning_rate": 4.285212279973166e-06,
      "loss": 0.9369,
      "step": 25565
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9271838665008545,
      "learning_rate": 4.284573363575376e-06,
      "loss": 0.7267,
      "step": 25566
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.4575624465942383,
      "learning_rate": 4.283934447177587e-06,
      "loss": 0.4486,
      "step": 25567
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6546472311019897,
      "learning_rate": 4.283295530779798e-06,
      "loss": 0.8005,
      "step": 25568
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6887353658676147,
      "learning_rate": 4.2826566143820085e-06,
      "loss": 0.8501,
      "step": 25569
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0713307857513428,
      "learning_rate": 4.282017697984219e-06,
      "loss": 1.1409,
      "step": 25570
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5976839065551758,
      "learning_rate": 4.28137878158643e-06,
      "loss": 0.9636,
      "step": 25571
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.21010684967041,
      "learning_rate": 4.280739865188641e-06,
      "loss": 0.7065,
      "step": 25572
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0981690883636475,
      "learning_rate": 4.280100948790851e-06,
      "loss": 0.8794,
      "step": 25573
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7046828269958496,
      "learning_rate": 4.279462032393062e-06,
      "loss": 0.6913,
      "step": 25574
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1243703365325928,
      "learning_rate": 4.278823115995273e-06,
      "loss": 0.9252,
      "step": 25575
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6152600049972534,
      "learning_rate": 4.278184199597483e-06,
      "loss": 0.8839,
      "step": 25576
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0194947719573975,
      "learning_rate": 4.277545283199694e-06,
      "loss": 0.9266,
      "step": 25577
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9662221670150757,
      "learning_rate": 4.276906366801905e-06,
      "loss": 0.9274,
      "step": 25578
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8304035663604736,
      "learning_rate": 4.2762674504041155e-06,
      "loss": 0.8456,
      "step": 25579
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.798365592956543,
      "learning_rate": 4.275628534006326e-06,
      "loss": 0.7466,
      "step": 25580
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.123775005340576,
      "learning_rate": 4.274989617608536e-06,
      "loss": 1.0379,
      "step": 25581
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8771790266036987,
      "learning_rate": 4.274350701210747e-06,
      "loss": 0.836,
      "step": 25582
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.168508529663086,
      "learning_rate": 4.2737117848129575e-06,
      "loss": 0.7666,
      "step": 25583
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3016197681427,
      "learning_rate": 4.273072868415168e-06,
      "loss": 0.7415,
      "step": 25584
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.658126950263977,
      "learning_rate": 4.272433952017379e-06,
      "loss": 0.8129,
      "step": 25585
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.649132490158081,
      "learning_rate": 4.2717950356195896e-06,
      "loss": 1.1413,
      "step": 25586
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.14064884185791,
      "learning_rate": 4.2711561192218e-06,
      "loss": 0.8568,
      "step": 25587
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.6154396533966064,
      "learning_rate": 4.270517202824011e-06,
      "loss": 0.8753,
      "step": 25588
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6231162548065186,
      "learning_rate": 4.269878286426222e-06,
      "loss": 0.7727,
      "step": 25589
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6160216331481934,
      "learning_rate": 4.269239370028432e-06,
      "loss": 0.8506,
      "step": 25590
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7947018146514893,
      "learning_rate": 4.268600453630643e-06,
      "loss": 0.7779,
      "step": 25591
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6288151741027832,
      "learning_rate": 4.267961537232854e-06,
      "loss": 0.8366,
      "step": 25592
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.02809739112854,
      "learning_rate": 4.2673226208350645e-06,
      "loss": 0.9732,
      "step": 25593
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.726278305053711,
      "learning_rate": 4.266683704437275e-06,
      "loss": 0.731,
      "step": 25594
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7572473287582397,
      "learning_rate": 4.266044788039486e-06,
      "loss": 0.8316,
      "step": 25595
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5619451999664307,
      "learning_rate": 4.2654058716416966e-06,
      "loss": 0.7714,
      "step": 25596
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.2080488204956055,
      "learning_rate": 4.264766955243907e-06,
      "loss": 1.5401,
      "step": 25597
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9785207509994507,
      "learning_rate": 4.264128038846118e-06,
      "loss": 0.9449,
      "step": 25598
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.50606107711792,
      "learning_rate": 4.263489122448328e-06,
      "loss": 0.9249,
      "step": 25599
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.395040035247803,
      "learning_rate": 4.2628502060505385e-06,
      "loss": 1.408,
      "step": 25600
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3304009437561035,
      "learning_rate": 4.262211289652749e-06,
      "loss": 0.9811,
      "step": 25601
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5625457763671875,
      "learning_rate": 4.26157237325496e-06,
      "loss": 0.8463,
      "step": 25602
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.013325214385986,
      "learning_rate": 4.260933456857171e-06,
      "loss": 0.8108,
      "step": 25603
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7525441646575928,
      "learning_rate": 4.260294540459381e-06,
      "loss": 0.7562,
      "step": 25604
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.66135573387146,
      "learning_rate": 4.259655624061592e-06,
      "loss": 0.8223,
      "step": 25605
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6128615140914917,
      "learning_rate": 4.259016707663802e-06,
      "loss": 0.7781,
      "step": 25606
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.612015724182129,
      "learning_rate": 4.2583777912660126e-06,
      "loss": 0.8577,
      "step": 25607
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.5753049850463867,
      "learning_rate": 4.257738874868223e-06,
      "loss": 0.8745,
      "step": 25608
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.798299312591553,
      "learning_rate": 4.257099958470434e-06,
      "loss": 1.1676,
      "step": 25609
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9221593141555786,
      "learning_rate": 4.256461042072645e-06,
      "loss": 0.3618,
      "step": 25610
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.645809531211853,
      "learning_rate": 4.255822125674855e-06,
      "loss": 1.0689,
      "step": 25611
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2026174068450928,
      "learning_rate": 4.255183209277066e-06,
      "loss": 0.9485,
      "step": 25612
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.2754549980163574,
      "learning_rate": 4.254544292879277e-06,
      "loss": 0.8749,
      "step": 25613
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8772741556167603,
      "learning_rate": 4.2539053764814875e-06,
      "loss": 0.8806,
      "step": 25614
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.662396192550659,
      "learning_rate": 4.253266460083698e-06,
      "loss": 0.4619,
      "step": 25615
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.386184811592102,
      "learning_rate": 4.252627543685909e-06,
      "loss": 0.8364,
      "step": 25616
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.784035325050354,
      "learning_rate": 4.2519886272881196e-06,
      "loss": 0.8052,
      "step": 25617
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8772239685058594,
      "learning_rate": 4.25134971089033e-06,
      "loss": 0.8274,
      "step": 25618
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.874964475631714,
      "learning_rate": 4.250710794492541e-06,
      "loss": 0.9925,
      "step": 25619
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.643629312515259,
      "learning_rate": 4.250071878094752e-06,
      "loss": 1.1836,
      "step": 25620
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.091130256652832,
      "learning_rate": 4.249432961696962e-06,
      "loss": 0.7802,
      "step": 25621
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.508445739746094,
      "learning_rate": 4.248794045299173e-06,
      "loss": 1.0489,
      "step": 25622
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.037553787231445,
      "learning_rate": 4.248155128901384e-06,
      "loss": 1.158,
      "step": 25623
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.53104567527771,
      "learning_rate": 4.247516212503594e-06,
      "loss": 0.6876,
      "step": 25624
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.05124044418335,
      "learning_rate": 4.246877296105804e-06,
      "loss": 0.7032,
      "step": 25625
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.907565712928772,
      "learning_rate": 4.246238379708015e-06,
      "loss": 0.7397,
      "step": 25626
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7488515377044678,
      "learning_rate": 4.245599463310226e-06,
      "loss": 0.8576,
      "step": 25627
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.444064736366272,
      "learning_rate": 4.2449605469124364e-06,
      "loss": 0.3879,
      "step": 25628
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.540416955947876,
      "learning_rate": 4.244321630514647e-06,
      "loss": 0.8172,
      "step": 25629
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.815215826034546,
      "learning_rate": 4.243682714116858e-06,
      "loss": 0.9526,
      "step": 25630
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9170751571655273,
      "learning_rate": 4.2430437977190685e-06,
      "loss": 0.8462,
      "step": 25631
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7772940397262573,
      "learning_rate": 4.242404881321279e-06,
      "loss": 0.6273,
      "step": 25632
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.561890602111816,
      "learning_rate": 4.24176596492349e-06,
      "loss": 0.9923,
      "step": 25633
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9712717533111572,
      "learning_rate": 4.241127048525701e-06,
      "loss": 0.8846,
      "step": 25634
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9678757190704346,
      "learning_rate": 4.240488132127911e-06,
      "loss": 1.0104,
      "step": 25635
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.054313659667969,
      "learning_rate": 4.239849215730122e-06,
      "loss": 1.3006,
      "step": 25636
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9536033868789673,
      "learning_rate": 4.239210299332333e-06,
      "loss": 0.8255,
      "step": 25637
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7402859926223755,
      "learning_rate": 4.2385713829345434e-06,
      "loss": 0.8429,
      "step": 25638
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.220792055130005,
      "learning_rate": 4.237932466536754e-06,
      "loss": 0.8112,
      "step": 25639
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0084011554718018,
      "learning_rate": 4.237293550138965e-06,
      "loss": 0.7663,
      "step": 25640
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5399932861328125,
      "learning_rate": 4.2366546337411755e-06,
      "loss": 0.3303,
      "step": 25641
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.961351990699768,
      "learning_rate": 4.236015717343385e-06,
      "loss": 0.8751,
      "step": 25642
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8002443313598633,
      "learning_rate": 4.235376800945596e-06,
      "loss": 0.8341,
      "step": 25643
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9696202278137207,
      "learning_rate": 4.234737884547807e-06,
      "loss": 0.8097,
      "step": 25644
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.966586947441101,
      "learning_rate": 4.2340989681500175e-06,
      "loss": 0.7992,
      "step": 25645
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5342764854431152,
      "learning_rate": 4.233460051752228e-06,
      "loss": 0.3922,
      "step": 25646
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8728103637695312,
      "learning_rate": 4.232821135354439e-06,
      "loss": 1.1499,
      "step": 25647
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6648625135421753,
      "learning_rate": 4.23218221895665e-06,
      "loss": 0.8861,
      "step": 25648
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6892335414886475,
      "learning_rate": 4.23154330255886e-06,
      "loss": 0.7643,
      "step": 25649
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.990536093711853,
      "learning_rate": 4.230904386161071e-06,
      "loss": 0.854,
      "step": 25650
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7900484800338745,
      "learning_rate": 4.230265469763282e-06,
      "loss": 0.7772,
      "step": 25651
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.399596929550171,
      "learning_rate": 4.229626553365492e-06,
      "loss": 0.4733,
      "step": 25652
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.526364803314209,
      "learning_rate": 4.228987636967703e-06,
      "loss": 0.8247,
      "step": 25653
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7671802043914795,
      "learning_rate": 4.228348720569914e-06,
      "loss": 0.8042,
      "step": 25654
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.225670337677002,
      "learning_rate": 4.2277098041721245e-06,
      "loss": 0.8652,
      "step": 25655
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7365224361419678,
      "learning_rate": 4.227070887774335e-06,
      "loss": 0.8268,
      "step": 25656
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.602810859680176,
      "learning_rate": 4.226431971376546e-06,
      "loss": 1.1744,
      "step": 25657
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.882064700126648,
      "learning_rate": 4.225793054978757e-06,
      "loss": 0.8983,
      "step": 25658
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.696826696395874,
      "learning_rate": 4.225154138580967e-06,
      "loss": 0.8507,
      "step": 25659
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.473767638206482,
      "learning_rate": 4.224515222183177e-06,
      "loss": 0.7011,
      "step": 25660
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5598968267440796,
      "learning_rate": 4.223876305785388e-06,
      "loss": 0.7562,
      "step": 25661
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9456164836883545,
      "learning_rate": 4.2232373893875985e-06,
      "loss": 0.7758,
      "step": 25662
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.153541326522827,
      "learning_rate": 4.222598472989809e-06,
      "loss": 0.8877,
      "step": 25663
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6119464635849,
      "learning_rate": 4.22195955659202e-06,
      "loss": 0.7464,
      "step": 25664
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7631787061691284,
      "learning_rate": 4.221320640194231e-06,
      "loss": 0.7633,
      "step": 25665
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.76510751247406,
      "learning_rate": 4.220681723796441e-06,
      "loss": 0.869,
      "step": 25666
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6493767499923706,
      "learning_rate": 4.220042807398652e-06,
      "loss": 0.8931,
      "step": 25667
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5652867555618286,
      "learning_rate": 4.219403891000863e-06,
      "loss": 0.8291,
      "step": 25668
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.62330162525177,
      "learning_rate": 4.2187649746030734e-06,
      "loss": 0.8046,
      "step": 25669
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2572569847106934,
      "learning_rate": 4.218126058205284e-06,
      "loss": 0.9773,
      "step": 25670
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.8175368309020996,
      "learning_rate": 4.217487141807495e-06,
      "loss": 0.9746,
      "step": 25671
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.040903091430664,
      "learning_rate": 4.2168482254097055e-06,
      "loss": 0.9065,
      "step": 25672
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.292843818664551,
      "learning_rate": 4.216209309011916e-06,
      "loss": 1.3767,
      "step": 25673
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5789097547531128,
      "learning_rate": 4.215570392614127e-06,
      "loss": 0.3819,
      "step": 25674
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9091997146606445,
      "learning_rate": 4.214931476216338e-06,
      "loss": 1.0447,
      "step": 25675
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.188610553741455,
      "learning_rate": 4.214292559818548e-06,
      "loss": 1.5814,
      "step": 25676
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3717135190963745,
      "learning_rate": 4.213653643420759e-06,
      "loss": 0.788,
      "step": 25677
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9323756694793701,
      "learning_rate": 4.21301472702297e-06,
      "loss": 0.8867,
      "step": 25678
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0058579444885254,
      "learning_rate": 4.21237581062518e-06,
      "loss": 0.9359,
      "step": 25679
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7541311979293823,
      "learning_rate": 4.21173689422739e-06,
      "loss": 0.9408,
      "step": 25680
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9881170988082886,
      "learning_rate": 4.211097977829601e-06,
      "loss": 0.9858,
      "step": 25681
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.101922035217285,
      "learning_rate": 4.210459061431812e-06,
      "loss": 0.9926,
      "step": 25682
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.230086326599121,
      "learning_rate": 4.209820145034022e-06,
      "loss": 0.944,
      "step": 25683
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4450100660324097,
      "learning_rate": 4.209181228636233e-06,
      "loss": 0.7888,
      "step": 25684
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.998173475265503,
      "learning_rate": 4.208542312238444e-06,
      "loss": 0.7707,
      "step": 25685
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0562031269073486,
      "learning_rate": 4.2079033958406545e-06,
      "loss": 0.8563,
      "step": 25686
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3403868675231934,
      "learning_rate": 4.207264479442865e-06,
      "loss": 0.8338,
      "step": 25687
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.701834797859192,
      "learning_rate": 4.206625563045076e-06,
      "loss": 0.7987,
      "step": 25688
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9168360233306885,
      "learning_rate": 4.205986646647287e-06,
      "loss": 1.2507,
      "step": 25689
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.660566806793213,
      "learning_rate": 4.205347730249497e-06,
      "loss": 1.3008,
      "step": 25690
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.3356516361236572,
      "learning_rate": 4.204708813851708e-06,
      "loss": 1.1939,
      "step": 25691
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.277492880821228,
      "learning_rate": 4.204069897453919e-06,
      "loss": 0.7248,
      "step": 25692
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.506877899169922,
      "learning_rate": 4.203430981056129e-06,
      "loss": 0.8525,
      "step": 25693
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.4328572750091553,
      "learning_rate": 4.20279206465834e-06,
      "loss": 1.1398,
      "step": 25694
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8753852844238281,
      "learning_rate": 4.202153148260551e-06,
      "loss": 0.7865,
      "step": 25695
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9096605777740479,
      "learning_rate": 4.2015142318627615e-06,
      "loss": 0.8384,
      "step": 25696
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.045771360397339,
      "learning_rate": 4.200875315464971e-06,
      "loss": 0.8427,
      "step": 25697
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5592087507247925,
      "learning_rate": 4.200236399067182e-06,
      "loss": 0.7535,
      "step": 25698
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.106614351272583,
      "learning_rate": 4.199597482669393e-06,
      "loss": 0.6519,
      "step": 25699
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9692528247833252,
      "learning_rate": 4.1989585662716034e-06,
      "loss": 0.7625,
      "step": 25700
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3011815547943115,
      "learning_rate": 4.198319649873814e-06,
      "loss": 0.3786,
      "step": 25701
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9495298862457275,
      "learning_rate": 4.197680733476025e-06,
      "loss": 0.9863,
      "step": 25702
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1648032665252686,
      "learning_rate": 4.1970418170782355e-06,
      "loss": 0.8223,
      "step": 25703
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8854306936264038,
      "learning_rate": 4.196402900680446e-06,
      "loss": 0.9221,
      "step": 25704
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3797831535339355,
      "learning_rate": 4.195763984282657e-06,
      "loss": 0.397,
      "step": 25705
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7730276584625244,
      "learning_rate": 4.195125067884868e-06,
      "loss": 0.9656,
      "step": 25706
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0425946712493896,
      "learning_rate": 4.194486151487078e-06,
      "loss": 0.9051,
      "step": 25707
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1610119342803955,
      "learning_rate": 4.193847235089289e-06,
      "loss": 0.7382,
      "step": 25708
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8568918704986572,
      "learning_rate": 4.1932083186915e-06,
      "loss": 0.8079,
      "step": 25709
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8392330408096313,
      "learning_rate": 4.1925694022937104e-06,
      "loss": 0.9607,
      "step": 25710
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0828475952148438,
      "learning_rate": 4.191930485895921e-06,
      "loss": 0.8341,
      "step": 25711
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.1072349548339844,
      "learning_rate": 4.191291569498132e-06,
      "loss": 0.8688,
      "step": 25712
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.984825849533081,
      "learning_rate": 4.1906526531003425e-06,
      "loss": 0.8463,
      "step": 25713
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8102558851242065,
      "learning_rate": 4.190013736702553e-06,
      "loss": 0.8854,
      "step": 25714
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5368739366531372,
      "learning_rate": 4.189374820304763e-06,
      "loss": 0.8791,
      "step": 25715
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.978934645652771,
      "learning_rate": 4.188735903906974e-06,
      "loss": 0.7689,
      "step": 25716
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9064658880233765,
      "learning_rate": 4.1880969875091845e-06,
      "loss": 1.2042,
      "step": 25717
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.033231258392334,
      "learning_rate": 4.187458071111395e-06,
      "loss": 1.1869,
      "step": 25718
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2868123054504395,
      "learning_rate": 4.186819154713606e-06,
      "loss": 1.3446,
      "step": 25719
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0911989212036133,
      "learning_rate": 4.186180238315817e-06,
      "loss": 0.7638,
      "step": 25720
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.012697219848633,
      "learning_rate": 4.185541321918027e-06,
      "loss": 1.3587,
      "step": 25721
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.4747116565704346,
      "learning_rate": 4.184902405520238e-06,
      "loss": 0.9334,
      "step": 25722
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7172844409942627,
      "learning_rate": 4.184263489122449e-06,
      "loss": 0.8111,
      "step": 25723
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7057955265045166,
      "learning_rate": 4.183624572724659e-06,
      "loss": 0.8056,
      "step": 25724
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9126479625701904,
      "learning_rate": 4.18298565632687e-06,
      "loss": 0.8857,
      "step": 25725
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4982410669326782,
      "learning_rate": 4.182346739929081e-06,
      "loss": 0.4051,
      "step": 25726
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.523098349571228,
      "learning_rate": 4.1817078235312915e-06,
      "loss": 0.3694,
      "step": 25727
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2098724842071533,
      "learning_rate": 4.181068907133502e-06,
      "loss": 1.0477,
      "step": 25728
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4438610076904297,
      "learning_rate": 4.180429990735713e-06,
      "loss": 0.8779,
      "step": 25729
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9283965826034546,
      "learning_rate": 4.179791074337924e-06,
      "loss": 0.7885,
      "step": 25730
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.023075580596924,
      "learning_rate": 4.179152157940134e-06,
      "loss": 1.6919,
      "step": 25731
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9445106983184814,
      "learning_rate": 4.178513241542345e-06,
      "loss": 1.2315,
      "step": 25732
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.145817995071411,
      "learning_rate": 4.177874325144555e-06,
      "loss": 0.9374,
      "step": 25733
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9804441928863525,
      "learning_rate": 4.1772354087467655e-06,
      "loss": 0.6524,
      "step": 25734
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.233578681945801,
      "learning_rate": 4.176596492348976e-06,
      "loss": 1.5176,
      "step": 25735
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7568655014038086,
      "learning_rate": 4.175957575951187e-06,
      "loss": 0.8025,
      "step": 25736
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8428568840026855,
      "learning_rate": 4.175318659553398e-06,
      "loss": 0.8087,
      "step": 25737
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5957844257354736,
      "learning_rate": 4.174679743155608e-06,
      "loss": 0.9447,
      "step": 25738
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.571724534034729,
      "learning_rate": 4.174040826757819e-06,
      "loss": 0.8634,
      "step": 25739
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.631168007850647,
      "learning_rate": 4.17340191036003e-06,
      "loss": 0.7564,
      "step": 25740
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7741191387176514,
      "learning_rate": 4.1727629939622404e-06,
      "loss": 0.8573,
      "step": 25741
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.4217865467071533,
      "learning_rate": 4.172124077564451e-06,
      "loss": 0.571,
      "step": 25742
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6043468713760376,
      "learning_rate": 4.171485161166662e-06,
      "loss": 0.8375,
      "step": 25743
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7993814945220947,
      "learning_rate": 4.1708462447688725e-06,
      "loss": 0.8342,
      "step": 25744
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.095428943634033,
      "learning_rate": 4.170207328371083e-06,
      "loss": 1.6146,
      "step": 25745
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5493299961090088,
      "learning_rate": 4.169568411973294e-06,
      "loss": 0.458,
      "step": 25746
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.007645845413208,
      "learning_rate": 4.168929495575505e-06,
      "loss": 0.8183,
      "step": 25747
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7512491941452026,
      "learning_rate": 4.168290579177715e-06,
      "loss": 0.7798,
      "step": 25748
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6883561611175537,
      "learning_rate": 4.167651662779926e-06,
      "loss": 0.8793,
      "step": 25749
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.008204460144043,
      "learning_rate": 4.167012746382137e-06,
      "loss": 0.8824,
      "step": 25750
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.8296921253204346,
      "learning_rate": 4.1663738299843474e-06,
      "loss": 1.2084,
      "step": 25751
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6248300075531006,
      "learning_rate": 4.165734913586557e-06,
      "loss": 0.8237,
      "step": 25752
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7237493991851807,
      "learning_rate": 4.165095997188768e-06,
      "loss": 0.7997,
      "step": 25753
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.96584153175354,
      "learning_rate": 4.164457080790979e-06,
      "loss": 0.7584,
      "step": 25754
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.7238872051239014,
      "learning_rate": 4.163818164393189e-06,
      "loss": 0.6786,
      "step": 25755
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.41951584815979,
      "learning_rate": 4.1631792479954e-06,
      "loss": 0.6783,
      "step": 25756
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7584936618804932,
      "learning_rate": 4.162540331597611e-06,
      "loss": 0.8298,
      "step": 25757
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.730811357498169,
      "learning_rate": 4.1619014151998215e-06,
      "loss": 0.7812,
      "step": 25758
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.178061008453369,
      "learning_rate": 4.161262498802032e-06,
      "loss": 0.8449,
      "step": 25759
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4351067543029785,
      "learning_rate": 4.160623582404243e-06,
      "loss": 0.7304,
      "step": 25760
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.038592576980591,
      "learning_rate": 4.159984666006454e-06,
      "loss": 0.8454,
      "step": 25761
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1921472549438477,
      "learning_rate": 4.159345749608664e-06,
      "loss": 0.9272,
      "step": 25762
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4831231832504272,
      "learning_rate": 4.158706833210875e-06,
      "loss": 0.8301,
      "step": 25763
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2878830432891846,
      "learning_rate": 4.158067916813086e-06,
      "loss": 0.99,
      "step": 25764
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9031949043273926,
      "learning_rate": 4.157429000415296e-06,
      "loss": 0.8891,
      "step": 25765
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5625877380371094,
      "learning_rate": 4.156790084017507e-06,
      "loss": 0.8222,
      "step": 25766
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7038215398788452,
      "learning_rate": 4.156151167619718e-06,
      "loss": 0.8448,
      "step": 25767
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5806529521942139,
      "learning_rate": 4.1555122512219285e-06,
      "loss": 0.8726,
      "step": 25768
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.4380109310150146,
      "learning_rate": 4.154873334824139e-06,
      "loss": 1.1809,
      "step": 25769
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.643759250640869,
      "learning_rate": 4.154234418426349e-06,
      "loss": 0.7127,
      "step": 25770
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9703528881073,
      "learning_rate": 4.15359550202856e-06,
      "loss": 0.5465,
      "step": 25771
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2717177867889404,
      "learning_rate": 4.1529565856307704e-06,
      "loss": 0.8831,
      "step": 25772
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.8128395080566406,
      "learning_rate": 4.152317669232981e-06,
      "loss": 0.7788,
      "step": 25773
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.0936906337738037,
      "learning_rate": 4.151678752835192e-06,
      "loss": 0.9655,
      "step": 25774
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2909932136535645,
      "learning_rate": 4.1510398364374025e-06,
      "loss": 0.8975,
      "step": 25775
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6775469779968262,
      "learning_rate": 4.150400920039613e-06,
      "loss": 0.8469,
      "step": 25776
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5393528938293457,
      "learning_rate": 4.149762003641824e-06,
      "loss": 0.8472,
      "step": 25777
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7865058183670044,
      "learning_rate": 4.149123087244035e-06,
      "loss": 0.7569,
      "step": 25778
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.866057276725769,
      "learning_rate": 4.148484170846245e-06,
      "loss": 0.8222,
      "step": 25779
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6041524410247803,
      "learning_rate": 4.147845254448456e-06,
      "loss": 0.7632,
      "step": 25780
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4650707244873047,
      "learning_rate": 4.147206338050667e-06,
      "loss": 0.8226,
      "step": 25781
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.058761119842529,
      "learning_rate": 4.1465674216528774e-06,
      "loss": 0.9893,
      "step": 25782
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0405070781707764,
      "learning_rate": 4.145928505255088e-06,
      "loss": 1.148,
      "step": 25783
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3730521202087402,
      "learning_rate": 4.145289588857299e-06,
      "loss": 0.5339,
      "step": 25784
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3785271644592285,
      "learning_rate": 4.1446506724595095e-06,
      "loss": 0.4136,
      "step": 25785
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1775426864624023,
      "learning_rate": 4.14401175606172e-06,
      "loss": 0.7692,
      "step": 25786
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8719594478607178,
      "learning_rate": 4.143372839663931e-06,
      "loss": 0.7899,
      "step": 25787
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.967437744140625,
      "learning_rate": 4.142733923266141e-06,
      "loss": 1.4302,
      "step": 25788
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.767830729484558,
      "learning_rate": 4.1420950068683515e-06,
      "loss": 0.6535,
      "step": 25789
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3244653940200806,
      "learning_rate": 4.141456090470562e-06,
      "loss": 0.3769,
      "step": 25790
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6481465101242065,
      "learning_rate": 4.140817174072773e-06,
      "loss": 0.7954,
      "step": 25791
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.103567361831665,
      "learning_rate": 4.140178257674984e-06,
      "loss": 0.9682,
      "step": 25792
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8271628618240356,
      "learning_rate": 4.139539341277194e-06,
      "loss": 0.8561,
      "step": 25793
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4997341632843018,
      "learning_rate": 4.138900424879405e-06,
      "loss": 0.7743,
      "step": 25794
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.751082420349121,
      "learning_rate": 4.138261508481616e-06,
      "loss": 0.8524,
      "step": 25795
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5850133895874023,
      "learning_rate": 4.137622592083826e-06,
      "loss": 0.8298,
      "step": 25796
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2347981929779053,
      "learning_rate": 4.136983675686037e-06,
      "loss": 0.838,
      "step": 25797
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3920565843582153,
      "learning_rate": 4.136344759288248e-06,
      "loss": 0.7172,
      "step": 25798
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8964000940322876,
      "learning_rate": 4.135705842890458e-06,
      "loss": 0.7711,
      "step": 25799
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.088857412338257,
      "learning_rate": 4.135066926492668e-06,
      "loss": 0.8383,
      "step": 25800
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.912553071975708,
      "learning_rate": 4.134428010094879e-06,
      "loss": 0.8948,
      "step": 25801
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.142357110977173,
      "learning_rate": 4.13378909369709e-06,
      "loss": 1.1633,
      "step": 25802
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5653564929962158,
      "learning_rate": 4.1331501772993005e-06,
      "loss": 0.799,
      "step": 25803
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0449464321136475,
      "learning_rate": 4.132511260901511e-06,
      "loss": 0.8576,
      "step": 25804
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9833848476409912,
      "learning_rate": 4.131872344503722e-06,
      "loss": 0.8286,
      "step": 25805
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8629140853881836,
      "learning_rate": 4.1312334281059326e-06,
      "loss": 0.789,
      "step": 25806
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.43151593208313,
      "learning_rate": 4.130594511708143e-06,
      "loss": 0.8897,
      "step": 25807
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7134380340576172,
      "learning_rate": 4.129955595310354e-06,
      "loss": 0.8302,
      "step": 25808
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9490572214126587,
      "learning_rate": 4.129316678912565e-06,
      "loss": 0.8655,
      "step": 25809
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.118556499481201,
      "learning_rate": 4.128677762514775e-06,
      "loss": 0.6785,
      "step": 25810
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1318411827087402,
      "learning_rate": 4.128038846116986e-06,
      "loss": 0.8582,
      "step": 25811
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.858355164527893,
      "learning_rate": 4.127399929719197e-06,
      "loss": 0.8349,
      "step": 25812
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0748610496520996,
      "learning_rate": 4.126761013321407e-06,
      "loss": 0.9882,
      "step": 25813
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.4279181957244873,
      "learning_rate": 4.126122096923617e-06,
      "loss": 1.1277,
      "step": 25814
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5902822017669678,
      "learning_rate": 4.125483180525828e-06,
      "loss": 0.7433,
      "step": 25815
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9311861991882324,
      "learning_rate": 4.124844264128039e-06,
      "loss": 0.8724,
      "step": 25816
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.4213645458221436,
      "learning_rate": 4.124205347730249e-06,
      "loss": 0.7342,
      "step": 25817
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6175400018692017,
      "learning_rate": 4.12356643133246e-06,
      "loss": 0.4673,
      "step": 25818
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1743686199188232,
      "learning_rate": 4.122927514934671e-06,
      "loss": 1.0249,
      "step": 25819
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0300748348236084,
      "learning_rate": 4.1222885985368815e-06,
      "loss": 0.9325,
      "step": 25820
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5787829160690308,
      "learning_rate": 4.121649682139092e-06,
      "loss": 0.8536,
      "step": 25821
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.47216534614563,
      "learning_rate": 4.121010765741303e-06,
      "loss": 0.8174,
      "step": 25822
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.0660390853881836,
      "learning_rate": 4.120371849343514e-06,
      "loss": 0.8187,
      "step": 25823
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7651069164276123,
      "learning_rate": 4.119732932945724e-06,
      "loss": 0.773,
      "step": 25824
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1606881618499756,
      "learning_rate": 4.119094016547935e-06,
      "loss": 0.8642,
      "step": 25825
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6852049827575684,
      "learning_rate": 4.118455100150146e-06,
      "loss": 0.7842,
      "step": 25826
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.034757137298584,
      "learning_rate": 4.117816183752356e-06,
      "loss": 0.9693,
      "step": 25827
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.114502429962158,
      "learning_rate": 4.117177267354567e-06,
      "loss": 0.8902,
      "step": 25828
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.781692385673523,
      "learning_rate": 4.116538350956778e-06,
      "loss": 0.8671,
      "step": 25829
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.6075100898742676,
      "learning_rate": 4.1158994345589885e-06,
      "loss": 0.6387,
      "step": 25830
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.136930465698242,
      "learning_rate": 4.115260518161198e-06,
      "loss": 0.8746,
      "step": 25831
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.507539749145508,
      "learning_rate": 4.114621601763409e-06,
      "loss": 1.141,
      "step": 25832
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.454862356185913,
      "learning_rate": 4.11398268536562e-06,
      "loss": 0.7248,
      "step": 25833
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.939336895942688,
      "learning_rate": 4.1133437689678305e-06,
      "loss": 0.8669,
      "step": 25834
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.888932943344116,
      "learning_rate": 4.112704852570041e-06,
      "loss": 0.7265,
      "step": 25835
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.104667901992798,
      "learning_rate": 4.112065936172252e-06,
      "loss": 0.9906,
      "step": 25836
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.800817489624023,
      "learning_rate": 4.1114270197744626e-06,
      "loss": 1.0787,
      "step": 25837
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8000890016555786,
      "learning_rate": 4.110788103376673e-06,
      "loss": 0.7029,
      "step": 25838
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.627230405807495,
      "learning_rate": 4.110149186978884e-06,
      "loss": 0.972,
      "step": 25839
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4817498922348022,
      "learning_rate": 4.109510270581095e-06,
      "loss": 0.3498,
      "step": 25840
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.112780809402466,
      "learning_rate": 4.108871354183305e-06,
      "loss": 0.6882,
      "step": 25841
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3358750343322754,
      "learning_rate": 4.108232437785516e-06,
      "loss": 0.4193,
      "step": 25842
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9593944549560547,
      "learning_rate": 4.107593521387727e-06,
      "loss": 0.8496,
      "step": 25843
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.823760509490967,
      "learning_rate": 4.1069546049899375e-06,
      "loss": 0.7422,
      "step": 25844
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5105586051940918,
      "learning_rate": 4.106315688592148e-06,
      "loss": 0.8523,
      "step": 25845
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.101480722427368,
      "learning_rate": 4.105676772194359e-06,
      "loss": 0.4775,
      "step": 25846
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.6635634899139404,
      "learning_rate": 4.1050378557965696e-06,
      "loss": 0.8606,
      "step": 25847
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.914851188659668,
      "learning_rate": 4.10439893939878e-06,
      "loss": 0.6306,
      "step": 25848
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.530308723449707,
      "learning_rate": 4.10376002300099e-06,
      "loss": 0.8663,
      "step": 25849
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.932041049003601,
      "learning_rate": 4.103121106603201e-06,
      "loss": 0.8765,
      "step": 25850
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9511140584945679,
      "learning_rate": 4.1024821902054115e-06,
      "loss": 0.9673,
      "step": 25851
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7247231006622314,
      "learning_rate": 4.101843273807622e-06,
      "loss": 0.8599,
      "step": 25852
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7432841062545776,
      "learning_rate": 4.101204357409833e-06,
      "loss": 0.8277,
      "step": 25853
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.100823402404785,
      "learning_rate": 4.100565441012044e-06,
      "loss": 0.6389,
      "step": 25854
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.629474401473999,
      "learning_rate": 4.099926524614254e-06,
      "loss": 0.827,
      "step": 25855
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.130379676818848,
      "learning_rate": 4.099287608216465e-06,
      "loss": 1.0997,
      "step": 25856
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3029303550720215,
      "learning_rate": 4.098648691818676e-06,
      "loss": 0.7139,
      "step": 25857
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.972582459449768,
      "learning_rate": 4.098009775420886e-06,
      "loss": 0.8435,
      "step": 25858
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.114995241165161,
      "learning_rate": 4.097370859023097e-06,
      "loss": 0.6787,
      "step": 25859
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6068040132522583,
      "learning_rate": 4.096731942625308e-06,
      "loss": 0.8083,
      "step": 25860
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9447031021118164,
      "learning_rate": 4.0960930262275185e-06,
      "loss": 1.0894,
      "step": 25861
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.780451774597168,
      "learning_rate": 4.095454109829729e-06,
      "loss": 1.0417,
      "step": 25862
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9117026329040527,
      "learning_rate": 4.09481519343194e-06,
      "loss": 0.9616,
      "step": 25863
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.675463080406189,
      "learning_rate": 4.094176277034151e-06,
      "loss": 0.8159,
      "step": 25864
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2246313095092773,
      "learning_rate": 4.093537360636361e-06,
      "loss": 0.9204,
      "step": 25865
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.261192798614502,
      "learning_rate": 4.092898444238572e-06,
      "loss": 1.0135,
      "step": 25866
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7274420261383057,
      "learning_rate": 4.092259527840783e-06,
      "loss": 0.8212,
      "step": 25867
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5654038190841675,
      "learning_rate": 4.0916206114429926e-06,
      "loss": 0.8574,
      "step": 25868
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.972205638885498,
      "learning_rate": 4.090981695045203e-06,
      "loss": 0.527,
      "step": 25869
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8370369672775269,
      "learning_rate": 4.090342778647414e-06,
      "loss": 0.8116,
      "step": 25870
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2822226285934448,
      "learning_rate": 4.089703862249625e-06,
      "loss": 0.3564,
      "step": 25871
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1052346229553223,
      "learning_rate": 4.089064945851835e-06,
      "loss": 0.816,
      "step": 25872
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4746534824371338,
      "learning_rate": 4.088426029454046e-06,
      "loss": 0.3798,
      "step": 25873
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7003223896026611,
      "learning_rate": 4.087787113056257e-06,
      "loss": 0.7978,
      "step": 25874
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.797418236732483,
      "learning_rate": 4.0871481966584675e-06,
      "loss": 0.98,
      "step": 25875
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3771015405654907,
      "learning_rate": 4.086509280260678e-06,
      "loss": 0.7358,
      "step": 25876
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6047827005386353,
      "learning_rate": 4.085870363862889e-06,
      "loss": 0.8338,
      "step": 25877
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7993545532226562,
      "learning_rate": 4.0852314474650996e-06,
      "loss": 0.7592,
      "step": 25878
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8923567533493042,
      "learning_rate": 4.08459253106731e-06,
      "loss": 0.9031,
      "step": 25879
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.00441575050354,
      "learning_rate": 4.083953614669521e-06,
      "loss": 1.1303,
      "step": 25880
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8003711700439453,
      "learning_rate": 4.083314698271732e-06,
      "loss": 0.8138,
      "step": 25881
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.926514148712158,
      "learning_rate": 4.082675781873942e-06,
      "loss": 0.6086,
      "step": 25882
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.179258108139038,
      "learning_rate": 4.082036865476153e-06,
      "loss": 0.7639,
      "step": 25883
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.083532810211182,
      "learning_rate": 4.081397949078364e-06,
      "loss": 1.0209,
      "step": 25884
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2681174278259277,
      "learning_rate": 4.0807590326805745e-06,
      "loss": 0.9037,
      "step": 25885
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.595253348350525,
      "learning_rate": 4.080120116282784e-06,
      "loss": 0.8207,
      "step": 25886
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4534987211227417,
      "learning_rate": 4.079481199884995e-06,
      "loss": 0.3591,
      "step": 25887
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.437009811401367,
      "learning_rate": 4.078842283487206e-06,
      "loss": 0.8871,
      "step": 25888
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6651359796524048,
      "learning_rate": 4.078203367089416e-06,
      "loss": 0.829,
      "step": 25889
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.156940460205078,
      "learning_rate": 4.077564450691627e-06,
      "loss": 0.8367,
      "step": 25890
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.176542282104492,
      "learning_rate": 4.076925534293838e-06,
      "loss": 0.9471,
      "step": 25891
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8159502744674683,
      "learning_rate": 4.0762866178960485e-06,
      "loss": 0.8898,
      "step": 25892
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5813921689987183,
      "learning_rate": 4.075647701498259e-06,
      "loss": 0.7579,
      "step": 25893
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.105625152587891,
      "learning_rate": 4.07500878510047e-06,
      "loss": 1.0217,
      "step": 25894
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.419456481933594,
      "learning_rate": 4.074369868702681e-06,
      "loss": 0.9105,
      "step": 25895
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.207279682159424,
      "learning_rate": 4.073730952304891e-06,
      "loss": 0.7341,
      "step": 25896
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3160452842712402,
      "learning_rate": 4.073092035907102e-06,
      "loss": 0.8209,
      "step": 25897
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.885352373123169,
      "learning_rate": 4.072453119509313e-06,
      "loss": 0.752,
      "step": 25898
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.4945225715637207,
      "learning_rate": 4.071814203111523e-06,
      "loss": 0.9231,
      "step": 25899
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1380739212036133,
      "learning_rate": 4.071175286713734e-06,
      "loss": 0.9041,
      "step": 25900
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.218986988067627,
      "learning_rate": 4.070536370315945e-06,
      "loss": 0.973,
      "step": 25901
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.907017707824707,
      "learning_rate": 4.0698974539181555e-06,
      "loss": 0.8742,
      "step": 25902
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.212172508239746,
      "learning_rate": 4.069258537520366e-06,
      "loss": 1.0011,
      "step": 25903
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0954666137695312,
      "learning_rate": 4.068619621122576e-06,
      "loss": 0.8357,
      "step": 25904
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.127029895782471,
      "learning_rate": 4.067980704724787e-06,
      "loss": 1.679,
      "step": 25905
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8973503112792969,
      "learning_rate": 4.0673417883269975e-06,
      "loss": 0.9841,
      "step": 25906
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.511460304260254,
      "learning_rate": 4.066702871929208e-06,
      "loss": 0.3554,
      "step": 25907
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.8183445930480957,
      "learning_rate": 4.066063955531419e-06,
      "loss": 0.9074,
      "step": 25908
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8803545236587524,
      "learning_rate": 4.0654250391336296e-06,
      "loss": 0.9022,
      "step": 25909
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9024509191513062,
      "learning_rate": 4.06478612273584e-06,
      "loss": 0.8109,
      "step": 25910
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4752203226089478,
      "learning_rate": 4.064147206338051e-06,
      "loss": 0.7171,
      "step": 25911
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9599231481552124,
      "learning_rate": 4.063508289940262e-06,
      "loss": 0.8669,
      "step": 25912
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.722615361213684,
      "learning_rate": 4.062869373542472e-06,
      "loss": 0.8483,
      "step": 25913
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.34332537651062,
      "learning_rate": 4.062230457144683e-06,
      "loss": 0.8677,
      "step": 25914
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.4566729068756104,
      "learning_rate": 4.061591540746894e-06,
      "loss": 0.8417,
      "step": 25915
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9276974201202393,
      "learning_rate": 4.0609526243491045e-06,
      "loss": 0.776,
      "step": 25916
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.131732940673828,
      "learning_rate": 4.060313707951315e-06,
      "loss": 0.9852,
      "step": 25917
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8305456638336182,
      "learning_rate": 4.059674791553526e-06,
      "loss": 1.1906,
      "step": 25918
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.545075535774231,
      "learning_rate": 4.0590358751557366e-06,
      "loss": 0.8221,
      "step": 25919
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6131335496902466,
      "learning_rate": 4.058396958757947e-06,
      "loss": 0.9655,
      "step": 25920
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.091449499130249,
      "learning_rate": 4.057758042360158e-06,
      "loss": 0.8501,
      "step": 25921
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.906665325164795,
      "learning_rate": 4.057119125962368e-06,
      "loss": 1.1443,
      "step": 25922
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.80802321434021,
      "learning_rate": 4.0564802095645785e-06,
      "loss": 0.8142,
      "step": 25923
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6187002658843994,
      "learning_rate": 4.055841293166789e-06,
      "loss": 0.8252,
      "step": 25924
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.0870537757873535,
      "learning_rate": 4.055202376769e-06,
      "loss": 0.5529,
      "step": 25925
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4350053071975708,
      "learning_rate": 4.054563460371211e-06,
      "loss": 0.768,
      "step": 25926
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9622719287872314,
      "learning_rate": 4.053924543973421e-06,
      "loss": 0.8233,
      "step": 25927
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1755266189575195,
      "learning_rate": 4.053285627575632e-06,
      "loss": 0.9167,
      "step": 25928
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3987715244293213,
      "learning_rate": 4.052646711177843e-06,
      "loss": 0.8817,
      "step": 25929
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.965112328529358,
      "learning_rate": 4.052007794780053e-06,
      "loss": 0.7533,
      "step": 25930
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.147540807723999,
      "learning_rate": 4.051368878382264e-06,
      "loss": 0.9643,
      "step": 25931
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6464248895645142,
      "learning_rate": 4.050729961984475e-06,
      "loss": 0.8687,
      "step": 25932
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9908556938171387,
      "learning_rate": 4.0500910455866855e-06,
      "loss": 1.0345,
      "step": 25933
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.459604263305664,
      "learning_rate": 4.049452129188896e-06,
      "loss": 0.7344,
      "step": 25934
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6706085205078125,
      "learning_rate": 4.048813212791107e-06,
      "loss": 0.8204,
      "step": 25935
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.2427430152893066,
      "learning_rate": 4.048174296393318e-06,
      "loss": 1.0298,
      "step": 25936
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.61482572555542,
      "learning_rate": 4.047535379995528e-06,
      "loss": 0.7664,
      "step": 25937
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3604141473770142,
      "learning_rate": 4.046896463597739e-06,
      "loss": 0.4042,
      "step": 25938
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1329495906829834,
      "learning_rate": 4.04625754719995e-06,
      "loss": 0.6894,
      "step": 25939
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7174152135849,
      "learning_rate": 4.04561863080216e-06,
      "loss": 0.8873,
      "step": 25940
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.6724154949188232,
      "learning_rate": 4.04497971440437e-06,
      "loss": 0.7256,
      "step": 25941
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8724191188812256,
      "learning_rate": 4.044340798006581e-06,
      "loss": 0.8649,
      "step": 25942
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.4206674098968506,
      "learning_rate": 4.043701881608792e-06,
      "loss": 0.9198,
      "step": 25943
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.194112777709961,
      "learning_rate": 4.043062965211002e-06,
      "loss": 0.9619,
      "step": 25944
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.782224655151367,
      "learning_rate": 4.042424048813213e-06,
      "loss": 1.3325,
      "step": 25945
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.004920482635498,
      "learning_rate": 4.041785132415424e-06,
      "loss": 0.7418,
      "step": 25946
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.784394383430481,
      "learning_rate": 4.0411462160176345e-06,
      "loss": 0.494,
      "step": 25947
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7396035194396973,
      "learning_rate": 4.040507299619845e-06,
      "loss": 0.8952,
      "step": 25948
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8090277910232544,
      "learning_rate": 4.039868383222056e-06,
      "loss": 0.8142,
      "step": 25949
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.612706661224365,
      "learning_rate": 4.0392294668242666e-06,
      "loss": 1.1705,
      "step": 25950
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.125682353973389,
      "learning_rate": 4.038590550426477e-06,
      "loss": 1.029,
      "step": 25951
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7492166757583618,
      "learning_rate": 4.037951634028688e-06,
      "loss": 0.86,
      "step": 25952
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.576674699783325,
      "learning_rate": 4.037312717630899e-06,
      "loss": 0.5048,
      "step": 25953
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7511589527130127,
      "learning_rate": 4.036673801233109e-06,
      "loss": 0.7397,
      "step": 25954
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.240313768386841,
      "learning_rate": 4.03603488483532e-06,
      "loss": 0.9378,
      "step": 25955
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9000791311264038,
      "learning_rate": 4.035395968437531e-06,
      "loss": 0.8569,
      "step": 25956
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.944081425666809,
      "learning_rate": 4.0347570520397415e-06,
      "loss": 0.7552,
      "step": 25957
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4122837781906128,
      "learning_rate": 4.034118135641952e-06,
      "loss": 0.3642,
      "step": 25958
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7755588293075562,
      "learning_rate": 4.033479219244162e-06,
      "loss": 1.2104,
      "step": 25959
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.2349765300750732,
      "learning_rate": 4.032840302846373e-06,
      "loss": 0.6132,
      "step": 25960
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2133448123931885,
      "learning_rate": 4.032201386448583e-06,
      "loss": 0.3951,
      "step": 25961
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.7815394401550293,
      "learning_rate": 4.031562470050794e-06,
      "loss": 1.2851,
      "step": 25962
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4547451734542847,
      "learning_rate": 4.030923553653005e-06,
      "loss": 0.3858,
      "step": 25963
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.427581548690796,
      "learning_rate": 4.0302846372552155e-06,
      "loss": 0.7187,
      "step": 25964
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0487890243530273,
      "learning_rate": 4.029645720857426e-06,
      "loss": 0.8341,
      "step": 25965
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9818456172943115,
      "learning_rate": 4.029006804459637e-06,
      "loss": 0.8313,
      "step": 25966
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.588737964630127,
      "learning_rate": 4.028367888061848e-06,
      "loss": 0.7876,
      "step": 25967
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9216760396957397,
      "learning_rate": 4.027728971664058e-06,
      "loss": 1.2253,
      "step": 25968
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0897622108459473,
      "learning_rate": 4.027090055266269e-06,
      "loss": 0.8652,
      "step": 25969
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8683149814605713,
      "learning_rate": 4.02645113886848e-06,
      "loss": 0.9215,
      "step": 25970
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8944364786148071,
      "learning_rate": 4.02581222247069e-06,
      "loss": 0.8648,
      "step": 25971
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2925167083740234,
      "learning_rate": 4.025173306072901e-06,
      "loss": 0.8081,
      "step": 25972
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8086793422698975,
      "learning_rate": 4.024534389675112e-06,
      "loss": 0.8698,
      "step": 25973
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.666475772857666,
      "learning_rate": 4.0238954732773225e-06,
      "loss": 1.0591,
      "step": 25974
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.438014507293701,
      "learning_rate": 4.023256556879533e-06,
      "loss": 1.3183,
      "step": 25975
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5001754760742188,
      "learning_rate": 4.022617640481744e-06,
      "loss": 0.8662,
      "step": 25976
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7809911966323853,
      "learning_rate": 4.021978724083954e-06,
      "loss": 0.8898,
      "step": 25977
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9108132123947144,
      "learning_rate": 4.0213398076861645e-06,
      "loss": 1.1685,
      "step": 25978
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2370026111602783,
      "learning_rate": 4.020700891288375e-06,
      "loss": 0.8487,
      "step": 25979
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.849269151687622,
      "learning_rate": 4.020061974890586e-06,
      "loss": 1.2195,
      "step": 25980
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6729921102523804,
      "learning_rate": 4.0194230584927966e-06,
      "loss": 0.9201,
      "step": 25981
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.897300362586975,
      "learning_rate": 4.018784142095007e-06,
      "loss": 0.8424,
      "step": 25982
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8937773704528809,
      "learning_rate": 4.018145225697218e-06,
      "loss": 0.8108,
      "step": 25983
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.716910719871521,
      "learning_rate": 4.017506309299429e-06,
      "loss": 0.7064,
      "step": 25984
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9500250816345215,
      "learning_rate": 4.016867392901639e-06,
      "loss": 0.7786,
      "step": 25985
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9831997156143188,
      "learning_rate": 4.01622847650385e-06,
      "loss": 0.759,
      "step": 25986
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.26330304145813,
      "learning_rate": 4.015589560106061e-06,
      "loss": 0.8067,
      "step": 25987
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.963092803955078,
      "learning_rate": 4.0149506437082715e-06,
      "loss": 0.6991,
      "step": 25988
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7078449726104736,
      "learning_rate": 4.014311727310482e-06,
      "loss": 0.8731,
      "step": 25989
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.86416757106781,
      "learning_rate": 4.013672810912693e-06,
      "loss": 0.9004,
      "step": 25990
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.070488452911377,
      "learning_rate": 4.0130338945149036e-06,
      "loss": 1.1274,
      "step": 25991
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.6065499782562256,
      "learning_rate": 4.012394978117114e-06,
      "loss": 0.7836,
      "step": 25992
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.968813419342041,
      "learning_rate": 4.011756061719324e-06,
      "loss": 0.5033,
      "step": 25993
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.565908670425415,
      "learning_rate": 4.011117145321535e-06,
      "loss": 0.384,
      "step": 25994
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.379270315170288,
      "learning_rate": 4.0104782289237455e-06,
      "loss": 0.7625,
      "step": 25995
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.710331678390503,
      "learning_rate": 4.009839312525956e-06,
      "loss": 0.8842,
      "step": 25996
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.811785340309143,
      "learning_rate": 4.009200396128167e-06,
      "loss": 0.797,
      "step": 25997
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.979602336883545,
      "learning_rate": 4.008561479730378e-06,
      "loss": 1.1321,
      "step": 25998
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3213250637054443,
      "learning_rate": 4.007922563332588e-06,
      "loss": 0.7513,
      "step": 25999
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.6443965435028076,
      "learning_rate": 4.007283646934799e-06,
      "loss": 1.108,
      "step": 26000
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7005419731140137,
      "learning_rate": 4.00664473053701e-06,
      "loss": 0.9959,
      "step": 26001
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.3626954555511475,
      "learning_rate": 4.00600581413922e-06,
      "loss": 0.9932,
      "step": 26002
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.763993501663208,
      "learning_rate": 4.00536689774143e-06,
      "loss": 0.7688,
      "step": 26003
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.904897689819336,
      "learning_rate": 4.004727981343641e-06,
      "loss": 1.3356,
      "step": 26004
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7975285053253174,
      "learning_rate": 4.004089064945852e-06,
      "loss": 1.0531,
      "step": 26005
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5416896343231201,
      "learning_rate": 4.003450148548062e-06,
      "loss": 0.7634,
      "step": 26006
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.349487543106079,
      "learning_rate": 4.002811232150273e-06,
      "loss": 0.3863,
      "step": 26007
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.378936529159546,
      "learning_rate": 4.002172315752484e-06,
      "loss": 0.635,
      "step": 26008
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5691676139831543,
      "learning_rate": 4.0015333993546945e-06,
      "loss": 0.837,
      "step": 26009
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6667771339416504,
      "learning_rate": 4.000894482956905e-06,
      "loss": 0.6879,
      "step": 26010
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6771680116653442,
      "learning_rate": 4.000255566559116e-06,
      "loss": 0.4413,
      "step": 26011
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.493303060531616,
      "learning_rate": 3.999616650161327e-06,
      "loss": 1.2553,
      "step": 26012
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8799859285354614,
      "learning_rate": 3.998977733763537e-06,
      "loss": 1.1801,
      "step": 26013
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.5633115768432617,
      "learning_rate": 3.998338817365748e-06,
      "loss": 1.1929,
      "step": 26014
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.231113433837891,
      "learning_rate": 3.997699900967959e-06,
      "loss": 0.9489,
      "step": 26015
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.024590492248535,
      "learning_rate": 3.997060984570169e-06,
      "loss": 1.1342,
      "step": 26016
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0511977672576904,
      "learning_rate": 3.99642206817238e-06,
      "loss": 1.1242,
      "step": 26017
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.153260231018066,
      "learning_rate": 3.995783151774591e-06,
      "loss": 1.198,
      "step": 26018
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5335562229156494,
      "learning_rate": 3.9951442353768015e-06,
      "loss": 0.687,
      "step": 26019
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6015230417251587,
      "learning_rate": 3.994505318979011e-06,
      "loss": 0.8245,
      "step": 26020
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.960716962814331,
      "learning_rate": 3.993866402581222e-06,
      "loss": 0.8352,
      "step": 26021
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.771343469619751,
      "learning_rate": 3.993227486183433e-06,
      "loss": 0.6966,
      "step": 26022
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5541021823883057,
      "learning_rate": 3.9925885697856434e-06,
      "loss": 0.7348,
      "step": 26023
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5915124416351318,
      "learning_rate": 3.991949653387854e-06,
      "loss": 0.838,
      "step": 26024
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.054807186126709,
      "learning_rate": 3.991310736990065e-06,
      "loss": 1.1879,
      "step": 26025
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8909187316894531,
      "learning_rate": 3.9906718205922755e-06,
      "loss": 0.7831,
      "step": 26026
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6738609075546265,
      "learning_rate": 3.990032904194486e-06,
      "loss": 0.7214,
      "step": 26027
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9475144147872925,
      "learning_rate": 3.989393987796697e-06,
      "loss": 1.1738,
      "step": 26028
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5116088390350342,
      "learning_rate": 3.988755071398908e-06,
      "loss": 0.47,
      "step": 26029
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7217943668365479,
      "learning_rate": 3.988116155001118e-06,
      "loss": 0.8049,
      "step": 26030
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2243428230285645,
      "learning_rate": 3.987477238603329e-06,
      "loss": 0.9137,
      "step": 26031
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5099868774414062,
      "learning_rate": 3.98683832220554e-06,
      "loss": 0.7295,
      "step": 26032
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.40026593208313,
      "learning_rate": 3.9861994058077504e-06,
      "loss": 0.9129,
      "step": 26033
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.51265287399292,
      "learning_rate": 3.985560489409961e-06,
      "loss": 0.6865,
      "step": 26034
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.234159469604492,
      "learning_rate": 3.984921573012172e-06,
      "loss": 0.9085,
      "step": 26035
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8928965330123901,
      "learning_rate": 3.9842826566143825e-06,
      "loss": 0.857,
      "step": 26036
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6396888494491577,
      "learning_rate": 3.983643740216593e-06,
      "loss": 0.7661,
      "step": 26037
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.960967779159546,
      "learning_rate": 3.983004823818803e-06,
      "loss": 0.986,
      "step": 26038
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9397900104522705,
      "learning_rate": 3.982365907421014e-06,
      "loss": 0.7982,
      "step": 26039
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6647100448608398,
      "learning_rate": 3.9817269910232245e-06,
      "loss": 0.7413,
      "step": 26040
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.612684726715088,
      "learning_rate": 3.981088074625435e-06,
      "loss": 0.9556,
      "step": 26041
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.208458423614502,
      "learning_rate": 3.980449158227646e-06,
      "loss": 0.9575,
      "step": 26042
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7575726509094238,
      "learning_rate": 3.979810241829857e-06,
      "loss": 0.9098,
      "step": 26043
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.5277276039123535,
      "learning_rate": 3.979171325432067e-06,
      "loss": 1.0245,
      "step": 26044
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.925961971282959,
      "learning_rate": 3.978532409034278e-06,
      "loss": 0.8901,
      "step": 26045
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9515764713287354,
      "learning_rate": 3.977893492636489e-06,
      "loss": 0.7011,
      "step": 26046
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.140886068344116,
      "learning_rate": 3.977254576238699e-06,
      "loss": 0.7886,
      "step": 26047
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1911392211914062,
      "learning_rate": 3.97661565984091e-06,
      "loss": 0.9396,
      "step": 26048
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9481806755065918,
      "learning_rate": 3.975976743443121e-06,
      "loss": 1.0667,
      "step": 26049
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7086822986602783,
      "learning_rate": 3.9753378270453315e-06,
      "loss": 1.0092,
      "step": 26050
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2618870735168457,
      "learning_rate": 3.974698910647542e-06,
      "loss": 0.8848,
      "step": 26051
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1469485759735107,
      "learning_rate": 3.974059994249753e-06,
      "loss": 0.9321,
      "step": 26052
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8037372827529907,
      "learning_rate": 3.973421077851964e-06,
      "loss": 0.8443,
      "step": 26053
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5734351873397827,
      "learning_rate": 3.972782161454174e-06,
      "loss": 0.6784,
      "step": 26054
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7059355974197388,
      "learning_rate": 3.972143245056385e-06,
      "loss": 0.9076,
      "step": 26055
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6024879217147827,
      "learning_rate": 3.971504328658595e-06,
      "loss": 0.9323,
      "step": 26056
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.090867519378662,
      "learning_rate": 3.9708654122608055e-06,
      "loss": 1.1755,
      "step": 26057
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4149242639541626,
      "learning_rate": 3.970226495863016e-06,
      "loss": 0.8327,
      "step": 26058
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5482962131500244,
      "learning_rate": 3.969587579465227e-06,
      "loss": 0.7858,
      "step": 26059
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8479888439178467,
      "learning_rate": 3.968948663067438e-06,
      "loss": 0.8039,
      "step": 26060
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.426586627960205,
      "learning_rate": 3.968309746669648e-06,
      "loss": 1.0142,
      "step": 26061
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.927303671836853,
      "learning_rate": 3.967670830271859e-06,
      "loss": 0.7102,
      "step": 26062
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3962911367416382,
      "learning_rate": 3.96703191387407e-06,
      "loss": 0.7851,
      "step": 26063
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.9079155921936035,
      "learning_rate": 3.9663929974762804e-06,
      "loss": 0.768,
      "step": 26064
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.244771957397461,
      "learning_rate": 3.965754081078491e-06,
      "loss": 0.9372,
      "step": 26065
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.914644241333008,
      "learning_rate": 3.965115164680702e-06,
      "loss": 0.7044,
      "step": 26066
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.16413950920105,
      "learning_rate": 3.9644762482829125e-06,
      "loss": 0.818,
      "step": 26067
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8026609420776367,
      "learning_rate": 3.963837331885123e-06,
      "loss": 0.8107,
      "step": 26068
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5903735160827637,
      "learning_rate": 3.963198415487334e-06,
      "loss": 0.8533,
      "step": 26069
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.667370319366455,
      "learning_rate": 3.962559499089545e-06,
      "loss": 0.8739,
      "step": 26070
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9898959398269653,
      "learning_rate": 3.961920582691755e-06,
      "loss": 0.7731,
      "step": 26071
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.438448905944824,
      "learning_rate": 3.961281666293966e-06,
      "loss": 1.126,
      "step": 26072
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.174881935119629,
      "learning_rate": 3.960642749896177e-06,
      "loss": 1.1227,
      "step": 26073
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.4796252250671387,
      "learning_rate": 3.9600038334983874e-06,
      "loss": 0.9991,
      "step": 26074
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.29681921005249,
      "learning_rate": 3.959364917100597e-06,
      "loss": 1.3246,
      "step": 26075
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.0957555770874023,
      "learning_rate": 3.958726000702808e-06,
      "loss": 1.2032,
      "step": 26076
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.407378673553467,
      "learning_rate": 3.958087084305019e-06,
      "loss": 0.7697,
      "step": 26077
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6795824766159058,
      "learning_rate": 3.957448167907229e-06,
      "loss": 0.7482,
      "step": 26078
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8077880144119263,
      "learning_rate": 3.95680925150944e-06,
      "loss": 0.6894,
      "step": 26079
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5298725366592407,
      "learning_rate": 3.956170335111651e-06,
      "loss": 0.7953,
      "step": 26080
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9746170043945312,
      "learning_rate": 3.9555314187138615e-06,
      "loss": 0.8267,
      "step": 26081
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9436049461364746,
      "learning_rate": 3.954892502316072e-06,
      "loss": 0.9059,
      "step": 26082
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.118314743041992,
      "learning_rate": 3.954253585918283e-06,
      "loss": 1.3368,
      "step": 26083
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6363511085510254,
      "learning_rate": 3.953614669520494e-06,
      "loss": 0.7351,
      "step": 26084
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.881285667419434,
      "learning_rate": 3.952975753122704e-06,
      "loss": 1.3929,
      "step": 26085
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7219704389572144,
      "learning_rate": 3.952336836724915e-06,
      "loss": 0.9087,
      "step": 26086
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.914583444595337,
      "learning_rate": 3.951697920327126e-06,
      "loss": 0.8916,
      "step": 26087
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5196276903152466,
      "learning_rate": 3.951059003929336e-06,
      "loss": 0.7772,
      "step": 26088
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.8636624813079834,
      "learning_rate": 3.950420087531547e-06,
      "loss": 0.8033,
      "step": 26089
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.889512062072754,
      "learning_rate": 3.949781171133758e-06,
      "loss": 0.8463,
      "step": 26090
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7390989065170288,
      "learning_rate": 3.9491422547359685e-06,
      "loss": 0.8595,
      "step": 26091
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7978719472885132,
      "learning_rate": 3.948503338338179e-06,
      "loss": 0.8704,
      "step": 26092
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.8588345050811768,
      "learning_rate": 3.947864421940389e-06,
      "loss": 1.0416,
      "step": 26093
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.57954740524292,
      "learning_rate": 3.9472255055426e-06,
      "loss": 1.7931,
      "step": 26094
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.090724468231201,
      "learning_rate": 3.9465865891448104e-06,
      "loss": 0.9214,
      "step": 26095
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7021303176879883,
      "learning_rate": 3.945947672747021e-06,
      "loss": 0.7697,
      "step": 26096
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7373710870742798,
      "learning_rate": 3.945308756349232e-06,
      "loss": 0.6943,
      "step": 26097
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6186408996582031,
      "learning_rate": 3.9446698399514425e-06,
      "loss": 0.7368,
      "step": 26098
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.737872838973999,
      "learning_rate": 3.944030923553653e-06,
      "loss": 0.8934,
      "step": 26099
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5846481323242188,
      "learning_rate": 3.943392007155864e-06,
      "loss": 0.7929,
      "step": 26100
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9183346033096313,
      "learning_rate": 3.942753090758075e-06,
      "loss": 0.881,
      "step": 26101
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8506267070770264,
      "learning_rate": 3.942114174360285e-06,
      "loss": 1.0169,
      "step": 26102
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.422287940979004,
      "learning_rate": 3.941475257962496e-06,
      "loss": 0.4664,
      "step": 26103
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.683509111404419,
      "learning_rate": 3.940836341564707e-06,
      "loss": 0.8449,
      "step": 26104
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.695544958114624,
      "learning_rate": 3.9401974251669174e-06,
      "loss": 0.9872,
      "step": 26105
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.108943223953247,
      "learning_rate": 3.939558508769128e-06,
      "loss": 0.6622,
      "step": 26106
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8251749277114868,
      "learning_rate": 3.938919592371339e-06,
      "loss": 0.8453,
      "step": 26107
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5365941524505615,
      "learning_rate": 3.9382806759735495e-06,
      "loss": 0.7071,
      "step": 26108
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8226252794265747,
      "learning_rate": 3.93764175957576e-06,
      "loss": 0.824,
      "step": 26109
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.120120048522949,
      "learning_rate": 3.937002843177971e-06,
      "loss": 1.0631,
      "step": 26110
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.545479416847229,
      "learning_rate": 3.936363926780181e-06,
      "loss": 0.7203,
      "step": 26111
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2157630920410156,
      "learning_rate": 3.9357250103823915e-06,
      "loss": 0.8456,
      "step": 26112
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.790462851524353,
      "learning_rate": 3.935086093984602e-06,
      "loss": 0.8183,
      "step": 26113
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7466208934783936,
      "learning_rate": 3.934447177586813e-06,
      "loss": 0.919,
      "step": 26114
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.8466687202453613,
      "learning_rate": 3.933808261189024e-06,
      "loss": 0.9726,
      "step": 26115
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.308198928833008,
      "learning_rate": 3.933169344791234e-06,
      "loss": 1.1693,
      "step": 26116
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3363616466522217,
      "learning_rate": 3.932530428393445e-06,
      "loss": 0.8079,
      "step": 26117
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.942264437675476,
      "learning_rate": 3.931891511995656e-06,
      "loss": 0.7961,
      "step": 26118
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6466106176376343,
      "learning_rate": 3.931252595597866e-06,
      "loss": 0.7455,
      "step": 26119
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8797993659973145,
      "learning_rate": 3.930613679200077e-06,
      "loss": 0.7993,
      "step": 26120
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.655216097831726,
      "learning_rate": 3.929974762802288e-06,
      "loss": 1.115,
      "step": 26121
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.050051689147949,
      "learning_rate": 3.9293358464044985e-06,
      "loss": 1.0814,
      "step": 26122
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.163940668106079,
      "learning_rate": 3.928696930006709e-06,
      "loss": 1.1547,
      "step": 26123
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.17969012260437,
      "learning_rate": 3.92805801360892e-06,
      "loss": 0.9446,
      "step": 26124
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.692617654800415,
      "learning_rate": 3.927419097211131e-06,
      "loss": 0.9389,
      "step": 26125
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.596799373626709,
      "learning_rate": 3.926780180813341e-06,
      "loss": 0.8443,
      "step": 26126
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8493691682815552,
      "learning_rate": 3.926141264415552e-06,
      "loss": 0.9164,
      "step": 26127
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.692466378211975,
      "learning_rate": 3.925502348017763e-06,
      "loss": 0.9419,
      "step": 26128
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0505762100219727,
      "learning_rate": 3.9248634316199725e-06,
      "loss": 1.0636,
      "step": 26129
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5868560075759888,
      "learning_rate": 3.924224515222183e-06,
      "loss": 0.7908,
      "step": 26130
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.436568260192871,
      "learning_rate": 3.923585598824394e-06,
      "loss": 0.9988,
      "step": 26131
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5509932041168213,
      "learning_rate": 3.922946682426605e-06,
      "loss": 0.7695,
      "step": 26132
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6288655996322632,
      "learning_rate": 3.922307766028815e-06,
      "loss": 0.9046,
      "step": 26133
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4259181022644043,
      "learning_rate": 3.921668849631026e-06,
      "loss": 0.7425,
      "step": 26134
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.618528127670288,
      "learning_rate": 3.921029933233237e-06,
      "loss": 0.8845,
      "step": 26135
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7546048164367676,
      "learning_rate": 3.9203910168354474e-06,
      "loss": 1.2111,
      "step": 26136
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.2420387268066406,
      "learning_rate": 3.919752100437658e-06,
      "loss": 0.7421,
      "step": 26137
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.843316078186035,
      "learning_rate": 3.919113184039869e-06,
      "loss": 0.8463,
      "step": 26138
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6522997617721558,
      "learning_rate": 3.9184742676420795e-06,
      "loss": 0.8167,
      "step": 26139
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7491947412490845,
      "learning_rate": 3.91783535124429e-06,
      "loss": 0.8882,
      "step": 26140
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.8542346954345703,
      "learning_rate": 3.917196434846501e-06,
      "loss": 1.2925,
      "step": 26141
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.4161272048950195,
      "learning_rate": 3.916557518448712e-06,
      "loss": 1.0274,
      "step": 26142
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.094355583190918,
      "learning_rate": 3.915918602050922e-06,
      "loss": 0.808,
      "step": 26143
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.61697518825531,
      "learning_rate": 3.915279685653133e-06,
      "loss": 0.8906,
      "step": 26144
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4327013492584229,
      "learning_rate": 3.914640769255344e-06,
      "loss": 0.4019,
      "step": 26145
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7144770622253418,
      "learning_rate": 3.9140018528575544e-06,
      "loss": 0.8503,
      "step": 26146
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7734342813491821,
      "learning_rate": 3.913362936459765e-06,
      "loss": 0.838,
      "step": 26147
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9440191984176636,
      "learning_rate": 3.912724020061975e-06,
      "loss": 0.902,
      "step": 26148
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8970911502838135,
      "learning_rate": 3.912085103664186e-06,
      "loss": 0.8733,
      "step": 26149
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.567765712738037,
      "learning_rate": 3.911446187266396e-06,
      "loss": 0.993,
      "step": 26150
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8813129663467407,
      "learning_rate": 3.910807270868607e-06,
      "loss": 0.8682,
      "step": 26151
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6785999536514282,
      "learning_rate": 3.910168354470818e-06,
      "loss": 0.9209,
      "step": 26152
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.9085922241210938,
      "learning_rate": 3.9095294380730285e-06,
      "loss": 0.5775,
      "step": 26153
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8629018068313599,
      "learning_rate": 3.908890521675239e-06,
      "loss": 1.178,
      "step": 26154
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.496309518814087,
      "learning_rate": 3.90825160527745e-06,
      "loss": 0.4775,
      "step": 26155
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5924694538116455,
      "learning_rate": 3.907612688879661e-06,
      "loss": 0.8305,
      "step": 26156
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.780556321144104,
      "learning_rate": 3.906973772481871e-06,
      "loss": 0.6651,
      "step": 26157
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.205256223678589,
      "learning_rate": 3.906334856084082e-06,
      "loss": 0.6776,
      "step": 26158
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4846667051315308,
      "learning_rate": 3.905695939686293e-06,
      "loss": 0.766,
      "step": 26159
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.9641647338867188,
      "learning_rate": 3.905057023288503e-06,
      "loss": 1.1258,
      "step": 26160
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9762330055236816,
      "learning_rate": 3.904418106890714e-06,
      "loss": 0.9537,
      "step": 26161
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7043254375457764,
      "learning_rate": 3.903779190492925e-06,
      "loss": 1.0227,
      "step": 26162
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.799870491027832,
      "learning_rate": 3.9031402740951355e-06,
      "loss": 1.0039,
      "step": 26163
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7522350549697876,
      "learning_rate": 3.902501357697346e-06,
      "loss": 0.8448,
      "step": 26164
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.499941110610962,
      "learning_rate": 3.901862441299557e-06,
      "loss": 0.7958,
      "step": 26165
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1026220321655273,
      "learning_rate": 3.901223524901767e-06,
      "loss": 0.9715,
      "step": 26166
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7095683813095093,
      "learning_rate": 3.9005846085039774e-06,
      "loss": 0.8423,
      "step": 26167
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4030845165252686,
      "learning_rate": 3.899945692106188e-06,
      "loss": 0.4228,
      "step": 26168
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.713954210281372,
      "learning_rate": 3.899306775708399e-06,
      "loss": 1.1777,
      "step": 26169
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.67667818069458,
      "learning_rate": 3.8986678593106095e-06,
      "loss": 0.9509,
      "step": 26170
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0820438861846924,
      "learning_rate": 3.89802894291282e-06,
      "loss": 0.8032,
      "step": 26171
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7694107294082642,
      "learning_rate": 3.897390026515031e-06,
      "loss": 0.8439,
      "step": 26172
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.3751368522644043,
      "learning_rate": 3.896751110117242e-06,
      "loss": 0.8451,
      "step": 26173
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5191973447799683,
      "learning_rate": 3.896112193719452e-06,
      "loss": 0.7825,
      "step": 26174
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.2461133003234863,
      "learning_rate": 3.895473277321663e-06,
      "loss": 0.8516,
      "step": 26175
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7901124954223633,
      "learning_rate": 3.894834360923874e-06,
      "loss": 0.9568,
      "step": 26176
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.343920946121216,
      "learning_rate": 3.8941954445260844e-06,
      "loss": 0.8684,
      "step": 26177
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7668401002883911,
      "learning_rate": 3.893556528128295e-06,
      "loss": 0.8835,
      "step": 26178
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.823373794555664,
      "learning_rate": 3.892917611730506e-06,
      "loss": 0.7568,
      "step": 26179
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.6270596981048584,
      "learning_rate": 3.8922786953327165e-06,
      "loss": 0.8631,
      "step": 26180
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.54227876663208,
      "learning_rate": 3.891639778934927e-06,
      "loss": 0.7944,
      "step": 26181
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.6590960025787354,
      "learning_rate": 3.891000862537138e-06,
      "loss": 0.8087,
      "step": 26182
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.0917093753814697,
      "learning_rate": 3.890361946139349e-06,
      "loss": 0.6284,
      "step": 26183
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8613154888153076,
      "learning_rate": 3.8897230297415585e-06,
      "loss": 0.9513,
      "step": 26184
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.224639892578125,
      "learning_rate": 3.889084113343769e-06,
      "loss": 1.3801,
      "step": 26185
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.7587366104125977,
      "learning_rate": 3.88844519694598e-06,
      "loss": 0.85,
      "step": 26186
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.752997636795044,
      "learning_rate": 3.887806280548191e-06,
      "loss": 0.7064,
      "step": 26187
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3835042715072632,
      "learning_rate": 3.887167364150401e-06,
      "loss": 1.0592,
      "step": 26188
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0883090496063232,
      "learning_rate": 3.886528447752612e-06,
      "loss": 0.5975,
      "step": 26189
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5379550457000732,
      "learning_rate": 3.885889531354822e-06,
      "loss": 0.8374,
      "step": 26190
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.66127610206604,
      "learning_rate": 3.8852506149570326e-06,
      "loss": 0.822,
      "step": 26191
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8670154809951782,
      "learning_rate": 3.884611698559243e-06,
      "loss": 0.7988,
      "step": 26192
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.942722201347351,
      "learning_rate": 3.883972782161454e-06,
      "loss": 0.848,
      "step": 26193
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.181903600692749,
      "learning_rate": 3.883333865763665e-06,
      "loss": 0.9684,
      "step": 26194
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.517664909362793,
      "learning_rate": 3.882694949365875e-06,
      "loss": 1.6674,
      "step": 26195
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0247819423675537,
      "learning_rate": 3.882056032968086e-06,
      "loss": 1.0422,
      "step": 26196
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.3477418422698975,
      "learning_rate": 3.881417116570297e-06,
      "loss": 1.0542,
      "step": 26197
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6190279722213745,
      "learning_rate": 3.8807782001725075e-06,
      "loss": 0.7394,
      "step": 26198
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5577399730682373,
      "learning_rate": 3.880139283774718e-06,
      "loss": 0.7985,
      "step": 26199
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.582521915435791,
      "learning_rate": 3.879500367376929e-06,
      "loss": 0.8808,
      "step": 26200
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3996347188949585,
      "learning_rate": 3.8788614509791396e-06,
      "loss": 0.8501,
      "step": 26201
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4533603191375732,
      "learning_rate": 3.87822253458135e-06,
      "loss": 0.7738,
      "step": 26202
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1407928466796875,
      "learning_rate": 3.877583618183561e-06,
      "loss": 0.957,
      "step": 26203
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.368194341659546,
      "learning_rate": 3.876944701785772e-06,
      "loss": 1.1333,
      "step": 26204
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.43228816986084,
      "learning_rate": 3.876305785387982e-06,
      "loss": 0.8831,
      "step": 26205
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.097241163253784,
      "learning_rate": 3.875666868990193e-06,
      "loss": 0.7367,
      "step": 26206
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.877149224281311,
      "learning_rate": 3.875027952592404e-06,
      "loss": 0.7588,
      "step": 26207
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7170109748840332,
      "learning_rate": 3.8743890361946145e-06,
      "loss": 0.7562,
      "step": 26208
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9094957113265991,
      "learning_rate": 3.873750119796824e-06,
      "loss": 0.9219,
      "step": 26209
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.626297950744629,
      "learning_rate": 3.873111203399035e-06,
      "loss": 0.748,
      "step": 26210
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6282824277877808,
      "learning_rate": 3.872472287001246e-06,
      "loss": 0.8312,
      "step": 26211
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.893458604812622,
      "learning_rate": 3.871833370603456e-06,
      "loss": 0.8544,
      "step": 26212
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.3307418823242188,
      "learning_rate": 3.871194454205667e-06,
      "loss": 0.8493,
      "step": 26213
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.4768381118774414,
      "learning_rate": 3.870555537807878e-06,
      "loss": 0.7267,
      "step": 26214
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.106025218963623,
      "learning_rate": 3.8699166214100885e-06,
      "loss": 0.85,
      "step": 26215
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.873787760734558,
      "learning_rate": 3.869277705012299e-06,
      "loss": 0.869,
      "step": 26216
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.04537296295166,
      "learning_rate": 3.86863878861451e-06,
      "loss": 1.128,
      "step": 26217
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7522304058074951,
      "learning_rate": 3.867999872216721e-06,
      "loss": 0.7488,
      "step": 26218
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1405134201049805,
      "learning_rate": 3.867360955818931e-06,
      "loss": 0.4837,
      "step": 26219
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.223231792449951,
      "learning_rate": 3.866722039421142e-06,
      "loss": 0.8842,
      "step": 26220
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6455916166305542,
      "learning_rate": 3.866083123023353e-06,
      "loss": 0.8508,
      "step": 26221
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.25002384185791,
      "learning_rate": 3.865444206625563e-06,
      "loss": 0.6981,
      "step": 26222
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7940709590911865,
      "learning_rate": 3.864805290227774e-06,
      "loss": 0.7981,
      "step": 26223
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.988445520401001,
      "learning_rate": 3.864166373829985e-06,
      "loss": 1.0719,
      "step": 26224
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.284902334213257,
      "learning_rate": 3.8635274574321955e-06,
      "loss": 0.9023,
      "step": 26225
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7268447875976562,
      "learning_rate": 3.862888541034406e-06,
      "loss": 0.8139,
      "step": 26226
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5568861961364746,
      "learning_rate": 3.862249624636616e-06,
      "loss": 0.6446,
      "step": 26227
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.746246099472046,
      "learning_rate": 3.861610708238827e-06,
      "loss": 1.1832,
      "step": 26228
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.418353796005249,
      "learning_rate": 3.8609717918410375e-06,
      "loss": 0.839,
      "step": 26229
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.79695725440979,
      "learning_rate": 3.860332875443248e-06,
      "loss": 0.7814,
      "step": 26230
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8913495540618896,
      "learning_rate": 3.859693959045459e-06,
      "loss": 0.9685,
      "step": 26231
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7222917079925537,
      "learning_rate": 3.8590550426476696e-06,
      "loss": 0.9086,
      "step": 26232
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.562353253364563,
      "learning_rate": 3.85841612624988e-06,
      "loss": 0.8504,
      "step": 26233
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.6256332397460938,
      "learning_rate": 3.857777209852091e-06,
      "loss": 1.0119,
      "step": 26234
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7557909488677979,
      "learning_rate": 3.857138293454302e-06,
      "loss": 0.7762,
      "step": 26235
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.803778886795044,
      "learning_rate": 3.856499377056512e-06,
      "loss": 0.3972,
      "step": 26236
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0947039127349854,
      "learning_rate": 3.855860460658723e-06,
      "loss": 0.829,
      "step": 26237
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5702943801879883,
      "learning_rate": 3.855221544260934e-06,
      "loss": 0.8,
      "step": 26238
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.2260642051696777,
      "learning_rate": 3.8545826278631445e-06,
      "loss": 0.7258,
      "step": 26239
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0393717288970947,
      "learning_rate": 3.853943711465355e-06,
      "loss": 0.7151,
      "step": 26240
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.251799583435059,
      "learning_rate": 3.853304795067566e-06,
      "loss": 1.0185,
      "step": 26241
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5583035945892334,
      "learning_rate": 3.8526658786697766e-06,
      "loss": 0.7979,
      "step": 26242
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5068628787994385,
      "learning_rate": 3.852026962271987e-06,
      "loss": 1.053,
      "step": 26243
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9714967012405396,
      "learning_rate": 3.851388045874198e-06,
      "loss": 1.1643,
      "step": 26244
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6543936729431152,
      "learning_rate": 3.850749129476408e-06,
      "loss": 0.9354,
      "step": 26245
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6544735431671143,
      "learning_rate": 3.8501102130786185e-06,
      "loss": 0.7881,
      "step": 26246
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.019038677215576,
      "learning_rate": 3.849471296680829e-06,
      "loss": 0.7642,
      "step": 26247
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6491954326629639,
      "learning_rate": 3.84883238028304e-06,
      "loss": 0.7201,
      "step": 26248
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6863104104995728,
      "learning_rate": 3.848193463885251e-06,
      "loss": 0.7973,
      "step": 26249
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1126911640167236,
      "learning_rate": 3.847554547487461e-06,
      "loss": 0.8506,
      "step": 26250
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.356292247772217,
      "learning_rate": 3.846915631089672e-06,
      "loss": 0.996,
      "step": 26251
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.945764422416687,
      "learning_rate": 3.846276714691883e-06,
      "loss": 0.7816,
      "step": 26252
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9060657024383545,
      "learning_rate": 3.845637798294093e-06,
      "loss": 0.7733,
      "step": 26253
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.674355387687683,
      "learning_rate": 3.844998881896304e-06,
      "loss": 0.8005,
      "step": 26254
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.371448993682861,
      "learning_rate": 3.844359965498515e-06,
      "loss": 1.1225,
      "step": 26255
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.7750043869018555,
      "learning_rate": 3.8437210491007255e-06,
      "loss": 0.6881,
      "step": 26256
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5786269903182983,
      "learning_rate": 3.843082132702936e-06,
      "loss": 0.8789,
      "step": 26257
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.151228666305542,
      "learning_rate": 3.842443216305147e-06,
      "loss": 0.9035,
      "step": 26258
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.195132732391357,
      "learning_rate": 3.841804299907358e-06,
      "loss": 0.8694,
      "step": 26259
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.378669500350952,
      "learning_rate": 3.841165383509568e-06,
      "loss": 0.7039,
      "step": 26260
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.009490489959717,
      "learning_rate": 3.840526467111779e-06,
      "loss": 0.8671,
      "step": 26261
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6918835639953613,
      "learning_rate": 3.83988755071399e-06,
      "loss": 0.7105,
      "step": 26262
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4848037958145142,
      "learning_rate": 3.8392486343161996e-06,
      "loss": 0.7616,
      "step": 26263
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.414034366607666,
      "learning_rate": 3.83860971791841e-06,
      "loss": 0.8407,
      "step": 26264
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4509053230285645,
      "learning_rate": 3.837970801520621e-06,
      "loss": 0.3993,
      "step": 26265
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.175437927246094,
      "learning_rate": 3.837331885122832e-06,
      "loss": 1.0829,
      "step": 26266
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.4103410243988037,
      "learning_rate": 3.836692968725042e-06,
      "loss": 0.8204,
      "step": 26267
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7979066371917725,
      "learning_rate": 3.836054052327253e-06,
      "loss": 0.8319,
      "step": 26268
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.098884582519531,
      "learning_rate": 3.835415135929464e-06,
      "loss": 1.3546,
      "step": 26269
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7358441352844238,
      "learning_rate": 3.8347762195316745e-06,
      "loss": 0.718,
      "step": 26270
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.419023871421814,
      "learning_rate": 3.834137303133885e-06,
      "loss": 0.2989,
      "step": 26271
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8042433261871338,
      "learning_rate": 3.833498386736096e-06,
      "loss": 0.8745,
      "step": 26272
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9760035276412964,
      "learning_rate": 3.8328594703383066e-06,
      "loss": 0.8829,
      "step": 26273
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7591716051101685,
      "learning_rate": 3.832220553940517e-06,
      "loss": 0.8068,
      "step": 26274
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5677285194396973,
      "learning_rate": 3.831581637542728e-06,
      "loss": 0.9681,
      "step": 26275
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9561946392059326,
      "learning_rate": 3.830942721144939e-06,
      "loss": 1.0566,
      "step": 26276
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.311143159866333,
      "learning_rate": 3.830303804747149e-06,
      "loss": 0.8508,
      "step": 26277
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.026157379150391,
      "learning_rate": 3.82966488834936e-06,
      "loss": 1.1668,
      "step": 26278
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.719313621520996,
      "learning_rate": 3.829025971951571e-06,
      "loss": 0.8421,
      "step": 26279
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.531353235244751,
      "learning_rate": 3.8283870555537815e-06,
      "loss": 0.7683,
      "step": 26280
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.5564358234405518,
      "learning_rate": 3.827748139155992e-06,
      "loss": 1.0249,
      "step": 26281
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0134940147399902,
      "learning_rate": 3.827109222758202e-06,
      "loss": 0.7964,
      "step": 26282
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6729001998901367,
      "learning_rate": 3.826470306360413e-06,
      "loss": 0.7964,
      "step": 26283
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9022719860076904,
      "learning_rate": 3.825831389962623e-06,
      "loss": 0.8777,
      "step": 26284
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5369906425476074,
      "learning_rate": 3.825192473564834e-06,
      "loss": 0.8761,
      "step": 26285
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4851500988006592,
      "learning_rate": 3.824553557167045e-06,
      "loss": 0.6982,
      "step": 26286
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0298287868499756,
      "learning_rate": 3.8239146407692555e-06,
      "loss": 0.8446,
      "step": 26287
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8283796310424805,
      "learning_rate": 3.823275724371466e-06,
      "loss": 0.8708,
      "step": 26288
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1668484210968018,
      "learning_rate": 3.822636807973677e-06,
      "loss": 0.8064,
      "step": 26289
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9672894477844238,
      "learning_rate": 3.821997891575888e-06,
      "loss": 0.8374,
      "step": 26290
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7372050285339355,
      "learning_rate": 3.821358975178098e-06,
      "loss": 0.8809,
      "step": 26291
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.5529704093933105,
      "learning_rate": 3.820720058780309e-06,
      "loss": 1.2011,
      "step": 26292
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.089054822921753,
      "learning_rate": 3.82008114238252e-06,
      "loss": 0.9393,
      "step": 26293
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5341849327087402,
      "learning_rate": 3.81944222598473e-06,
      "loss": 1.0005,
      "step": 26294
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.4180407524108887,
      "learning_rate": 3.818803309586941e-06,
      "loss": 0.7079,
      "step": 26295
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.049899101257324,
      "learning_rate": 3.818164393189152e-06,
      "loss": 1.0152,
      "step": 26296
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.922282338142395,
      "learning_rate": 3.8175254767913625e-06,
      "loss": 0.9984,
      "step": 26297
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6739495992660522,
      "learning_rate": 3.816886560393573e-06,
      "loss": 0.8427,
      "step": 26298
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.836095094680786,
      "learning_rate": 3.816247643995784e-06,
      "loss": 0.7685,
      "step": 26299
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.064441204071045,
      "learning_rate": 3.815608727597994e-06,
      "loss": 0.7283,
      "step": 26300
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.474214553833008,
      "learning_rate": 3.8149698112002045e-06,
      "loss": 0.825,
      "step": 26301
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7469457387924194,
      "learning_rate": 3.8143308948024156e-06,
      "loss": 0.7983,
      "step": 26302
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.523641586303711,
      "learning_rate": 3.8136919784046263e-06,
      "loss": 1.1141,
      "step": 26303
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9163981676101685,
      "learning_rate": 3.813053062006837e-06,
      "loss": 0.7207,
      "step": 26304
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8902482986450195,
      "learning_rate": 3.8124141456090473e-06,
      "loss": 0.9044,
      "step": 26305
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6661224365234375,
      "learning_rate": 3.811775229211258e-06,
      "loss": 0.8845,
      "step": 26306
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0920560359954834,
      "learning_rate": 3.8111363128134687e-06,
      "loss": 0.824,
      "step": 26307
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9674333333969116,
      "learning_rate": 3.8104973964156794e-06,
      "loss": 0.9883,
      "step": 26308
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5513980388641357,
      "learning_rate": 3.80985848001789e-06,
      "loss": 0.8609,
      "step": 26309
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6601815223693848,
      "learning_rate": 3.8092195636201008e-06,
      "loss": 0.8522,
      "step": 26310
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9198477268218994,
      "learning_rate": 3.8085806472223115e-06,
      "loss": 0.9244,
      "step": 26311
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9341487884521484,
      "learning_rate": 3.807941730824522e-06,
      "loss": 0.8495,
      "step": 26312
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7263710498809814,
      "learning_rate": 3.807302814426733e-06,
      "loss": 0.7465,
      "step": 26313
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.109349489212036,
      "learning_rate": 3.806663898028943e-06,
      "loss": 1.0455,
      "step": 26314
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8621180057525635,
      "learning_rate": 3.806024981631154e-06,
      "loss": 0.8535,
      "step": 26315
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4464317560195923,
      "learning_rate": 3.8053860652333645e-06,
      "loss": 0.4093,
      "step": 26316
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9827537536621094,
      "learning_rate": 3.8047471488355752e-06,
      "loss": 0.8888,
      "step": 26317
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.036311149597168,
      "learning_rate": 3.804108232437786e-06,
      "loss": 1.412,
      "step": 26318
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.916881799697876,
      "learning_rate": 3.8034693160399966e-06,
      "loss": 1.0391,
      "step": 26319
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.782569169998169,
      "learning_rate": 3.8028303996422073e-06,
      "loss": 0.7106,
      "step": 26320
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7620151042938232,
      "learning_rate": 3.802191483244418e-06,
      "loss": 0.8274,
      "step": 26321
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5376250743865967,
      "learning_rate": 3.8015525668466287e-06,
      "loss": 0.3245,
      "step": 26322
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6820586919784546,
      "learning_rate": 3.800913650448839e-06,
      "loss": 0.7325,
      "step": 26323
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8853834867477417,
      "learning_rate": 3.8002747340510497e-06,
      "loss": 1.0243,
      "step": 26324
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8351902961730957,
      "learning_rate": 3.7996358176532604e-06,
      "loss": 0.7878,
      "step": 26325
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.26747465133667,
      "learning_rate": 3.798996901255471e-06,
      "loss": 0.9552,
      "step": 26326
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7768505811691284,
      "learning_rate": 3.798357984857682e-06,
      "loss": 0.8288,
      "step": 26327
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6318025588989258,
      "learning_rate": 3.7977190684598925e-06,
      "loss": 0.8312,
      "step": 26328
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.510533094406128,
      "learning_rate": 3.7970801520621032e-06,
      "loss": 0.8241,
      "step": 26329
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2273181676864624,
      "learning_rate": 3.796441235664314e-06,
      "loss": 0.3842,
      "step": 26330
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.0498297214508057,
      "learning_rate": 3.7958023192665246e-06,
      "loss": 1.0256,
      "step": 26331
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.777922511100769,
      "learning_rate": 3.795163402868735e-06,
      "loss": 0.8344,
      "step": 26332
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7797000408172607,
      "learning_rate": 3.7945244864709456e-06,
      "loss": 0.8285,
      "step": 26333
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.58115553855896,
      "learning_rate": 3.7938855700731563e-06,
      "loss": 0.7683,
      "step": 26334
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.104174852371216,
      "learning_rate": 3.793246653675367e-06,
      "loss": 1.0198,
      "step": 26335
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.101709365844727,
      "learning_rate": 3.7926077372775777e-06,
      "loss": 1.0862,
      "step": 26336
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4521474838256836,
      "learning_rate": 3.7919688208797884e-06,
      "loss": 0.8798,
      "step": 26337
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.6335465908050537,
      "learning_rate": 3.791329904481999e-06,
      "loss": 1.1453,
      "step": 26338
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.116363763809204,
      "learning_rate": 3.79069098808421e-06,
      "loss": 1.0168,
      "step": 26339
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.715529680252075,
      "learning_rate": 3.7900520716864205e-06,
      "loss": 1.0828,
      "step": 26340
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7130532264709473,
      "learning_rate": 3.7894131552886308e-06,
      "loss": 0.8175,
      "step": 26341
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7593473196029663,
      "learning_rate": 3.7887742388908415e-06,
      "loss": 0.8598,
      "step": 26342
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6713030338287354,
      "learning_rate": 3.788135322493052e-06,
      "loss": 0.8407,
      "step": 26343
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.367704391479492,
      "learning_rate": 3.787496406095263e-06,
      "loss": 0.8667,
      "step": 26344
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7559032440185547,
      "learning_rate": 3.7868574896974736e-06,
      "loss": 0.8545,
      "step": 26345
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5293047428131104,
      "learning_rate": 3.7862185732996843e-06,
      "loss": 0.8428,
      "step": 26346
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.472035527229309,
      "learning_rate": 3.785579656901895e-06,
      "loss": 0.8772,
      "step": 26347
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9775927066802979,
      "learning_rate": 3.7849407405041057e-06,
      "loss": 0.8881,
      "step": 26348
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.528219223022461,
      "learning_rate": 3.7843018241063164e-06,
      "loss": 0.7283,
      "step": 26349
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.3929121494293213,
      "learning_rate": 3.7836629077085266e-06,
      "loss": 0.6081,
      "step": 26350
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.842143177986145,
      "learning_rate": 3.7830239913107373e-06,
      "loss": 0.8046,
      "step": 26351
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8354157209396362,
      "learning_rate": 3.782385074912948e-06,
      "loss": 0.8983,
      "step": 26352
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0548431873321533,
      "learning_rate": 3.7817461585151587e-06,
      "loss": 0.9793,
      "step": 26353
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.00758957862854,
      "learning_rate": 3.7811072421173694e-06,
      "loss": 0.6603,
      "step": 26354
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.158870220184326,
      "learning_rate": 3.78046832571958e-06,
      "loss": 0.7673,
      "step": 26355
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.2432005405426025,
      "learning_rate": 3.779829409321791e-06,
      "loss": 1.2553,
      "step": 26356
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.3264334201812744,
      "learning_rate": 3.7791904929240015e-06,
      "loss": 0.8689,
      "step": 26357
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8218309879302979,
      "learning_rate": 3.7785515765262122e-06,
      "loss": 0.8436,
      "step": 26358
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9370900392532349,
      "learning_rate": 3.7779126601284225e-06,
      "loss": 0.8898,
      "step": 26359
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8022658824920654,
      "learning_rate": 3.7772737437306332e-06,
      "loss": 0.831,
      "step": 26360
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6095781326293945,
      "learning_rate": 3.776634827332844e-06,
      "loss": 0.8704,
      "step": 26361
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.1091034412384033,
      "learning_rate": 3.7759959109350546e-06,
      "loss": 0.8087,
      "step": 26362
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2154664993286133,
      "learning_rate": 3.7753569945372653e-06,
      "loss": 0.8171,
      "step": 26363
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.8574795722961426,
      "learning_rate": 3.774718078139476e-06,
      "loss": 0.9034,
      "step": 26364
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.821932792663574,
      "learning_rate": 3.7740791617416867e-06,
      "loss": 1.1144,
      "step": 26365
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.973888635635376,
      "learning_rate": 3.7734402453438974e-06,
      "loss": 0.8985,
      "step": 26366
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5153613090515137,
      "learning_rate": 3.772801328946108e-06,
      "loss": 0.7192,
      "step": 26367
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7429667711257935,
      "learning_rate": 3.7721624125483184e-06,
      "loss": 0.8549,
      "step": 26368
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4024122953414917,
      "learning_rate": 3.771523496150529e-06,
      "loss": 0.3808,
      "step": 26369
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6633692979812622,
      "learning_rate": 3.77088457975274e-06,
      "loss": 0.8975,
      "step": 26370
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.344673752784729,
      "learning_rate": 3.7702456633549505e-06,
      "loss": 0.2989,
      "step": 26371
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.292022466659546,
      "learning_rate": 3.769606746957161e-06,
      "loss": 0.6335,
      "step": 26372
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9654664993286133,
      "learning_rate": 3.768967830559372e-06,
      "loss": 0.7454,
      "step": 26373
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8999533653259277,
      "learning_rate": 3.7683289141615826e-06,
      "loss": 0.7303,
      "step": 26374
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.5415642261505127,
      "learning_rate": 3.7676899977637933e-06,
      "loss": 0.8129,
      "step": 26375
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.5386996269226074,
      "learning_rate": 3.767051081366004e-06,
      "loss": 0.8778,
      "step": 26376
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8085254430770874,
      "learning_rate": 3.7664121649682147e-06,
      "loss": 0.7414,
      "step": 26377
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7822242975234985,
      "learning_rate": 3.765773248570425e-06,
      "loss": 0.8389,
      "step": 26378
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.978726387023926,
      "learning_rate": 3.7651343321726353e-06,
      "loss": 0.9721,
      "step": 26379
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.681195259094238,
      "learning_rate": 3.764495415774846e-06,
      "loss": 1.3187,
      "step": 26380
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.697460412979126,
      "learning_rate": 3.7638564993770567e-06,
      "loss": 1.0732,
      "step": 26381
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4359865188598633,
      "learning_rate": 3.7632175829792674e-06,
      "loss": 0.7401,
      "step": 26382
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8082376718521118,
      "learning_rate": 3.762578666581478e-06,
      "loss": 0.8008,
      "step": 26383
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6521579027175903,
      "learning_rate": 3.7619397501836883e-06,
      "loss": 0.8508,
      "step": 26384
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6946669816970825,
      "learning_rate": 3.761300833785899e-06,
      "loss": 0.7371,
      "step": 26385
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.269620895385742,
      "learning_rate": 3.7606619173881097e-06,
      "loss": 1.1969,
      "step": 26386
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.31024169921875,
      "learning_rate": 3.7600230009903204e-06,
      "loss": 0.8008,
      "step": 26387
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7520719766616821,
      "learning_rate": 3.759384084592531e-06,
      "loss": 0.8452,
      "step": 26388
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7634477615356445,
      "learning_rate": 3.758745168194742e-06,
      "loss": 0.8239,
      "step": 26389
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.715865135192871,
      "learning_rate": 3.7581062517969525e-06,
      "loss": 1.2494,
      "step": 26390
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.263909101486206,
      "learning_rate": 3.7574673353991632e-06,
      "loss": 0.7606,
      "step": 26391
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9023572206497192,
      "learning_rate": 3.756828419001374e-06,
      "loss": 0.8392,
      "step": 26392
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7880053520202637,
      "learning_rate": 3.756189502603584e-06,
      "loss": 0.9452,
      "step": 26393
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9991453886032104,
      "learning_rate": 3.755550586205795e-06,
      "loss": 0.7459,
      "step": 26394
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.757697105407715,
      "learning_rate": 3.7549116698080056e-06,
      "loss": 1.0679,
      "step": 26395
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0685153007507324,
      "learning_rate": 3.7542727534102163e-06,
      "loss": 0.9211,
      "step": 26396
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6623181104660034,
      "learning_rate": 3.753633837012427e-06,
      "loss": 0.7692,
      "step": 26397
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8716416358947754,
      "learning_rate": 3.7529949206146377e-06,
      "loss": 0.9499,
      "step": 26398
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0464553833007812,
      "learning_rate": 3.7523560042168484e-06,
      "loss": 0.8235,
      "step": 26399
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.107642889022827,
      "learning_rate": 3.751717087819059e-06,
      "loss": 0.9554,
      "step": 26400
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6270794868469238,
      "learning_rate": 3.75107817142127e-06,
      "loss": 0.6642,
      "step": 26401
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9551775455474854,
      "learning_rate": 3.75043925502348e-06,
      "loss": 0.8461,
      "step": 26402
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5180034637451172,
      "learning_rate": 3.7498003386256908e-06,
      "loss": 0.7967,
      "step": 26403
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5635716915130615,
      "learning_rate": 3.7491614222279015e-06,
      "loss": 0.7982,
      "step": 26404
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1519625186920166,
      "learning_rate": 3.748522505830112e-06,
      "loss": 1.0536,
      "step": 26405
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2438764572143555,
      "learning_rate": 3.747883589432323e-06,
      "loss": 0.8313,
      "step": 26406
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8220314979553223,
      "learning_rate": 3.7472446730345336e-06,
      "loss": 0.8415,
      "step": 26407
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.750605583190918,
      "learning_rate": 3.7466057566367443e-06,
      "loss": 0.8555,
      "step": 26408
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8963978290557861,
      "learning_rate": 3.745966840238955e-06,
      "loss": 0.8399,
      "step": 26409
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6783686876296997,
      "learning_rate": 3.7453279238411657e-06,
      "loss": 0.8128,
      "step": 26410
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.53464937210083,
      "learning_rate": 3.744689007443376e-06,
      "loss": 1.0499,
      "step": 26411
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4401135444641113,
      "learning_rate": 3.7440500910455867e-06,
      "loss": 0.7698,
      "step": 26412
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.4935994148254395,
      "learning_rate": 3.7434111746477974e-06,
      "loss": 1.1524,
      "step": 26413
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.111659526824951,
      "learning_rate": 3.742772258250008e-06,
      "loss": 0.8963,
      "step": 26414
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4396851062774658,
      "learning_rate": 3.7421333418522188e-06,
      "loss": 0.3822,
      "step": 26415
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.051862955093384,
      "learning_rate": 3.7414944254544295e-06,
      "loss": 1.1692,
      "step": 26416
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6470108032226562,
      "learning_rate": 3.74085550905664e-06,
      "loss": 1.1123,
      "step": 26417
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3174419403076172,
      "learning_rate": 3.740216592658851e-06,
      "loss": 0.3773,
      "step": 26418
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4806228876113892,
      "learning_rate": 3.7395776762610616e-06,
      "loss": 0.6422,
      "step": 26419
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.333733320236206,
      "learning_rate": 3.738938759863272e-06,
      "loss": 1.0531,
      "step": 26420
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7603259086608887,
      "learning_rate": 3.7382998434654825e-06,
      "loss": 0.9143,
      "step": 26421
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9176135063171387,
      "learning_rate": 3.7376609270676932e-06,
      "loss": 0.8786,
      "step": 26422
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7103697061538696,
      "learning_rate": 3.737022010669904e-06,
      "loss": 0.8315,
      "step": 26423
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6777061223983765,
      "learning_rate": 3.7363830942721146e-06,
      "loss": 0.8498,
      "step": 26424
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6511750221252441,
      "learning_rate": 3.7357441778743253e-06,
      "loss": 0.8397,
      "step": 26425
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8774995803833008,
      "learning_rate": 3.735105261476536e-06,
      "loss": 0.831,
      "step": 26426
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.160396099090576,
      "learning_rate": 3.7344663450787467e-06,
      "loss": 0.9842,
      "step": 26427
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8135648965835571,
      "learning_rate": 3.7338274286809574e-06,
      "loss": 0.6754,
      "step": 26428
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7106508016586304,
      "learning_rate": 3.733188512283168e-06,
      "loss": 0.8856,
      "step": 26429
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.763858437538147,
      "learning_rate": 3.7325495958853784e-06,
      "loss": 0.7538,
      "step": 26430
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8520716428756714,
      "learning_rate": 3.731910679487589e-06,
      "loss": 0.8114,
      "step": 26431
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.021223783493042,
      "learning_rate": 3.7312717630898e-06,
      "loss": 0.7673,
      "step": 26432
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.905477285385132,
      "learning_rate": 3.7306328466920105e-06,
      "loss": 1.2972,
      "step": 26433
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6187200546264648,
      "learning_rate": 3.729993930294221e-06,
      "loss": 0.7643,
      "step": 26434
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.28081476688385,
      "learning_rate": 3.729355013896432e-06,
      "loss": 0.3477,
      "step": 26435
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0723533630371094,
      "learning_rate": 3.7287160974986426e-06,
      "loss": 1.1359,
      "step": 26436
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4916305541992188,
      "learning_rate": 3.7280771811008533e-06,
      "loss": 0.7398,
      "step": 26437
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.895501971244812,
      "learning_rate": 3.727438264703064e-06,
      "loss": 0.7584,
      "step": 26438
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7927870750427246,
      "learning_rate": 3.7267993483052743e-06,
      "loss": 0.8998,
      "step": 26439
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9881027936935425,
      "learning_rate": 3.726160431907485e-06,
      "loss": 0.8236,
      "step": 26440
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.38509464263916,
      "learning_rate": 3.7255215155096957e-06,
      "loss": 1.2109,
      "step": 26441
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.725348711013794,
      "learning_rate": 3.7248825991119064e-06,
      "loss": 0.764,
      "step": 26442
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9756689071655273,
      "learning_rate": 3.724243682714117e-06,
      "loss": 0.9286,
      "step": 26443
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5070505142211914,
      "learning_rate": 3.7236047663163278e-06,
      "loss": 0.845,
      "step": 26444
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.76322340965271,
      "learning_rate": 3.7229658499185385e-06,
      "loss": 0.9301,
      "step": 26445
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.822047233581543,
      "learning_rate": 3.722326933520749e-06,
      "loss": 1.1396,
      "step": 26446
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5295683145523071,
      "learning_rate": 3.72168801712296e-06,
      "loss": 0.3964,
      "step": 26447
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2207844257354736,
      "learning_rate": 3.72104910072517e-06,
      "loss": 0.8879,
      "step": 26448
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7955105304718018,
      "learning_rate": 3.720410184327381e-06,
      "loss": 0.7282,
      "step": 26449
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.840954065322876,
      "learning_rate": 3.7197712679295916e-06,
      "loss": 0.7456,
      "step": 26450
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4509954452514648,
      "learning_rate": 3.7191323515318023e-06,
      "loss": 0.401,
      "step": 26451
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0330708026885986,
      "learning_rate": 3.718493435134013e-06,
      "loss": 0.8226,
      "step": 26452
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5163276195526123,
      "learning_rate": 3.7178545187362237e-06,
      "loss": 0.9632,
      "step": 26453
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9164605140686035,
      "learning_rate": 3.7172156023384344e-06,
      "loss": 0.9494,
      "step": 26454
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5047131776809692,
      "learning_rate": 3.716576685940645e-06,
      "loss": 0.8718,
      "step": 26455
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.69456148147583,
      "learning_rate": 3.7159377695428558e-06,
      "loss": 0.8172,
      "step": 26456
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9109793901443481,
      "learning_rate": 3.715298853145066e-06,
      "loss": 1.1833,
      "step": 26457
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5835567712783813,
      "learning_rate": 3.7146599367472767e-06,
      "loss": 0.8389,
      "step": 26458
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7809135913848877,
      "learning_rate": 3.7140210203494874e-06,
      "loss": 0.8518,
      "step": 26459
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7898001670837402,
      "learning_rate": 3.713382103951698e-06,
      "loss": 0.7428,
      "step": 26460
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9943970441818237,
      "learning_rate": 3.712743187553909e-06,
      "loss": 0.7749,
      "step": 26461
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7468608617782593,
      "learning_rate": 3.7121042711561195e-06,
      "loss": 0.7812,
      "step": 26462
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.621683955192566,
      "learning_rate": 3.7114653547583302e-06,
      "loss": 0.8527,
      "step": 26463
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.085444927215576,
      "learning_rate": 3.710826438360541e-06,
      "loss": 1.2607,
      "step": 26464
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7281450033187866,
      "learning_rate": 3.7101875219627516e-06,
      "loss": 0.8549,
      "step": 26465
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8776352405548096,
      "learning_rate": 3.709548605564962e-06,
      "loss": 0.9194,
      "step": 26466
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.040825605392456,
      "learning_rate": 3.7089096891671726e-06,
      "loss": 0.9401,
      "step": 26467
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.5513112545013428,
      "learning_rate": 3.7082707727693833e-06,
      "loss": 0.6845,
      "step": 26468
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.660296082496643,
      "learning_rate": 3.707631856371594e-06,
      "loss": 0.851,
      "step": 26469
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5988095998764038,
      "learning_rate": 3.7069929399738047e-06,
      "loss": 0.7974,
      "step": 26470
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.3477461338043213,
      "learning_rate": 3.7063540235760154e-06,
      "loss": 0.7785,
      "step": 26471
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.63407826423645,
      "learning_rate": 3.705715107178226e-06,
      "loss": 0.8058,
      "step": 26472
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.0898919105529785,
      "learning_rate": 3.705076190780437e-06,
      "loss": 0.9193,
      "step": 26473
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8984036445617676,
      "learning_rate": 3.7044372743826475e-06,
      "loss": 0.8523,
      "step": 26474
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.970871925354004,
      "learning_rate": 3.703798357984858e-06,
      "loss": 0.8592,
      "step": 26475
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.269491672515869,
      "learning_rate": 3.7031594415870685e-06,
      "loss": 0.7882,
      "step": 26476
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.733506917953491,
      "learning_rate": 3.702520525189279e-06,
      "loss": 0.899,
      "step": 26477
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.609395503997803,
      "learning_rate": 3.70188160879149e-06,
      "loss": 1.4633,
      "step": 26478
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7855198383331299,
      "learning_rate": 3.7012426923937006e-06,
      "loss": 0.7803,
      "step": 26479
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1962974071502686,
      "learning_rate": 3.7006037759959113e-06,
      "loss": 0.9144,
      "step": 26480
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.593133568763733,
      "learning_rate": 3.699964859598122e-06,
      "loss": 0.7483,
      "step": 26481
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.797465801239014,
      "learning_rate": 3.6993259432003327e-06,
      "loss": 1.4453,
      "step": 26482
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6005165576934814,
      "learning_rate": 3.6986870268025434e-06,
      "loss": 0.874,
      "step": 26483
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4151360988616943,
      "learning_rate": 3.6980481104047537e-06,
      "loss": 1.0364,
      "step": 26484
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.200753927230835,
      "learning_rate": 3.6974091940069644e-06,
      "loss": 0.9945,
      "step": 26485
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0077362060546875,
      "learning_rate": 3.696770277609175e-06,
      "loss": 0.89,
      "step": 26486
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6027424335479736,
      "learning_rate": 3.6961313612113858e-06,
      "loss": 0.849,
      "step": 26487
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.769446849822998,
      "learning_rate": 3.6954924448135965e-06,
      "loss": 0.8208,
      "step": 26488
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7971123456954956,
      "learning_rate": 3.694853528415807e-06,
      "loss": 0.7571,
      "step": 26489
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.962053894996643,
      "learning_rate": 3.694214612018018e-06,
      "loss": 0.9545,
      "step": 26490
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.706824779510498,
      "learning_rate": 3.6935756956202286e-06,
      "loss": 0.506,
      "step": 26491
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6344327926635742,
      "learning_rate": 3.6929367792224393e-06,
      "loss": 0.8192,
      "step": 26492
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7665717601776123,
      "learning_rate": 3.6922978628246495e-06,
      "loss": 0.9092,
      "step": 26493
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.738863468170166,
      "learning_rate": 3.6916589464268602e-06,
      "loss": 0.9819,
      "step": 26494
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.854929804801941,
      "learning_rate": 3.691020030029071e-06,
      "loss": 0.889,
      "step": 26495
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4919971227645874,
      "learning_rate": 3.6903811136312816e-06,
      "loss": 0.3129,
      "step": 26496
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8288968801498413,
      "learning_rate": 3.6897421972334923e-06,
      "loss": 0.7766,
      "step": 26497
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.992671251296997,
      "learning_rate": 3.689103280835703e-06,
      "loss": 1.2147,
      "step": 26498
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.039850950241089,
      "learning_rate": 3.6884643644379137e-06,
      "loss": 0.5463,
      "step": 26499
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6425585746765137,
      "learning_rate": 3.6878254480401244e-06,
      "loss": 0.8113,
      "step": 26500
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1648497581481934,
      "learning_rate": 3.687186531642335e-06,
      "loss": 0.8678,
      "step": 26501
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7161952257156372,
      "learning_rate": 3.686547615244546e-06,
      "loss": 0.657,
      "step": 26502
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.720266342163086,
      "learning_rate": 3.685908698846756e-06,
      "loss": 0.9254,
      "step": 26503
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9065563678741455,
      "learning_rate": 3.685269782448967e-06,
      "loss": 0.9511,
      "step": 26504
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.133155822753906,
      "learning_rate": 3.6846308660511775e-06,
      "loss": 1.0517,
      "step": 26505
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7373713254928589,
      "learning_rate": 3.6839919496533882e-06,
      "loss": 0.8829,
      "step": 26506
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9170684814453125,
      "learning_rate": 3.683353033255599e-06,
      "loss": 0.854,
      "step": 26507
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.431661605834961,
      "learning_rate": 3.6827141168578096e-06,
      "loss": 0.7333,
      "step": 26508
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.397591590881348,
      "learning_rate": 3.6820752004600203e-06,
      "loss": 1.8528,
      "step": 26509
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.3236804008483887,
      "learning_rate": 3.681436284062231e-06,
      "loss": 0.8942,
      "step": 26510
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0827114582061768,
      "learning_rate": 3.6807973676644417e-06,
      "loss": 0.8245,
      "step": 26511
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.823929786682129,
      "learning_rate": 3.680158451266652e-06,
      "loss": 1.336,
      "step": 26512
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.489828586578369,
      "learning_rate": 3.6795195348688627e-06,
      "loss": 0.731,
      "step": 26513
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9955337047576904,
      "learning_rate": 3.6788806184710734e-06,
      "loss": 1.0873,
      "step": 26514
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1424949169158936,
      "learning_rate": 3.678241702073284e-06,
      "loss": 0.8939,
      "step": 26515
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6949704885482788,
      "learning_rate": 3.677602785675495e-06,
      "loss": 0.8878,
      "step": 26516
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.373575448989868,
      "learning_rate": 3.6769638692777055e-06,
      "loss": 0.5765,
      "step": 26517
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.096951961517334,
      "learning_rate": 3.676324952879916e-06,
      "loss": 0.4741,
      "step": 26518
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.956775188446045,
      "learning_rate": 3.675686036482127e-06,
      "loss": 1.1818,
      "step": 26519
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0330069065093994,
      "learning_rate": 3.6750471200843376e-06,
      "loss": 0.8416,
      "step": 26520
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.7107598781585693,
      "learning_rate": 3.674408203686548e-06,
      "loss": 0.4757,
      "step": 26521
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.762691617012024,
      "learning_rate": 3.6737692872887586e-06,
      "loss": 0.9525,
      "step": 26522
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6556856632232666,
      "learning_rate": 3.6731303708909693e-06,
      "loss": 0.9263,
      "step": 26523
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4047632217407227,
      "learning_rate": 3.67249145449318e-06,
      "loss": 0.3887,
      "step": 26524
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.786068320274353,
      "learning_rate": 3.6718525380953907e-06,
      "loss": 0.9079,
      "step": 26525
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9323687553405762,
      "learning_rate": 3.6712136216976014e-06,
      "loss": 0.9685,
      "step": 26526
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5736162662506104,
      "learning_rate": 3.670574705299812e-06,
      "loss": 0.8149,
      "step": 26527
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.219120025634766,
      "learning_rate": 3.6699357889020228e-06,
      "loss": 1.142,
      "step": 26528
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8618910312652588,
      "learning_rate": 3.6692968725042335e-06,
      "loss": 0.8721,
      "step": 26529
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.318619251251221,
      "learning_rate": 3.6686579561064437e-06,
      "loss": 0.7669,
      "step": 26530
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3931159973144531,
      "learning_rate": 3.6680190397086544e-06,
      "loss": 0.368,
      "step": 26531
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4082293510437012,
      "learning_rate": 3.667380123310865e-06,
      "loss": 0.3583,
      "step": 26532
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8666372299194336,
      "learning_rate": 3.666741206913076e-06,
      "loss": 0.7877,
      "step": 26533
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8325837850570679,
      "learning_rate": 3.6661022905152865e-06,
      "loss": 0.8731,
      "step": 26534
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.7531018257141113,
      "learning_rate": 3.6654633741174972e-06,
      "loss": 0.8088,
      "step": 26535
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4164173603057861,
      "learning_rate": 3.664824457719708e-06,
      "loss": 0.8212,
      "step": 26536
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8662775754928589,
      "learning_rate": 3.6641855413219186e-06,
      "loss": 0.7544,
      "step": 26537
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.4691076278686523,
      "learning_rate": 3.6635466249241293e-06,
      "loss": 0.9573,
      "step": 26538
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.057185649871826,
      "learning_rate": 3.6629077085263396e-06,
      "loss": 1.0239,
      "step": 26539
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.351045608520508,
      "learning_rate": 3.6622687921285503e-06,
      "loss": 1.1839,
      "step": 26540
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9101277589797974,
      "learning_rate": 3.661629875730761e-06,
      "loss": 0.7758,
      "step": 26541
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7091302871704102,
      "learning_rate": 3.6609909593329717e-06,
      "loss": 0.7354,
      "step": 26542
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.865098476409912,
      "learning_rate": 3.6603520429351824e-06,
      "loss": 0.7274,
      "step": 26543
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8162761926651,
      "learning_rate": 3.659713126537393e-06,
      "loss": 0.9276,
      "step": 26544
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.5423214435577393,
      "learning_rate": 3.659074210139604e-06,
      "loss": 1.1759,
      "step": 26545
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.241102933883667,
      "learning_rate": 3.6584352937418145e-06,
      "loss": 0.9684,
      "step": 26546
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0142781734466553,
      "learning_rate": 3.6577963773440252e-06,
      "loss": 0.8684,
      "step": 26547
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9102379083633423,
      "learning_rate": 3.6571574609462355e-06,
      "loss": 0.8584,
      "step": 26548
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5533900260925293,
      "learning_rate": 3.656518544548446e-06,
      "loss": 0.842,
      "step": 26549
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8937879800796509,
      "learning_rate": 3.655879628150657e-06,
      "loss": 0.8506,
      "step": 26550
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.8976428508758545,
      "learning_rate": 3.6552407117528676e-06,
      "loss": 1.2097,
      "step": 26551
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6609464883804321,
      "learning_rate": 3.6546017953550783e-06,
      "loss": 0.8845,
      "step": 26552
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8301613330841064,
      "learning_rate": 3.653962878957289e-06,
      "loss": 0.8812,
      "step": 26553
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.112637996673584,
      "learning_rate": 3.6533239625594997e-06,
      "loss": 0.922,
      "step": 26554
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0686452388763428,
      "learning_rate": 3.6526850461617104e-06,
      "loss": 0.8552,
      "step": 26555
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.174159526824951,
      "learning_rate": 3.652046129763921e-06,
      "loss": 0.8746,
      "step": 26556
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6890650987625122,
      "learning_rate": 3.6514072133661314e-06,
      "loss": 1.226,
      "step": 26557
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8025656938552856,
      "learning_rate": 3.650768296968342e-06,
      "loss": 0.8333,
      "step": 26558
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7413978576660156,
      "learning_rate": 3.6501293805705528e-06,
      "loss": 0.9095,
      "step": 26559
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.123366355895996,
      "learning_rate": 3.6494904641727635e-06,
      "loss": 0.8647,
      "step": 26560
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.18780517578125,
      "learning_rate": 3.648851547774974e-06,
      "loss": 0.8671,
      "step": 26561
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.309378623962402,
      "learning_rate": 3.648212631377185e-06,
      "loss": 0.6038,
      "step": 26562
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.050292730331421,
      "learning_rate": 3.6475737149793956e-06,
      "loss": 0.8746,
      "step": 26563
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.634498953819275,
      "learning_rate": 3.6469347985816063e-06,
      "loss": 0.8614,
      "step": 26564
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0101242065429688,
      "learning_rate": 3.646295882183817e-06,
      "loss": 0.7613,
      "step": 26565
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7875598669052124,
      "learning_rate": 3.6456569657860272e-06,
      "loss": 0.8668,
      "step": 26566
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.953944206237793,
      "learning_rate": 3.645018049388238e-06,
      "loss": 0.734,
      "step": 26567
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6372263431549072,
      "learning_rate": 3.6443791329904486e-06,
      "loss": 0.9951,
      "step": 26568
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.577124834060669,
      "learning_rate": 3.6437402165926593e-06,
      "loss": 0.7958,
      "step": 26569
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.887515068054199,
      "learning_rate": 3.64310130019487e-06,
      "loss": 0.98,
      "step": 26570
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9187594652175903,
      "learning_rate": 3.6424623837970807e-06,
      "loss": 0.7182,
      "step": 26571
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9460837841033936,
      "learning_rate": 3.641823467399291e-06,
      "loss": 0.8583,
      "step": 26572
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.3684840202331543,
      "learning_rate": 3.6411845510015013e-06,
      "loss": 0.9319,
      "step": 26573
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.5107808113098145,
      "learning_rate": 3.640545634603712e-06,
      "loss": 0.5565,
      "step": 26574
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.45962381362915,
      "learning_rate": 3.6399067182059227e-06,
      "loss": 1.0343,
      "step": 26575
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.6780309677124023,
      "learning_rate": 3.6392678018081334e-06,
      "loss": 0.8277,
      "step": 26576
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.270248889923096,
      "learning_rate": 3.638628885410344e-06,
      "loss": 1.1629,
      "step": 26577
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.5822713375091553,
      "learning_rate": 3.637989969012555e-06,
      "loss": 0.7613,
      "step": 26578
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.612295389175415,
      "learning_rate": 3.6373510526147655e-06,
      "loss": 0.8386,
      "step": 26579
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4449855089187622,
      "learning_rate": 3.636712136216976e-06,
      "loss": 0.4168,
      "step": 26580
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9456602334976196,
      "learning_rate": 3.636073219819187e-06,
      "loss": 0.8491,
      "step": 26581
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.07277250289917,
      "learning_rate": 3.635434303421397e-06,
      "loss": 0.8829,
      "step": 26582
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.393407821655273,
      "learning_rate": 3.634795387023608e-06,
      "loss": 1.2268,
      "step": 26583
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0822055339813232,
      "learning_rate": 3.6341564706258186e-06,
      "loss": 0.7347,
      "step": 26584
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.331825852394104,
      "learning_rate": 3.6335175542280293e-06,
      "loss": 0.3413,
      "step": 26585
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6978776454925537,
      "learning_rate": 3.63287863783024e-06,
      "loss": 0.7803,
      "step": 26586
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0230367183685303,
      "learning_rate": 3.6322397214324507e-06,
      "loss": 0.5219,
      "step": 26587
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.083883762359619,
      "learning_rate": 3.6316008050346614e-06,
      "loss": 0.7861,
      "step": 26588
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.006480693817139,
      "learning_rate": 3.630961888636872e-06,
      "loss": 0.8962,
      "step": 26589
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8065624237060547,
      "learning_rate": 3.6303229722390828e-06,
      "loss": 0.8395,
      "step": 26590
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5145280361175537,
      "learning_rate": 3.629684055841293e-06,
      "loss": 0.4517,
      "step": 26591
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.3504128456115723,
      "learning_rate": 3.6290451394435038e-06,
      "loss": 1.0866,
      "step": 26592
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1181211471557617,
      "learning_rate": 3.6284062230457145e-06,
      "loss": 0.8474,
      "step": 26593
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.5166804790496826,
      "learning_rate": 3.627767306647925e-06,
      "loss": 0.5279,
      "step": 26594
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.353628158569336,
      "learning_rate": 3.627128390250136e-06,
      "loss": 0.9379,
      "step": 26595
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8891834020614624,
      "learning_rate": 3.6264894738523466e-06,
      "loss": 0.8934,
      "step": 26596
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9352113008499146,
      "learning_rate": 3.6258505574545573e-06,
      "loss": 0.8087,
      "step": 26597
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7532259225845337,
      "learning_rate": 3.625211641056768e-06,
      "loss": 0.7963,
      "step": 26598
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9772518873214722,
      "learning_rate": 3.6245727246589787e-06,
      "loss": 0.9432,
      "step": 26599
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9588359594345093,
      "learning_rate": 3.623933808261189e-06,
      "loss": 0.8828,
      "step": 26600
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6917829513549805,
      "learning_rate": 3.6232948918633996e-06,
      "loss": 0.8449,
      "step": 26601
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5872211456298828,
      "learning_rate": 3.6226559754656103e-06,
      "loss": 0.7002,
      "step": 26602
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.945128917694092,
      "learning_rate": 3.622017059067821e-06,
      "loss": 0.4836,
      "step": 26603
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8653395175933838,
      "learning_rate": 3.6213781426700317e-06,
      "loss": 0.8607,
      "step": 26604
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.646632432937622,
      "learning_rate": 3.6207392262722424e-06,
      "loss": 0.8688,
      "step": 26605
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6630388498306274,
      "learning_rate": 3.620100309874453e-06,
      "loss": 0.7733,
      "step": 26606
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8130955696105957,
      "learning_rate": 3.619461393476664e-06,
      "loss": 0.803,
      "step": 26607
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9069011211395264,
      "learning_rate": 3.6188224770788745e-06,
      "loss": 0.7959,
      "step": 26608
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6934642791748047,
      "learning_rate": 3.618183560681085e-06,
      "loss": 0.6523,
      "step": 26609
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9264888763427734,
      "learning_rate": 3.6175446442832955e-06,
      "loss": 1.0277,
      "step": 26610
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6957483291625977,
      "learning_rate": 3.616905727885506e-06,
      "loss": 0.7834,
      "step": 26611
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.334712982177734,
      "learning_rate": 3.616266811487717e-06,
      "loss": 1.1856,
      "step": 26612
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.872756004333496,
      "learning_rate": 3.6156278950899276e-06,
      "loss": 0.5987,
      "step": 26613
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.882483959197998,
      "learning_rate": 3.6149889786921383e-06,
      "loss": 1.5732,
      "step": 26614
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8709850311279297,
      "learning_rate": 3.614350062294349e-06,
      "loss": 0.6314,
      "step": 26615
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6857162714004517,
      "learning_rate": 3.6137111458965597e-06,
      "loss": 0.8455,
      "step": 26616
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.599807620048523,
      "learning_rate": 3.6130722294987704e-06,
      "loss": 0.8388,
      "step": 26617
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8446959257125854,
      "learning_rate": 3.6124333131009807e-06,
      "loss": 0.8693,
      "step": 26618
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7506977319717407,
      "learning_rate": 3.6117943967031914e-06,
      "loss": 0.9004,
      "step": 26619
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.6981496810913086,
      "learning_rate": 3.611155480305402e-06,
      "loss": 0.9338,
      "step": 26620
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.501271367073059,
      "learning_rate": 3.6105165639076128e-06,
      "loss": 0.8442,
      "step": 26621
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.078970193862915,
      "learning_rate": 3.6098776475098235e-06,
      "loss": 0.9159,
      "step": 26622
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1755943298339844,
      "learning_rate": 3.609238731112034e-06,
      "loss": 0.5459,
      "step": 26623
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.663817048072815,
      "learning_rate": 3.608599814714245e-06,
      "loss": 0.8335,
      "step": 26624
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0972156524658203,
      "learning_rate": 3.6079608983164556e-06,
      "loss": 1.057,
      "step": 26625
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8940839767456055,
      "learning_rate": 3.6073219819186663e-06,
      "loss": 0.992,
      "step": 26626
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7482984066009521,
      "learning_rate": 3.6066830655208766e-06,
      "loss": 0.478,
      "step": 26627
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6218916177749634,
      "learning_rate": 3.6060441491230873e-06,
      "loss": 0.7475,
      "step": 26628
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3069117069244385,
      "learning_rate": 3.605405232725298e-06,
      "loss": 0.8025,
      "step": 26629
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9582022428512573,
      "learning_rate": 3.6047663163275087e-06,
      "loss": 0.7668,
      "step": 26630
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.894230604171753,
      "learning_rate": 3.6041273999297194e-06,
      "loss": 0.7676,
      "step": 26631
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.744520664215088,
      "learning_rate": 3.60348848353193e-06,
      "loss": 0.8317,
      "step": 26632
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.733869552612305,
      "learning_rate": 3.6028495671341408e-06,
      "loss": 1.1452,
      "step": 26633
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.1459784507751465,
      "learning_rate": 3.6022106507363515e-06,
      "loss": 0.7192,
      "step": 26634
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4533581733703613,
      "learning_rate": 3.601571734338562e-06,
      "loss": 0.6964,
      "step": 26635
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1764886379241943,
      "learning_rate": 3.600932817940773e-06,
      "loss": 0.8366,
      "step": 26636
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6692700386047363,
      "learning_rate": 3.600293901542983e-06,
      "loss": 0.8707,
      "step": 26637
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0405707359313965,
      "learning_rate": 3.599654985145194e-06,
      "loss": 1.0549,
      "step": 26638
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.5016045570373535,
      "learning_rate": 3.5990160687474045e-06,
      "loss": 0.9677,
      "step": 26639
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3138465881347656,
      "learning_rate": 3.5983771523496152e-06,
      "loss": 0.8028,
      "step": 26640
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.557298183441162,
      "learning_rate": 3.597738235951826e-06,
      "loss": 0.7606,
      "step": 26641
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4508109092712402,
      "learning_rate": 3.5970993195540366e-06,
      "loss": 0.7743,
      "step": 26642
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6874314546585083,
      "learning_rate": 3.5964604031562473e-06,
      "loss": 0.8051,
      "step": 26643
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.613637089729309,
      "learning_rate": 3.595821486758458e-06,
      "loss": 0.82,
      "step": 26644
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4026503562927246,
      "learning_rate": 3.5951825703606687e-06,
      "loss": 0.4096,
      "step": 26645
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.559852123260498,
      "learning_rate": 3.594543653962879e-06,
      "loss": 0.4641,
      "step": 26646
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.6993391513824463,
      "learning_rate": 3.5939047375650897e-06,
      "loss": 0.7092,
      "step": 26647
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0766441822052,
      "learning_rate": 3.5932658211673004e-06,
      "loss": 0.82,
      "step": 26648
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.916583776473999,
      "learning_rate": 3.592626904769511e-06,
      "loss": 0.9235,
      "step": 26649
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.089357614517212,
      "learning_rate": 3.591987988371722e-06,
      "loss": 0.888,
      "step": 26650
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.871927738189697,
      "learning_rate": 3.5913490719739325e-06,
      "loss": 0.7409,
      "step": 26651
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1505091190338135,
      "learning_rate": 3.590710155576143e-06,
      "loss": 0.9107,
      "step": 26652
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.827842354774475,
      "learning_rate": 3.590071239178354e-06,
      "loss": 0.7459,
      "step": 26653
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5045809745788574,
      "learning_rate": 3.5894323227805646e-06,
      "loss": 0.8029,
      "step": 26654
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.897907018661499,
      "learning_rate": 3.588793406382775e-06,
      "loss": 0.8792,
      "step": 26655
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.7613890171051025,
      "learning_rate": 3.5881544899849856e-06,
      "loss": 1.2394,
      "step": 26656
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4328680038452148,
      "learning_rate": 3.5875155735871963e-06,
      "loss": 0.7471,
      "step": 26657
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6309367418289185,
      "learning_rate": 3.586876657189407e-06,
      "loss": 0.8028,
      "step": 26658
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9828110933303833,
      "learning_rate": 3.5862377407916177e-06,
      "loss": 0.7689,
      "step": 26659
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5063532590866089,
      "learning_rate": 3.5855988243938284e-06,
      "loss": 0.8012,
      "step": 26660
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2422995567321777,
      "learning_rate": 3.584959907996039e-06,
      "loss": 0.4286,
      "step": 26661
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9528847932815552,
      "learning_rate": 3.5843209915982498e-06,
      "loss": 0.5421,
      "step": 26662
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.592246651649475,
      "learning_rate": 3.5836820752004605e-06,
      "loss": 0.8227,
      "step": 26663
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.560397148132324,
      "learning_rate": 3.5830431588026708e-06,
      "loss": 1.142,
      "step": 26664
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7390505075454712,
      "learning_rate": 3.5824042424048815e-06,
      "loss": 0.9134,
      "step": 26665
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7910654544830322,
      "learning_rate": 3.581765326007092e-06,
      "loss": 0.7761,
      "step": 26666
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9459596872329712,
      "learning_rate": 3.581126409609303e-06,
      "loss": 0.885,
      "step": 26667
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.3246452808380127,
      "learning_rate": 3.5804874932115136e-06,
      "loss": 0.9564,
      "step": 26668
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3690983057022095,
      "learning_rate": 3.5798485768137243e-06,
      "loss": 0.3918,
      "step": 26669
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6289438009262085,
      "learning_rate": 3.579209660415935e-06,
      "loss": 0.8951,
      "step": 26670
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.393453359603882,
      "learning_rate": 3.5785707440181457e-06,
      "loss": 0.8011,
      "step": 26671
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7215313911437988,
      "learning_rate": 3.5779318276203564e-06,
      "loss": 0.872,
      "step": 26672
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7032840251922607,
      "learning_rate": 3.5772929112225666e-06,
      "loss": 0.718,
      "step": 26673
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.729393720626831,
      "learning_rate": 3.5766539948247773e-06,
      "loss": 0.7639,
      "step": 26674
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.255040645599365,
      "learning_rate": 3.576015078426988e-06,
      "loss": 1.4835,
      "step": 26675
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.383826732635498,
      "learning_rate": 3.5753761620291987e-06,
      "loss": 1.4723,
      "step": 26676
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9208664894104004,
      "learning_rate": 3.5747372456314094e-06,
      "loss": 0.7733,
      "step": 26677
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8866046667099,
      "learning_rate": 3.57409832923362e-06,
      "loss": 0.8644,
      "step": 26678
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.1770706176757812,
      "learning_rate": 3.573459412835831e-06,
      "loss": 0.8441,
      "step": 26679
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4584916830062866,
      "learning_rate": 3.5728204964380415e-06,
      "loss": 0.79,
      "step": 26680
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9349852800369263,
      "learning_rate": 3.5721815800402522e-06,
      "loss": 0.9054,
      "step": 26681
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6884536743164062,
      "learning_rate": 3.5715426636424625e-06,
      "loss": 1.048,
      "step": 26682
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.74419367313385,
      "learning_rate": 3.5709037472446732e-06,
      "loss": 0.8958,
      "step": 26683
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.00483775138855,
      "learning_rate": 3.570264830846884e-06,
      "loss": 0.661,
      "step": 26684
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7204538583755493,
      "learning_rate": 3.5696259144490946e-06,
      "loss": 0.8265,
      "step": 26685
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8704071044921875,
      "learning_rate": 3.5689869980513053e-06,
      "loss": 0.7454,
      "step": 26686
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.1377503871917725,
      "learning_rate": 3.568348081653516e-06,
      "loss": 1.1467,
      "step": 26687
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.53887677192688,
      "learning_rate": 3.5677091652557267e-06,
      "loss": 0.5628,
      "step": 26688
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.029920816421509,
      "learning_rate": 3.5670702488579374e-06,
      "loss": 0.8156,
      "step": 26689
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4765689373016357,
      "learning_rate": 3.566431332460148e-06,
      "loss": 0.9595,
      "step": 26690
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7217397689819336,
      "learning_rate": 3.5657924160623584e-06,
      "loss": 0.8182,
      "step": 26691
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.6519124507904053,
      "learning_rate": 3.565153499664569e-06,
      "loss": 0.719,
      "step": 26692
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4490931034088135,
      "learning_rate": 3.56451458326678e-06,
      "loss": 0.7698,
      "step": 26693
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.692819595336914,
      "learning_rate": 3.5638756668689905e-06,
      "loss": 0.9222,
      "step": 26694
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.044405221939087,
      "learning_rate": 3.563236750471201e-06,
      "loss": 0.883,
      "step": 26695
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.6546335220336914,
      "learning_rate": 3.562597834073412e-06,
      "loss": 1.067,
      "step": 26696
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9621808528900146,
      "learning_rate": 3.5619589176756226e-06,
      "loss": 0.8107,
      "step": 26697
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8030025959014893,
      "learning_rate": 3.5613200012778333e-06,
      "loss": 0.8006,
      "step": 26698
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.524287223815918,
      "learning_rate": 3.560681084880044e-06,
      "loss": 1.1143,
      "step": 26699
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.7673583030700684,
      "learning_rate": 3.5600421684822543e-06,
      "loss": 0.6695,
      "step": 26700
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9768784046173096,
      "learning_rate": 3.559403252084465e-06,
      "loss": 0.9474,
      "step": 26701
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.187906742095947,
      "learning_rate": 3.5587643356866757e-06,
      "loss": 1.2995,
      "step": 26702
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.0986886024475098,
      "learning_rate": 3.5581254192888864e-06,
      "loss": 0.8755,
      "step": 26703
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7630983591079712,
      "learning_rate": 3.557486502891097e-06,
      "loss": 0.8168,
      "step": 26704
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.099355936050415,
      "learning_rate": 3.5568475864933078e-06,
      "loss": 0.9922,
      "step": 26705
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8983240127563477,
      "learning_rate": 3.5562086700955185e-06,
      "loss": 1.1585,
      "step": 26706
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.560746192932129,
      "learning_rate": 3.555569753697729e-06,
      "loss": 0.7567,
      "step": 26707
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4676227569580078,
      "learning_rate": 3.55493083729994e-06,
      "loss": 0.6407,
      "step": 26708
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.9787702560424805,
      "learning_rate": 3.5542919209021506e-06,
      "loss": 0.7637,
      "step": 26709
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.011098861694336,
      "learning_rate": 3.553653004504361e-06,
      "loss": 1.3321,
      "step": 26710
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8031264543533325,
      "learning_rate": 3.5530140881065715e-06,
      "loss": 0.7138,
      "step": 26711
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9470394849777222,
      "learning_rate": 3.5523751717087822e-06,
      "loss": 0.7443,
      "step": 26712
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8513606786727905,
      "learning_rate": 3.551736255310993e-06,
      "loss": 1.0176,
      "step": 26713
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.918705701828003,
      "learning_rate": 3.5510973389132036e-06,
      "loss": 0.8118,
      "step": 26714
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9422200918197632,
      "learning_rate": 3.5504584225154143e-06,
      "loss": 0.9055,
      "step": 26715
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8630781173706055,
      "learning_rate": 3.549819506117625e-06,
      "loss": 0.8582,
      "step": 26716
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9782381057739258,
      "learning_rate": 3.5491805897198357e-06,
      "loss": 0.8337,
      "step": 26717
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.3159143924713135,
      "learning_rate": 3.5485416733220464e-06,
      "loss": 0.9908,
      "step": 26718
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8824094533920288,
      "learning_rate": 3.5479027569242567e-06,
      "loss": 0.8602,
      "step": 26719
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.7679567337036133,
      "learning_rate": 3.5472638405264674e-06,
      "loss": 0.9046,
      "step": 26720
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.001890182495117,
      "learning_rate": 3.546624924128678e-06,
      "loss": 0.8762,
      "step": 26721
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7614859342575073,
      "learning_rate": 3.545986007730889e-06,
      "loss": 0.7556,
      "step": 26722
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6565502882003784,
      "learning_rate": 3.5453470913330995e-06,
      "loss": 0.73,
      "step": 26723
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9843016862869263,
      "learning_rate": 3.5447081749353102e-06,
      "loss": 0.9761,
      "step": 26724
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7129331827163696,
      "learning_rate": 3.544069258537521e-06,
      "loss": 0.7924,
      "step": 26725
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7861690521240234,
      "learning_rate": 3.5434303421397316e-06,
      "loss": 0.9334,
      "step": 26726
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.65100359916687,
      "learning_rate": 3.5427914257419423e-06,
      "loss": 0.7784,
      "step": 26727
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8462114334106445,
      "learning_rate": 3.5421525093441526e-06,
      "loss": 0.9611,
      "step": 26728
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8424071073532104,
      "learning_rate": 3.5415135929463633e-06,
      "loss": 0.7792,
      "step": 26729
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.164078950881958,
      "learning_rate": 3.540874676548574e-06,
      "loss": 0.9085,
      "step": 26730
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.839451551437378,
      "learning_rate": 3.5402357601507847e-06,
      "loss": 1.0066,
      "step": 26731
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.68092942237854,
      "learning_rate": 3.5395968437529954e-06,
      "loss": 0.8502,
      "step": 26732
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7067654132843018,
      "learning_rate": 3.538957927355206e-06,
      "loss": 0.7644,
      "step": 26733
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9017752408981323,
      "learning_rate": 3.538319010957417e-06,
      "loss": 0.8433,
      "step": 26734
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8402318954467773,
      "learning_rate": 3.5376800945596275e-06,
      "loss": 0.8801,
      "step": 26735
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.014922618865967,
      "learning_rate": 3.537041178161838e-06,
      "loss": 1.5464,
      "step": 26736
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9868186712265015,
      "learning_rate": 3.5364022617640485e-06,
      "loss": 1.041,
      "step": 26737
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4967246055603027,
      "learning_rate": 3.535763345366259e-06,
      "loss": 0.409,
      "step": 26738
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3236231803894043,
      "learning_rate": 3.53512442896847e-06,
      "loss": 0.841,
      "step": 26739
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.862622022628784,
      "learning_rate": 3.5344855125706806e-06,
      "loss": 0.8009,
      "step": 26740
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8171935081481934,
      "learning_rate": 3.5338465961728913e-06,
      "loss": 0.9176,
      "step": 26741
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.786497950553894,
      "learning_rate": 3.533207679775102e-06,
      "loss": 0.779,
      "step": 26742
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7859797477722168,
      "learning_rate": 3.5325687633773127e-06,
      "loss": 0.8511,
      "step": 26743
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1790215969085693,
      "learning_rate": 3.5319298469795234e-06,
      "loss": 0.438,
      "step": 26744
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.5325927734375,
      "learning_rate": 3.531290930581734e-06,
      "loss": 1.6125,
      "step": 26745
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.0350210666656494,
      "learning_rate": 3.5306520141839443e-06,
      "loss": 1.0175,
      "step": 26746
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.004004955291748,
      "learning_rate": 3.530013097786155e-06,
      "loss": 1.0412,
      "step": 26747
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.206187725067139,
      "learning_rate": 3.5293741813883657e-06,
      "loss": 0.8823,
      "step": 26748
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.104292392730713,
      "learning_rate": 3.5287352649905764e-06,
      "loss": 0.8373,
      "step": 26749
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.3356993198394775,
      "learning_rate": 3.528096348592787e-06,
      "loss": 1.0891,
      "step": 26750
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.296187400817871,
      "learning_rate": 3.527457432194998e-06,
      "loss": 1.1161,
      "step": 26751
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6636024713516235,
      "learning_rate": 3.5268185157972085e-06,
      "loss": 0.7721,
      "step": 26752
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8148165941238403,
      "learning_rate": 3.5261795993994192e-06,
      "loss": 0.9766,
      "step": 26753
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8053101301193237,
      "learning_rate": 3.52554068300163e-06,
      "loss": 0.9039,
      "step": 26754
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5378438234329224,
      "learning_rate": 3.5249017666038402e-06,
      "loss": 0.8453,
      "step": 26755
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.24302077293396,
      "learning_rate": 3.524262850206051e-06,
      "loss": 0.698,
      "step": 26756
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.797090768814087,
      "learning_rate": 3.5236239338082616e-06,
      "loss": 1.2043,
      "step": 26757
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.171184778213501,
      "learning_rate": 3.5229850174104723e-06,
      "loss": 1.1705,
      "step": 26758
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.125399112701416,
      "learning_rate": 3.522346101012683e-06,
      "loss": 0.6498,
      "step": 26759
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5876617431640625,
      "learning_rate": 3.5217071846148937e-06,
      "loss": 0.7922,
      "step": 26760
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.440354824066162,
      "learning_rate": 3.5210682682171044e-06,
      "loss": 0.7738,
      "step": 26761
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8837085962295532,
      "learning_rate": 3.520429351819315e-06,
      "loss": 0.7993,
      "step": 26762
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.8809502124786377,
      "learning_rate": 3.519790435421526e-06,
      "loss": 0.8445,
      "step": 26763
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7492854595184326,
      "learning_rate": 3.519151519023736e-06,
      "loss": 0.7659,
      "step": 26764
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3498153686523438,
      "learning_rate": 3.5185126026259464e-06,
      "loss": 0.9074,
      "step": 26765
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8063889741897583,
      "learning_rate": 3.517873686228157e-06,
      "loss": 0.9308,
      "step": 26766
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.887696623802185,
      "learning_rate": 3.5172347698303678e-06,
      "loss": 0.8649,
      "step": 26767
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.188239097595215,
      "learning_rate": 3.5165958534325785e-06,
      "loss": 1.0095,
      "step": 26768
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5024559497833252,
      "learning_rate": 3.515956937034789e-06,
      "loss": 0.754,
      "step": 26769
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.094743013381958,
      "learning_rate": 3.515318020637e-06,
      "loss": 1.1997,
      "step": 26770
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.82244873046875,
      "learning_rate": 3.51467910423921e-06,
      "loss": 0.6382,
      "step": 26771
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4780216217041016,
      "learning_rate": 3.514040187841421e-06,
      "loss": 0.5181,
      "step": 26772
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7649400234222412,
      "learning_rate": 3.5134012714436316e-06,
      "loss": 0.7452,
      "step": 26773
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.925326943397522,
      "learning_rate": 3.5127623550458423e-06,
      "loss": 0.8896,
      "step": 26774
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9662487506866455,
      "learning_rate": 3.512123438648053e-06,
      "loss": 0.9787,
      "step": 26775
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7144529819488525,
      "learning_rate": 3.5114845222502637e-06,
      "loss": 0.8162,
      "step": 26776
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5694257020950317,
      "learning_rate": 3.5108456058524744e-06,
      "loss": 0.7377,
      "step": 26777
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.030405044555664,
      "learning_rate": 3.510206689454685e-06,
      "loss": 1.1123,
      "step": 26778
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8305219411849976,
      "learning_rate": 3.5095677730568958e-06,
      "loss": 0.8597,
      "step": 26779
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5248792171478271,
      "learning_rate": 3.508928856659106e-06,
      "loss": 0.7019,
      "step": 26780
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.300159215927124,
      "learning_rate": 3.5082899402613167e-06,
      "loss": 1.0203,
      "step": 26781
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6439483165740967,
      "learning_rate": 3.5076510238635274e-06,
      "loss": 0.8151,
      "step": 26782
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.376733422279358,
      "learning_rate": 3.507012107465738e-06,
      "loss": 0.4605,
      "step": 26783
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.051782608032227,
      "learning_rate": 3.506373191067949e-06,
      "loss": 1.065,
      "step": 26784
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.553741455078125,
      "learning_rate": 3.5057342746701595e-06,
      "loss": 0.7502,
      "step": 26785
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.688358783721924,
      "learning_rate": 3.5050953582723702e-06,
      "loss": 1.1418,
      "step": 26786
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.936854124069214,
      "learning_rate": 3.504456441874581e-06,
      "loss": 0.9699,
      "step": 26787
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.308656692504883,
      "learning_rate": 3.5038175254767916e-06,
      "loss": 0.5766,
      "step": 26788
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5590265989303589,
      "learning_rate": 3.503178609079002e-06,
      "loss": 0.8021,
      "step": 26789
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7941384315490723,
      "learning_rate": 3.5025396926812126e-06,
      "loss": 0.6658,
      "step": 26790
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6823110580444336,
      "learning_rate": 3.5019007762834233e-06,
      "loss": 0.7738,
      "step": 26791
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.000281810760498,
      "learning_rate": 3.501261859885634e-06,
      "loss": 0.8279,
      "step": 26792
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.5257718563079834,
      "learning_rate": 3.5006229434878447e-06,
      "loss": 1.0961,
      "step": 26793
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9666104316711426,
      "learning_rate": 3.4999840270900554e-06,
      "loss": 0.8267,
      "step": 26794
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3489643335342407,
      "learning_rate": 3.499345110692266e-06,
      "loss": 0.3984,
      "step": 26795
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.089891195297241,
      "learning_rate": 3.498706194294477e-06,
      "loss": 0.8375,
      "step": 26796
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3778955936431885,
      "learning_rate": 3.4980672778966875e-06,
      "loss": 0.8812,
      "step": 26797
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6068085432052612,
      "learning_rate": 3.4974283614988978e-06,
      "loss": 0.7577,
      "step": 26798
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.635792851448059,
      "learning_rate": 3.4967894451011085e-06,
      "loss": 0.8314,
      "step": 26799
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2350857257843018,
      "learning_rate": 3.496150528703319e-06,
      "loss": 0.7794,
      "step": 26800
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8432010412216187,
      "learning_rate": 3.49551161230553e-06,
      "loss": 0.7368,
      "step": 26801
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5575774908065796,
      "learning_rate": 3.4948726959077406e-06,
      "loss": 0.8125,
      "step": 26802
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7764173746109009,
      "learning_rate": 3.4942337795099513e-06,
      "loss": 0.9219,
      "step": 26803
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.5992043018341064,
      "learning_rate": 3.493594863112162e-06,
      "loss": 0.9063,
      "step": 26804
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.061598777770996,
      "learning_rate": 3.4929559467143727e-06,
      "loss": 0.907,
      "step": 26805
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8122663497924805,
      "learning_rate": 3.4923170303165834e-06,
      "loss": 0.8748,
      "step": 26806
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5679184198379517,
      "learning_rate": 3.4916781139187937e-06,
      "loss": 0.7871,
      "step": 26807
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0011024475097656,
      "learning_rate": 3.4910391975210044e-06,
      "loss": 0.873,
      "step": 26808
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6569678783416748,
      "learning_rate": 3.490400281123215e-06,
      "loss": 0.9016,
      "step": 26809
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.736953854560852,
      "learning_rate": 3.4897613647254258e-06,
      "loss": 0.9844,
      "step": 26810
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.6727094650268555,
      "learning_rate": 3.4891224483276365e-06,
      "loss": 0.79,
      "step": 26811
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.002826452255249,
      "learning_rate": 3.488483531929847e-06,
      "loss": 0.9321,
      "step": 26812
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9560492038726807,
      "learning_rate": 3.487844615532058e-06,
      "loss": 0.6576,
      "step": 26813
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.8196611404418945,
      "learning_rate": 3.4872056991342686e-06,
      "loss": 1.1047,
      "step": 26814
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0006918907165527,
      "learning_rate": 3.4865667827364793e-06,
      "loss": 1.0892,
      "step": 26815
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9876971244812012,
      "learning_rate": 3.4859278663386895e-06,
      "loss": 0.8801,
      "step": 26816
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6601841449737549,
      "learning_rate": 3.4852889499409002e-06,
      "loss": 0.808,
      "step": 26817
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6896603107452393,
      "learning_rate": 3.484650033543111e-06,
      "loss": 0.6816,
      "step": 26818
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.959397554397583,
      "learning_rate": 3.4840111171453216e-06,
      "loss": 0.9004,
      "step": 26819
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7022491693496704,
      "learning_rate": 3.4833722007475323e-06,
      "loss": 0.8722,
      "step": 26820
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.313217282295227,
      "learning_rate": 3.482733284349743e-06,
      "loss": 0.3447,
      "step": 26821
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.619801998138428,
      "learning_rate": 3.4820943679519537e-06,
      "loss": 1.5005,
      "step": 26822
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.6996817588806152,
      "learning_rate": 3.4814554515541644e-06,
      "loss": 0.7325,
      "step": 26823
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.536362648010254,
      "learning_rate": 3.480816535156375e-06,
      "loss": 0.7412,
      "step": 26824
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9434661865234375,
      "learning_rate": 3.4801776187585854e-06,
      "loss": 0.9036,
      "step": 26825
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6838412284851074,
      "learning_rate": 3.479538702360796e-06,
      "loss": 0.7648,
      "step": 26826
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2679901123046875,
      "learning_rate": 3.478899785963007e-06,
      "loss": 0.4477,
      "step": 26827
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8426731824874878,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 0.6472,
      "step": 26828
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0949039459228516,
      "learning_rate": 3.477621953167428e-06,
      "loss": 0.5259,
      "step": 26829
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.003018856048584,
      "learning_rate": 3.476983036769639e-06,
      "loss": 0.8467,
      "step": 26830
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9718060493469238,
      "learning_rate": 3.4763441203718496e-06,
      "loss": 0.8454,
      "step": 26831
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.1747896671295166,
      "learning_rate": 3.4757052039740603e-06,
      "loss": 1.2219,
      "step": 26832
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9227299690246582,
      "learning_rate": 3.475066287576271e-06,
      "loss": 0.8495,
      "step": 26833
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.039116621017456,
      "learning_rate": 3.4744273711784817e-06,
      "loss": 0.5973,
      "step": 26834
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.758737087249756,
      "learning_rate": 3.473788454780692e-06,
      "loss": 1.0389,
      "step": 26835
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.129871129989624,
      "learning_rate": 3.4731495383829027e-06,
      "loss": 0.8883,
      "step": 26836
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8618597984313965,
      "learning_rate": 3.4725106219851134e-06,
      "loss": 0.7815,
      "step": 26837
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.7200517654418945,
      "learning_rate": 3.471871705587324e-06,
      "loss": 0.9298,
      "step": 26838
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7346504926681519,
      "learning_rate": 3.4712327891895348e-06,
      "loss": 0.8764,
      "step": 26839
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4469293355941772,
      "learning_rate": 3.4705938727917455e-06,
      "loss": 0.3988,
      "step": 26840
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.510047435760498,
      "learning_rate": 3.469954956393956e-06,
      "loss": 0.7102,
      "step": 26841
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.1757493019104,
      "learning_rate": 3.469316039996167e-06,
      "loss": 1.2618,
      "step": 26842
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.898977756500244,
      "learning_rate": 3.4686771235983776e-06,
      "loss": 1.0398,
      "step": 26843
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.236088752746582,
      "learning_rate": 3.468038207200588e-06,
      "loss": 0.6861,
      "step": 26844
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.817989468574524,
      "learning_rate": 3.4673992908027986e-06,
      "loss": 0.7234,
      "step": 26845
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.4713056087493896,
      "learning_rate": 3.4667603744050093e-06,
      "loss": 0.7843,
      "step": 26846
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.074157476425171,
      "learning_rate": 3.46612145800722e-06,
      "loss": 0.8502,
      "step": 26847
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.999022960662842,
      "learning_rate": 3.4654825416094307e-06,
      "loss": 1.473,
      "step": 26848
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8644126653671265,
      "learning_rate": 3.4648436252116414e-06,
      "loss": 0.9188,
      "step": 26849
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5661076307296753,
      "learning_rate": 3.464204708813852e-06,
      "loss": 0.3951,
      "step": 26850
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2631330490112305,
      "learning_rate": 3.4635657924160628e-06,
      "loss": 0.8618,
      "step": 26851
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.75780725479126,
      "learning_rate": 3.4629268760182735e-06,
      "loss": 1.265,
      "step": 26852
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.871437907218933,
      "learning_rate": 3.4622879596204837e-06,
      "loss": 0.9219,
      "step": 26853
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.819884181022644,
      "learning_rate": 3.4616490432226944e-06,
      "loss": 0.7273,
      "step": 26854
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1309375762939453,
      "learning_rate": 3.461010126824905e-06,
      "loss": 0.9454,
      "step": 26855
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9883228540420532,
      "learning_rate": 3.460371210427116e-06,
      "loss": 1.1659,
      "step": 26856
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.967061996459961,
      "learning_rate": 3.4597322940293265e-06,
      "loss": 0.9134,
      "step": 26857
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.022671699523926,
      "learning_rate": 3.4590933776315372e-06,
      "loss": 1.2363,
      "step": 26858
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.824916124343872,
      "learning_rate": 3.458454461233748e-06,
      "loss": 0.8097,
      "step": 26859
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3209002017974854,
      "learning_rate": 3.4578155448359586e-06,
      "loss": 1.0179,
      "step": 26860
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.848562479019165,
      "learning_rate": 3.4571766284381693e-06,
      "loss": 0.7559,
      "step": 26861
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7120559215545654,
      "learning_rate": 3.4565377120403796e-06,
      "loss": 0.8144,
      "step": 26862
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.4505414962768555,
      "learning_rate": 3.4558987956425903e-06,
      "loss": 1.356,
      "step": 26863
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2069473266601562,
      "learning_rate": 3.455259879244801e-06,
      "loss": 1.1332,
      "step": 26864
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.4908804893493652,
      "learning_rate": 3.4546209628470117e-06,
      "loss": 1.1222,
      "step": 26865
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.863398551940918,
      "learning_rate": 3.4539820464492224e-06,
      "loss": 0.8062,
      "step": 26866
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8468847274780273,
      "learning_rate": 3.453343130051433e-06,
      "loss": 0.8669,
      "step": 26867
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3096394538879395,
      "learning_rate": 3.452704213653644e-06,
      "loss": 0.7251,
      "step": 26868
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7696112394332886,
      "learning_rate": 3.4520652972558545e-06,
      "loss": 0.8319,
      "step": 26869
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.70601224899292,
      "learning_rate": 3.4514263808580652e-06,
      "loss": 0.7367,
      "step": 26870
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.451894760131836,
      "learning_rate": 3.4507874644602755e-06,
      "loss": 1.2702,
      "step": 26871
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.023148775100708,
      "learning_rate": 3.450148548062486e-06,
      "loss": 0.8616,
      "step": 26872
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5621354579925537,
      "learning_rate": 3.449509631664697e-06,
      "loss": 0.7675,
      "step": 26873
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7105557918548584,
      "learning_rate": 3.4488707152669076e-06,
      "loss": 0.9023,
      "step": 26874
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.4466004371643066,
      "learning_rate": 3.4482317988691183e-06,
      "loss": 0.8039,
      "step": 26875
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4578046798706055,
      "learning_rate": 3.447592882471329e-06,
      "loss": 0.8956,
      "step": 26876
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3535096645355225,
      "learning_rate": 3.4469539660735397e-06,
      "loss": 0.4227,
      "step": 26877
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.849050760269165,
      "learning_rate": 3.4463150496757504e-06,
      "loss": 0.8105,
      "step": 26878
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8861356973648071,
      "learning_rate": 3.445676133277961e-06,
      "loss": 0.8736,
      "step": 26879
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.394268751144409,
      "learning_rate": 3.4450372168801714e-06,
      "loss": 0.877,
      "step": 26880
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.318574905395508,
      "learning_rate": 3.444398300482382e-06,
      "loss": 1.0229,
      "step": 26881
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5879133939743042,
      "learning_rate": 3.4437593840845928e-06,
      "loss": 0.8954,
      "step": 26882
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.564009428024292,
      "learning_rate": 3.4431204676868035e-06,
      "loss": 1.0613,
      "step": 26883
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.41684627532959,
      "learning_rate": 3.442481551289014e-06,
      "loss": 0.7876,
      "step": 26884
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5423836708068848,
      "learning_rate": 3.441842634891225e-06,
      "loss": 0.8363,
      "step": 26885
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4503846168518066,
      "learning_rate": 3.4412037184934356e-06,
      "loss": 0.7173,
      "step": 26886
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.875774621963501,
      "learning_rate": 3.4405648020956463e-06,
      "loss": 0.8059,
      "step": 26887
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1149208545684814,
      "learning_rate": 3.439925885697857e-06,
      "loss": 0.8332,
      "step": 26888
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9662208557128906,
      "learning_rate": 3.4392869693000672e-06,
      "loss": 0.9002,
      "step": 26889
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8805938959121704,
      "learning_rate": 3.438648052902278e-06,
      "loss": 0.954,
      "step": 26890
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.2579023838043213,
      "learning_rate": 3.4380091365044886e-06,
      "loss": 0.8855,
      "step": 26891
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.014021158218384,
      "learning_rate": 3.4373702201066993e-06,
      "loss": 1.1449,
      "step": 26892
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7234182357788086,
      "learning_rate": 3.43673130370891e-06,
      "loss": 0.8558,
      "step": 26893
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.929524302482605,
      "learning_rate": 3.4360923873111207e-06,
      "loss": 0.821,
      "step": 26894
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.6351521015167236,
      "learning_rate": 3.4354534709133314e-06,
      "loss": 1.1265,
      "step": 26895
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.306903123855591,
      "learning_rate": 3.434814554515542e-06,
      "loss": 0.5119,
      "step": 26896
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.118539571762085,
      "learning_rate": 3.434175638117753e-06,
      "loss": 0.8678,
      "step": 26897
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.953918218612671,
      "learning_rate": 3.433536721719963e-06,
      "loss": 0.8126,
      "step": 26898
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.51594877243042,
      "learning_rate": 3.432897805322174e-06,
      "loss": 1.1257,
      "step": 26899
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1114134788513184,
      "learning_rate": 3.4322588889243845e-06,
      "loss": 1.0313,
      "step": 26900
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.898146152496338,
      "learning_rate": 3.4316199725265952e-06,
      "loss": 0.8905,
      "step": 26901
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3103456497192383,
      "learning_rate": 3.430981056128806e-06,
      "loss": 0.8238,
      "step": 26902
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.5573246479034424,
      "learning_rate": 3.4303421397310166e-06,
      "loss": 0.9423,
      "step": 26903
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0270235538482666,
      "learning_rate": 3.4297032233332273e-06,
      "loss": 0.9288,
      "step": 26904
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0442922115325928,
      "learning_rate": 3.429064306935438e-06,
      "loss": 0.8509,
      "step": 26905
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5699485540390015,
      "learning_rate": 3.4284253905376487e-06,
      "loss": 0.7456,
      "step": 26906
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7486224174499512,
      "learning_rate": 3.4277864741398594e-06,
      "loss": 0.8439,
      "step": 26907
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.640166997909546,
      "learning_rate": 3.4271475577420697e-06,
      "loss": 0.8231,
      "step": 26908
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6379361152648926,
      "learning_rate": 3.4265086413442804e-06,
      "loss": 0.8183,
      "step": 26909
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5300675630569458,
      "learning_rate": 3.425869724946491e-06,
      "loss": 0.766,
      "step": 26910
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.818951964378357,
      "learning_rate": 3.425230808548702e-06,
      "loss": 0.886,
      "step": 26911
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3951051235198975,
      "learning_rate": 3.4245918921509125e-06,
      "loss": 0.7456,
      "step": 26912
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.9219393730163574,
      "learning_rate": 3.423952975753123e-06,
      "loss": 1.0817,
      "step": 26913
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.2236223220825195,
      "learning_rate": 3.423314059355334e-06,
      "loss": 1.0708,
      "step": 26914
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.654500126838684,
      "learning_rate": 3.4226751429575446e-06,
      "loss": 0.9025,
      "step": 26915
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.2984113693237305,
      "learning_rate": 3.4220362265597553e-06,
      "loss": 0.9006,
      "step": 26916
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7913033962249756,
      "learning_rate": 3.4213973101619656e-06,
      "loss": 0.7149,
      "step": 26917
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7882417440414429,
      "learning_rate": 3.4207583937641763e-06,
      "loss": 0.8715,
      "step": 26918
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.111638069152832,
      "learning_rate": 3.420119477366387e-06,
      "loss": 0.8546,
      "step": 26919
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.413336992263794,
      "learning_rate": 3.4194805609685977e-06,
      "loss": 0.8353,
      "step": 26920
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.227081775665283,
      "learning_rate": 3.4188416445708084e-06,
      "loss": 0.6411,
      "step": 26921
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.519598960876465,
      "learning_rate": 3.418202728173019e-06,
      "loss": 0.8663,
      "step": 26922
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.5368032455444336,
      "learning_rate": 3.4175638117752298e-06,
      "loss": 1.1754,
      "step": 26923
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0791127681732178,
      "learning_rate": 3.4169248953774405e-06,
      "loss": 0.8431,
      "step": 26924
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1271188259124756,
      "learning_rate": 3.416285978979651e-06,
      "loss": 0.9285,
      "step": 26925
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.730922818183899,
      "learning_rate": 3.4156470625818614e-06,
      "loss": 0.8731,
      "step": 26926
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6675807237625122,
      "learning_rate": 3.415008146184072e-06,
      "loss": 0.7849,
      "step": 26927
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0550696849823,
      "learning_rate": 3.414369229786283e-06,
      "loss": 1.1751,
      "step": 26928
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7716186046600342,
      "learning_rate": 3.4137303133884935e-06,
      "loss": 0.7982,
      "step": 26929
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.6175174713134766,
      "learning_rate": 3.4130913969907042e-06,
      "loss": 0.8683,
      "step": 26930
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.00298810005188,
      "learning_rate": 3.412452480592915e-06,
      "loss": 0.91,
      "step": 26931
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.124748945236206,
      "learning_rate": 3.4118135641951256e-06,
      "loss": 0.7983,
      "step": 26932
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3429834842681885,
      "learning_rate": 3.4111746477973363e-06,
      "loss": 0.6879,
      "step": 26933
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8404043912887573,
      "learning_rate": 3.410535731399547e-06,
      "loss": 0.8144,
      "step": 26934
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.707619309425354,
      "learning_rate": 3.4098968150017573e-06,
      "loss": 0.8456,
      "step": 26935
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8802282810211182,
      "learning_rate": 3.409257898603968e-06,
      "loss": 0.8028,
      "step": 26936
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.4426093101501465,
      "learning_rate": 3.4086189822061787e-06,
      "loss": 1.4036,
      "step": 26937
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7791701555252075,
      "learning_rate": 3.4079800658083894e-06,
      "loss": 0.8089,
      "step": 26938
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0122766494750977,
      "learning_rate": 3.4073411494106e-06,
      "loss": 0.8154,
      "step": 26939
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.283251762390137,
      "learning_rate": 3.406702233012811e-06,
      "loss": 0.9357,
      "step": 26940
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1493372917175293,
      "learning_rate": 3.4060633166150215e-06,
      "loss": 0.9116,
      "step": 26941
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6771388053894043,
      "learning_rate": 3.4054244002172322e-06,
      "loss": 0.8412,
      "step": 26942
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.162342071533203,
      "learning_rate": 3.404785483819443e-06,
      "loss": 0.7508,
      "step": 26943
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8245763778686523,
      "learning_rate": 3.404146567421653e-06,
      "loss": 1.1638,
      "step": 26944
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.347262978553772,
      "learning_rate": 3.403507651023864e-06,
      "loss": 0.2917,
      "step": 26945
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.218038558959961,
      "learning_rate": 3.4028687346260746e-06,
      "loss": 0.4349,
      "step": 26946
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6840800046920776,
      "learning_rate": 3.4022298182282853e-06,
      "loss": 0.8509,
      "step": 26947
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.053569793701172,
      "learning_rate": 3.401590901830496e-06,
      "loss": 1.505,
      "step": 26948
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.3885345458984375,
      "learning_rate": 3.4009519854327067e-06,
      "loss": 0.6938,
      "step": 26949
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.0570478439331055,
      "learning_rate": 3.4003130690349174e-06,
      "loss": 1.1262,
      "step": 26950
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8630900382995605,
      "learning_rate": 3.399674152637128e-06,
      "loss": 0.9692,
      "step": 26951
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.918062686920166,
      "learning_rate": 3.399035236239339e-06,
      "loss": 0.7414,
      "step": 26952
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.8250555992126465,
      "learning_rate": 3.398396319841549e-06,
      "loss": 1.1061,
      "step": 26953
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0503790378570557,
      "learning_rate": 3.3977574034437598e-06,
      "loss": 0.696,
      "step": 26954
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9521360397338867,
      "learning_rate": 3.3971184870459705e-06,
      "loss": 0.6333,
      "step": 26955
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.693380117416382,
      "learning_rate": 3.396479570648181e-06,
      "loss": 1.0258,
      "step": 26956
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.664413928985596,
      "learning_rate": 3.395840654250392e-06,
      "loss": 0.8176,
      "step": 26957
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6322516202926636,
      "learning_rate": 3.3952017378526026e-06,
      "loss": 0.8143,
      "step": 26958
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.011337995529175,
      "learning_rate": 3.3945628214548124e-06,
      "loss": 1.1806,
      "step": 26959
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9220459461212158,
      "learning_rate": 3.393923905057023e-06,
      "loss": 0.883,
      "step": 26960
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2345223426818848,
      "learning_rate": 3.393284988659234e-06,
      "loss": 1.2704,
      "step": 26961
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8317110538482666,
      "learning_rate": 3.3926460722614445e-06,
      "loss": 0.9122,
      "step": 26962
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7140718698501587,
      "learning_rate": 3.3920071558636552e-06,
      "loss": 0.8484,
      "step": 26963
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.9353315830230713,
      "learning_rate": 3.391368239465866e-06,
      "loss": 0.7952,
      "step": 26964
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.208853721618652,
      "learning_rate": 3.3907293230680766e-06,
      "loss": 1.0814,
      "step": 26965
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7160388231277466,
      "learning_rate": 3.3900904066702873e-06,
      "loss": 0.822,
      "step": 26966
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2765204906463623,
      "learning_rate": 3.389451490272498e-06,
      "loss": 0.7189,
      "step": 26967
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8505913019180298,
      "learning_rate": 3.3888125738747087e-06,
      "loss": 0.8052,
      "step": 26968
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.0008299350738525,
      "learning_rate": 3.388173657476919e-06,
      "loss": 0.5515,
      "step": 26969
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.312983751296997,
      "learning_rate": 3.3875347410791297e-06,
      "loss": 1.1285,
      "step": 26970
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.783350944519043,
      "learning_rate": 3.3868958246813404e-06,
      "loss": 0.7974,
      "step": 26971
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8913897275924683,
      "learning_rate": 3.386256908283551e-06,
      "loss": 0.7443,
      "step": 26972
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7609810829162598,
      "learning_rate": 3.385617991885762e-06,
      "loss": 0.8395,
      "step": 26973
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5418683290481567,
      "learning_rate": 3.3849790754879725e-06,
      "loss": 0.8499,
      "step": 26974
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.966211199760437,
      "learning_rate": 3.384340159090183e-06,
      "loss": 1.2344,
      "step": 26975
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.1705827713012695,
      "learning_rate": 3.383701242692394e-06,
      "loss": 1.1168,
      "step": 26976
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5574543476104736,
      "learning_rate": 3.3830623262946046e-06,
      "loss": 1.0233,
      "step": 26977
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6406185626983643,
      "learning_rate": 3.382423409896815e-06,
      "loss": 0.687,
      "step": 26978
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.761988878250122,
      "learning_rate": 3.3817844934990256e-06,
      "loss": 0.856,
      "step": 26979
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9799515008926392,
      "learning_rate": 3.3811455771012363e-06,
      "loss": 0.9155,
      "step": 26980
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.538885116577148,
      "learning_rate": 3.380506660703447e-06,
      "loss": 0.9735,
      "step": 26981
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4787954092025757,
      "learning_rate": 3.3798677443056577e-06,
      "loss": 0.7319,
      "step": 26982
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4126179218292236,
      "learning_rate": 3.3792288279078684e-06,
      "loss": 0.7234,
      "step": 26983
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5580132007598877,
      "learning_rate": 3.378589911510079e-06,
      "loss": 1.1099,
      "step": 26984
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.295121908187866,
      "learning_rate": 3.3779509951122898e-06,
      "loss": 0.8055,
      "step": 26985
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.925966739654541,
      "learning_rate": 3.3773120787145005e-06,
      "loss": 1.3817,
      "step": 26986
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.5139684677124023,
      "learning_rate": 3.3766731623167108e-06,
      "loss": 0.6555,
      "step": 26987
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0421030521392822,
      "learning_rate": 3.3760342459189215e-06,
      "loss": 0.8696,
      "step": 26988
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8782540559768677,
      "learning_rate": 3.375395329521132e-06,
      "loss": 0.9245,
      "step": 26989
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.597158908843994,
      "learning_rate": 3.374756413123343e-06,
      "loss": 0.9289,
      "step": 26990
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8875712156295776,
      "learning_rate": 3.3741174967255536e-06,
      "loss": 0.9598,
      "step": 26991
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.756528615951538,
      "learning_rate": 3.3734785803277643e-06,
      "loss": 0.984,
      "step": 26992
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8145349025726318,
      "learning_rate": 3.372839663929975e-06,
      "loss": 0.7866,
      "step": 26993
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.575270891189575,
      "learning_rate": 3.3722007475321857e-06,
      "loss": 0.9104,
      "step": 26994
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.060068130493164,
      "learning_rate": 3.3715618311343964e-06,
      "loss": 0.7644,
      "step": 26995
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.388006210327148,
      "learning_rate": 3.3709229147366066e-06,
      "loss": 0.9823,
      "step": 26996
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4902921915054321,
      "learning_rate": 3.3702839983388173e-06,
      "loss": 0.7194,
      "step": 26997
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.0198631286621094,
      "learning_rate": 3.369645081941028e-06,
      "loss": 0.8158,
      "step": 26998
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7205294370651245,
      "learning_rate": 3.3690061655432387e-06,
      "loss": 0.84,
      "step": 26999
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.632457256317139,
      "learning_rate": 3.3683672491454494e-06,
      "loss": 1.0996,
      "step": 27000
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5506218671798706,
      "learning_rate": 3.36772833274766e-06,
      "loss": 0.8104,
      "step": 27001
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6318798065185547,
      "learning_rate": 3.367089416349871e-06,
      "loss": 0.3814,
      "step": 27002
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.28926944732666,
      "learning_rate": 3.3664504999520815e-06,
      "loss": 0.6852,
      "step": 27003
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4627844095230103,
      "learning_rate": 3.3658115835542922e-06,
      "loss": 0.9135,
      "step": 27004
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.004695177078247,
      "learning_rate": 3.3651726671565025e-06,
      "loss": 0.8579,
      "step": 27005
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.356865406036377,
      "learning_rate": 3.364533750758713e-06,
      "loss": 0.7423,
      "step": 27006
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.1635448932647705,
      "learning_rate": 3.363894834360924e-06,
      "loss": 1.0344,
      "step": 27007
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8260059356689453,
      "learning_rate": 3.3632559179631346e-06,
      "loss": 0.8355,
      "step": 27008
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.73329758644104,
      "learning_rate": 3.3626170015653453e-06,
      "loss": 0.7558,
      "step": 27009
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6688017845153809,
      "learning_rate": 3.361978085167556e-06,
      "loss": 0.9458,
      "step": 27010
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1098694801330566,
      "learning_rate": 3.3613391687697667e-06,
      "loss": 0.8056,
      "step": 27011
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.491482138633728,
      "learning_rate": 3.3607002523719774e-06,
      "loss": 0.411,
      "step": 27012
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.707846760749817,
      "learning_rate": 3.360061335974188e-06,
      "loss": 0.876,
      "step": 27013
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.3647449016571045,
      "learning_rate": 3.3594224195763984e-06,
      "loss": 0.5171,
      "step": 27014
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5898979902267456,
      "learning_rate": 3.358783503178609e-06,
      "loss": 0.8174,
      "step": 27015
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0581233501434326,
      "learning_rate": 3.3581445867808198e-06,
      "loss": 0.8541,
      "step": 27016
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6730653047561646,
      "learning_rate": 3.3575056703830305e-06,
      "loss": 0.756,
      "step": 27017
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9966142177581787,
      "learning_rate": 3.356866753985241e-06,
      "loss": 0.9535,
      "step": 27018
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.951900601387024,
      "learning_rate": 3.356227837587452e-06,
      "loss": 0.8432,
      "step": 27019
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5353854894638062,
      "learning_rate": 3.3555889211896626e-06,
      "loss": 0.8685,
      "step": 27020
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1024882793426514,
      "learning_rate": 3.3549500047918733e-06,
      "loss": 0.847,
      "step": 27021
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6460059881210327,
      "learning_rate": 3.354311088394084e-06,
      "loss": 0.8225,
      "step": 27022
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9829432964324951,
      "learning_rate": 3.3536721719962943e-06,
      "loss": 0.7964,
      "step": 27023
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5909397602081299,
      "learning_rate": 3.353033255598505e-06,
      "loss": 0.842,
      "step": 27024
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.196663856506348,
      "learning_rate": 3.3523943392007157e-06,
      "loss": 1.1399,
      "step": 27025
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8811231851577759,
      "learning_rate": 3.3517554228029264e-06,
      "loss": 1.1133,
      "step": 27026
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6113511323928833,
      "learning_rate": 3.351116506405137e-06,
      "loss": 0.9095,
      "step": 27027
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0161492824554443,
      "learning_rate": 3.3504775900073478e-06,
      "loss": 0.965,
      "step": 27028
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5030461549758911,
      "learning_rate": 3.3498386736095585e-06,
      "loss": 0.8071,
      "step": 27029
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1813621520996094,
      "learning_rate": 3.349199757211769e-06,
      "loss": 0.6914,
      "step": 27030
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9780652523040771,
      "learning_rate": 3.34856084081398e-06,
      "loss": 0.9084,
      "step": 27031
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.425901889801025,
      "learning_rate": 3.3479219244161906e-06,
      "loss": 1.7065,
      "step": 27032
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7418651580810547,
      "learning_rate": 3.347283008018401e-06,
      "loss": 0.8413,
      "step": 27033
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.98635995388031,
      "learning_rate": 3.3466440916206115e-06,
      "loss": 1.1967,
      "step": 27034
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.200636863708496,
      "learning_rate": 3.3460051752228222e-06,
      "loss": 0.8246,
      "step": 27035
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.402988910675049,
      "learning_rate": 3.345366258825033e-06,
      "loss": 0.7108,
      "step": 27036
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.138209819793701,
      "learning_rate": 3.3447273424272436e-06,
      "loss": 0.7728,
      "step": 27037
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.791511058807373,
      "learning_rate": 3.3440884260294543e-06,
      "loss": 0.871,
      "step": 27038
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5948325395584106,
      "learning_rate": 3.343449509631665e-06,
      "loss": 0.836,
      "step": 27039
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.141174793243408,
      "learning_rate": 3.3428105932338757e-06,
      "loss": 1.0032,
      "step": 27040
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.461313486099243,
      "learning_rate": 3.3421716768360864e-06,
      "loss": 0.8522,
      "step": 27041
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.9621286392211914,
      "learning_rate": 3.3415327604382967e-06,
      "loss": 0.7869,
      "step": 27042
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8205032348632812,
      "learning_rate": 3.3408938440405074e-06,
      "loss": 0.8082,
      "step": 27043
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7842036485671997,
      "learning_rate": 3.340254927642718e-06,
      "loss": 0.8567,
      "step": 27044
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.771467924118042,
      "learning_rate": 3.339616011244929e-06,
      "loss": 0.8869,
      "step": 27045
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.710078239440918,
      "learning_rate": 3.3389770948471395e-06,
      "loss": 0.7841,
      "step": 27046
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0363287925720215,
      "learning_rate": 3.3383381784493502e-06,
      "loss": 0.8366,
      "step": 27047
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.566574811935425,
      "learning_rate": 3.337699262051561e-06,
      "loss": 1.0828,
      "step": 27048
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.588130235671997,
      "learning_rate": 3.3370603456537716e-06,
      "loss": 0.7411,
      "step": 27049
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.3570446968078613,
      "learning_rate": 3.3364214292559823e-06,
      "loss": 0.8404,
      "step": 27050
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9239747524261475,
      "learning_rate": 3.3357825128581926e-06,
      "loss": 0.7744,
      "step": 27051
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.074990749359131,
      "learning_rate": 3.3351435964604033e-06,
      "loss": 0.768,
      "step": 27052
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.605615496635437,
      "learning_rate": 3.334504680062614e-06,
      "loss": 0.8568,
      "step": 27053
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.135011911392212,
      "learning_rate": 3.3338657636648247e-06,
      "loss": 0.931,
      "step": 27054
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5123376846313477,
      "learning_rate": 3.3332268472670354e-06,
      "loss": 1.1544,
      "step": 27055
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.472813367843628,
      "learning_rate": 3.332587930869246e-06,
      "loss": 0.3989,
      "step": 27056
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6859647035598755,
      "learning_rate": 3.331949014471457e-06,
      "loss": 0.8616,
      "step": 27057
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.72627592086792,
      "learning_rate": 3.3313100980736675e-06,
      "loss": 1.4309,
      "step": 27058
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7094959020614624,
      "learning_rate": 3.330671181675878e-06,
      "loss": 0.833,
      "step": 27059
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.084569215774536,
      "learning_rate": 3.3300322652780885e-06,
      "loss": 0.7743,
      "step": 27060
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5570292472839355,
      "learning_rate": 3.329393348880299e-06,
      "loss": 0.5135,
      "step": 27061
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2073404788970947,
      "learning_rate": 3.32875443248251e-06,
      "loss": 0.8721,
      "step": 27062
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.010054588317871,
      "learning_rate": 3.3281155160847206e-06,
      "loss": 1.3203,
      "step": 27063
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2375952005386353,
      "learning_rate": 3.3274765996869313e-06,
      "loss": 0.3333,
      "step": 27064
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4978668689727783,
      "learning_rate": 3.326837683289142e-06,
      "loss": 0.7705,
      "step": 27065
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6471449136734009,
      "learning_rate": 3.3261987668913527e-06,
      "loss": 0.7895,
      "step": 27066
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.5204412937164307,
      "learning_rate": 3.3255598504935634e-06,
      "loss": 1.0608,
      "step": 27067
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5634381771087646,
      "learning_rate": 3.324920934095774e-06,
      "loss": 0.7658,
      "step": 27068
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.41560697555542,
      "learning_rate": 3.3242820176979843e-06,
      "loss": 1.0969,
      "step": 27069
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.6872479915618896,
      "learning_rate": 3.323643101300195e-06,
      "loss": 0.6595,
      "step": 27070
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1893608570098877,
      "learning_rate": 3.3230041849024057e-06,
      "loss": 0.8775,
      "step": 27071
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7542725801467896,
      "learning_rate": 3.3223652685046164e-06,
      "loss": 0.9171,
      "step": 27072
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.96518874168396,
      "learning_rate": 3.321726352106827e-06,
      "loss": 0.8621,
      "step": 27073
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6145517826080322,
      "learning_rate": 3.321087435709038e-06,
      "loss": 0.9324,
      "step": 27074
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0116288661956787,
      "learning_rate": 3.3204485193112485e-06,
      "loss": 0.7887,
      "step": 27075
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5279157161712646,
      "learning_rate": 3.3198096029134592e-06,
      "loss": 0.7366,
      "step": 27076
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8559938669204712,
      "learning_rate": 3.31917068651567e-06,
      "loss": 0.8122,
      "step": 27077
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0282325744628906,
      "learning_rate": 3.3185317701178802e-06,
      "loss": 0.7876,
      "step": 27078
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6759166717529297,
      "learning_rate": 3.317892853720091e-06,
      "loss": 0.8102,
      "step": 27079
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.139848232269287,
      "learning_rate": 3.3172539373223016e-06,
      "loss": 1.3158,
      "step": 27080
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.0587992668151855,
      "learning_rate": 3.3166150209245123e-06,
      "loss": 0.7543,
      "step": 27081
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.1983838081359863,
      "learning_rate": 3.315976104526723e-06,
      "loss": 0.8071,
      "step": 27082
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5881454944610596,
      "learning_rate": 3.3153371881289337e-06,
      "loss": 1.077,
      "step": 27083
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7117637395858765,
      "learning_rate": 3.3146982717311444e-06,
      "loss": 0.8404,
      "step": 27084
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9827922582626343,
      "learning_rate": 3.314059355333355e-06,
      "loss": 0.8314,
      "step": 27085
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7198964357376099,
      "learning_rate": 3.313420438935566e-06,
      "loss": 0.9078,
      "step": 27086
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.565657377243042,
      "learning_rate": 3.312781522537776e-06,
      "loss": 0.8577,
      "step": 27087
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.4588816165924072,
      "learning_rate": 3.312142606139987e-06,
      "loss": 0.837,
      "step": 27088
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.801780343055725,
      "learning_rate": 3.3115036897421975e-06,
      "loss": 0.7799,
      "step": 27089
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0127546787261963,
      "learning_rate": 3.310864773344408e-06,
      "loss": 0.8577,
      "step": 27090
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.469893455505371,
      "learning_rate": 3.310225856946619e-06,
      "loss": 0.7541,
      "step": 27091
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7914320230484009,
      "learning_rate": 3.3095869405488296e-06,
      "loss": 0.8756,
      "step": 27092
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.399627923965454,
      "learning_rate": 3.3089480241510403e-06,
      "loss": 0.3604,
      "step": 27093
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.382338523864746,
      "learning_rate": 3.308309107753251e-06,
      "loss": 0.8358,
      "step": 27094
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5167139768600464,
      "learning_rate": 3.3076701913554617e-06,
      "loss": 0.3635,
      "step": 27095
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.3014883995056152,
      "learning_rate": 3.307031274957672e-06,
      "loss": 1.0103,
      "step": 27096
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5500330924987793,
      "learning_rate": 3.3063923585598827e-06,
      "loss": 0.8299,
      "step": 27097
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.346916198730469,
      "learning_rate": 3.3057534421620934e-06,
      "loss": 1.1223,
      "step": 27098
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.176072597503662,
      "learning_rate": 3.305114525764304e-06,
      "loss": 0.843,
      "step": 27099
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.803809642791748,
      "learning_rate": 3.3044756093665148e-06,
      "loss": 0.7491,
      "step": 27100
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.499040126800537,
      "learning_rate": 3.3038366929687255e-06,
      "loss": 0.8155,
      "step": 27101
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8102749586105347,
      "learning_rate": 3.303197776570936e-06,
      "loss": 0.8241,
      "step": 27102
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6604340076446533,
      "learning_rate": 3.302558860173147e-06,
      "loss": 0.8712,
      "step": 27103
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4530209302902222,
      "learning_rate": 3.3019199437753576e-06,
      "loss": 0.7611,
      "step": 27104
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8880102634429932,
      "learning_rate": 3.3012810273775683e-06,
      "loss": 0.7699,
      "step": 27105
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.719433307647705,
      "learning_rate": 3.3006421109797785e-06,
      "loss": 0.8172,
      "step": 27106
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5464427471160889,
      "learning_rate": 3.3000031945819892e-06,
      "loss": 0.7185,
      "step": 27107
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.781819462776184,
      "learning_rate": 3.2993642781842e-06,
      "loss": 0.8351,
      "step": 27108
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.633484959602356,
      "learning_rate": 3.2987253617864106e-06,
      "loss": 0.7514,
      "step": 27109
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.904834270477295,
      "learning_rate": 3.2980864453886213e-06,
      "loss": 1.6242,
      "step": 27110
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0020766258239746,
      "learning_rate": 3.297447528990832e-06,
      "loss": 1.145,
      "step": 27111
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.060312032699585,
      "learning_rate": 3.2968086125930427e-06,
      "loss": 0.9055,
      "step": 27112
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6254323720932007,
      "learning_rate": 3.2961696961952534e-06,
      "loss": 0.7792,
      "step": 27113
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.45245885848999,
      "learning_rate": 3.295530779797464e-06,
      "loss": 1.3202,
      "step": 27114
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1098296642303467,
      "learning_rate": 3.2948918633996744e-06,
      "loss": 0.9178,
      "step": 27115
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.094505786895752,
      "learning_rate": 3.294252947001885e-06,
      "loss": 1.2815,
      "step": 27116
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4736889600753784,
      "learning_rate": 3.293614030604096e-06,
      "loss": 0.7779,
      "step": 27117
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4263360500335693,
      "learning_rate": 3.2929751142063065e-06,
      "loss": 0.7309,
      "step": 27118
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4385935068130493,
      "learning_rate": 3.2923361978085172e-06,
      "loss": 0.8405,
      "step": 27119
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.579611897468567,
      "learning_rate": 3.291697281410728e-06,
      "loss": 0.6558,
      "step": 27120
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.231626510620117,
      "learning_rate": 3.2910583650129386e-06,
      "loss": 1.1508,
      "step": 27121
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8002665042877197,
      "learning_rate": 3.2904194486151493e-06,
      "loss": 0.8335,
      "step": 27122
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5414481163024902,
      "learning_rate": 3.28978053221736e-06,
      "loss": 0.8854,
      "step": 27123
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7009400129318237,
      "learning_rate": 3.2891416158195703e-06,
      "loss": 0.7903,
      "step": 27124
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6028896570205688,
      "learning_rate": 3.288502699421781e-06,
      "loss": 0.779,
      "step": 27125
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0102462768554688,
      "learning_rate": 3.2878637830239917e-06,
      "loss": 0.7532,
      "step": 27126
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.495532035827637,
      "learning_rate": 3.2872248666262024e-06,
      "loss": 1.2069,
      "step": 27127
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4894462823867798,
      "learning_rate": 3.286585950228413e-06,
      "loss": 0.8509,
      "step": 27128
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6877832412719727,
      "learning_rate": 3.285947033830624e-06,
      "loss": 0.9026,
      "step": 27129
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.1417899131774902,
      "learning_rate": 3.2853081174328345e-06,
      "loss": 0.8285,
      "step": 27130
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.05490779876709,
      "learning_rate": 3.284669201035045e-06,
      "loss": 1.2934,
      "step": 27131
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.5635433197021484,
      "learning_rate": 3.284030284637256e-06,
      "loss": 0.7457,
      "step": 27132
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4755051136016846,
      "learning_rate": 3.283391368239466e-06,
      "loss": 0.8207,
      "step": 27133
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.4489567279815674,
      "learning_rate": 3.282752451841677e-06,
      "loss": 1.1162,
      "step": 27134
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7611210346221924,
      "learning_rate": 3.2821135354438876e-06,
      "loss": 0.8247,
      "step": 27135
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7756993770599365,
      "learning_rate": 3.2814746190460983e-06,
      "loss": 0.7437,
      "step": 27136
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.994043231010437,
      "learning_rate": 3.280835702648309e-06,
      "loss": 0.8696,
      "step": 27137
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.017632246017456,
      "learning_rate": 3.2801967862505197e-06,
      "loss": 0.8355,
      "step": 27138
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.507085800170898,
      "learning_rate": 3.2795578698527304e-06,
      "loss": 1.2033,
      "step": 27139
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.558309555053711,
      "learning_rate": 3.278918953454941e-06,
      "loss": 1.2937,
      "step": 27140
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.208183765411377,
      "learning_rate": 3.2782800370571518e-06,
      "loss": 0.6708,
      "step": 27141
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.3320462703704834,
      "learning_rate": 3.277641120659362e-06,
      "loss": 0.9155,
      "step": 27142
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1166281700134277,
      "learning_rate": 3.2770022042615728e-06,
      "loss": 0.9131,
      "step": 27143
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.01971173286438,
      "learning_rate": 3.2763632878637835e-06,
      "loss": 1.1872,
      "step": 27144
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9827032089233398,
      "learning_rate": 3.275724371465994e-06,
      "loss": 0.8049,
      "step": 27145
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.341243743896484,
      "learning_rate": 3.275085455068205e-06,
      "loss": 1.2031,
      "step": 27146
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6832637786865234,
      "learning_rate": 3.2744465386704156e-06,
      "loss": 0.8174,
      "step": 27147
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.816199779510498,
      "learning_rate": 3.2738076222726263e-06,
      "loss": 0.8368,
      "step": 27148
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0587189197540283,
      "learning_rate": 3.273168705874837e-06,
      "loss": 0.8224,
      "step": 27149
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8224812746047974,
      "learning_rate": 3.2725297894770477e-06,
      "loss": 0.8959,
      "step": 27150
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.851752758026123,
      "learning_rate": 3.271890873079258e-06,
      "loss": 1.3417,
      "step": 27151
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.975766897201538,
      "learning_rate": 3.271251956681468e-06,
      "loss": 0.8301,
      "step": 27152
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.965030550956726,
      "learning_rate": 3.270613040283679e-06,
      "loss": 0.4966,
      "step": 27153
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1070170402526855,
      "learning_rate": 3.2699741238858896e-06,
      "loss": 0.8903,
      "step": 27154
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4497445821762085,
      "learning_rate": 3.2693352074881003e-06,
      "loss": 0.3801,
      "step": 27155
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6835912466049194,
      "learning_rate": 3.268696291090311e-06,
      "loss": 0.7835,
      "step": 27156
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5028820037841797,
      "learning_rate": 3.2680573746925213e-06,
      "loss": 1.1162,
      "step": 27157
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5489683151245117,
      "learning_rate": 3.267418458294732e-06,
      "loss": 0.7246,
      "step": 27158
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.408440113067627,
      "learning_rate": 3.2667795418969427e-06,
      "loss": 0.6569,
      "step": 27159
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.3643198013305664,
      "learning_rate": 3.2661406254991534e-06,
      "loss": 1.1648,
      "step": 27160
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.7792160511016846,
      "learning_rate": 3.265501709101364e-06,
      "loss": 0.5976,
      "step": 27161
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4842332601547241,
      "learning_rate": 3.2648627927035748e-06,
      "loss": 0.846,
      "step": 27162
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.571974515914917,
      "learning_rate": 3.2642238763057855e-06,
      "loss": 0.8728,
      "step": 27163
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7885513305664062,
      "learning_rate": 3.263584959907996e-06,
      "loss": 0.7965,
      "step": 27164
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5162849426269531,
      "learning_rate": 3.262946043510207e-06,
      "loss": 0.3317,
      "step": 27165
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.3046860694885254,
      "learning_rate": 3.2623071271124176e-06,
      "loss": 0.7996,
      "step": 27166
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0611071586608887,
      "learning_rate": 3.261668210714628e-06,
      "loss": 0.8993,
      "step": 27167
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1629276275634766,
      "learning_rate": 3.2610292943168386e-06,
      "loss": 0.9634,
      "step": 27168
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7677501440048218,
      "learning_rate": 3.2603903779190493e-06,
      "loss": 0.8402,
      "step": 27169
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1708788871765137,
      "learning_rate": 3.25975146152126e-06,
      "loss": 0.8796,
      "step": 27170
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7680920362472534,
      "learning_rate": 3.2591125451234707e-06,
      "loss": 0.7728,
      "step": 27171
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.1413819789886475,
      "learning_rate": 3.2584736287256814e-06,
      "loss": 0.8078,
      "step": 27172
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.008113384246826,
      "learning_rate": 3.257834712327892e-06,
      "loss": 1.2236,
      "step": 27173
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6184569597244263,
      "learning_rate": 3.2571957959301028e-06,
      "loss": 0.8179,
      "step": 27174
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9749243259429932,
      "learning_rate": 3.2565568795323135e-06,
      "loss": 0.8251,
      "step": 27175
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2553870677948,
      "learning_rate": 3.2559179631345237e-06,
      "loss": 0.846,
      "step": 27176
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9344161748886108,
      "learning_rate": 3.2552790467367344e-06,
      "loss": 0.8685,
      "step": 27177
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7222609519958496,
      "learning_rate": 3.254640130338945e-06,
      "loss": 0.901,
      "step": 27178
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.1736695766448975,
      "learning_rate": 3.254001213941156e-06,
      "loss": 0.8206,
      "step": 27179
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.621605634689331,
      "learning_rate": 3.2533622975433665e-06,
      "loss": 0.8702,
      "step": 27180
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5991146564483643,
      "learning_rate": 3.2527233811455772e-06,
      "loss": 0.8314,
      "step": 27181
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8964364528656006,
      "learning_rate": 3.252084464747788e-06,
      "loss": 0.7532,
      "step": 27182
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.611470341682434,
      "learning_rate": 3.2514455483499986e-06,
      "loss": 0.6952,
      "step": 27183
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4908546209335327,
      "learning_rate": 3.2508066319522093e-06,
      "loss": 0.7767,
      "step": 27184
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4182939529418945,
      "learning_rate": 3.2501677155544196e-06,
      "loss": 1.0288,
      "step": 27185
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8268285989761353,
      "learning_rate": 3.2495287991566303e-06,
      "loss": 0.8849,
      "step": 27186
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3734612464904785,
      "learning_rate": 3.248889882758841e-06,
      "loss": 0.398,
      "step": 27187
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0967650413513184,
      "learning_rate": 3.2482509663610517e-06,
      "loss": 0.9605,
      "step": 27188
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.350495338439941,
      "learning_rate": 3.2476120499632624e-06,
      "loss": 1.5712,
      "step": 27189
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0381968021392822,
      "learning_rate": 3.246973133565473e-06,
      "loss": 0.9043,
      "step": 27190
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.37007212638855,
      "learning_rate": 3.246334217167684e-06,
      "loss": 0.9044,
      "step": 27191
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8553526401519775,
      "learning_rate": 3.2456953007698945e-06,
      "loss": 0.9018,
      "step": 27192
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8941997289657593,
      "learning_rate": 3.245056384372105e-06,
      "loss": 0.7118,
      "step": 27193
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7263532876968384,
      "learning_rate": 3.2444174679743155e-06,
      "loss": 0.8,
      "step": 27194
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6987223625183105,
      "learning_rate": 3.243778551576526e-06,
      "loss": 0.8595,
      "step": 27195
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.481553077697754,
      "learning_rate": 3.243139635178737e-06,
      "loss": 0.3428,
      "step": 27196
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4978636503219604,
      "learning_rate": 3.2425007187809476e-06,
      "loss": 0.3272,
      "step": 27197
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4682369232177734,
      "learning_rate": 3.2418618023831583e-06,
      "loss": 0.8936,
      "step": 27198
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.93343448638916,
      "learning_rate": 3.241222885985369e-06,
      "loss": 1.0667,
      "step": 27199
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6454219818115234,
      "learning_rate": 3.2405839695875797e-06,
      "loss": 0.8645,
      "step": 27200
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.598433256149292,
      "learning_rate": 3.2399450531897904e-06,
      "loss": 1.0682,
      "step": 27201
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0691871643066406,
      "learning_rate": 3.239306136792001e-06,
      "loss": 0.8652,
      "step": 27202
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0781304836273193,
      "learning_rate": 3.2386672203942114e-06,
      "loss": 0.8282,
      "step": 27203
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4696872234344482,
      "learning_rate": 3.238028303996422e-06,
      "loss": 0.959,
      "step": 27204
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2037672996520996,
      "learning_rate": 3.2373893875986328e-06,
      "loss": 1.0628,
      "step": 27205
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.804245948791504,
      "learning_rate": 3.2367504712008435e-06,
      "loss": 0.7894,
      "step": 27206
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9954110383987427,
      "learning_rate": 3.236111554803054e-06,
      "loss": 0.9162,
      "step": 27207
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.177902936935425,
      "learning_rate": 3.235472638405265e-06,
      "loss": 1.101,
      "step": 27208
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0595381259918213,
      "learning_rate": 3.2348337220074756e-06,
      "loss": 0.817,
      "step": 27209
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8036965131759644,
      "learning_rate": 3.2341948056096863e-06,
      "loss": 0.8139,
      "step": 27210
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8145463466644287,
      "learning_rate": 3.233555889211897e-06,
      "loss": 0.9005,
      "step": 27211
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.697112560272217,
      "learning_rate": 3.2329169728141072e-06,
      "loss": 0.8234,
      "step": 27212
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7812237739562988,
      "learning_rate": 3.232278056416318e-06,
      "loss": 0.8568,
      "step": 27213
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.071365356445312,
      "learning_rate": 3.2316391400185286e-06,
      "loss": 1.5479,
      "step": 27214
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2857933044433594,
      "learning_rate": 3.2310002236207393e-06,
      "loss": 0.9809,
      "step": 27215
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0960354804992676,
      "learning_rate": 3.23036130722295e-06,
      "loss": 0.6501,
      "step": 27216
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.885498046875,
      "learning_rate": 3.2297223908251607e-06,
      "loss": 0.8394,
      "step": 27217
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.152254343032837,
      "learning_rate": 3.2290834744273714e-06,
      "loss": 1.0534,
      "step": 27218
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.763474464416504,
      "learning_rate": 3.228444558029582e-06,
      "loss": 0.8581,
      "step": 27219
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9663867950439453,
      "learning_rate": 3.227805641631793e-06,
      "loss": 0.8857,
      "step": 27220
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6546863317489624,
      "learning_rate": 3.227166725234003e-06,
      "loss": 0.8322,
      "step": 27221
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6288049221038818,
      "learning_rate": 3.226527808836214e-06,
      "loss": 0.8262,
      "step": 27222
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.398818016052246,
      "learning_rate": 3.2258888924384245e-06,
      "loss": 0.357,
      "step": 27223
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.185149669647217,
      "learning_rate": 3.2252499760406352e-06,
      "loss": 1.1656,
      "step": 27224
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.064115285873413,
      "learning_rate": 3.224611059642846e-06,
      "loss": 0.8748,
      "step": 27225
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.5355560779571533,
      "learning_rate": 3.2239721432450566e-06,
      "loss": 0.9098,
      "step": 27226
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.229545831680298,
      "learning_rate": 3.2233332268472673e-06,
      "loss": 0.9497,
      "step": 27227
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0310328006744385,
      "learning_rate": 3.222694310449478e-06,
      "loss": 0.8436,
      "step": 27228
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4395703077316284,
      "learning_rate": 3.2220553940516887e-06,
      "loss": 0.77,
      "step": 27229
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6392662525177002,
      "learning_rate": 3.221416477653899e-06,
      "loss": 0.8535,
      "step": 27230
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7119667530059814,
      "learning_rate": 3.2207775612561097e-06,
      "loss": 0.8301,
      "step": 27231
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1713321208953857,
      "learning_rate": 3.2201386448583204e-06,
      "loss": 0.9542,
      "step": 27232
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6265450716018677,
      "learning_rate": 3.219499728460531e-06,
      "loss": 0.7261,
      "step": 27233
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.3662872314453125,
      "learning_rate": 3.2188608120627418e-06,
      "loss": 1.7584,
      "step": 27234
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.253147840499878,
      "learning_rate": 3.2182218956649525e-06,
      "loss": 0.9152,
      "step": 27235
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3277597427368164,
      "learning_rate": 3.217582979267163e-06,
      "loss": 0.8034,
      "step": 27236
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.2285003662109375,
      "learning_rate": 3.216944062869374e-06,
      "loss": 0.6034,
      "step": 27237
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5939775705337524,
      "learning_rate": 3.2163051464715846e-06,
      "loss": 0.7884,
      "step": 27238
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0146842002868652,
      "learning_rate": 3.2156662300737953e-06,
      "loss": 1.2305,
      "step": 27239
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.802443742752075,
      "learning_rate": 3.2150273136760056e-06,
      "loss": 1.0993,
      "step": 27240
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1738359928131104,
      "learning_rate": 3.2143883972782163e-06,
      "loss": 0.8269,
      "step": 27241
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.5097150802612305,
      "learning_rate": 3.213749480880427e-06,
      "loss": 0.8372,
      "step": 27242
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.852585792541504,
      "learning_rate": 3.2131105644826377e-06,
      "loss": 0.8566,
      "step": 27243
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4901692867279053,
      "learning_rate": 3.2124716480848484e-06,
      "loss": 0.3952,
      "step": 27244
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.1153223514556885,
      "learning_rate": 3.211832731687059e-06,
      "loss": 0.7703,
      "step": 27245
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6303744316101074,
      "learning_rate": 3.2111938152892698e-06,
      "loss": 0.7986,
      "step": 27246
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4686787128448486,
      "learning_rate": 3.2105548988914805e-06,
      "loss": 0.7601,
      "step": 27247
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.732951283454895,
      "learning_rate": 3.209915982493691e-06,
      "loss": 1.0003,
      "step": 27248
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.656909704208374,
      "learning_rate": 3.2092770660959014e-06,
      "loss": 0.7708,
      "step": 27249
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6388022899627686,
      "learning_rate": 3.208638149698112e-06,
      "loss": 0.6997,
      "step": 27250
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0692977905273438,
      "learning_rate": 3.207999233300323e-06,
      "loss": 0.8849,
      "step": 27251
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.4172520637512207,
      "learning_rate": 3.2073603169025335e-06,
      "loss": 1.0834,
      "step": 27252
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6133073568344116,
      "learning_rate": 3.2067214005047442e-06,
      "loss": 0.851,
      "step": 27253
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8146854639053345,
      "learning_rate": 3.206082484106955e-06,
      "loss": 0.7505,
      "step": 27254
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6396708488464355,
      "learning_rate": 3.2054435677091656e-06,
      "loss": 0.8495,
      "step": 27255
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4824085235595703,
      "learning_rate": 3.2048046513113763e-06,
      "loss": 0.8908,
      "step": 27256
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.739238977432251,
      "learning_rate": 3.204165734913587e-06,
      "loss": 1.0732,
      "step": 27257
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4252164363861084,
      "learning_rate": 3.2035268185157973e-06,
      "loss": 0.6984,
      "step": 27258
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8102060556411743,
      "learning_rate": 3.202887902118008e-06,
      "loss": 0.8353,
      "step": 27259
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.3075883388519287,
      "learning_rate": 3.2022489857202187e-06,
      "loss": 0.9745,
      "step": 27260
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.7086915969848633,
      "learning_rate": 3.2016100693224294e-06,
      "loss": 1.1087,
      "step": 27261
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.369410514831543,
      "learning_rate": 3.20097115292464e-06,
      "loss": 1.2287,
      "step": 27262
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.362689256668091,
      "learning_rate": 3.200332236526851e-06,
      "loss": 0.8029,
      "step": 27263
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.304131031036377,
      "learning_rate": 3.1996933201290615e-06,
      "loss": 0.916,
      "step": 27264
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.96280038356781,
      "learning_rate": 3.1990544037312722e-06,
      "loss": 1.127,
      "step": 27265
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8121618032455444,
      "learning_rate": 3.198415487333483e-06,
      "loss": 0.8844,
      "step": 27266
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.00885009765625,
      "learning_rate": 3.197776570935693e-06,
      "loss": 0.9367,
      "step": 27267
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8007842302322388,
      "learning_rate": 3.197137654537904e-06,
      "loss": 0.8944,
      "step": 27268
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9697465896606445,
      "learning_rate": 3.1964987381401146e-06,
      "loss": 0.7246,
      "step": 27269
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8403645753860474,
      "learning_rate": 3.1958598217423253e-06,
      "loss": 0.9105,
      "step": 27270
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6005756855010986,
      "learning_rate": 3.195220905344536e-06,
      "loss": 0.4523,
      "step": 27271
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.914999008178711,
      "learning_rate": 3.1945819889467467e-06,
      "loss": 0.8128,
      "step": 27272
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.413751602172852,
      "learning_rate": 3.1939430725489574e-06,
      "loss": 1.393,
      "step": 27273
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.693885326385498,
      "learning_rate": 3.193304156151168e-06,
      "loss": 0.8454,
      "step": 27274
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.31056547164917,
      "learning_rate": 3.192665239753379e-06,
      "loss": 0.7521,
      "step": 27275
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.824274063110352,
      "learning_rate": 3.192026323355589e-06,
      "loss": 1.1754,
      "step": 27276
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5619378089904785,
      "learning_rate": 3.1913874069577998e-06,
      "loss": 0.8362,
      "step": 27277
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.090662956237793,
      "learning_rate": 3.1907484905600105e-06,
      "loss": 0.6789,
      "step": 27278
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.277175188064575,
      "learning_rate": 3.190109574162221e-06,
      "loss": 0.8973,
      "step": 27279
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0130884647369385,
      "learning_rate": 3.189470657764432e-06,
      "loss": 0.9409,
      "step": 27280
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.8112101554870605,
      "learning_rate": 3.1888317413666426e-06,
      "loss": 0.8477,
      "step": 27281
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.464334487915039,
      "learning_rate": 3.1881928249688533e-06,
      "loss": 0.4359,
      "step": 27282
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5036122798919678,
      "learning_rate": 3.187553908571064e-06,
      "loss": 0.8043,
      "step": 27283
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.004119873046875,
      "learning_rate": 3.1869149921732747e-06,
      "loss": 0.7306,
      "step": 27284
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.622852087020874,
      "learning_rate": 3.186276075775485e-06,
      "loss": 0.8739,
      "step": 27285
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6219860315322876,
      "learning_rate": 3.1856371593776956e-06,
      "loss": 0.8114,
      "step": 27286
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.301611065864563,
      "learning_rate": 3.1849982429799063e-06,
      "loss": 0.4029,
      "step": 27287
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.8223936557769775,
      "learning_rate": 3.184359326582117e-06,
      "loss": 0.7231,
      "step": 27288
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.001274824142456,
      "learning_rate": 3.1837204101843277e-06,
      "loss": 0.79,
      "step": 27289
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.602973461151123,
      "learning_rate": 3.1830814937865384e-06,
      "loss": 0.7813,
      "step": 27290
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0001730918884277,
      "learning_rate": 3.182442577388749e-06,
      "loss": 0.7627,
      "step": 27291
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.905713677406311,
      "learning_rate": 3.18180366099096e-06,
      "loss": 0.7873,
      "step": 27292
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.237001419067383,
      "learning_rate": 3.1811647445931705e-06,
      "loss": 0.9641,
      "step": 27293
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.955899953842163,
      "learning_rate": 3.180525828195381e-06,
      "loss": 0.8023,
      "step": 27294
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7887828350067139,
      "learning_rate": 3.1798869117975915e-06,
      "loss": 0.4963,
      "step": 27295
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.086811065673828,
      "learning_rate": 3.1792479953998022e-06,
      "loss": 1.2635,
      "step": 27296
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2410151958465576,
      "learning_rate": 3.178609079002013e-06,
      "loss": 0.8395,
      "step": 27297
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.640830397605896,
      "learning_rate": 3.1779701626042236e-06,
      "loss": 0.7207,
      "step": 27298
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.7358832359313965,
      "learning_rate": 3.1773312462064343e-06,
      "loss": 1.652,
      "step": 27299
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5631719827651978,
      "learning_rate": 3.176692329808645e-06,
      "loss": 1.158,
      "step": 27300
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.728315591812134,
      "learning_rate": 3.1760534134108557e-06,
      "loss": 1.0205,
      "step": 27301
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.279135227203369,
      "learning_rate": 3.1754144970130664e-06,
      "loss": 0.6921,
      "step": 27302
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.045105457305908,
      "learning_rate": 3.1747755806152767e-06,
      "loss": 1.0294,
      "step": 27303
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1735994815826416,
      "learning_rate": 3.1741366642174874e-06,
      "loss": 0.9348,
      "step": 27304
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.360342502593994,
      "learning_rate": 3.173497747819698e-06,
      "loss": 0.5431,
      "step": 27305
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7351356744766235,
      "learning_rate": 3.172858831421909e-06,
      "loss": 0.8912,
      "step": 27306
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.25152325630188,
      "learning_rate": 3.1722199150241195e-06,
      "loss": 0.6904,
      "step": 27307
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.009810447692871,
      "learning_rate": 3.17158099862633e-06,
      "loss": 0.8656,
      "step": 27308
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5348161458969116,
      "learning_rate": 3.170942082228541e-06,
      "loss": 0.8308,
      "step": 27309
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.830621361732483,
      "learning_rate": 3.1703031658307516e-06,
      "loss": 0.7373,
      "step": 27310
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5978989601135254,
      "learning_rate": 3.1696642494329623e-06,
      "loss": 0.4103,
      "step": 27311
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8330129384994507,
      "learning_rate": 3.169025333035173e-06,
      "loss": 0.742,
      "step": 27312
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4302482604980469,
      "learning_rate": 3.1683864166373833e-06,
      "loss": 0.818,
      "step": 27313
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9684101343154907,
      "learning_rate": 3.167747500239594e-06,
      "loss": 0.7911,
      "step": 27314
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6473373174667358,
      "learning_rate": 3.1671085838418047e-06,
      "loss": 0.7373,
      "step": 27315
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7687171697616577,
      "learning_rate": 3.1664696674440154e-06,
      "loss": 0.7939,
      "step": 27316
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6159441471099854,
      "learning_rate": 3.165830751046226e-06,
      "loss": 0.8157,
      "step": 27317
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.941719651222229,
      "learning_rate": 3.1651918346484368e-06,
      "loss": 0.9532,
      "step": 27318
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0976994037628174,
      "learning_rate": 3.1645529182506475e-06,
      "loss": 0.7462,
      "step": 27319
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.686582088470459,
      "learning_rate": 3.163914001852858e-06,
      "loss": 0.83,
      "step": 27320
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9613059759140015,
      "learning_rate": 3.163275085455069e-06,
      "loss": 0.8628,
      "step": 27321
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8586792945861816,
      "learning_rate": 3.162636169057279e-06,
      "loss": 0.6914,
      "step": 27322
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5037283897399902,
      "learning_rate": 3.16199725265949e-06,
      "loss": 0.7386,
      "step": 27323
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.0845561027526855,
      "learning_rate": 3.1613583362617005e-06,
      "loss": 1.2096,
      "step": 27324
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1268091201782227,
      "learning_rate": 3.1607194198639112e-06,
      "loss": 0.8619,
      "step": 27325
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2053611278533936,
      "learning_rate": 3.160080503466122e-06,
      "loss": 0.3996,
      "step": 27326
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8816462755203247,
      "learning_rate": 3.1594415870683326e-06,
      "loss": 0.8247,
      "step": 27327
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8632590770721436,
      "learning_rate": 3.1588026706705433e-06,
      "loss": 0.7895,
      "step": 27328
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6359050273895264,
      "learning_rate": 3.158163754272754e-06,
      "loss": 0.72,
      "step": 27329
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6412116289138794,
      "learning_rate": 3.1575248378749647e-06,
      "loss": 0.7716,
      "step": 27330
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.282585382461548,
      "learning_rate": 3.156885921477175e-06,
      "loss": 1.0196,
      "step": 27331
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0556788444519043,
      "learning_rate": 3.1562470050793857e-06,
      "loss": 0.7644,
      "step": 27332
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.094731569290161,
      "learning_rate": 3.1556080886815964e-06,
      "loss": 0.527,
      "step": 27333
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.19826602935791,
      "learning_rate": 3.154969172283807e-06,
      "loss": 0.5947,
      "step": 27334
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6660163402557373,
      "learning_rate": 3.154330255886018e-06,
      "loss": 0.7613,
      "step": 27335
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7626391649246216,
      "learning_rate": 3.1536913394882285e-06,
      "loss": 0.8026,
      "step": 27336
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5945841073989868,
      "learning_rate": 3.1530524230904392e-06,
      "loss": 0.853,
      "step": 27337
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.184983491897583,
      "learning_rate": 3.15241350669265e-06,
      "loss": 0.9525,
      "step": 27338
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0442421436309814,
      "learning_rate": 3.1517745902948606e-06,
      "loss": 0.8332,
      "step": 27339
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.689825415611267,
      "learning_rate": 3.151135673897071e-06,
      "loss": 0.8079,
      "step": 27340
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1368422508239746,
      "learning_rate": 3.1504967574992816e-06,
      "loss": 0.9047,
      "step": 27341
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8648900985717773,
      "learning_rate": 3.1498578411014923e-06,
      "loss": 0.8053,
      "step": 27342
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8578877449035645,
      "learning_rate": 3.149218924703703e-06,
      "loss": 0.9302,
      "step": 27343
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3487749099731445,
      "learning_rate": 3.1485800083059137e-06,
      "loss": 0.3296,
      "step": 27344
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6057029962539673,
      "learning_rate": 3.147941091908124e-06,
      "loss": 0.8213,
      "step": 27345
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0476300716400146,
      "learning_rate": 3.1473021755103343e-06,
      "loss": 0.7411,
      "step": 27346
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.848783016204834,
      "learning_rate": 3.146663259112545e-06,
      "loss": 0.8983,
      "step": 27347
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9548652172088623,
      "learning_rate": 3.1460243427147557e-06,
      "loss": 0.8249,
      "step": 27348
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8128349781036377,
      "learning_rate": 3.1453854263169664e-06,
      "loss": 0.8939,
      "step": 27349
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.3694350719451904,
      "learning_rate": 3.144746509919177e-06,
      "loss": 1.1287,
      "step": 27350
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6934833526611328,
      "learning_rate": 3.1441075935213878e-06,
      "loss": 0.8322,
      "step": 27351
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.238482713699341,
      "learning_rate": 3.1434686771235985e-06,
      "loss": 0.8215,
      "step": 27352
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.486748218536377,
      "learning_rate": 3.142829760725809e-06,
      "loss": 1.0941,
      "step": 27353
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6310769319534302,
      "learning_rate": 3.14219084432802e-06,
      "loss": 0.7992,
      "step": 27354
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.553106427192688,
      "learning_rate": 3.14155192793023e-06,
      "loss": 0.8288,
      "step": 27355
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.860648274421692,
      "learning_rate": 3.140913011532441e-06,
      "loss": 0.7936,
      "step": 27356
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7634080648422241,
      "learning_rate": 3.1402740951346515e-06,
      "loss": 0.8792,
      "step": 27357
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4724513292312622,
      "learning_rate": 3.1396351787368622e-06,
      "loss": 0.42,
      "step": 27358
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.92177414894104,
      "learning_rate": 3.138996262339073e-06,
      "loss": 1.287,
      "step": 27359
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5469764471054077,
      "learning_rate": 3.1383573459412836e-06,
      "loss": 0.8057,
      "step": 27360
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.633524179458618,
      "learning_rate": 3.1377184295434943e-06,
      "loss": 0.8683,
      "step": 27361
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6025569438934326,
      "learning_rate": 3.137079513145705e-06,
      "loss": 0.8292,
      "step": 27362
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9048024415969849,
      "learning_rate": 3.1364405967479157e-06,
      "loss": 0.9044,
      "step": 27363
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7272800207138062,
      "learning_rate": 3.1358016803501264e-06,
      "loss": 0.7395,
      "step": 27364
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4464387893676758,
      "learning_rate": 3.1351627639523367e-06,
      "loss": 0.7691,
      "step": 27365
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9267374277114868,
      "learning_rate": 3.1345238475545474e-06,
      "loss": 0.8542,
      "step": 27366
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1739017963409424,
      "learning_rate": 3.133884931156758e-06,
      "loss": 1.0221,
      "step": 27367
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4809529781341553,
      "learning_rate": 3.133246014758969e-06,
      "loss": 0.6771,
      "step": 27368
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.5511868000030518,
      "learning_rate": 3.1326070983611795e-06,
      "loss": 0.642,
      "step": 27369
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.1590137481689453,
      "learning_rate": 3.13196818196339e-06,
      "loss": 0.7867,
      "step": 27370
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6615253686904907,
      "learning_rate": 3.131329265565601e-06,
      "loss": 0.7573,
      "step": 27371
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4404327869415283,
      "learning_rate": 3.1306903491678116e-06,
      "loss": 0.5398,
      "step": 27372
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.181725263595581,
      "learning_rate": 3.1300514327700223e-06,
      "loss": 0.5976,
      "step": 27373
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.038337469100952,
      "learning_rate": 3.1294125163722326e-06,
      "loss": 0.8592,
      "step": 27374
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9615696668624878,
      "learning_rate": 3.1287735999744433e-06,
      "loss": 0.8342,
      "step": 27375
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1003966331481934,
      "learning_rate": 3.128134683576654e-06,
      "loss": 0.5926,
      "step": 27376
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9330071210861206,
      "learning_rate": 3.1274957671788647e-06,
      "loss": 1.2316,
      "step": 27377
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.173168420791626,
      "learning_rate": 3.1268568507810754e-06,
      "loss": 0.6816,
      "step": 27378
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8232861757278442,
      "learning_rate": 3.126217934383286e-06,
      "loss": 0.7327,
      "step": 27379
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9946560859680176,
      "learning_rate": 3.1255790179854968e-06,
      "loss": 0.871,
      "step": 27380
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.016226053237915,
      "learning_rate": 3.1249401015877075e-06,
      "loss": 0.8041,
      "step": 27381
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.685649871826172,
      "learning_rate": 3.124301185189918e-06,
      "loss": 0.6401,
      "step": 27382
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7983375787734985,
      "learning_rate": 3.1236622687921285e-06,
      "loss": 0.8195,
      "step": 27383
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8273520469665527,
      "learning_rate": 3.123023352394339e-06,
      "loss": 0.8792,
      "step": 27384
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1600964069366455,
      "learning_rate": 3.12238443599655e-06,
      "loss": 1.0243,
      "step": 27385
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9041037559509277,
      "learning_rate": 3.1217455195987606e-06,
      "loss": 0.8348,
      "step": 27386
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8627500534057617,
      "learning_rate": 3.1211066032009713e-06,
      "loss": 0.9129,
      "step": 27387
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.1431195735931396,
      "learning_rate": 3.120467686803182e-06,
      "loss": 1.1363,
      "step": 27388
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.9162442684173584,
      "learning_rate": 3.1198287704053927e-06,
      "loss": 1.0719,
      "step": 27389
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.845761775970459,
      "learning_rate": 3.1191898540076034e-06,
      "loss": 0.4585,
      "step": 27390
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8850889205932617,
      "learning_rate": 3.118550937609814e-06,
      "loss": 0.8501,
      "step": 27391
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6788908243179321,
      "learning_rate": 3.1179120212120243e-06,
      "loss": 0.7666,
      "step": 27392
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1722795963287354,
      "learning_rate": 3.117273104814235e-06,
      "loss": 0.7742,
      "step": 27393
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.943200707435608,
      "learning_rate": 3.1166341884164457e-06,
      "loss": 0.8455,
      "step": 27394
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4050514698028564,
      "learning_rate": 3.1159952720186564e-06,
      "loss": 0.8498,
      "step": 27395
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7329853773117065,
      "learning_rate": 3.115356355620867e-06,
      "loss": 0.7686,
      "step": 27396
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.441812515258789,
      "learning_rate": 3.114717439223078e-06,
      "loss": 0.9891,
      "step": 27397
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8781206607818604,
      "learning_rate": 3.1140785228252885e-06,
      "loss": 0.8551,
      "step": 27398
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7291347980499268,
      "learning_rate": 3.1134396064274992e-06,
      "loss": 0.8154,
      "step": 27399
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.1775007247924805,
      "learning_rate": 3.11280069002971e-06,
      "loss": 1.1487,
      "step": 27400
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1938068866729736,
      "learning_rate": 3.11216177363192e-06,
      "loss": 0.8989,
      "step": 27401
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5482127666473389,
      "learning_rate": 3.111522857234131e-06,
      "loss": 0.8294,
      "step": 27402
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2008070945739746,
      "learning_rate": 3.1108839408363416e-06,
      "loss": 0.911,
      "step": 27403
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8738255500793457,
      "learning_rate": 3.1102450244385523e-06,
      "loss": 0.7239,
      "step": 27404
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0595521926879883,
      "learning_rate": 3.109606108040763e-06,
      "loss": 0.9237,
      "step": 27405
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2562956809997559,
      "learning_rate": 3.1089671916429737e-06,
      "loss": 0.3939,
      "step": 27406
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.679618000984192,
      "learning_rate": 3.1083282752451844e-06,
      "loss": 0.8107,
      "step": 27407
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0634543895721436,
      "learning_rate": 3.107689358847395e-06,
      "loss": 0.9918,
      "step": 27408
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6556310653686523,
      "learning_rate": 3.107050442449606e-06,
      "loss": 0.7939,
      "step": 27409
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5249569416046143,
      "learning_rate": 3.106411526051816e-06,
      "loss": 0.7603,
      "step": 27410
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0016188621520996,
      "learning_rate": 3.1057726096540268e-06,
      "loss": 0.7931,
      "step": 27411
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.354076862335205,
      "learning_rate": 3.1051336932562375e-06,
      "loss": 1.1246,
      "step": 27412
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.882047176361084,
      "learning_rate": 3.104494776858448e-06,
      "loss": 0.8345,
      "step": 27413
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.782674312591553,
      "learning_rate": 3.103855860460659e-06,
      "loss": 0.9294,
      "step": 27414
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6713987588882446,
      "learning_rate": 3.1032169440628696e-06,
      "loss": 0.7682,
      "step": 27415
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4299497604370117,
      "learning_rate": 3.1025780276650803e-06,
      "loss": 0.8644,
      "step": 27416
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5852134227752686,
      "learning_rate": 3.101939111267291e-06,
      "loss": 0.7942,
      "step": 27417
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.454775094985962,
      "learning_rate": 3.1013001948695017e-06,
      "loss": 0.3296,
      "step": 27418
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7792538404464722,
      "learning_rate": 3.100661278471712e-06,
      "loss": 0.8356,
      "step": 27419
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9094812870025635,
      "learning_rate": 3.1000223620739227e-06,
      "loss": 0.8119,
      "step": 27420
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7915114164352417,
      "learning_rate": 3.0993834456761334e-06,
      "loss": 0.8222,
      "step": 27421
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.637519359588623,
      "learning_rate": 3.098744529278344e-06,
      "loss": 0.9505,
      "step": 27422
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.076117515563965,
      "learning_rate": 3.0981056128805548e-06,
      "loss": 1.0228,
      "step": 27423
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8349326848983765,
      "learning_rate": 3.0974666964827655e-06,
      "loss": 0.72,
      "step": 27424
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5208710432052612,
      "learning_rate": 3.096827780084976e-06,
      "loss": 0.7439,
      "step": 27425
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.220187187194824,
      "learning_rate": 3.096188863687187e-06,
      "loss": 1.5749,
      "step": 27426
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.61067533493042,
      "learning_rate": 3.0955499472893976e-06,
      "loss": 0.7787,
      "step": 27427
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9597795009613037,
      "learning_rate": 3.094911030891608e-06,
      "loss": 0.8706,
      "step": 27428
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.5240068435668945,
      "learning_rate": 3.0942721144938185e-06,
      "loss": 1.5174,
      "step": 27429
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.835575103759766,
      "learning_rate": 3.0936331980960292e-06,
      "loss": 1.1299,
      "step": 27430
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.103780746459961,
      "learning_rate": 3.09299428169824e-06,
      "loss": 1.2377,
      "step": 27431
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.3899102210998535,
      "learning_rate": 3.0923553653004506e-06,
      "loss": 0.922,
      "step": 27432
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3364559412002563,
      "learning_rate": 3.0917164489026613e-06,
      "loss": 0.4052,
      "step": 27433
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.075991630554199,
      "learning_rate": 3.091077532504872e-06,
      "loss": 0.9457,
      "step": 27434
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6412770748138428,
      "learning_rate": 3.0904386161070827e-06,
      "loss": 0.9082,
      "step": 27435
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9853843450546265,
      "learning_rate": 3.0897996997092934e-06,
      "loss": 0.7761,
      "step": 27436
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.670773506164551,
      "learning_rate": 3.089160783311504e-06,
      "loss": 0.8891,
      "step": 27437
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.820304274559021,
      "learning_rate": 3.0885218669137144e-06,
      "loss": 0.8632,
      "step": 27438
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6181232929229736,
      "learning_rate": 3.087882950515925e-06,
      "loss": 0.8057,
      "step": 27439
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6103054285049438,
      "learning_rate": 3.087244034118136e-06,
      "loss": 0.8111,
      "step": 27440
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.7622649669647217,
      "learning_rate": 3.0866051177203465e-06,
      "loss": 0.7612,
      "step": 27441
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.303483963012695,
      "learning_rate": 3.0859662013225572e-06,
      "loss": 0.9545,
      "step": 27442
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.851625680923462,
      "learning_rate": 3.085327284924768e-06,
      "loss": 0.8288,
      "step": 27443
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.5013391971588135,
      "learning_rate": 3.0846883685269786e-06,
      "loss": 0.8811,
      "step": 27444
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7153294086456299,
      "learning_rate": 3.0840494521291893e-06,
      "loss": 0.8332,
      "step": 27445
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6196837425231934,
      "learning_rate": 3.0834105357314e-06,
      "loss": 0.9704,
      "step": 27446
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.3449630737304688,
      "learning_rate": 3.0827716193336103e-06,
      "loss": 0.7773,
      "step": 27447
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6965214014053345,
      "learning_rate": 3.082132702935821e-06,
      "loss": 0.7983,
      "step": 27448
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.884850263595581,
      "learning_rate": 3.0814937865380317e-06,
      "loss": 0.7925,
      "step": 27449
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.114867687225342,
      "learning_rate": 3.0808548701402424e-06,
      "loss": 0.8724,
      "step": 27450
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.495394229888916,
      "learning_rate": 3.080215953742453e-06,
      "loss": 1.1086,
      "step": 27451
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6062755584716797,
      "learning_rate": 3.079577037344664e-06,
      "loss": 0.8453,
      "step": 27452
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.756840467453003,
      "learning_rate": 3.0789381209468745e-06,
      "loss": 0.9112,
      "step": 27453
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.759995698928833,
      "learning_rate": 3.078299204549085e-06,
      "loss": 1.171,
      "step": 27454
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1698970794677734,
      "learning_rate": 3.077660288151296e-06,
      "loss": 1.0538,
      "step": 27455
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.290783166885376,
      "learning_rate": 3.077021371753506e-06,
      "loss": 1.0007,
      "step": 27456
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.95360267162323,
      "learning_rate": 3.076382455355717e-06,
      "loss": 0.8089,
      "step": 27457
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7699493169784546,
      "learning_rate": 3.0757435389579276e-06,
      "loss": 0.9246,
      "step": 27458
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8084170818328857,
      "learning_rate": 3.0751046225601383e-06,
      "loss": 0.9153,
      "step": 27459
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9235939979553223,
      "learning_rate": 3.074465706162349e-06,
      "loss": 0.914,
      "step": 27460
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.4158098697662354,
      "learning_rate": 3.0738267897645597e-06,
      "loss": 1.1777,
      "step": 27461
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.3475286960601807,
      "learning_rate": 3.0731878733667704e-06,
      "loss": 0.7545,
      "step": 27462
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7294678688049316,
      "learning_rate": 3.072548956968981e-06,
      "loss": 0.7568,
      "step": 27463
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6017780303955078,
      "learning_rate": 3.0719100405711918e-06,
      "loss": 0.836,
      "step": 27464
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0179789066314697,
      "learning_rate": 3.071271124173402e-06,
      "loss": 0.7944,
      "step": 27465
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8285614252090454,
      "learning_rate": 3.0706322077756127e-06,
      "loss": 0.75,
      "step": 27466
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.356309175491333,
      "learning_rate": 3.0699932913778234e-06,
      "loss": 0.9218,
      "step": 27467
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8108110427856445,
      "learning_rate": 3.069354374980034e-06,
      "loss": 0.76,
      "step": 27468
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7349357604980469,
      "learning_rate": 3.068715458582245e-06,
      "loss": 0.8401,
      "step": 27469
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0398166179656982,
      "learning_rate": 3.0680765421844555e-06,
      "loss": 0.838,
      "step": 27470
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9429739713668823,
      "learning_rate": 3.0674376257866662e-06,
      "loss": 0.8325,
      "step": 27471
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4999704360961914,
      "learning_rate": 3.066798709388877e-06,
      "loss": 0.7839,
      "step": 27472
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8273699283599854,
      "learning_rate": 3.0661597929910876e-06,
      "loss": 0.843,
      "step": 27473
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.21267032623291,
      "learning_rate": 3.065520876593298e-06,
      "loss": 1.3275,
      "step": 27474
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5044959783554077,
      "learning_rate": 3.0648819601955086e-06,
      "loss": 0.4113,
      "step": 27475
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5575982332229614,
      "learning_rate": 3.0642430437977193e-06,
      "loss": 0.8054,
      "step": 27476
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.508143424987793,
      "learning_rate": 3.06360412739993e-06,
      "loss": 0.8519,
      "step": 27477
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3203449249267578,
      "learning_rate": 3.0629652110021407e-06,
      "loss": 0.7776,
      "step": 27478
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5453379154205322,
      "learning_rate": 3.0623262946043514e-06,
      "loss": 0.7761,
      "step": 27479
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9915670156478882,
      "learning_rate": 3.061687378206562e-06,
      "loss": 0.8792,
      "step": 27480
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.935426712036133,
      "learning_rate": 3.061048461808773e-06,
      "loss": 1.0748,
      "step": 27481
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.2143263816833496,
      "learning_rate": 3.0604095454109835e-06,
      "loss": 0.8452,
      "step": 27482
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.5631580352783203,
      "learning_rate": 3.059770629013194e-06,
      "loss": 0.8886,
      "step": 27483
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.566756010055542,
      "learning_rate": 3.0591317126154045e-06,
      "loss": 0.8243,
      "step": 27484
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.604231119155884,
      "learning_rate": 3.058492796217615e-06,
      "loss": 1.0011,
      "step": 27485
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9386919736862183,
      "learning_rate": 3.057853879819826e-06,
      "loss": 0.7947,
      "step": 27486
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6124337911605835,
      "learning_rate": 3.0572149634220366e-06,
      "loss": 0.7923,
      "step": 27487
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7269870042800903,
      "learning_rate": 3.0565760470242473e-06,
      "loss": 0.9361,
      "step": 27488
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.686878204345703,
      "learning_rate": 3.055937130626458e-06,
      "loss": 1.2083,
      "step": 27489
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.7854790687561035,
      "learning_rate": 3.0552982142286687e-06,
      "loss": 1.6802,
      "step": 27490
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.515159249305725,
      "learning_rate": 3.0546592978308794e-06,
      "loss": 0.8127,
      "step": 27491
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3989862203598022,
      "learning_rate": 3.0540203814330897e-06,
      "loss": 0.3794,
      "step": 27492
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4677298069000244,
      "learning_rate": 3.0533814650353004e-06,
      "loss": 0.8273,
      "step": 27493
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2753937244415283,
      "learning_rate": 3.052742548637511e-06,
      "loss": 0.3967,
      "step": 27494
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.3059334754943848,
      "learning_rate": 3.0521036322397218e-06,
      "loss": 0.7819,
      "step": 27495
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4086312055587769,
      "learning_rate": 3.0514647158419325e-06,
      "loss": 0.3684,
      "step": 27496
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7342382669448853,
      "learning_rate": 3.050825799444143e-06,
      "loss": 0.7832,
      "step": 27497
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.662337303161621,
      "learning_rate": 3.050186883046354e-06,
      "loss": 0.8593,
      "step": 27498
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8573192358016968,
      "learning_rate": 3.0495479666485646e-06,
      "loss": 0.7912,
      "step": 27499
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.033094882965088,
      "learning_rate": 3.0489090502507753e-06,
      "loss": 1.064,
      "step": 27500
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0073893070220947,
      "learning_rate": 3.0482701338529855e-06,
      "loss": 1.0463,
      "step": 27501
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7101677656173706,
      "learning_rate": 3.0476312174551962e-06,
      "loss": 0.7622,
      "step": 27502
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0510010719299316,
      "learning_rate": 3.046992301057407e-06,
      "loss": 0.8586,
      "step": 27503
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5378812551498413,
      "learning_rate": 3.0463533846596176e-06,
      "loss": 0.8164,
      "step": 27504
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6490615606307983,
      "learning_rate": 3.0457144682618283e-06,
      "loss": 0.7916,
      "step": 27505
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.9512579441070557,
      "learning_rate": 3.045075551864039e-06,
      "loss": 1.0819,
      "step": 27506
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.5204732418060303,
      "learning_rate": 3.0444366354662497e-06,
      "loss": 0.7689,
      "step": 27507
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.44390869140625,
      "learning_rate": 3.0437977190684604e-06,
      "loss": 0.4603,
      "step": 27508
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9931544065475464,
      "learning_rate": 3.043158802670671e-06,
      "loss": 0.7977,
      "step": 27509
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6909024715423584,
      "learning_rate": 3.042519886272882e-06,
      "loss": 0.7603,
      "step": 27510
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.8684515953063965,
      "learning_rate": 3.041880969875092e-06,
      "loss": 0.5796,
      "step": 27511
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4627318382263184,
      "learning_rate": 3.041242053477303e-06,
      "loss": 0.7183,
      "step": 27512
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.186523914337158,
      "learning_rate": 3.0406031370795135e-06,
      "loss": 1.497,
      "step": 27513
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0382909774780273,
      "learning_rate": 3.0399642206817242e-06,
      "loss": 0.7098,
      "step": 27514
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.706508159637451,
      "learning_rate": 3.039325304283935e-06,
      "loss": 0.4171,
      "step": 27515
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5705918073654175,
      "learning_rate": 3.0386863878861456e-06,
      "loss": 0.4123,
      "step": 27516
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.746906280517578,
      "learning_rate": 3.0380474714883563e-06,
      "loss": 1.104,
      "step": 27517
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.592775583267212,
      "learning_rate": 3.037408555090567e-06,
      "loss": 0.7108,
      "step": 27518
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.5110247135162354,
      "learning_rate": 3.0367696386927777e-06,
      "loss": 0.8006,
      "step": 27519
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.645380973815918,
      "learning_rate": 3.036130722294988e-06,
      "loss": 0.5758,
      "step": 27520
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.974840760231018,
      "learning_rate": 3.0354918058971987e-06,
      "loss": 0.9366,
      "step": 27521
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1625301837921143,
      "learning_rate": 3.0348528894994094e-06,
      "loss": 0.8274,
      "step": 27522
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7686783075332642,
      "learning_rate": 3.03421397310162e-06,
      "loss": 0.8855,
      "step": 27523
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.8997669219970703,
      "learning_rate": 3.033575056703831e-06,
      "loss": 0.8294,
      "step": 27524
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6726065874099731,
      "learning_rate": 3.0329361403060415e-06,
      "loss": 0.7594,
      "step": 27525
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1577632427215576,
      "learning_rate": 3.032297223908252e-06,
      "loss": 0.9546,
      "step": 27526
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1630923748016357,
      "learning_rate": 3.031658307510463e-06,
      "loss": 0.8524,
      "step": 27527
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7807016372680664,
      "learning_rate": 3.0310193911126736e-06,
      "loss": 0.7943,
      "step": 27528
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.512336254119873,
      "learning_rate": 3.030380474714884e-06,
      "loss": 1.4003,
      "step": 27529
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8467183113098145,
      "learning_rate": 3.0297415583170946e-06,
      "loss": 1.1418,
      "step": 27530
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7829526662826538,
      "learning_rate": 3.0291026419193053e-06,
      "loss": 0.8335,
      "step": 27531
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.574297308921814,
      "learning_rate": 3.028463725521516e-06,
      "loss": 0.9422,
      "step": 27532
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9468811750411987,
      "learning_rate": 3.0278248091237267e-06,
      "loss": 0.7834,
      "step": 27533
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7824656963348389,
      "learning_rate": 3.0271858927259374e-06,
      "loss": 0.7181,
      "step": 27534
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.804901123046875,
      "learning_rate": 3.026546976328148e-06,
      "loss": 0.7748,
      "step": 27535
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.634848713874817,
      "learning_rate": 3.0259080599303588e-06,
      "loss": 0.7418,
      "step": 27536
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.819211721420288,
      "learning_rate": 3.0252691435325695e-06,
      "loss": 1.0467,
      "step": 27537
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4558080434799194,
      "learning_rate": 3.0246302271347793e-06,
      "loss": 0.382,
      "step": 27538
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.040757656097412,
      "learning_rate": 3.02399131073699e-06,
      "loss": 0.9869,
      "step": 27539
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9434566497802734,
      "learning_rate": 3.0233523943392007e-06,
      "loss": 0.8872,
      "step": 27540
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.08349347114563,
      "learning_rate": 3.0227134779414114e-06,
      "loss": 0.8804,
      "step": 27541
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.768380880355835,
      "learning_rate": 3.022074561543622e-06,
      "loss": 0.8377,
      "step": 27542
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6260719299316406,
      "learning_rate": 3.021435645145833e-06,
      "loss": 0.7765,
      "step": 27543
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5836879014968872,
      "learning_rate": 3.020796728748043e-06,
      "loss": 0.8161,
      "step": 27544
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9637439250946045,
      "learning_rate": 3.020157812350254e-06,
      "loss": 0.7523,
      "step": 27545
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7982442378997803,
      "learning_rate": 3.0195188959524645e-06,
      "loss": 0.8925,
      "step": 27546
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.641007900238037,
      "learning_rate": 3.018879979554675e-06,
      "loss": 0.9994,
      "step": 27547
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.712036371231079,
      "learning_rate": 3.018241063156886e-06,
      "loss": 0.845,
      "step": 27548
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8386198282241821,
      "learning_rate": 3.0176021467590966e-06,
      "loss": 0.9013,
      "step": 27549
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.759892225265503,
      "learning_rate": 3.0169632303613073e-06,
      "loss": 0.9144,
      "step": 27550
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.3717987537384033,
      "learning_rate": 3.016324313963518e-06,
      "loss": 0.9592,
      "step": 27551
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0747358798980713,
      "learning_rate": 3.0156853975657287e-06,
      "loss": 0.9179,
      "step": 27552
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0937397480010986,
      "learning_rate": 3.015046481167939e-06,
      "loss": 0.6306,
      "step": 27553
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.9433295726776123,
      "learning_rate": 3.0144075647701497e-06,
      "loss": 0.7596,
      "step": 27554
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.804826259613037,
      "learning_rate": 3.0137686483723604e-06,
      "loss": 0.8766,
      "step": 27555
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0737555027008057,
      "learning_rate": 3.013129731974571e-06,
      "loss": 0.8597,
      "step": 27556
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7724945545196533,
      "learning_rate": 3.0124908155767818e-06,
      "loss": 0.8733,
      "step": 27557
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.068042755126953,
      "learning_rate": 3.0118518991789925e-06,
      "loss": 0.8522,
      "step": 27558
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.5027332305908203,
      "learning_rate": 3.011212982781203e-06,
      "loss": 1.2044,
      "step": 27559
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.931222677230835,
      "learning_rate": 3.010574066383414e-06,
      "loss": 0.8044,
      "step": 27560
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7344576120376587,
      "learning_rate": 3.0099351499856246e-06,
      "loss": 1.1483,
      "step": 27561
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.802483081817627,
      "learning_rate": 3.0092962335878353e-06,
      "loss": 0.927,
      "step": 27562
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6691970825195312,
      "learning_rate": 3.0086573171900456e-06,
      "loss": 0.8238,
      "step": 27563
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9095563888549805,
      "learning_rate": 3.0080184007922563e-06,
      "loss": 0.8031,
      "step": 27564
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.645962119102478,
      "learning_rate": 3.007379484394467e-06,
      "loss": 0.8036,
      "step": 27565
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.750633955001831,
      "learning_rate": 3.0067405679966777e-06,
      "loss": 0.8806,
      "step": 27566
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6596568822860718,
      "learning_rate": 3.0061016515988884e-06,
      "loss": 0.8724,
      "step": 27567
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.196622848510742,
      "learning_rate": 3.005462735201099e-06,
      "loss": 0.8443,
      "step": 27568
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7005666494369507,
      "learning_rate": 3.0048238188033098e-06,
      "loss": 0.8877,
      "step": 27569
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.670514702796936,
      "learning_rate": 3.0041849024055205e-06,
      "loss": 0.8503,
      "step": 27570
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.5816569328308105,
      "learning_rate": 3.003545986007731e-06,
      "loss": 0.9481,
      "step": 27571
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0542616844177246,
      "learning_rate": 3.0029070696099414e-06,
      "loss": 0.9456,
      "step": 27572
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.403557777404785,
      "learning_rate": 3.002268153212152e-06,
      "loss": 1.0861,
      "step": 27573
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6185376644134521,
      "learning_rate": 3.001629236814363e-06,
      "loss": 0.8653,
      "step": 27574
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.679203987121582,
      "learning_rate": 3.0009903204165735e-06,
      "loss": 0.8418,
      "step": 27575
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5161771774291992,
      "learning_rate": 3.0003514040187842e-06,
      "loss": 0.855,
      "step": 27576
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.7756755352020264,
      "learning_rate": 2.999712487620995e-06,
      "loss": 0.4699,
      "step": 27577
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7180218696594238,
      "learning_rate": 2.9990735712232056e-06,
      "loss": 0.8904,
      "step": 27578
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0748441219329834,
      "learning_rate": 2.9984346548254163e-06,
      "loss": 0.8343,
      "step": 27579
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7848055362701416,
      "learning_rate": 2.997795738427627e-06,
      "loss": 0.8102,
      "step": 27580
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.225670099258423,
      "learning_rate": 2.9971568220298373e-06,
      "loss": 0.7654,
      "step": 27581
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.012173652648926,
      "learning_rate": 2.996517905632048e-06,
      "loss": 0.7158,
      "step": 27582
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.999082088470459,
      "learning_rate": 2.9958789892342587e-06,
      "loss": 0.8316,
      "step": 27583
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9258966445922852,
      "learning_rate": 2.9952400728364694e-06,
      "loss": 1.1845,
      "step": 27584
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7782022953033447,
      "learning_rate": 2.99460115643868e-06,
      "loss": 0.9425,
      "step": 27585
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7974112033843994,
      "learning_rate": 2.993962240040891e-06,
      "loss": 0.7781,
      "step": 27586
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6013625860214233,
      "learning_rate": 2.9933233236431015e-06,
      "loss": 0.716,
      "step": 27587
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7020829916000366,
      "learning_rate": 2.992684407245312e-06,
      "loss": 0.9398,
      "step": 27588
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.2420663833618164,
      "learning_rate": 2.992045490847523e-06,
      "loss": 1.3639,
      "step": 27589
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4890058040618896,
      "learning_rate": 2.991406574449733e-06,
      "loss": 0.9128,
      "step": 27590
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.998992919921875,
      "learning_rate": 2.990767658051944e-06,
      "loss": 0.8642,
      "step": 27591
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8645333051681519,
      "learning_rate": 2.9901287416541546e-06,
      "loss": 0.8691,
      "step": 27592
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1334140300750732,
      "learning_rate": 2.9894898252563653e-06,
      "loss": 0.8367,
      "step": 27593
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8532063961029053,
      "learning_rate": 2.988850908858576e-06,
      "loss": 0.6794,
      "step": 27594
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.251413106918335,
      "learning_rate": 2.9882119924607867e-06,
      "loss": 0.5413,
      "step": 27595
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.015511989593506,
      "learning_rate": 2.9875730760629974e-06,
      "loss": 1.1538,
      "step": 27596
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.176712989807129,
      "learning_rate": 2.986934159665208e-06,
      "loss": 0.9222,
      "step": 27597
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.641532063484192,
      "learning_rate": 2.9862952432674188e-06,
      "loss": 0.7861,
      "step": 27598
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5756323337554932,
      "learning_rate": 2.985656326869629e-06,
      "loss": 0.8517,
      "step": 27599
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.854419708251953,
      "learning_rate": 2.9850174104718398e-06,
      "loss": 1.048,
      "step": 27600
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.413525342941284,
      "learning_rate": 2.9843784940740505e-06,
      "loss": 0.7637,
      "step": 27601
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3706721067428589,
      "learning_rate": 2.983739577676261e-06,
      "loss": 0.8154,
      "step": 27602
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4847588539123535,
      "learning_rate": 2.983100661278472e-06,
      "loss": 0.7813,
      "step": 27603
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5293238162994385,
      "learning_rate": 2.9824617448806826e-06,
      "loss": 0.7866,
      "step": 27604
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5986851453781128,
      "learning_rate": 2.9818228284828933e-06,
      "loss": 0.7345,
      "step": 27605
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0162763595581055,
      "learning_rate": 2.981183912085104e-06,
      "loss": 0.8329,
      "step": 27606
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.056197166442871,
      "learning_rate": 2.9805449956873147e-06,
      "loss": 1.3499,
      "step": 27607
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4790962934494019,
      "learning_rate": 2.979906079289525e-06,
      "loss": 0.7136,
      "step": 27608
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.6337268352508545,
      "learning_rate": 2.9792671628917356e-06,
      "loss": 0.9207,
      "step": 27609
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.283689022064209,
      "learning_rate": 2.9786282464939463e-06,
      "loss": 0.8313,
      "step": 27610
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.278951644897461,
      "learning_rate": 2.977989330096157e-06,
      "loss": 0.9837,
      "step": 27611
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.724503993988037,
      "learning_rate": 2.9773504136983677e-06,
      "loss": 0.7336,
      "step": 27612
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.2671122550964355,
      "learning_rate": 2.9767114973005784e-06,
      "loss": 1.2796,
      "step": 27613
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.275510549545288,
      "learning_rate": 2.976072580902789e-06,
      "loss": 0.3485,
      "step": 27614
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6228564977645874,
      "learning_rate": 2.975433664505e-06,
      "loss": 0.7379,
      "step": 27615
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6541011333465576,
      "learning_rate": 2.9747947481072105e-06,
      "loss": 0.8234,
      "step": 27616
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.4620964527130127,
      "learning_rate": 2.974155831709421e-06,
      "loss": 1.0031,
      "step": 27617
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.148050308227539,
      "learning_rate": 2.9735169153116315e-06,
      "loss": 0.9806,
      "step": 27618
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9409875869750977,
      "learning_rate": 2.9728779989138422e-06,
      "loss": 0.8583,
      "step": 27619
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.1544973850250244,
      "learning_rate": 2.972239082516053e-06,
      "loss": 1.0565,
      "step": 27620
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5782898664474487,
      "learning_rate": 2.9716001661182636e-06,
      "loss": 0.7729,
      "step": 27621
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.474799633026123,
      "learning_rate": 2.9709612497204743e-06,
      "loss": 0.8275,
      "step": 27622
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.094876766204834,
      "learning_rate": 2.970322333322685e-06,
      "loss": 1.2362,
      "step": 27623
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.0958194732666016,
      "learning_rate": 2.9696834169248957e-06,
      "loss": 0.8362,
      "step": 27624
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.536926031112671,
      "learning_rate": 2.9690445005271064e-06,
      "loss": 0.7381,
      "step": 27625
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7098978757858276,
      "learning_rate": 2.9684055841293167e-06,
      "loss": 0.8473,
      "step": 27626
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.858864426612854,
      "learning_rate": 2.9677666677315274e-06,
      "loss": 0.7707,
      "step": 27627
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8898850679397583,
      "learning_rate": 2.967127751333738e-06,
      "loss": 0.9027,
      "step": 27628
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.384268283843994,
      "learning_rate": 2.966488834935949e-06,
      "loss": 0.8035,
      "step": 27629
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.347530484199524,
      "learning_rate": 2.9658499185381595e-06,
      "loss": 0.3434,
      "step": 27630
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8055721521377563,
      "learning_rate": 2.96521100214037e-06,
      "loss": 0.8103,
      "step": 27631
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.061422824859619,
      "learning_rate": 2.964572085742581e-06,
      "loss": 0.9887,
      "step": 27632
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.883873462677002,
      "learning_rate": 2.9639331693447916e-06,
      "loss": 1.1367,
      "step": 27633
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6159241199493408,
      "learning_rate": 2.9632942529470023e-06,
      "loss": 0.7694,
      "step": 27634
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8070998191833496,
      "learning_rate": 2.962655336549213e-06,
      "loss": 0.9047,
      "step": 27635
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.460479974746704,
      "learning_rate": 2.9620164201514233e-06,
      "loss": 0.7585,
      "step": 27636
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.378485918045044,
      "learning_rate": 2.961377503753634e-06,
      "loss": 0.4734,
      "step": 27637
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7651869058609009,
      "learning_rate": 2.9607385873558447e-06,
      "loss": 0.7952,
      "step": 27638
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6817902326583862,
      "learning_rate": 2.9600996709580554e-06,
      "loss": 0.7172,
      "step": 27639
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.123460292816162,
      "learning_rate": 2.959460754560266e-06,
      "loss": 0.9523,
      "step": 27640
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.591617584228516,
      "learning_rate": 2.9588218381624768e-06,
      "loss": 0.9769,
      "step": 27641
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8484313488006592,
      "learning_rate": 2.9581829217646875e-06,
      "loss": 0.9232,
      "step": 27642
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2713617086410522,
      "learning_rate": 2.957544005366898e-06,
      "loss": 0.752,
      "step": 27643
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1638150215148926,
      "learning_rate": 2.956905088969109e-06,
      "loss": 0.7834,
      "step": 27644
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.746619462966919,
      "learning_rate": 2.956266172571319e-06,
      "loss": 0.9632,
      "step": 27645
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.637584686279297,
      "learning_rate": 2.95562725617353e-06,
      "loss": 1.5026,
      "step": 27646
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.857885718345642,
      "learning_rate": 2.9549883397757405e-06,
      "loss": 0.6803,
      "step": 27647
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0123403072357178,
      "learning_rate": 2.9543494233779512e-06,
      "loss": 0.8094,
      "step": 27648
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9669743776321411,
      "learning_rate": 2.953710506980162e-06,
      "loss": 0.7437,
      "step": 27649
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.71811842918396,
      "learning_rate": 2.9530715905823726e-06,
      "loss": 0.5999,
      "step": 27650
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.575944423675537,
      "learning_rate": 2.9524326741845833e-06,
      "loss": 0.667,
      "step": 27651
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3432416915893555,
      "learning_rate": 2.951793757786794e-06,
      "loss": 0.8846,
      "step": 27652
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4293253421783447,
      "learning_rate": 2.9511548413890047e-06,
      "loss": 0.9112,
      "step": 27653
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9010353088378906,
      "learning_rate": 2.950515924991215e-06,
      "loss": 0.6308,
      "step": 27654
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8942664861679077,
      "learning_rate": 2.9498770085934257e-06,
      "loss": 1.1217,
      "step": 27655
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5432950258255005,
      "learning_rate": 2.9492380921956364e-06,
      "loss": 0.89,
      "step": 27656
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.014387369155884,
      "learning_rate": 2.948599175797847e-06,
      "loss": 0.9965,
      "step": 27657
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5625529289245605,
      "learning_rate": 2.947960259400058e-06,
      "loss": 0.4254,
      "step": 27658
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.622758388519287,
      "learning_rate": 2.9473213430022685e-06,
      "loss": 0.9283,
      "step": 27659
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7560802698135376,
      "learning_rate": 2.9466824266044792e-06,
      "loss": 0.9176,
      "step": 27660
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9126707315444946,
      "learning_rate": 2.94604351020669e-06,
      "loss": 0.8381,
      "step": 27661
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.493755578994751,
      "learning_rate": 2.9454045938089006e-06,
      "loss": 0.7864,
      "step": 27662
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.128145217895508,
      "learning_rate": 2.944765677411111e-06,
      "loss": 1.1379,
      "step": 27663
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4814685583114624,
      "learning_rate": 2.9441267610133216e-06,
      "loss": 0.3563,
      "step": 27664
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6718769073486328,
      "learning_rate": 2.9434878446155323e-06,
      "loss": 0.7614,
      "step": 27665
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8062713146209717,
      "learning_rate": 2.942848928217743e-06,
      "loss": 0.8346,
      "step": 27666
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6400058269500732,
      "learning_rate": 2.9422100118199537e-06,
      "loss": 0.8291,
      "step": 27667
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2024712562561035,
      "learning_rate": 2.9415710954221644e-06,
      "loss": 0.8868,
      "step": 27668
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6885488033294678,
      "learning_rate": 2.940932179024375e-06,
      "loss": 0.8574,
      "step": 27669
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6518735885620117,
      "learning_rate": 2.940293262626586e-06,
      "loss": 0.8192,
      "step": 27670
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.153841972351074,
      "learning_rate": 2.9396543462287965e-06,
      "loss": 0.8654,
      "step": 27671
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.682659864425659,
      "learning_rate": 2.9390154298310068e-06,
      "loss": 0.7111,
      "step": 27672
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.50553560256958,
      "learning_rate": 2.9383765134332175e-06,
      "loss": 0.729,
      "step": 27673
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1103944778442383,
      "learning_rate": 2.937737597035428e-06,
      "loss": 0.7553,
      "step": 27674
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3676559925079346,
      "learning_rate": 2.937098680637639e-06,
      "loss": 0.8151,
      "step": 27675
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.755539655685425,
      "learning_rate": 2.9364597642398496e-06,
      "loss": 0.8085,
      "step": 27676
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.173771858215332,
      "learning_rate": 2.9358208478420603e-06,
      "loss": 0.5679,
      "step": 27677
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6784039735794067,
      "learning_rate": 2.935181931444271e-06,
      "loss": 0.8952,
      "step": 27678
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5532678365707397,
      "learning_rate": 2.9345430150464817e-06,
      "loss": 1.0825,
      "step": 27679
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.7596120834350586,
      "learning_rate": 2.9339040986486924e-06,
      "loss": 0.678,
      "step": 27680
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.8670411109924316,
      "learning_rate": 2.9332651822509026e-06,
      "loss": 0.9395,
      "step": 27681
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6452382802963257,
      "learning_rate": 2.9326262658531133e-06,
      "loss": 0.8459,
      "step": 27682
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2094013690948486,
      "learning_rate": 2.931987349455324e-06,
      "loss": 0.9589,
      "step": 27683
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.3086812496185303,
      "learning_rate": 2.9313484330575347e-06,
      "loss": 1.1156,
      "step": 27684
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.442990779876709,
      "learning_rate": 2.9307095166597454e-06,
      "loss": 0.8453,
      "step": 27685
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4366182088851929,
      "learning_rate": 2.930070600261956e-06,
      "loss": 0.7918,
      "step": 27686
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.759986639022827,
      "learning_rate": 2.929431683864167e-06,
      "loss": 0.8554,
      "step": 27687
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.485302209854126,
      "learning_rate": 2.9287927674663775e-06,
      "loss": 0.9447,
      "step": 27688
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.189598560333252,
      "learning_rate": 2.9281538510685882e-06,
      "loss": 0.8797,
      "step": 27689
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3230695724487305,
      "learning_rate": 2.9275149346707985e-06,
      "loss": 0.6447,
      "step": 27690
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6233211755752563,
      "learning_rate": 2.9268760182730092e-06,
      "loss": 0.8412,
      "step": 27691
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.31813645362854,
      "learning_rate": 2.92623710187522e-06,
      "loss": 1.4681,
      "step": 27692
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8549729585647583,
      "learning_rate": 2.9255981854774306e-06,
      "loss": 0.9114,
      "step": 27693
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7252731323242188,
      "learning_rate": 2.9249592690796413e-06,
      "loss": 0.8076,
      "step": 27694
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.195482015609741,
      "learning_rate": 2.924320352681852e-06,
      "loss": 0.7677,
      "step": 27695
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.179720878601074,
      "learning_rate": 2.9236814362840627e-06,
      "loss": 1.245,
      "step": 27696
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.829801321029663,
      "learning_rate": 2.9230425198862734e-06,
      "loss": 1.2443,
      "step": 27697
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.457000494003296,
      "learning_rate": 2.922403603488484e-06,
      "loss": 0.8587,
      "step": 27698
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.607785940170288,
      "learning_rate": 2.9217646870906944e-06,
      "loss": 0.5323,
      "step": 27699
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.181652545928955,
      "learning_rate": 2.921125770692905e-06,
      "loss": 0.8372,
      "step": 27700
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.856942057609558,
      "learning_rate": 2.920486854295116e-06,
      "loss": 0.4925,
      "step": 27701
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.39488685131073,
      "learning_rate": 2.9198479378973265e-06,
      "loss": 0.758,
      "step": 27702
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9388086795806885,
      "learning_rate": 2.919209021499537e-06,
      "loss": 1.1566,
      "step": 27703
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7843025922775269,
      "learning_rate": 2.918570105101748e-06,
      "loss": 0.3619,
      "step": 27704
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.226149320602417,
      "learning_rate": 2.9179311887039586e-06,
      "loss": 0.8,
      "step": 27705
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.747858762741089,
      "learning_rate": 2.9172922723061693e-06,
      "loss": 1.2305,
      "step": 27706
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7231738567352295,
      "learning_rate": 2.91665335590838e-06,
      "loss": 0.7849,
      "step": 27707
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9946434497833252,
      "learning_rate": 2.9160144395105907e-06,
      "loss": 0.9335,
      "step": 27708
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6919703483581543,
      "learning_rate": 2.915375523112801e-06,
      "loss": 0.6604,
      "step": 27709
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2354307174682617,
      "learning_rate": 2.9147366067150117e-06,
      "loss": 0.8989,
      "step": 27710
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.855683445930481,
      "learning_rate": 2.9140976903172224e-06,
      "loss": 0.9828,
      "step": 27711
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9277451038360596,
      "learning_rate": 2.913458773919433e-06,
      "loss": 0.9658,
      "step": 27712
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6841509342193604,
      "learning_rate": 2.9128198575216438e-06,
      "loss": 0.7884,
      "step": 27713
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6257017850875854,
      "learning_rate": 2.9121809411238545e-06,
      "loss": 0.3492,
      "step": 27714
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.754683256149292,
      "learning_rate": 2.911542024726065e-06,
      "loss": 0.8292,
      "step": 27715
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.60006856918335,
      "learning_rate": 2.910903108328276e-06,
      "loss": 1.0749,
      "step": 27716
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7807708978652954,
      "learning_rate": 2.9102641919304866e-06,
      "loss": 0.6963,
      "step": 27717
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.8182454109191895,
      "learning_rate": 2.909625275532697e-06,
      "loss": 1.9047,
      "step": 27718
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7784109115600586,
      "learning_rate": 2.9089863591349076e-06,
      "loss": 0.8895,
      "step": 27719
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.8596014976501465,
      "learning_rate": 2.9083474427371183e-06,
      "loss": 0.9371,
      "step": 27720
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2453935146331787,
      "learning_rate": 2.907708526339329e-06,
      "loss": 0.7991,
      "step": 27721
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9457428455352783,
      "learning_rate": 2.9070696099415397e-06,
      "loss": 0.7497,
      "step": 27722
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0583736896514893,
      "learning_rate": 2.9064306935437504e-06,
      "loss": 0.8519,
      "step": 27723
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.631742238998413,
      "learning_rate": 2.905791777145961e-06,
      "loss": 0.7546,
      "step": 27724
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6000131368637085,
      "learning_rate": 2.9051528607481718e-06,
      "loss": 0.8164,
      "step": 27725
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7964709997177124,
      "learning_rate": 2.9045139443503825e-06,
      "loss": 0.9126,
      "step": 27726
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7386374473571777,
      "learning_rate": 2.9038750279525927e-06,
      "loss": 0.7714,
      "step": 27727
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3171465396881104,
      "learning_rate": 2.9032361115548034e-06,
      "loss": 0.8767,
      "step": 27728
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0816140174865723,
      "learning_rate": 2.902597195157014e-06,
      "loss": 1.2714,
      "step": 27729
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.105454444885254,
      "learning_rate": 2.901958278759225e-06,
      "loss": 1.2365,
      "step": 27730
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.300971746444702,
      "learning_rate": 2.9013193623614355e-06,
      "loss": 0.8457,
      "step": 27731
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.388779640197754,
      "learning_rate": 2.900680445963646e-06,
      "loss": 1.0444,
      "step": 27732
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6952197551727295,
      "learning_rate": 2.900041529565856e-06,
      "loss": 0.8361,
      "step": 27733
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.1748313903808594,
      "learning_rate": 2.8994026131680668e-06,
      "loss": 0.7596,
      "step": 27734
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.903882622718811,
      "learning_rate": 2.8987636967702775e-06,
      "loss": 0.817,
      "step": 27735
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1426165103912354,
      "learning_rate": 2.898124780372488e-06,
      "loss": 1.0587,
      "step": 27736
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8509736061096191,
      "learning_rate": 2.897485863974699e-06,
      "loss": 0.8157,
      "step": 27737
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2026214599609375,
      "learning_rate": 2.8968469475769096e-06,
      "loss": 1.0158,
      "step": 27738
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4597231149673462,
      "learning_rate": 2.8962080311791203e-06,
      "loss": 0.7937,
      "step": 27739
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.343782663345337,
      "learning_rate": 2.895569114781331e-06,
      "loss": 0.8442,
      "step": 27740
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2847299575805664,
      "learning_rate": 2.8949301983835417e-06,
      "loss": 0.9288,
      "step": 27741
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.651557445526123,
      "learning_rate": 2.894291281985752e-06,
      "loss": 0.7604,
      "step": 27742
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9695193767547607,
      "learning_rate": 2.8936523655879627e-06,
      "loss": 0.7669,
      "step": 27743
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0240824222564697,
      "learning_rate": 2.8930134491901734e-06,
      "loss": 0.836,
      "step": 27744
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.487596869468689,
      "learning_rate": 2.892374532792384e-06,
      "loss": 0.816,
      "step": 27745
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.2309188842773438,
      "learning_rate": 2.8917356163945948e-06,
      "loss": 0.5104,
      "step": 27746
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.4403257369995117,
      "learning_rate": 2.8910966999968055e-06,
      "loss": 0.8038,
      "step": 27747
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.064718008041382,
      "learning_rate": 2.890457783599016e-06,
      "loss": 0.8648,
      "step": 27748
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.199908971786499,
      "learning_rate": 2.889818867201227e-06,
      "loss": 0.8189,
      "step": 27749
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0749166011810303,
      "learning_rate": 2.8891799508034376e-06,
      "loss": 0.8152,
      "step": 27750
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9973328113555908,
      "learning_rate": 2.888541034405648e-06,
      "loss": 0.8822,
      "step": 27751
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.467924118041992,
      "learning_rate": 2.8879021180078585e-06,
      "loss": 1.1001,
      "step": 27752
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4700337648391724,
      "learning_rate": 2.8872632016100692e-06,
      "loss": 0.6735,
      "step": 27753
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.913789987564087,
      "learning_rate": 2.88662428521228e-06,
      "loss": 0.7918,
      "step": 27754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9322675466537476,
      "learning_rate": 2.8859853688144906e-06,
      "loss": 0.8673,
      "step": 27755
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9850826263427734,
      "learning_rate": 2.8853464524167013e-06,
      "loss": 0.8952,
      "step": 27756
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.258704662322998,
      "learning_rate": 2.884707536018912e-06,
      "loss": 0.8177,
      "step": 27757
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.860995292663574,
      "learning_rate": 2.8840686196211227e-06,
      "loss": 0.6917,
      "step": 27758
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0232350826263428,
      "learning_rate": 2.8834297032233334e-06,
      "loss": 0.7792,
      "step": 27759
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8895764350891113,
      "learning_rate": 2.8827907868255437e-06,
      "loss": 0.8817,
      "step": 27760
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.283951282501221,
      "learning_rate": 2.8821518704277544e-06,
      "loss": 1.2375,
      "step": 27761
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.801772952079773,
      "learning_rate": 2.881512954029965e-06,
      "loss": 0.8993,
      "step": 27762
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.938746690750122,
      "learning_rate": 2.880874037632176e-06,
      "loss": 0.787,
      "step": 27763
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0011227130889893,
      "learning_rate": 2.8802351212343865e-06,
      "loss": 0.8149,
      "step": 27764
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9543839693069458,
      "learning_rate": 2.879596204836597e-06,
      "loss": 1.0933,
      "step": 27765
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3824334144592285,
      "learning_rate": 2.878957288438808e-06,
      "loss": 0.9418,
      "step": 27766
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.451240301132202,
      "learning_rate": 2.8783183720410186e-06,
      "loss": 0.5966,
      "step": 27767
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.617815375328064,
      "learning_rate": 2.8776794556432293e-06,
      "loss": 0.7621,
      "step": 27768
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.689008712768555,
      "learning_rate": 2.87704053924544e-06,
      "loss": 1.269,
      "step": 27769
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6764070987701416,
      "learning_rate": 2.8764016228476503e-06,
      "loss": 0.8307,
      "step": 27770
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0123698711395264,
      "learning_rate": 2.875762706449861e-06,
      "loss": 0.7918,
      "step": 27771
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.133890390396118,
      "learning_rate": 2.8751237900520717e-06,
      "loss": 0.835,
      "step": 27772
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.087663650512695,
      "learning_rate": 2.8744848736542824e-06,
      "loss": 1.327,
      "step": 27773
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8977292776107788,
      "learning_rate": 2.873845957256493e-06,
      "loss": 0.9064,
      "step": 27774
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6324865818023682,
      "learning_rate": 2.8732070408587038e-06,
      "loss": 0.7098,
      "step": 27775
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5099471807479858,
      "learning_rate": 2.8725681244609145e-06,
      "loss": 0.7868,
      "step": 27776
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.147679090499878,
      "learning_rate": 2.871929208063125e-06,
      "loss": 0.9383,
      "step": 27777
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.860189437866211,
      "learning_rate": 2.871290291665336e-06,
      "loss": 1.0089,
      "step": 27778
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.3382322788238525,
      "learning_rate": 2.870651375267546e-06,
      "loss": 0.4573,
      "step": 27779
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.560784935951233,
      "learning_rate": 2.870012458869757e-06,
      "loss": 0.8559,
      "step": 27780
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.416929006576538,
      "learning_rate": 2.8693735424719676e-06,
      "loss": 0.6286,
      "step": 27781
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9572408199310303,
      "learning_rate": 2.8687346260741783e-06,
      "loss": 0.5593,
      "step": 27782
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.796160340309143,
      "learning_rate": 2.868095709676389e-06,
      "loss": 0.8171,
      "step": 27783
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.801396369934082,
      "learning_rate": 2.8674567932785997e-06,
      "loss": 0.9687,
      "step": 27784
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2607524394989014,
      "learning_rate": 2.8668178768808104e-06,
      "loss": 0.8266,
      "step": 27785
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.464808940887451,
      "learning_rate": 2.866178960483021e-06,
      "loss": 0.9298,
      "step": 27786
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.880958080291748,
      "learning_rate": 2.8655400440852318e-06,
      "loss": 0.7676,
      "step": 27787
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.911405324935913,
      "learning_rate": 2.864901127687442e-06,
      "loss": 0.952,
      "step": 27788
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7362433671951294,
      "learning_rate": 2.8642622112896527e-06,
      "loss": 0.7848,
      "step": 27789
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8339909315109253,
      "learning_rate": 2.8636232948918634e-06,
      "loss": 0.8543,
      "step": 27790
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6195268630981445,
      "learning_rate": 2.862984378494074e-06,
      "loss": 0.8507,
      "step": 27791
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.391196608543396,
      "learning_rate": 2.862345462096285e-06,
      "loss": 0.814,
      "step": 27792
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7035458087921143,
      "learning_rate": 2.8617065456984955e-06,
      "loss": 0.8966,
      "step": 27793
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7481026649475098,
      "learning_rate": 2.8610676293007062e-06,
      "loss": 0.8721,
      "step": 27794
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8848263025283813,
      "learning_rate": 2.860428712902917e-06,
      "loss": 0.8464,
      "step": 27795
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.180493116378784,
      "learning_rate": 2.8597897965051276e-06,
      "loss": 0.8477,
      "step": 27796
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.020279407501221,
      "learning_rate": 2.859150880107338e-06,
      "loss": 1.0146,
      "step": 27797
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.3198628425598145,
      "learning_rate": 2.8585119637095486e-06,
      "loss": 0.8915,
      "step": 27798
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0203046798706055,
      "learning_rate": 2.8578730473117593e-06,
      "loss": 0.7879,
      "step": 27799
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0055453777313232,
      "learning_rate": 2.85723413091397e-06,
      "loss": 0.8112,
      "step": 27800
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.724414348602295,
      "learning_rate": 2.8565952145161807e-06,
      "loss": 0.5941,
      "step": 27801
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.783678650856018,
      "learning_rate": 2.8559562981183914e-06,
      "loss": 0.8284,
      "step": 27802
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4484755992889404,
      "learning_rate": 2.855317381720602e-06,
      "loss": 0.8105,
      "step": 27803
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8125264644622803,
      "learning_rate": 2.854678465322813e-06,
      "loss": 0.6975,
      "step": 27804
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5813159942626953,
      "learning_rate": 2.8540395489250235e-06,
      "loss": 0.8458,
      "step": 27805
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3428692817687988,
      "learning_rate": 2.853400632527234e-06,
      "loss": 0.4327,
      "step": 27806
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8468748331069946,
      "learning_rate": 2.8527617161294445e-06,
      "loss": 0.8732,
      "step": 27807
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2303617000579834,
      "learning_rate": 2.852122799731655e-06,
      "loss": 0.7658,
      "step": 27808
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.243222713470459,
      "learning_rate": 2.851483883333866e-06,
      "loss": 0.7022,
      "step": 27809
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.044065475463867,
      "learning_rate": 2.8508449669360766e-06,
      "loss": 0.7829,
      "step": 27810
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.468536376953125,
      "learning_rate": 2.8502060505382873e-06,
      "loss": 1.0024,
      "step": 27811
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.025244951248169,
      "learning_rate": 2.849567134140498e-06,
      "loss": 0.7793,
      "step": 27812
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.540454864501953,
      "learning_rate": 2.8489282177427087e-06,
      "loss": 0.8233,
      "step": 27813
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6458282470703125,
      "learning_rate": 2.8482893013449194e-06,
      "loss": 0.9381,
      "step": 27814
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.5970048904418945,
      "learning_rate": 2.8476503849471297e-06,
      "loss": 0.5065,
      "step": 27815
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7381385564804077,
      "learning_rate": 2.8470114685493404e-06,
      "loss": 0.9106,
      "step": 27816
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.7163901329040527,
      "learning_rate": 2.846372552151551e-06,
      "loss": 1.2093,
      "step": 27817
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8881531953811646,
      "learning_rate": 2.8457336357537618e-06,
      "loss": 0.8491,
      "step": 27818
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5748320817947388,
      "learning_rate": 2.8450947193559725e-06,
      "loss": 0.8324,
      "step": 27819
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6064417362213135,
      "learning_rate": 2.844455802958183e-06,
      "loss": 0.8636,
      "step": 27820
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3134913444519043,
      "learning_rate": 2.843816886560394e-06,
      "loss": 0.9012,
      "step": 27821
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0416524410247803,
      "learning_rate": 2.8431779701626046e-06,
      "loss": 0.7583,
      "step": 27822
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3876519203186035,
      "learning_rate": 2.8425390537648153e-06,
      "loss": 0.385,
      "step": 27823
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.19261360168457,
      "learning_rate": 2.8419001373670255e-06,
      "loss": 1.1748,
      "step": 27824
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5301774740219116,
      "learning_rate": 2.8412612209692362e-06,
      "loss": 0.7651,
      "step": 27825
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.049182176589966,
      "learning_rate": 2.840622304571447e-06,
      "loss": 0.9227,
      "step": 27826
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2252249717712402,
      "learning_rate": 2.8399833881736576e-06,
      "loss": 0.8429,
      "step": 27827
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9074223041534424,
      "learning_rate": 2.8393444717758683e-06,
      "loss": 0.5865,
      "step": 27828
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9908242225646973,
      "learning_rate": 2.838705555378079e-06,
      "loss": 1.2737,
      "step": 27829
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8493287563323975,
      "learning_rate": 2.8380666389802897e-06,
      "loss": 0.8574,
      "step": 27830
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1795876026153564,
      "learning_rate": 2.8374277225825004e-06,
      "loss": 0.8605,
      "step": 27831
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9415746927261353,
      "learning_rate": 2.836788806184711e-06,
      "loss": 0.8321,
      "step": 27832
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.27262020111084,
      "learning_rate": 2.8361498897869214e-06,
      "loss": 0.8015,
      "step": 27833
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8191592693328857,
      "learning_rate": 2.835510973389132e-06,
      "loss": 0.8319,
      "step": 27834
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.157888650894165,
      "learning_rate": 2.834872056991343e-06,
      "loss": 1.0712,
      "step": 27835
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.353667974472046,
      "learning_rate": 2.8342331405935535e-06,
      "loss": 0.9355,
      "step": 27836
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.7634658813476562,
      "learning_rate": 2.8335942241957642e-06,
      "loss": 0.6176,
      "step": 27837
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8032671213150024,
      "learning_rate": 2.832955307797975e-06,
      "loss": 0.8756,
      "step": 27838
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9059745073318481,
      "learning_rate": 2.8323163914001856e-06,
      "loss": 1.2043,
      "step": 27839
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6791739463806152,
      "learning_rate": 2.8316774750023963e-06,
      "loss": 0.7614,
      "step": 27840
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1102559566497803,
      "learning_rate": 2.831038558604607e-06,
      "loss": 0.8696,
      "step": 27841
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2536661624908447,
      "learning_rate": 2.8303996422068177e-06,
      "loss": 0.8774,
      "step": 27842
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.3402915000915527,
      "learning_rate": 2.829760725809028e-06,
      "loss": 0.8012,
      "step": 27843
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5113774538040161,
      "learning_rate": 2.8291218094112387e-06,
      "loss": 0.7467,
      "step": 27844
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1728079319000244,
      "learning_rate": 2.8284828930134494e-06,
      "loss": 1.0165,
      "step": 27845
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.355841040611267,
      "learning_rate": 2.82784397661566e-06,
      "loss": 0.3611,
      "step": 27846
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5188260078430176,
      "learning_rate": 2.827205060217871e-06,
      "loss": 0.8652,
      "step": 27847
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.839613437652588,
      "learning_rate": 2.8265661438200815e-06,
      "loss": 0.8344,
      "step": 27848
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6852227449417114,
      "learning_rate": 2.825927227422292e-06,
      "loss": 0.7673,
      "step": 27849
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.760243535041809,
      "learning_rate": 2.825288311024503e-06,
      "loss": 0.8669,
      "step": 27850
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7385039329528809,
      "learning_rate": 2.8246493946267136e-06,
      "loss": 0.7666,
      "step": 27851
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.22869348526001,
      "learning_rate": 2.824010478228924e-06,
      "loss": 1.1685,
      "step": 27852
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8247684240341187,
      "learning_rate": 2.8233715618311346e-06,
      "loss": 0.7517,
      "step": 27853
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7456696033477783,
      "learning_rate": 2.8227326454333453e-06,
      "loss": 0.79,
      "step": 27854
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.761263847351074,
      "learning_rate": 2.822093729035556e-06,
      "loss": 0.8938,
      "step": 27855
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7940270900726318,
      "learning_rate": 2.8214548126377667e-06,
      "loss": 0.7954,
      "step": 27856
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9435558319091797,
      "learning_rate": 2.8208158962399774e-06,
      "loss": 0.9456,
      "step": 27857
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4566725492477417,
      "learning_rate": 2.820176979842188e-06,
      "loss": 0.6673,
      "step": 27858
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5158312320709229,
      "learning_rate": 2.8195380634443988e-06,
      "loss": 0.7201,
      "step": 27859
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8458958864212036,
      "learning_rate": 2.8188991470466095e-06,
      "loss": 0.9037,
      "step": 27860
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1421443223953247,
      "learning_rate": 2.8182602306488197e-06,
      "loss": 0.3251,
      "step": 27861
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.910713791847229,
      "learning_rate": 2.8176213142510304e-06,
      "loss": 0.7968,
      "step": 27862
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7187446355819702,
      "learning_rate": 2.816982397853241e-06,
      "loss": 0.9418,
      "step": 27863
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.953082323074341,
      "learning_rate": 2.816343481455452e-06,
      "loss": 0.9452,
      "step": 27864
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.455850839614868,
      "learning_rate": 2.8157045650576625e-06,
      "loss": 0.9758,
      "step": 27865
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6347367763519287,
      "learning_rate": 2.8150656486598732e-06,
      "loss": 0.8743,
      "step": 27866
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.5623257160186768,
      "learning_rate": 2.814426732262084e-06,
      "loss": 0.5726,
      "step": 27867
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1058175563812256,
      "learning_rate": 2.8137878158642946e-06,
      "loss": 0.9488,
      "step": 27868
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.7731499671936035,
      "learning_rate": 2.8131488994665053e-06,
      "loss": 1.7189,
      "step": 27869
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9550657272338867,
      "learning_rate": 2.8125099830687156e-06,
      "loss": 0.9197,
      "step": 27870
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.031268835067749,
      "learning_rate": 2.8118710666709263e-06,
      "loss": 0.7647,
      "step": 27871
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0803916454315186,
      "learning_rate": 2.811232150273137e-06,
      "loss": 0.8382,
      "step": 27872
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6069955825805664,
      "learning_rate": 2.8105932338753477e-06,
      "loss": 0.8212,
      "step": 27873
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.015373468399048,
      "learning_rate": 2.8099543174775584e-06,
      "loss": 0.5262,
      "step": 27874
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.522730588912964,
      "learning_rate": 2.809315401079769e-06,
      "loss": 0.8347,
      "step": 27875
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8607832193374634,
      "learning_rate": 2.80867648468198e-06,
      "loss": 0.8128,
      "step": 27876
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6708028316497803,
      "learning_rate": 2.8080375682841905e-06,
      "loss": 0.9596,
      "step": 27877
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0208842754364014,
      "learning_rate": 2.8073986518864012e-06,
      "loss": 0.8735,
      "step": 27878
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6070507764816284,
      "learning_rate": 2.8067597354886115e-06,
      "loss": 0.8153,
      "step": 27879
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.935707449913025,
      "learning_rate": 2.806120819090822e-06,
      "loss": 1.1123,
      "step": 27880
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.649258017539978,
      "learning_rate": 2.805481902693033e-06,
      "loss": 0.7176,
      "step": 27881
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.702174186706543,
      "learning_rate": 2.8048429862952436e-06,
      "loss": 0.953,
      "step": 27882
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.616494059562683,
      "learning_rate": 2.8042040698974543e-06,
      "loss": 0.824,
      "step": 27883
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8207718133926392,
      "learning_rate": 2.803565153499665e-06,
      "loss": 0.8848,
      "step": 27884
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7309417724609375,
      "learning_rate": 2.8029262371018757e-06,
      "loss": 0.8156,
      "step": 27885
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.388350009918213,
      "learning_rate": 2.8022873207040864e-06,
      "loss": 0.6294,
      "step": 27886
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.501742362976074,
      "learning_rate": 2.801648404306297e-06,
      "loss": 1.2707,
      "step": 27887
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0248842239379883,
      "learning_rate": 2.8010094879085074e-06,
      "loss": 0.8462,
      "step": 27888
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.3985836505889893,
      "learning_rate": 2.800370571510718e-06,
      "loss": 0.7337,
      "step": 27889
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8158177137374878,
      "learning_rate": 2.7997316551129288e-06,
      "loss": 0.7627,
      "step": 27890
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9638416767120361,
      "learning_rate": 2.7990927387151395e-06,
      "loss": 0.758,
      "step": 27891
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.475259780883789,
      "learning_rate": 2.79845382231735e-06,
      "loss": 1.4521,
      "step": 27892
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.118459939956665,
      "learning_rate": 2.797814905919561e-06,
      "loss": 0.9467,
      "step": 27893
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4323129653930664,
      "learning_rate": 2.7971759895217716e-06,
      "loss": 0.3611,
      "step": 27894
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5808552503585815,
      "learning_rate": 2.7965370731239823e-06,
      "loss": 0.771,
      "step": 27895
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8819630146026611,
      "learning_rate": 2.795898156726193e-06,
      "loss": 0.8319,
      "step": 27896
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2432806491851807,
      "learning_rate": 2.7952592403284033e-06,
      "loss": 0.4388,
      "step": 27897
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.021883010864258,
      "learning_rate": 2.794620323930614e-06,
      "loss": 0.837,
      "step": 27898
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8517156839370728,
      "learning_rate": 2.7939814075328246e-06,
      "loss": 0.8464,
      "step": 27899
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.378023147583008,
      "learning_rate": 2.7933424911350353e-06,
      "loss": 1.1089,
      "step": 27900
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0043439865112305,
      "learning_rate": 2.792703574737246e-06,
      "loss": 0.9848,
      "step": 27901
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.6572105884552,
      "learning_rate": 2.7920646583394567e-06,
      "loss": 0.8304,
      "step": 27902
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1411495208740234,
      "learning_rate": 2.7914257419416674e-06,
      "loss": 0.9109,
      "step": 27903
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5024868249893188,
      "learning_rate": 2.790786825543878e-06,
      "loss": 0.6831,
      "step": 27904
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.81915020942688,
      "learning_rate": 2.790147909146089e-06,
      "loss": 0.85,
      "step": 27905
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0247035026550293,
      "learning_rate": 2.7895089927482995e-06,
      "loss": 0.8826,
      "step": 27906
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.642293930053711,
      "learning_rate": 2.78887007635051e-06,
      "loss": 0.7893,
      "step": 27907
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4823222160339355,
      "learning_rate": 2.7882311599527205e-06,
      "loss": 0.4064,
      "step": 27908
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9723992347717285,
      "learning_rate": 2.7875922435549312e-06,
      "loss": 0.9343,
      "step": 27909
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2527945041656494,
      "learning_rate": 2.786953327157142e-06,
      "loss": 0.8599,
      "step": 27910
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7065513134002686,
      "learning_rate": 2.7863144107593526e-06,
      "loss": 0.7858,
      "step": 27911
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9657610654830933,
      "learning_rate": 2.7856754943615633e-06,
      "loss": 0.8478,
      "step": 27912
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9532358646392822,
      "learning_rate": 2.785036577963774e-06,
      "loss": 1.17,
      "step": 27913
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7094637155532837,
      "learning_rate": 2.7843976615659847e-06,
      "loss": 0.8154,
      "step": 27914
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5138769149780273,
      "learning_rate": 2.7837587451681954e-06,
      "loss": 0.7681,
      "step": 27915
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8007124662399292,
      "learning_rate": 2.7831198287704057e-06,
      "loss": 1.047,
      "step": 27916
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9161732196807861,
      "learning_rate": 2.7824809123726164e-06,
      "loss": 0.7323,
      "step": 27917
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.0411314964294434,
      "learning_rate": 2.781841995974827e-06,
      "loss": 1.0826,
      "step": 27918
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8196907043457031,
      "learning_rate": 2.781203079577038e-06,
      "loss": 0.8956,
      "step": 27919
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6917279958724976,
      "learning_rate": 2.7805641631792485e-06,
      "loss": 0.7863,
      "step": 27920
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.3161375522613525,
      "learning_rate": 2.779925246781459e-06,
      "loss": 1.0384,
      "step": 27921
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.7884154319763184,
      "learning_rate": 2.77928633038367e-06,
      "loss": 0.8484,
      "step": 27922
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6030455827713013,
      "learning_rate": 2.7786474139858806e-06,
      "loss": 0.8359,
      "step": 27923
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7535947561264038,
      "learning_rate": 2.7780084975880913e-06,
      "loss": 0.8225,
      "step": 27924
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.809530735015869,
      "learning_rate": 2.777369581190301e-06,
      "loss": 1.391,
      "step": 27925
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.780734896659851,
      "learning_rate": 2.776730664792512e-06,
      "loss": 0.9263,
      "step": 27926
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2932119369506836,
      "learning_rate": 2.7760917483947226e-06,
      "loss": 0.3652,
      "step": 27927
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5187313556671143,
      "learning_rate": 2.7754528319969333e-06,
      "loss": 0.2751,
      "step": 27928
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.011502981185913,
      "learning_rate": 2.774813915599144e-06,
      "loss": 0.8044,
      "step": 27929
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.98340904712677,
      "learning_rate": 2.7741749992013547e-06,
      "loss": 0.8455,
      "step": 27930
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0992159843444824,
      "learning_rate": 2.773536082803565e-06,
      "loss": 0.8736,
      "step": 27931
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8174309730529785,
      "learning_rate": 2.7728971664057756e-06,
      "loss": 0.8673,
      "step": 27932
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8925585746765137,
      "learning_rate": 2.7722582500079863e-06,
      "loss": 1.0394,
      "step": 27933
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.5629208087921143,
      "learning_rate": 2.771619333610197e-06,
      "loss": 0.3744,
      "step": 27934
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.062802314758301,
      "learning_rate": 2.7709804172124077e-06,
      "loss": 0.7609,
      "step": 27935
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.620590686798096,
      "learning_rate": 2.7703415008146184e-06,
      "loss": 0.9505,
      "step": 27936
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.786341428756714,
      "learning_rate": 2.769702584416829e-06,
      "loss": 0.7839,
      "step": 27937
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.521651268005371,
      "learning_rate": 2.76906366801904e-06,
      "loss": 1.121,
      "step": 27938
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.4272565841674805,
      "learning_rate": 2.7684247516212505e-06,
      "loss": 1.0841,
      "step": 27939
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4936555624008179,
      "learning_rate": 2.767785835223461e-06,
      "loss": 0.814,
      "step": 27940
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.246302843093872,
      "learning_rate": 2.7671469188256715e-06,
      "loss": 0.8914,
      "step": 27941
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6346759796142578,
      "learning_rate": 2.766508002427882e-06,
      "loss": 0.7225,
      "step": 27942
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7554377317428589,
      "learning_rate": 2.765869086030093e-06,
      "loss": 0.854,
      "step": 27943
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.917602777481079,
      "learning_rate": 2.7652301696323036e-06,
      "loss": 0.7558,
      "step": 27944
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.4824724197387695,
      "learning_rate": 2.7645912532345143e-06,
      "loss": 1.5686,
      "step": 27945
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.967686414718628,
      "learning_rate": 2.763952336836725e-06,
      "loss": 0.8078,
      "step": 27946
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6317678689956665,
      "learning_rate": 2.7633134204389357e-06,
      "loss": 0.8103,
      "step": 27947
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.097020149230957,
      "learning_rate": 2.7626745040411464e-06,
      "loss": 0.8002,
      "step": 27948
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6015135049819946,
      "learning_rate": 2.7620355876433567e-06,
      "loss": 0.7595,
      "step": 27949
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5776385068893433,
      "learning_rate": 2.7613966712455674e-06,
      "loss": 0.8861,
      "step": 27950
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0796401500701904,
      "learning_rate": 2.760757754847778e-06,
      "loss": 0.8411,
      "step": 27951
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7357709407806396,
      "learning_rate": 2.7601188384499888e-06,
      "loss": 0.7554,
      "step": 27952
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6592004299163818,
      "learning_rate": 2.7594799220521995e-06,
      "loss": 0.918,
      "step": 27953
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.643976211547852,
      "learning_rate": 2.75884100565441e-06,
      "loss": 0.9582,
      "step": 27954
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0620715618133545,
      "learning_rate": 2.758202089256621e-06,
      "loss": 0.9684,
      "step": 27955
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.8468263149261475,
      "learning_rate": 2.7575631728588316e-06,
      "loss": 1.2376,
      "step": 27956
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1464555263519287,
      "learning_rate": 2.7569242564610423e-06,
      "loss": 0.7682,
      "step": 27957
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.2048001289367676,
      "learning_rate": 2.7562853400632526e-06,
      "loss": 0.8559,
      "step": 27958
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1500988006591797,
      "learning_rate": 2.7556464236654633e-06,
      "loss": 1.1399,
      "step": 27959
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9817243814468384,
      "learning_rate": 2.755007507267674e-06,
      "loss": 1.1446,
      "step": 27960
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7795064449310303,
      "learning_rate": 2.7543685908698847e-06,
      "loss": 0.7097,
      "step": 27961
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9703131914138794,
      "learning_rate": 2.7537296744720954e-06,
      "loss": 0.8148,
      "step": 27962
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.863475799560547,
      "learning_rate": 2.753090758074306e-06,
      "loss": 1.1414,
      "step": 27963
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9123728275299072,
      "learning_rate": 2.7524518416765168e-06,
      "loss": 0.9246,
      "step": 27964
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7287235260009766,
      "learning_rate": 2.7518129252787275e-06,
      "loss": 0.7848,
      "step": 27965
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8966490030288696,
      "learning_rate": 2.751174008880938e-06,
      "loss": 0.6653,
      "step": 27966
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5812619924545288,
      "learning_rate": 2.750535092483149e-06,
      "loss": 0.8446,
      "step": 27967
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.668290138244629,
      "learning_rate": 2.749896176085359e-06,
      "loss": 0.8072,
      "step": 27968
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.275766611099243,
      "learning_rate": 2.74925725968757e-06,
      "loss": 0.7635,
      "step": 27969
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9905685186386108,
      "learning_rate": 2.7486183432897805e-06,
      "loss": 0.9157,
      "step": 27970
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8922816514968872,
      "learning_rate": 2.7479794268919912e-06,
      "loss": 0.916,
      "step": 27971
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6044037342071533,
      "learning_rate": 2.747340510494202e-06,
      "loss": 0.8323,
      "step": 27972
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4165005683898926,
      "learning_rate": 2.7467015940964126e-06,
      "loss": 0.8227,
      "step": 27973
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7930747270584106,
      "learning_rate": 2.7460626776986233e-06,
      "loss": 0.8418,
      "step": 27974
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.0526857376098633,
      "learning_rate": 2.745423761300834e-06,
      "loss": 0.8351,
      "step": 27975
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5417922735214233,
      "learning_rate": 2.7447848449030447e-06,
      "loss": 0.7962,
      "step": 27976
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.679589867591858,
      "learning_rate": 2.744145928505255e-06,
      "loss": 0.8489,
      "step": 27977
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7871309518814087,
      "learning_rate": 2.7435070121074657e-06,
      "loss": 0.5073,
      "step": 27978
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9192143678665161,
      "learning_rate": 2.7428680957096764e-06,
      "loss": 0.7473,
      "step": 27979
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.636125922203064,
      "learning_rate": 2.742229179311887e-06,
      "loss": 0.8244,
      "step": 27980
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3068369626998901,
      "learning_rate": 2.741590262914098e-06,
      "loss": 0.3949,
      "step": 27981
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.011136770248413,
      "learning_rate": 2.7409513465163085e-06,
      "loss": 0.7165,
      "step": 27982
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.868882656097412,
      "learning_rate": 2.740312430118519e-06,
      "loss": 1.0468,
      "step": 27983
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6102416515350342,
      "learning_rate": 2.73967351372073e-06,
      "loss": 0.8145,
      "step": 27984
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6512333154678345,
      "learning_rate": 2.7390345973229406e-06,
      "loss": 0.7985,
      "step": 27985
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.928502082824707,
      "learning_rate": 2.738395680925151e-06,
      "loss": 0.9889,
      "step": 27986
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.01096248626709,
      "learning_rate": 2.7377567645273616e-06,
      "loss": 0.7894,
      "step": 27987
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7728548049926758,
      "learning_rate": 2.7371178481295723e-06,
      "loss": 0.8809,
      "step": 27988
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.7260982990264893,
      "learning_rate": 2.736478931731783e-06,
      "loss": 1.0452,
      "step": 27989
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.301708459854126,
      "learning_rate": 2.7358400153339937e-06,
      "loss": 0.4041,
      "step": 27990
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.945881962776184,
      "learning_rate": 2.7352010989362044e-06,
      "loss": 0.8298,
      "step": 27991
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.091489791870117,
      "learning_rate": 2.734562182538415e-06,
      "loss": 0.8339,
      "step": 27992
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6881917715072632,
      "learning_rate": 2.7339232661406258e-06,
      "loss": 0.8641,
      "step": 27993
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.967246174812317,
      "learning_rate": 2.7332843497428365e-06,
      "loss": 0.8191,
      "step": 27994
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.839366436004639,
      "learning_rate": 2.7326454333450468e-06,
      "loss": 1.1798,
      "step": 27995
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.652778387069702,
      "learning_rate": 2.7320065169472575e-06,
      "loss": 0.759,
      "step": 27996
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.3773579597473145,
      "learning_rate": 2.731367600549468e-06,
      "loss": 1.0435,
      "step": 27997
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9777579307556152,
      "learning_rate": 2.730728684151679e-06,
      "loss": 0.8594,
      "step": 27998
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6535545587539673,
      "learning_rate": 2.7300897677538896e-06,
      "loss": 0.8371,
      "step": 27999
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.955227017402649,
      "learning_rate": 2.7294508513561003e-06,
      "loss": 0.8159,
      "step": 28000
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.381869077682495,
      "learning_rate": 2.728811934958311e-06,
      "loss": 0.8393,
      "step": 28001
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.7275846004486084,
      "learning_rate": 2.7281730185605217e-06,
      "loss": 0.7492,
      "step": 28002
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9508249759674072,
      "learning_rate": 2.7275341021627324e-06,
      "loss": 0.7635,
      "step": 28003
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5268362760543823,
      "learning_rate": 2.7268951857649426e-06,
      "loss": 0.8352,
      "step": 28004
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9665478467941284,
      "learning_rate": 2.7262562693671533e-06,
      "loss": 0.7736,
      "step": 28005
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.030876636505127,
      "learning_rate": 2.725617352969364e-06,
      "loss": 0.8064,
      "step": 28006
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.8164079189300537,
      "learning_rate": 2.7249784365715747e-06,
      "loss": 0.6176,
      "step": 28007
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.6869986057281494,
      "learning_rate": 2.7243395201737854e-06,
      "loss": 0.9772,
      "step": 28008
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0190579891204834,
      "learning_rate": 2.723700603775996e-06,
      "loss": 0.8688,
      "step": 28009
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.362602472305298,
      "learning_rate": 2.723061687378207e-06,
      "loss": 0.8764,
      "step": 28010
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8868252038955688,
      "learning_rate": 2.7224227709804175e-06,
      "loss": 0.8354,
      "step": 28011
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6308475732803345,
      "learning_rate": 2.7217838545826282e-06,
      "loss": 0.7649,
      "step": 28012
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.997384548187256,
      "learning_rate": 2.7211449381848385e-06,
      "loss": 1.6597,
      "step": 28013
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9519823789596558,
      "learning_rate": 2.7205060217870492e-06,
      "loss": 0.7916,
      "step": 28014
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5276668071746826,
      "learning_rate": 2.71986710538926e-06,
      "loss": 0.8486,
      "step": 28015
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.33019757270813,
      "learning_rate": 2.7192281889914706e-06,
      "loss": 0.9812,
      "step": 28016
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.173810005187988,
      "learning_rate": 2.7185892725936813e-06,
      "loss": 1.3605,
      "step": 28017
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.568938732147217,
      "learning_rate": 2.717950356195892e-06,
      "loss": 1.0634,
      "step": 28018
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.587888240814209,
      "learning_rate": 2.7173114397981027e-06,
      "loss": 0.783,
      "step": 28019
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8763223886489868,
      "learning_rate": 2.7166725234003134e-06,
      "loss": 0.8265,
      "step": 28020
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1610071659088135,
      "learning_rate": 2.716033607002524e-06,
      "loss": 0.9434,
      "step": 28021
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.5386648178100586,
      "learning_rate": 2.7153946906047344e-06,
      "loss": 0.979,
      "step": 28022
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3529183864593506,
      "learning_rate": 2.714755774206945e-06,
      "loss": 0.3871,
      "step": 28023
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2174229621887207,
      "learning_rate": 2.714116857809156e-06,
      "loss": 0.8072,
      "step": 28024
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5942234992980957,
      "learning_rate": 2.7134779414113665e-06,
      "loss": 0.8504,
      "step": 28025
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.37495756149292,
      "learning_rate": 2.712839025013577e-06,
      "loss": 0.7562,
      "step": 28026
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.869962215423584,
      "learning_rate": 2.712200108615788e-06,
      "loss": 0.812,
      "step": 28027
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.438318967819214,
      "learning_rate": 2.7115611922179986e-06,
      "loss": 0.9825,
      "step": 28028
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6379278898239136,
      "learning_rate": 2.7109222758202093e-06,
      "loss": 0.7334,
      "step": 28029
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1126809120178223,
      "learning_rate": 2.71028335942242e-06,
      "loss": 1.1581,
      "step": 28030
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1634247303009033,
      "learning_rate": 2.7096444430246303e-06,
      "loss": 0.9876,
      "step": 28031
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9109266996383667,
      "learning_rate": 2.709005526626841e-06,
      "loss": 0.8231,
      "step": 28032
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8253886699676514,
      "learning_rate": 2.7083666102290517e-06,
      "loss": 0.8312,
      "step": 28033
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7610644102096558,
      "learning_rate": 2.7077276938312624e-06,
      "loss": 1.0007,
      "step": 28034
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.108814239501953,
      "learning_rate": 2.707088777433473e-06,
      "loss": 0.7002,
      "step": 28035
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1777963638305664,
      "learning_rate": 2.7064498610356838e-06,
      "loss": 0.679,
      "step": 28036
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9530301094055176,
      "learning_rate": 2.7058109446378945e-06,
      "loss": 1.1133,
      "step": 28037
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.78499436378479,
      "learning_rate": 2.705172028240105e-06,
      "loss": 1.1655,
      "step": 28038
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.408111095428467,
      "learning_rate": 2.704533111842316e-06,
      "loss": 1.2239,
      "step": 28039
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9931087493896484,
      "learning_rate": 2.7038941954445266e-06,
      "loss": 0.8904,
      "step": 28040
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7893215417861938,
      "learning_rate": 2.703255279046737e-06,
      "loss": 0.8216,
      "step": 28041
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0794873237609863,
      "learning_rate": 2.7026163626489475e-06,
      "loss": 0.894,
      "step": 28042
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9772260189056396,
      "learning_rate": 2.7019774462511582e-06,
      "loss": 0.8435,
      "step": 28043
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.320768356323242,
      "learning_rate": 2.701338529853369e-06,
      "loss": 0.7227,
      "step": 28044
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0456724166870117,
      "learning_rate": 2.7006996134555796e-06,
      "loss": 0.8729,
      "step": 28045
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6593587398529053,
      "learning_rate": 2.7000606970577903e-06,
      "loss": 0.775,
      "step": 28046
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.879986047744751,
      "learning_rate": 2.699421780660001e-06,
      "loss": 1.1582,
      "step": 28047
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.211677312850952,
      "learning_rate": 2.6987828642622117e-06,
      "loss": 0.8275,
      "step": 28048
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.558109998703003,
      "learning_rate": 2.6981439478644224e-06,
      "loss": 0.8607,
      "step": 28049
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9611390829086304,
      "learning_rate": 2.6975050314666327e-06,
      "loss": 0.7128,
      "step": 28050
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.175896406173706,
      "learning_rate": 2.6968661150688434e-06,
      "loss": 0.8701,
      "step": 28051
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6503394842147827,
      "learning_rate": 2.696227198671054e-06,
      "loss": 0.7791,
      "step": 28052
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.592165470123291,
      "learning_rate": 2.695588282273265e-06,
      "loss": 0.7908,
      "step": 28053
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5009236335754395,
      "learning_rate": 2.6949493658754755e-06,
      "loss": 0.8772,
      "step": 28054
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5024546384811401,
      "learning_rate": 2.6943104494776862e-06,
      "loss": 0.3577,
      "step": 28055
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7162290811538696,
      "learning_rate": 2.693671533079897e-06,
      "loss": 0.8178,
      "step": 28056
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1770718097686768,
      "learning_rate": 2.6930326166821076e-06,
      "loss": 0.5512,
      "step": 28057
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.269822597503662,
      "learning_rate": 2.6923937002843183e-06,
      "loss": 1.3327,
      "step": 28058
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0056514739990234,
      "learning_rate": 2.6917547838865286e-06,
      "loss": 0.8092,
      "step": 28059
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9120275974273682,
      "learning_rate": 2.6911158674887393e-06,
      "loss": 0.8092,
      "step": 28060
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3408441543579102,
      "learning_rate": 2.69047695109095e-06,
      "loss": 0.4028,
      "step": 28061
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.154972553253174,
      "learning_rate": 2.6898380346931607e-06,
      "loss": 0.7655,
      "step": 28062
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.7030718326568604,
      "learning_rate": 2.6891991182953714e-06,
      "loss": 1.1058,
      "step": 28063
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7792105674743652,
      "learning_rate": 2.688560201897582e-06,
      "loss": 0.4969,
      "step": 28064
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.016765832901001,
      "learning_rate": 2.687921285499793e-06,
      "loss": 1.0439,
      "step": 28065
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.5743303298950195,
      "learning_rate": 2.6872823691020035e-06,
      "loss": 0.641,
      "step": 28066
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6128603219985962,
      "learning_rate": 2.686643452704214e-06,
      "loss": 0.831,
      "step": 28067
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.6909122467041016,
      "learning_rate": 2.6860045363064245e-06,
      "loss": 0.6963,
      "step": 28068
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1581413745880127,
      "learning_rate": 2.685365619908635e-06,
      "loss": 0.7918,
      "step": 28069
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.6444084644317627,
      "learning_rate": 2.684726703510846e-06,
      "loss": 0.5716,
      "step": 28070
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6681561470031738,
      "learning_rate": 2.6840877871130566e-06,
      "loss": 0.8162,
      "step": 28071
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7944116592407227,
      "learning_rate": 2.6834488707152673e-06,
      "loss": 0.8089,
      "step": 28072
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0398263931274414,
      "learning_rate": 2.682809954317478e-06,
      "loss": 0.8392,
      "step": 28073
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6988880634307861,
      "learning_rate": 2.6821710379196887e-06,
      "loss": 0.776,
      "step": 28074
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.331813812255859,
      "learning_rate": 2.6815321215218994e-06,
      "loss": 1.3178,
      "step": 28075
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2064534425735474,
      "learning_rate": 2.68089320512411e-06,
      "loss": 0.4015,
      "step": 28076
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.155618667602539,
      "learning_rate": 2.6802542887263203e-06,
      "loss": 0.7868,
      "step": 28077
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.914860486984253,
      "learning_rate": 2.679615372328531e-06,
      "loss": 0.8235,
      "step": 28078
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.841554880142212,
      "learning_rate": 2.6789764559307417e-06,
      "loss": 0.7845,
      "step": 28079
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.30722188949585,
      "learning_rate": 2.6783375395329524e-06,
      "loss": 0.8904,
      "step": 28080
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7154052257537842,
      "learning_rate": 2.677698623135163e-06,
      "loss": 0.7546,
      "step": 28081
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.942136287689209,
      "learning_rate": 2.677059706737374e-06,
      "loss": 0.8475,
      "step": 28082
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3022260665893555,
      "learning_rate": 2.6764207903395845e-06,
      "loss": 0.9337,
      "step": 28083
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7574222087860107,
      "learning_rate": 2.6757818739417952e-06,
      "loss": 0.824,
      "step": 28084
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.656411051750183,
      "learning_rate": 2.675142957544006e-06,
      "loss": 0.8552,
      "step": 28085
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.2429680824279785,
      "learning_rate": 2.6745040411462162e-06,
      "loss": 0.8412,
      "step": 28086
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.377688407897949,
      "learning_rate": 2.673865124748427e-06,
      "loss": 0.6681,
      "step": 28087
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7535470724105835,
      "learning_rate": 2.6732262083506376e-06,
      "loss": 0.7424,
      "step": 28088
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.296781063079834,
      "learning_rate": 2.6725872919528483e-06,
      "loss": 0.3811,
      "step": 28089
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3608078956604004,
      "learning_rate": 2.671948375555059e-06,
      "loss": 0.9341,
      "step": 28090
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6499478816986084,
      "learning_rate": 2.6713094591572697e-06,
      "loss": 0.4434,
      "step": 28091
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2519941329956055,
      "learning_rate": 2.6706705427594804e-06,
      "loss": 0.7082,
      "step": 28092
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7893799543380737,
      "learning_rate": 2.670031626361691e-06,
      "loss": 0.8757,
      "step": 28093
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9929717779159546,
      "learning_rate": 2.669392709963902e-06,
      "loss": 1.0003,
      "step": 28094
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8546345233917236,
      "learning_rate": 2.668753793566112e-06,
      "loss": 0.7795,
      "step": 28095
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7707964181900024,
      "learning_rate": 2.668114877168323e-06,
      "loss": 0.8172,
      "step": 28096
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.105156183242798,
      "learning_rate": 2.6674759607705335e-06,
      "loss": 0.7953,
      "step": 28097
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6537681818008423,
      "learning_rate": 2.666837044372744e-06,
      "loss": 0.461,
      "step": 28098
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0564980506896973,
      "learning_rate": 2.666198127974955e-06,
      "loss": 0.9948,
      "step": 28099
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.210933208465576,
      "learning_rate": 2.6655592115771656e-06,
      "loss": 0.711,
      "step": 28100
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6923545598983765,
      "learning_rate": 2.6649202951793763e-06,
      "loss": 0.833,
      "step": 28101
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.002889633178711,
      "learning_rate": 2.664281378781587e-06,
      "loss": 0.9305,
      "step": 28102
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8748369216918945,
      "learning_rate": 2.6636424623837977e-06,
      "loss": 0.8422,
      "step": 28103
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8172458410263062,
      "learning_rate": 2.663003545986008e-06,
      "loss": 0.8607,
      "step": 28104
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9132088422775269,
      "learning_rate": 2.6623646295882187e-06,
      "loss": 1.0678,
      "step": 28105
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.448493003845215,
      "learning_rate": 2.6617257131904294e-06,
      "loss": 1.1575,
      "step": 28106
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.551053762435913,
      "learning_rate": 2.66108679679264e-06,
      "loss": 0.944,
      "step": 28107
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2151544094085693,
      "learning_rate": 2.6604478803948508e-06,
      "loss": 1.3189,
      "step": 28108
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.011348247528076,
      "learning_rate": 2.6598089639970615e-06,
      "loss": 0.838,
      "step": 28109
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3444039821624756,
      "learning_rate": 2.659170047599272e-06,
      "loss": 0.8488,
      "step": 28110
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8131924867630005,
      "learning_rate": 2.658531131201483e-06,
      "loss": 0.8789,
      "step": 28111
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.2929346561431885,
      "learning_rate": 2.6578922148036936e-06,
      "loss": 0.7159,
      "step": 28112
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.5608034133911133,
      "learning_rate": 2.6572532984059043e-06,
      "loss": 0.6681,
      "step": 28113
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5756803750991821,
      "learning_rate": 2.6566143820081146e-06,
      "loss": 0.7455,
      "step": 28114
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6695196628570557,
      "learning_rate": 2.6559754656103253e-06,
      "loss": 0.853,
      "step": 28115
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9888361692428589,
      "learning_rate": 2.655336549212536e-06,
      "loss": 0.8831,
      "step": 28116
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5277841091156006,
      "learning_rate": 2.6546976328147467e-06,
      "loss": 0.8135,
      "step": 28117
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6489851474761963,
      "learning_rate": 2.654058716416957e-06,
      "loss": 0.7098,
      "step": 28118
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3507914543151855,
      "learning_rate": 2.6534198000191676e-06,
      "loss": 0.8546,
      "step": 28119
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.6511032581329346,
      "learning_rate": 2.652780883621378e-06,
      "loss": 0.9026,
      "step": 28120
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.014018535614014,
      "learning_rate": 2.6521419672235886e-06,
      "loss": 0.9406,
      "step": 28121
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.894416332244873,
      "learning_rate": 2.6515030508257993e-06,
      "loss": 0.9402,
      "step": 28122
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5622817277908325,
      "learning_rate": 2.65086413442801e-06,
      "loss": 0.8211,
      "step": 28123
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7230957746505737,
      "learning_rate": 2.6502252180302207e-06,
      "loss": 0.8886,
      "step": 28124
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6440750360488892,
      "learning_rate": 2.6495863016324314e-06,
      "loss": 0.7768,
      "step": 28125
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.112917900085449,
      "learning_rate": 2.648947385234642e-06,
      "loss": 0.7489,
      "step": 28126
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6103770732879639,
      "learning_rate": 2.648308468836853e-06,
      "loss": 0.7218,
      "step": 28127
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.4118618965148926,
      "learning_rate": 2.6476695524390635e-06,
      "loss": 0.7985,
      "step": 28128
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.998581647872925,
      "learning_rate": 2.6470306360412738e-06,
      "loss": 0.5077,
      "step": 28129
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.603583812713623,
      "learning_rate": 2.6463917196434845e-06,
      "loss": 1.1414,
      "step": 28130
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.396059274673462,
      "learning_rate": 2.645752803245695e-06,
      "loss": 0.6052,
      "step": 28131
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.26132869720459,
      "learning_rate": 2.645113886847906e-06,
      "loss": 1.0149,
      "step": 28132
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.1704342365264893,
      "learning_rate": 2.6444749704501166e-06,
      "loss": 1.0589,
      "step": 28133
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.969598412513733,
      "learning_rate": 2.6438360540523273e-06,
      "loss": 0.8912,
      "step": 28134
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8358668088912964,
      "learning_rate": 2.643197137654538e-06,
      "loss": 0.7072,
      "step": 28135
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.976627230644226,
      "learning_rate": 2.6425582212567487e-06,
      "loss": 0.9832,
      "step": 28136
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.1579508781433105,
      "learning_rate": 2.6419193048589594e-06,
      "loss": 1.2379,
      "step": 28137
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.0127809047698975,
      "learning_rate": 2.6412803884611697e-06,
      "loss": 1.1848,
      "step": 28138
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4011688232421875,
      "learning_rate": 2.6406414720633804e-06,
      "loss": 0.9377,
      "step": 28139
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7787069082260132,
      "learning_rate": 2.640002555665591e-06,
      "loss": 0.8361,
      "step": 28140
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1890013217926025,
      "learning_rate": 2.6393636392678018e-06,
      "loss": 0.8034,
      "step": 28141
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.5391149520874023,
      "learning_rate": 2.6387247228700125e-06,
      "loss": 0.4717,
      "step": 28142
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6344687938690186,
      "learning_rate": 2.638085806472223e-06,
      "loss": 0.8517,
      "step": 28143
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7817779779434204,
      "learning_rate": 2.637446890074434e-06,
      "loss": 0.7242,
      "step": 28144
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9428569078445435,
      "learning_rate": 2.6368079736766446e-06,
      "loss": 0.8661,
      "step": 28145
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4768519401550293,
      "learning_rate": 2.6361690572788553e-06,
      "loss": 0.8039,
      "step": 28146
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8058032989501953,
      "learning_rate": 2.6355301408810655e-06,
      "loss": 0.9103,
      "step": 28147
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0300920009613037,
      "learning_rate": 2.6348912244832762e-06,
      "loss": 0.7092,
      "step": 28148
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.5420939922332764,
      "learning_rate": 2.634252308085487e-06,
      "loss": 0.6646,
      "step": 28149
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3104455471038818,
      "learning_rate": 2.6336133916876976e-06,
      "loss": 0.3711,
      "step": 28150
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7937685251235962,
      "learning_rate": 2.6329744752899083e-06,
      "loss": 0.8106,
      "step": 28151
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5261393785476685,
      "learning_rate": 2.632335558892119e-06,
      "loss": 0.4294,
      "step": 28152
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7119864225387573,
      "learning_rate": 2.6316966424943297e-06,
      "loss": 0.8821,
      "step": 28153
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7264126539230347,
      "learning_rate": 2.6310577260965404e-06,
      "loss": 0.7833,
      "step": 28154
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9220659732818604,
      "learning_rate": 2.630418809698751e-06,
      "loss": 0.8259,
      "step": 28155
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.189045429229736,
      "learning_rate": 2.6297798933009614e-06,
      "loss": 1.3114,
      "step": 28156
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.321870803833008,
      "learning_rate": 2.629140976903172e-06,
      "loss": 1.5914,
      "step": 28157
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6013377904891968,
      "learning_rate": 2.628502060505383e-06,
      "loss": 0.8108,
      "step": 28158
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.91369891166687,
      "learning_rate": 2.6278631441075935e-06,
      "loss": 1.3014,
      "step": 28159
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9951221942901611,
      "learning_rate": 2.627224227709804e-06,
      "loss": 0.7878,
      "step": 28160
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0067830085754395,
      "learning_rate": 2.626585311312015e-06,
      "loss": 0.8516,
      "step": 28161
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0846803188323975,
      "learning_rate": 2.6259463949142256e-06,
      "loss": 0.8713,
      "step": 28162
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2380077838897705,
      "learning_rate": 2.6253074785164363e-06,
      "loss": 0.8341,
      "step": 28163
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6903871297836304,
      "learning_rate": 2.624668562118647e-06,
      "loss": 0.8222,
      "step": 28164
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4407635927200317,
      "learning_rate": 2.6240296457208577e-06,
      "loss": 0.853,
      "step": 28165
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.6159727573394775,
      "learning_rate": 2.623390729323068e-06,
      "loss": 0.9069,
      "step": 28166
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.546661615371704,
      "learning_rate": 2.6227518129252787e-06,
      "loss": 0.5899,
      "step": 28167
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.6356089115142822,
      "learning_rate": 2.6221128965274894e-06,
      "loss": 1.0005,
      "step": 28168
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.279202938079834,
      "learning_rate": 2.6214739801297e-06,
      "loss": 0.944,
      "step": 28169
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6853723526000977,
      "learning_rate": 2.6208350637319108e-06,
      "loss": 0.6977,
      "step": 28170
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6610418558120728,
      "learning_rate": 2.6201961473341215e-06,
      "loss": 0.9095,
      "step": 28171
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.025367259979248,
      "learning_rate": 2.619557230936332e-06,
      "loss": 0.8669,
      "step": 28172
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6478772163391113,
      "learning_rate": 2.618918314538543e-06,
      "loss": 0.8932,
      "step": 28173
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8505383729934692,
      "learning_rate": 2.6182793981407536e-06,
      "loss": 0.8108,
      "step": 28174
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6503963470458984,
      "learning_rate": 2.617640481742964e-06,
      "loss": 0.8177,
      "step": 28175
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.641090989112854,
      "learning_rate": 2.6170015653451746e-06,
      "loss": 0.8092,
      "step": 28176
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6861361265182495,
      "learning_rate": 2.6163626489473853e-06,
      "loss": 0.7753,
      "step": 28177
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.53034508228302,
      "learning_rate": 2.615723732549596e-06,
      "loss": 0.7888,
      "step": 28178
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9700793027877808,
      "learning_rate": 2.6150848161518067e-06,
      "loss": 0.8945,
      "step": 28179
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.079042911529541,
      "learning_rate": 2.6144458997540174e-06,
      "loss": 0.8203,
      "step": 28180
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1244747638702393,
      "learning_rate": 2.613806983356228e-06,
      "loss": 0.7755,
      "step": 28181
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9760187864303589,
      "learning_rate": 2.6131680669584388e-06,
      "loss": 0.767,
      "step": 28182
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7243725061416626,
      "learning_rate": 2.6125291505606495e-06,
      "loss": 0.8465,
      "step": 28183
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.9463071823120117,
      "learning_rate": 2.6118902341628597e-06,
      "loss": 1.1222,
      "step": 28184
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7476799488067627,
      "learning_rate": 2.6112513177650704e-06,
      "loss": 0.9025,
      "step": 28185
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2242591381073,
      "learning_rate": 2.610612401367281e-06,
      "loss": 0.8089,
      "step": 28186
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6113307476043701,
      "learning_rate": 2.609973484969492e-06,
      "loss": 0.7914,
      "step": 28187
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.594858169555664,
      "learning_rate": 2.6093345685717025e-06,
      "loss": 0.7393,
      "step": 28188
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.022819995880127,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.8159,
      "step": 28189
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9879692792892456,
      "learning_rate": 2.608056735776124e-06,
      "loss": 0.776,
      "step": 28190
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0939629077911377,
      "learning_rate": 2.6074178193783346e-06,
      "loss": 1.1609,
      "step": 28191
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7292524576187134,
      "learning_rate": 2.6067789029805453e-06,
      "loss": 0.7445,
      "step": 28192
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.450850486755371,
      "learning_rate": 2.6061399865827556e-06,
      "loss": 0.5569,
      "step": 28193
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9025828838348389,
      "learning_rate": 2.6055010701849663e-06,
      "loss": 0.7773,
      "step": 28194
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7340790033340454,
      "learning_rate": 2.604862153787177e-06,
      "loss": 0.8374,
      "step": 28195
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4914332628250122,
      "learning_rate": 2.6042232373893877e-06,
      "loss": 0.7931,
      "step": 28196
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0764517784118652,
      "learning_rate": 2.6035843209915984e-06,
      "loss": 0.7977,
      "step": 28197
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7865241765975952,
      "learning_rate": 2.602945404593809e-06,
      "loss": 0.7959,
      "step": 28198
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6043850183486938,
      "learning_rate": 2.60230648819602e-06,
      "loss": 0.7814,
      "step": 28199
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.285475730895996,
      "learning_rate": 2.6016675717982305e-06,
      "loss": 0.8965,
      "step": 28200
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.794357180595398,
      "learning_rate": 2.6010286554004412e-06,
      "loss": 0.7335,
      "step": 28201
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4867727756500244,
      "learning_rate": 2.6003897390026515e-06,
      "loss": 0.4292,
      "step": 28202
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7130069732666016,
      "learning_rate": 2.599750822604862e-06,
      "loss": 0.7357,
      "step": 28203
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.339999198913574,
      "learning_rate": 2.599111906207073e-06,
      "loss": 0.8264,
      "step": 28204
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0086305141448975,
      "learning_rate": 2.5984729898092836e-06,
      "loss": 0.6683,
      "step": 28205
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4076725244522095,
      "learning_rate": 2.5978340734114943e-06,
      "loss": 0.3958,
      "step": 28206
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5469427108764648,
      "learning_rate": 2.597195157013705e-06,
      "loss": 0.4684,
      "step": 28207
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5148708820343018,
      "learning_rate": 2.5965562406159157e-06,
      "loss": 0.8504,
      "step": 28208
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.250568389892578,
      "learning_rate": 2.5959173242181264e-06,
      "loss": 0.561,
      "step": 28209
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.797039270401001,
      "learning_rate": 2.595278407820337e-06,
      "loss": 0.7204,
      "step": 28210
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0634095668792725,
      "learning_rate": 2.5946394914225474e-06,
      "loss": 0.8312,
      "step": 28211
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.8928706645965576,
      "learning_rate": 2.594000575024758e-06,
      "loss": 0.961,
      "step": 28212
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.4771766662597656,
      "learning_rate": 2.5933616586269688e-06,
      "loss": 1.0929,
      "step": 28213
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7903220653533936,
      "learning_rate": 2.5927227422291795e-06,
      "loss": 0.6448,
      "step": 28214
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.119741916656494,
      "learning_rate": 2.59208382583139e-06,
      "loss": 0.9398,
      "step": 28215
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.229238986968994,
      "learning_rate": 2.591444909433601e-06,
      "loss": 0.814,
      "step": 28216
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.2157528400421143,
      "learning_rate": 2.5908059930358116e-06,
      "loss": 0.9283,
      "step": 28217
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.9241397380828857,
      "learning_rate": 2.5901670766380223e-06,
      "loss": 0.6987,
      "step": 28218
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5715065002441406,
      "learning_rate": 2.589528160240233e-06,
      "loss": 0.8131,
      "step": 28219
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.007702112197876,
      "learning_rate": 2.5888892438424432e-06,
      "loss": 0.8063,
      "step": 28220
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.366157054901123,
      "learning_rate": 2.588250327444654e-06,
      "loss": 1.0439,
      "step": 28221
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.8537216186523438,
      "learning_rate": 2.5876114110468646e-06,
      "loss": 1.1372,
      "step": 28222
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8186568021774292,
      "learning_rate": 2.5869724946490753e-06,
      "loss": 0.8816,
      "step": 28223
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5873568058013916,
      "learning_rate": 2.586333578251286e-06,
      "loss": 0.7706,
      "step": 28224
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9789271354675293,
      "learning_rate": 2.5856946618534967e-06,
      "loss": 1.2395,
      "step": 28225
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.57586669921875,
      "learning_rate": 2.5850557454557074e-06,
      "loss": 1.248,
      "step": 28226
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9333173036575317,
      "learning_rate": 2.584416829057918e-06,
      "loss": 0.7657,
      "step": 28227
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.556011438369751,
      "learning_rate": 2.583777912660129e-06,
      "loss": 0.8537,
      "step": 28228
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.3511393070220947,
      "learning_rate": 2.583138996262339e-06,
      "loss": 0.9909,
      "step": 28229
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.5748131275177,
      "learning_rate": 2.58250007986455e-06,
      "loss": 0.8587,
      "step": 28230
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.079230308532715,
      "learning_rate": 2.5818611634667605e-06,
      "loss": 0.8298,
      "step": 28231
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7020957469940186,
      "learning_rate": 2.5812222470689712e-06,
      "loss": 0.7935,
      "step": 28232
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.767091155052185,
      "learning_rate": 2.580583330671182e-06,
      "loss": 0.8917,
      "step": 28233
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8130384683609009,
      "learning_rate": 2.5799444142733926e-06,
      "loss": 0.7561,
      "step": 28234
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.100152015686035,
      "learning_rate": 2.5793054978756033e-06,
      "loss": 1.1596,
      "step": 28235
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7959548234939575,
      "learning_rate": 2.578666581477814e-06,
      "loss": 0.8428,
      "step": 28236
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.101567029953003,
      "learning_rate": 2.5780276650800247e-06,
      "loss": 0.8437,
      "step": 28237
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2730488777160645,
      "learning_rate": 2.5773887486822354e-06,
      "loss": 1.2054,
      "step": 28238
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.151129961013794,
      "learning_rate": 2.5767498322844457e-06,
      "loss": 0.707,
      "step": 28239
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8840084075927734,
      "learning_rate": 2.5761109158866564e-06,
      "loss": 0.9656,
      "step": 28240
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.918488502502441,
      "learning_rate": 2.575471999488867e-06,
      "loss": 1.1104,
      "step": 28241
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.310204029083252,
      "learning_rate": 2.574833083091078e-06,
      "loss": 1.1254,
      "step": 28242
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7502477169036865,
      "learning_rate": 2.5741941666932885e-06,
      "loss": 1.0585,
      "step": 28243
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4956307411193848,
      "learning_rate": 2.573555250295499e-06,
      "loss": 0.9099,
      "step": 28244
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8432680368423462,
      "learning_rate": 2.57291633389771e-06,
      "loss": 0.8434,
      "step": 28245
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1543591022491455,
      "learning_rate": 2.5722774174999206e-06,
      "loss": 0.8387,
      "step": 28246
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8679195642471313,
      "learning_rate": 2.5716385011021313e-06,
      "loss": 0.9262,
      "step": 28247
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.546815276145935,
      "learning_rate": 2.5709995847043416e-06,
      "loss": 0.7658,
      "step": 28248
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7078818082809448,
      "learning_rate": 2.5703606683065523e-06,
      "loss": 0.7659,
      "step": 28249
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.673527717590332,
      "learning_rate": 2.569721751908763e-06,
      "loss": 0.8313,
      "step": 28250
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.337376594543457,
      "learning_rate": 2.5690828355109737e-06,
      "loss": 1.0481,
      "step": 28251
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6408729553222656,
      "learning_rate": 2.5684439191131844e-06,
      "loss": 0.8167,
      "step": 28252
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.938692569732666,
      "learning_rate": 2.567805002715395e-06,
      "loss": 0.7776,
      "step": 28253
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.3805150985717773,
      "learning_rate": 2.5671660863176058e-06,
      "loss": 0.9029,
      "step": 28254
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.2537059783935547,
      "learning_rate": 2.5665271699198165e-06,
      "loss": 0.7557,
      "step": 28255
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8340895175933838,
      "learning_rate": 2.565888253522027e-06,
      "loss": 0.851,
      "step": 28256
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.9582676887512207,
      "learning_rate": 2.5652493371242374e-06,
      "loss": 0.8171,
      "step": 28257
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5940914154052734,
      "learning_rate": 2.564610420726448e-06,
      "loss": 0.8959,
      "step": 28258
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8241941928863525,
      "learning_rate": 2.563971504328659e-06,
      "loss": 1.0404,
      "step": 28259
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2782280445098877,
      "learning_rate": 2.5633325879308695e-06,
      "loss": 0.9558,
      "step": 28260
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6969916820526123,
      "learning_rate": 2.5626936715330802e-06,
      "loss": 0.8375,
      "step": 28261
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7929960489273071,
      "learning_rate": 2.562054755135291e-06,
      "loss": 0.5782,
      "step": 28262
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.383624792098999,
      "learning_rate": 2.5614158387375016e-06,
      "loss": 0.5027,
      "step": 28263
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.035201072692871,
      "learning_rate": 2.5607769223397123e-06,
      "loss": 0.8346,
      "step": 28264
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1805038452148438,
      "learning_rate": 2.560138005941923e-06,
      "loss": 0.8615,
      "step": 28265
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.117147207260132,
      "learning_rate": 2.5594990895441333e-06,
      "loss": 0.8861,
      "step": 28266
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7058384418487549,
      "learning_rate": 2.558860173146344e-06,
      "loss": 0.7874,
      "step": 28267
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.105776786804199,
      "learning_rate": 2.5582212567485547e-06,
      "loss": 0.8551,
      "step": 28268
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.167957067489624,
      "learning_rate": 2.5575823403507654e-06,
      "loss": 0.9056,
      "step": 28269
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7288035154342651,
      "learning_rate": 2.556943423952976e-06,
      "loss": 0.7776,
      "step": 28270
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7370121479034424,
      "learning_rate": 2.556304507555187e-06,
      "loss": 0.7378,
      "step": 28271
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.982907772064209,
      "learning_rate": 2.5556655911573975e-06,
      "loss": 1.2158,
      "step": 28272
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7389832735061646,
      "learning_rate": 2.5550266747596082e-06,
      "loss": 0.684,
      "step": 28273
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8771029710769653,
      "learning_rate": 2.554387758361819e-06,
      "loss": 0.7682,
      "step": 28274
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6647427082061768,
      "learning_rate": 2.553748841964029e-06,
      "loss": 0.8852,
      "step": 28275
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4404487609863281,
      "learning_rate": 2.55310992556624e-06,
      "loss": 1.1516,
      "step": 28276
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.410911560058594,
      "learning_rate": 2.5524710091684506e-06,
      "loss": 1.0515,
      "step": 28277
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0165510177612305,
      "learning_rate": 2.5518320927706613e-06,
      "loss": 1.1508,
      "step": 28278
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.345155715942383,
      "learning_rate": 2.551193176372872e-06,
      "loss": 0.5625,
      "step": 28279
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9345688819885254,
      "learning_rate": 2.5505542599750827e-06,
      "loss": 0.785,
      "step": 28280
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.046050548553467,
      "learning_rate": 2.5499153435772934e-06,
      "loss": 0.7554,
      "step": 28281
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6690177917480469,
      "learning_rate": 2.549276427179504e-06,
      "loss": 0.9229,
      "step": 28282
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.712297797203064,
      "learning_rate": 2.548637510781715e-06,
      "loss": 0.9466,
      "step": 28283
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9934805631637573,
      "learning_rate": 2.547998594383925e-06,
      "loss": 0.7899,
      "step": 28284
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1071383953094482,
      "learning_rate": 2.5473596779861358e-06,
      "loss": 0.9623,
      "step": 28285
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9094542264938354,
      "learning_rate": 2.5467207615883465e-06,
      "loss": 0.6802,
      "step": 28286
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8265886306762695,
      "learning_rate": 2.546081845190557e-06,
      "loss": 0.7655,
      "step": 28287
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8504319190979004,
      "learning_rate": 2.545442928792768e-06,
      "loss": 0.7585,
      "step": 28288
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.35699200630188,
      "learning_rate": 2.5448040123949786e-06,
      "loss": 0.8372,
      "step": 28289
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.214106321334839,
      "learning_rate": 2.5441650959971893e-06,
      "loss": 0.8357,
      "step": 28290
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2175724506378174,
      "learning_rate": 2.5435261795994e-06,
      "loss": 0.8997,
      "step": 28291
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6469497680664062,
      "learning_rate": 2.5428872632016107e-06,
      "loss": 0.9245,
      "step": 28292
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.49375319480896,
      "learning_rate": 2.542248346803821e-06,
      "loss": 1.1109,
      "step": 28293
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.8669095039367676,
      "learning_rate": 2.5416094304060317e-06,
      "loss": 0.825,
      "step": 28294
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.042954444885254,
      "learning_rate": 2.5409705140082424e-06,
      "loss": 0.8041,
      "step": 28295
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9272767305374146,
      "learning_rate": 2.540331597610453e-06,
      "loss": 0.6448,
      "step": 28296
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7134047746658325,
      "learning_rate": 2.5396926812126638e-06,
      "loss": 0.7416,
      "step": 28297
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.323445796966553,
      "learning_rate": 2.5390537648148745e-06,
      "loss": 1.1008,
      "step": 28298
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6501139402389526,
      "learning_rate": 2.538414848417085e-06,
      "loss": 0.7921,
      "step": 28299
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.195991039276123,
      "learning_rate": 2.537775932019296e-06,
      "loss": 1.2184,
      "step": 28300
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.5933051109313965,
      "learning_rate": 2.5371370156215066e-06,
      "loss": 1.153,
      "step": 28301
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6934858560562134,
      "learning_rate": 2.536498099223717e-06,
      "loss": 0.8076,
      "step": 28302
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7500184774398804,
      "learning_rate": 2.5358591828259275e-06,
      "loss": 0.8779,
      "step": 28303
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.897239327430725,
      "learning_rate": 2.5352202664281382e-06,
      "loss": 0.9176,
      "step": 28304
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2883729934692383,
      "learning_rate": 2.534581350030349e-06,
      "loss": 0.5318,
      "step": 28305
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2170586585998535,
      "learning_rate": 2.5339424336325596e-06,
      "loss": 0.8551,
      "step": 28306
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8000718355178833,
      "learning_rate": 2.5333035172347703e-06,
      "loss": 1.1539,
      "step": 28307
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4816906452178955,
      "learning_rate": 2.532664600836981e-06,
      "loss": 0.7602,
      "step": 28308
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.8198885917663574,
      "learning_rate": 2.5320256844391917e-06,
      "loss": 0.9243,
      "step": 28309
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6710360050201416,
      "learning_rate": 2.5313867680414024e-06,
      "loss": 0.8297,
      "step": 28310
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.709488868713379,
      "learning_rate": 2.5307478516436123e-06,
      "loss": 0.8065,
      "step": 28311
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7607908248901367,
      "learning_rate": 2.530108935245823e-06,
      "loss": 0.8845,
      "step": 28312
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.6148595809936523,
      "learning_rate": 2.5294700188480337e-06,
      "loss": 1.0716,
      "step": 28313
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7059228420257568,
      "learning_rate": 2.5288311024502444e-06,
      "loss": 0.7623,
      "step": 28314
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0550014972686768,
      "learning_rate": 2.528192186052455e-06,
      "loss": 0.8303,
      "step": 28315
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.86453378200531,
      "learning_rate": 2.5275532696546658e-06,
      "loss": 0.8632,
      "step": 28316
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8083645105361938,
      "learning_rate": 2.5269143532568765e-06,
      "loss": 0.8148,
      "step": 28317
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9834917783737183,
      "learning_rate": 2.5262754368590868e-06,
      "loss": 1.1385,
      "step": 28318
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.812695026397705,
      "learning_rate": 2.5256365204612975e-06,
      "loss": 0.778,
      "step": 28319
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1097419261932373,
      "learning_rate": 2.524997604063508e-06,
      "loss": 0.8211,
      "step": 28320
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.622199535369873,
      "learning_rate": 2.524358687665719e-06,
      "loss": 0.7803,
      "step": 28321
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.066255569458008,
      "learning_rate": 2.5237197712679296e-06,
      "loss": 1.0492,
      "step": 28322
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.3966100215911865,
      "learning_rate": 2.5230808548701403e-06,
      "loss": 0.8498,
      "step": 28323
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8502860069274902,
      "learning_rate": 2.522441938472351e-06,
      "loss": 0.9508,
      "step": 28324
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.914165496826172,
      "learning_rate": 2.5218030220745617e-06,
      "loss": 0.8287,
      "step": 28325
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.1035566329956055,
      "learning_rate": 2.5211641056767724e-06,
      "loss": 0.7926,
      "step": 28326
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.148371458053589,
      "learning_rate": 2.5205251892789826e-06,
      "loss": 0.7805,
      "step": 28327
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.223484992980957,
      "learning_rate": 2.5198862728811933e-06,
      "loss": 0.8474,
      "step": 28328
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.0974838733673096,
      "learning_rate": 2.519247356483404e-06,
      "loss": 0.8345,
      "step": 28329
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7602043151855469,
      "learning_rate": 2.5186084400856147e-06,
      "loss": 0.7988,
      "step": 28330
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1535747051239014,
      "learning_rate": 2.5179695236878254e-06,
      "loss": 0.8749,
      "step": 28331
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.8132641315460205,
      "learning_rate": 2.517330607290036e-06,
      "loss": 1.0576,
      "step": 28332
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8637360334396362,
      "learning_rate": 2.516691690892247e-06,
      "loss": 0.7271,
      "step": 28333
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9801408052444458,
      "learning_rate": 2.5160527744944575e-06,
      "loss": 0.7971,
      "step": 28334
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7699902057647705,
      "learning_rate": 2.5154138580966682e-06,
      "loss": 0.842,
      "step": 28335
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7875096797943115,
      "learning_rate": 2.5147749416988785e-06,
      "loss": 0.8681,
      "step": 28336
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.517362117767334,
      "learning_rate": 2.514136025301089e-06,
      "loss": 1.4401,
      "step": 28337
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.003571033477783,
      "learning_rate": 2.5134971089033e-06,
      "loss": 0.8111,
      "step": 28338
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.036417007446289,
      "learning_rate": 2.5128581925055106e-06,
      "loss": 0.7353,
      "step": 28339
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2794601917266846,
      "learning_rate": 2.5122192761077213e-06,
      "loss": 0.7052,
      "step": 28340
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6942991018295288,
      "learning_rate": 2.511580359709932e-06,
      "loss": 0.7835,
      "step": 28341
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8579033613204956,
      "learning_rate": 2.5109414433121427e-06,
      "loss": 0.8629,
      "step": 28342
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.53445565700531,
      "learning_rate": 2.5103025269143534e-06,
      "loss": 0.812,
      "step": 28343
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.6719298362731934,
      "learning_rate": 2.509663610516564e-06,
      "loss": 0.6906,
      "step": 28344
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.134718418121338,
      "learning_rate": 2.5090246941187744e-06,
      "loss": 1.0051,
      "step": 28345
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.119779109954834,
      "learning_rate": 2.508385777720985e-06,
      "loss": 0.4483,
      "step": 28346
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2274731397628784,
      "learning_rate": 2.5077468613231958e-06,
      "loss": 0.4049,
      "step": 28347
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.196049690246582,
      "learning_rate": 2.5071079449254065e-06,
      "loss": 0.9872,
      "step": 28348
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.8369011878967285,
      "learning_rate": 2.506469028527617e-06,
      "loss": 1.3315,
      "step": 28349
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8697727918624878,
      "learning_rate": 2.505830112129828e-06,
      "loss": 0.8521,
      "step": 28350
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.804824948310852,
      "learning_rate": 2.5051911957320386e-06,
      "loss": 0.9599,
      "step": 28351
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.3839738368988037,
      "learning_rate": 2.5045522793342493e-06,
      "loss": 0.9469,
      "step": 28352
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8020739555358887,
      "learning_rate": 2.50391336293646e-06,
      "loss": 0.7497,
      "step": 28353
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.464094161987305,
      "learning_rate": 2.5032744465386703e-06,
      "loss": 1.3622,
      "step": 28354
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7961854934692383,
      "learning_rate": 2.502635530140881e-06,
      "loss": 0.8086,
      "step": 28355
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7516560554504395,
      "learning_rate": 2.5019966137430917e-06,
      "loss": 0.7204,
      "step": 28356
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8314073085784912,
      "learning_rate": 2.5013576973453024e-06,
      "loss": 0.8987,
      "step": 28357
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8148061037063599,
      "learning_rate": 2.500718780947513e-06,
      "loss": 0.772,
      "step": 28358
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7410318851470947,
      "learning_rate": 2.5000798645497238e-06,
      "loss": 0.8918,
      "step": 28359
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4598307609558105,
      "learning_rate": 2.4994409481519345e-06,
      "loss": 0.8132,
      "step": 28360
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9430042505264282,
      "learning_rate": 2.498802031754145e-06,
      "loss": 0.82,
      "step": 28361
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4515655040740967,
      "learning_rate": 2.498163115356356e-06,
      "loss": 0.7973,
      "step": 28362
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.1518938541412354,
      "learning_rate": 2.497524198958566e-06,
      "loss": 1.0043,
      "step": 28363
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1444625854492188,
      "learning_rate": 2.496885282560777e-06,
      "loss": 0.4011,
      "step": 28364
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.324769973754883,
      "learning_rate": 2.4962463661629875e-06,
      "loss": 1.2307,
      "step": 28365
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.283710241317749,
      "learning_rate": 2.4956074497651982e-06,
      "loss": 0.9008,
      "step": 28366
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7195086479187012,
      "learning_rate": 2.494968533367409e-06,
      "loss": 0.8679,
      "step": 28367
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.140235185623169,
      "learning_rate": 2.4943296169696196e-06,
      "loss": 0.7224,
      "step": 28368
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.774515986442566,
      "learning_rate": 2.4936907005718303e-06,
      "loss": 0.6619,
      "step": 28369
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1969645023345947,
      "learning_rate": 2.493051784174041e-06,
      "loss": 0.8503,
      "step": 28370
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9604246616363525,
      "learning_rate": 2.4924128677762517e-06,
      "loss": 0.8901,
      "step": 28371
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.978016972541809,
      "learning_rate": 2.4917739513784624e-06,
      "loss": 0.8852,
      "step": 28372
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.3724160194396973,
      "learning_rate": 2.4911350349806727e-06,
      "loss": 0.806,
      "step": 28373
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9792413711547852,
      "learning_rate": 2.4904961185828834e-06,
      "loss": 1.0972,
      "step": 28374
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.3435611724853516,
      "learning_rate": 2.489857202185094e-06,
      "loss": 0.8296,
      "step": 28375
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.6967620849609375,
      "learning_rate": 2.489218285787305e-06,
      "loss": 1.2426,
      "step": 28376
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4002678394317627,
      "learning_rate": 2.4885793693895155e-06,
      "loss": 0.715,
      "step": 28377
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1722826957702637,
      "learning_rate": 2.4879404529917262e-06,
      "loss": 0.9618,
      "step": 28378
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7433085441589355,
      "learning_rate": 2.487301536593937e-06,
      "loss": 0.7698,
      "step": 28379
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5279154777526855,
      "learning_rate": 2.4866626201961476e-06,
      "loss": 0.4194,
      "step": 28380
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.799415111541748,
      "learning_rate": 2.4860237037983583e-06,
      "loss": 0.9424,
      "step": 28381
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.005335569381714,
      "learning_rate": 2.4853847874005686e-06,
      "loss": 0.9435,
      "step": 28382
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.690567970275879,
      "learning_rate": 2.4847458710027793e-06,
      "loss": 0.9368,
      "step": 28383
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.584148645401001,
      "learning_rate": 2.48410695460499e-06,
      "loss": 1.01,
      "step": 28384
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1902830600738525,
      "learning_rate": 2.4834680382072007e-06,
      "loss": 0.7857,
      "step": 28385
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.8617665767669678,
      "learning_rate": 2.4828291218094114e-06,
      "loss": 0.7995,
      "step": 28386
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8558349609375,
      "learning_rate": 2.482190205411622e-06,
      "loss": 0.8357,
      "step": 28387
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8746627569198608,
      "learning_rate": 2.481551289013833e-06,
      "loss": 0.7915,
      "step": 28388
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8108782768249512,
      "learning_rate": 2.4809123726160435e-06,
      "loss": 0.9662,
      "step": 28389
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9843190908432007,
      "learning_rate": 2.480273456218254e-06,
      "loss": 0.8536,
      "step": 28390
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7318058013916016,
      "learning_rate": 2.4796345398204645e-06,
      "loss": 0.7903,
      "step": 28391
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9721835851669312,
      "learning_rate": 2.478995623422675e-06,
      "loss": 1.0303,
      "step": 28392
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.145240545272827,
      "learning_rate": 2.478356707024886e-06,
      "loss": 0.7387,
      "step": 28393
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8290396928787231,
      "learning_rate": 2.4777177906270966e-06,
      "loss": 0.7933,
      "step": 28394
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.735715627670288,
      "learning_rate": 2.4770788742293073e-06,
      "loss": 0.5766,
      "step": 28395
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6805726289749146,
      "learning_rate": 2.476439957831518e-06,
      "loss": 0.831,
      "step": 28396
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8600612878799438,
      "learning_rate": 2.4758010414337287e-06,
      "loss": 0.8093,
      "step": 28397
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4076722860336304,
      "learning_rate": 2.4751621250359394e-06,
      "loss": 0.7093,
      "step": 28398
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.018481969833374,
      "learning_rate": 2.47452320863815e-06,
      "loss": 0.8186,
      "step": 28399
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6467232704162598,
      "learning_rate": 2.4738842922403603e-06,
      "loss": 0.695,
      "step": 28400
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9627741575241089,
      "learning_rate": 2.473245375842571e-06,
      "loss": 0.8513,
      "step": 28401
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2709639072418213,
      "learning_rate": 2.4726064594447817e-06,
      "loss": 0.7673,
      "step": 28402
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.003066301345825,
      "learning_rate": 2.4719675430469924e-06,
      "loss": 0.5588,
      "step": 28403
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2191362380981445,
      "learning_rate": 2.471328626649203e-06,
      "loss": 0.732,
      "step": 28404
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.037508010864258,
      "learning_rate": 2.470689710251414e-06,
      "loss": 0.8464,
      "step": 28405
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6100401878356934,
      "learning_rate": 2.4700507938536245e-06,
      "loss": 0.8253,
      "step": 28406
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0991876125335693,
      "learning_rate": 2.4694118774558352e-06,
      "loss": 0.9906,
      "step": 28407
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.2769079208374023,
      "learning_rate": 2.468772961058046e-06,
      "loss": 0.7977,
      "step": 28408
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8589996099472046,
      "learning_rate": 2.4681340446602562e-06,
      "loss": 0.85,
      "step": 28409
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.327351450920105,
      "learning_rate": 2.467495128262467e-06,
      "loss": 0.7922,
      "step": 28410
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9573752880096436,
      "learning_rate": 2.4668562118646776e-06,
      "loss": 0.7714,
      "step": 28411
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6547763347625732,
      "learning_rate": 2.4662172954668883e-06,
      "loss": 0.8078,
      "step": 28412
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8728407621383667,
      "learning_rate": 2.465578379069099e-06,
      "loss": 0.8842,
      "step": 28413
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7800114154815674,
      "learning_rate": 2.4649394626713097e-06,
      "loss": 0.9276,
      "step": 28414
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7776979207992554,
      "learning_rate": 2.4643005462735204e-06,
      "loss": 0.7908,
      "step": 28415
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.704318881034851,
      "learning_rate": 2.463661629875731e-06,
      "loss": 0.7926,
      "step": 28416
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8797931671142578,
      "learning_rate": 2.463022713477942e-06,
      "loss": 0.8273,
      "step": 28417
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6665180921554565,
      "learning_rate": 2.462383797080152e-06,
      "loss": 0.7942,
      "step": 28418
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9692987203598022,
      "learning_rate": 2.461744880682363e-06,
      "loss": 0.8574,
      "step": 28419
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.117237091064453,
      "learning_rate": 2.4611059642845735e-06,
      "loss": 0.9731,
      "step": 28420
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.756225824356079,
      "learning_rate": 2.460467047886784e-06,
      "loss": 0.9958,
      "step": 28421
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.870575189590454,
      "learning_rate": 2.459828131488995e-06,
      "loss": 0.7988,
      "step": 28422
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.3680498600006104,
      "learning_rate": 2.4591892150912056e-06,
      "loss": 0.7552,
      "step": 28423
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6719001531600952,
      "learning_rate": 2.4585502986934163e-06,
      "loss": 0.6622,
      "step": 28424
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.611076831817627,
      "learning_rate": 2.457911382295627e-06,
      "loss": 0.8269,
      "step": 28425
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4233345985412598,
      "learning_rate": 2.4572724658978377e-06,
      "loss": 0.8947,
      "step": 28426
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9367872476577759,
      "learning_rate": 2.456633549500048e-06,
      "loss": 0.8362,
      "step": 28427
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.898645877838135,
      "learning_rate": 2.4559946331022587e-06,
      "loss": 0.8858,
      "step": 28428
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9570144414901733,
      "learning_rate": 2.4553557167044694e-06,
      "loss": 0.8633,
      "step": 28429
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.612724781036377,
      "learning_rate": 2.45471680030668e-06,
      "loss": 0.68,
      "step": 28430
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5266908407211304,
      "learning_rate": 2.4540778839088908e-06,
      "loss": 0.7993,
      "step": 28431
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8183248043060303,
      "learning_rate": 2.4534389675111015e-06,
      "loss": 0.8967,
      "step": 28432
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0381460189819336,
      "learning_rate": 2.452800051113312e-06,
      "loss": 1.1788,
      "step": 28433
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4935457706451416,
      "learning_rate": 2.452161134715523e-06,
      "loss": 0.7125,
      "step": 28434
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0845694541931152,
      "learning_rate": 2.4515222183177336e-06,
      "loss": 0.8194,
      "step": 28435
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.96589994430542,
      "learning_rate": 2.450883301919944e-06,
      "loss": 0.8876,
      "step": 28436
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.121683120727539,
      "learning_rate": 2.4502443855221545e-06,
      "loss": 0.9421,
      "step": 28437
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.643397331237793,
      "learning_rate": 2.4496054691243652e-06,
      "loss": 0.7247,
      "step": 28438
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.357067584991455,
      "learning_rate": 2.448966552726576e-06,
      "loss": 0.7833,
      "step": 28439
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7809295654296875,
      "learning_rate": 2.4483276363287866e-06,
      "loss": 1.1131,
      "step": 28440
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2237071990966797,
      "learning_rate": 2.4476887199309973e-06,
      "loss": 0.8757,
      "step": 28441
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.3354570865631104,
      "learning_rate": 2.447049803533208e-06,
      "loss": 0.8003,
      "step": 28442
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.707982063293457,
      "learning_rate": 2.4464108871354187e-06,
      "loss": 0.8025,
      "step": 28443
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8583920001983643,
      "learning_rate": 2.4457719707376294e-06,
      "loss": 0.97,
      "step": 28444
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0668997764587402,
      "learning_rate": 2.44513305433984e-06,
      "loss": 0.8072,
      "step": 28445
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9626038074493408,
      "learning_rate": 2.4444941379420504e-06,
      "loss": 0.7681,
      "step": 28446
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.347212314605713,
      "learning_rate": 2.443855221544261e-06,
      "loss": 0.6688,
      "step": 28447
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.330960273742676,
      "learning_rate": 2.443216305146472e-06,
      "loss": 0.98,
      "step": 28448
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.072779655456543,
      "learning_rate": 2.4425773887486825e-06,
      "loss": 0.8943,
      "step": 28449
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.299009084701538,
      "learning_rate": 2.4419384723508932e-06,
      "loss": 1.027,
      "step": 28450
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.931775689125061,
      "learning_rate": 2.441299555953104e-06,
      "loss": 0.9404,
      "step": 28451
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7437469959259033,
      "learning_rate": 2.4406606395553146e-06,
      "loss": 0.8059,
      "step": 28452
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.356818914413452,
      "learning_rate": 2.4400217231575253e-06,
      "loss": 0.8274,
      "step": 28453
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7452905178070068,
      "learning_rate": 2.439382806759736e-06,
      "loss": 0.7164,
      "step": 28454
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7836260795593262,
      "learning_rate": 2.4387438903619463e-06,
      "loss": 0.8068,
      "step": 28455
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.321305274963379,
      "learning_rate": 2.438104973964157e-06,
      "loss": 1.183,
      "step": 28456
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9663267135620117,
      "learning_rate": 2.4374660575663677e-06,
      "loss": 0.8712,
      "step": 28457
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6554070711135864,
      "learning_rate": 2.436827141168578e-06,
      "loss": 0.754,
      "step": 28458
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8683207035064697,
      "learning_rate": 2.4361882247707887e-06,
      "loss": 0.8985,
      "step": 28459
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6452937126159668,
      "learning_rate": 2.4355493083729994e-06,
      "loss": 1.0925,
      "step": 28460
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2401480674743652,
      "learning_rate": 2.43491039197521e-06,
      "loss": 0.8529,
      "step": 28461
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.338700771331787,
      "learning_rate": 2.4342714755774208e-06,
      "loss": 0.9324,
      "step": 28462
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5784039497375488,
      "learning_rate": 2.4336325591796315e-06,
      "loss": 0.7323,
      "step": 28463
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.018956422805786,
      "learning_rate": 2.432993642781842e-06,
      "loss": 0.9643,
      "step": 28464
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.812435269355774,
      "learning_rate": 2.432354726384053e-06,
      "loss": 0.7393,
      "step": 28465
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7448852062225342,
      "learning_rate": 2.4317158099862636e-06,
      "loss": 0.7919,
      "step": 28466
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.168896198272705,
      "learning_rate": 2.431076893588474e-06,
      "loss": 1.3502,
      "step": 28467
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7012704610824585,
      "learning_rate": 2.4304379771906846e-06,
      "loss": 0.8073,
      "step": 28468
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8907029628753662,
      "learning_rate": 2.4297990607928953e-06,
      "loss": 0.8953,
      "step": 28469
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.930002212524414,
      "learning_rate": 2.429160144395106e-06,
      "loss": 0.7881,
      "step": 28470
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6657699346542358,
      "learning_rate": 2.4285212279973167e-06,
      "loss": 0.8544,
      "step": 28471
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6503174304962158,
      "learning_rate": 2.4278823115995274e-06,
      "loss": 0.7711,
      "step": 28472
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9627480506896973,
      "learning_rate": 2.427243395201738e-06,
      "loss": 1.4568,
      "step": 28473
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6162269115448,
      "learning_rate": 2.4266044788039488e-06,
      "loss": 1.1237,
      "step": 28474
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.152358293533325,
      "learning_rate": 2.4259655624061595e-06,
      "loss": 1.1785,
      "step": 28475
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.2972679138183594,
      "learning_rate": 2.4253266460083697e-06,
      "loss": 0.9615,
      "step": 28476
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8753174543380737,
      "learning_rate": 2.4246877296105804e-06,
      "loss": 0.6944,
      "step": 28477
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.846779704093933,
      "learning_rate": 2.424048813212791e-06,
      "loss": 0.7445,
      "step": 28478
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.231207847595215,
      "learning_rate": 2.423409896815002e-06,
      "loss": 0.9279,
      "step": 28479
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9026533365249634,
      "learning_rate": 2.4227709804172125e-06,
      "loss": 0.7812,
      "step": 28480
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5671958923339844,
      "learning_rate": 2.4221320640194232e-06,
      "loss": 0.7821,
      "step": 28481
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.2805886268615723,
      "learning_rate": 2.421493147621634e-06,
      "loss": 0.9686,
      "step": 28482
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.14117169380188,
      "learning_rate": 2.4208542312238446e-06,
      "loss": 0.7892,
      "step": 28483
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.779553771018982,
      "learning_rate": 2.4202153148260553e-06,
      "loss": 0.692,
      "step": 28484
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9745028018951416,
      "learning_rate": 2.4195763984282656e-06,
      "loss": 1.1918,
      "step": 28485
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1408960819244385,
      "learning_rate": 2.4189374820304763e-06,
      "loss": 0.8057,
      "step": 28486
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2701830863952637,
      "learning_rate": 2.418298565632687e-06,
      "loss": 0.9369,
      "step": 28487
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6019831895828247,
      "learning_rate": 2.4176596492348977e-06,
      "loss": 0.835,
      "step": 28488
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8115570545196533,
      "learning_rate": 2.4170207328371084e-06,
      "loss": 0.993,
      "step": 28489
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7762681245803833,
      "learning_rate": 2.416381816439319e-06,
      "loss": 0.7633,
      "step": 28490
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6275110244750977,
      "learning_rate": 2.41574290004153e-06,
      "loss": 0.9036,
      "step": 28491
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8724929094314575,
      "learning_rate": 2.4151039836437405e-06,
      "loss": 0.814,
      "step": 28492
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.835093379020691,
      "learning_rate": 2.414465067245951e-06,
      "loss": 1.0669,
      "step": 28493
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7232425212860107,
      "learning_rate": 2.4138261508481615e-06,
      "loss": 0.8814,
      "step": 28494
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6822762489318848,
      "learning_rate": 2.413187234450372e-06,
      "loss": 0.8047,
      "step": 28495
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6175018548965454,
      "learning_rate": 2.412548318052583e-06,
      "loss": 0.789,
      "step": 28496
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9243735074996948,
      "learning_rate": 2.4119094016547936e-06,
      "loss": 1.2961,
      "step": 28497
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6750051975250244,
      "learning_rate": 2.4112704852570043e-06,
      "loss": 0.8092,
      "step": 28498
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.359429359436035,
      "learning_rate": 2.410631568859215e-06,
      "loss": 0.7708,
      "step": 28499
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.99251389503479,
      "learning_rate": 2.4099926524614257e-06,
      "loss": 0.7382,
      "step": 28500
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.346813440322876,
      "learning_rate": 2.4093537360636364e-06,
      "loss": 0.8996,
      "step": 28501
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8974857330322266,
      "learning_rate": 2.408714819665847e-06,
      "loss": 0.7914,
      "step": 28502
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.438682556152344,
      "learning_rate": 2.4080759032680578e-06,
      "loss": 1.0692,
      "step": 28503
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.441952109336853,
      "learning_rate": 2.407436986870268e-06,
      "loss": 0.4127,
      "step": 28504
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8730227947235107,
      "learning_rate": 2.4067980704724788e-06,
      "loss": 0.7968,
      "step": 28505
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9258756637573242,
      "learning_rate": 2.4061591540746895e-06,
      "loss": 0.8594,
      "step": 28506
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.687522292137146,
      "learning_rate": 2.4055202376769e-06,
      "loss": 0.7055,
      "step": 28507
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.0284810066223145,
      "learning_rate": 2.404881321279111e-06,
      "loss": 1.0238,
      "step": 28508
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7402307987213135,
      "learning_rate": 2.4042424048813216e-06,
      "loss": 0.8192,
      "step": 28509
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9559780359268188,
      "learning_rate": 2.4036034884835323e-06,
      "loss": 0.823,
      "step": 28510
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7633943557739258,
      "learning_rate": 2.402964572085743e-06,
      "loss": 0.76,
      "step": 28511
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7636245489120483,
      "learning_rate": 2.4023256556879537e-06,
      "loss": 0.7834,
      "step": 28512
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.932332754135132,
      "learning_rate": 2.401686739290164e-06,
      "loss": 0.8856,
      "step": 28513
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7264634370803833,
      "learning_rate": 2.4010478228923746e-06,
      "loss": 0.6646,
      "step": 28514
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6961679458618164,
      "learning_rate": 2.4004089064945853e-06,
      "loss": 0.7939,
      "step": 28515
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.362780570983887,
      "learning_rate": 2.399769990096796e-06,
      "loss": 1.2261,
      "step": 28516
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6590023040771484,
      "learning_rate": 2.3991310736990067e-06,
      "loss": 1.2222,
      "step": 28517
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.072751998901367,
      "learning_rate": 2.3984921573012174e-06,
      "loss": 0.9937,
      "step": 28518
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1165530681610107,
      "learning_rate": 2.397853240903428e-06,
      "loss": 0.864,
      "step": 28519
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.144209861755371,
      "learning_rate": 2.397214324505639e-06,
      "loss": 0.9094,
      "step": 28520
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4669512510299683,
      "learning_rate": 2.3965754081078495e-06,
      "loss": 0.8105,
      "step": 28521
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.772252082824707,
      "learning_rate": 2.39593649171006e-06,
      "loss": 0.9291,
      "step": 28522
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.883207082748413,
      "learning_rate": 2.3952975753122705e-06,
      "loss": 0.8966,
      "step": 28523
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.016038656234741,
      "learning_rate": 2.394658658914481e-06,
      "loss": 1.0053,
      "step": 28524
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9685962200164795,
      "learning_rate": 2.394019742516692e-06,
      "loss": 0.8522,
      "step": 28525
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4597578048706055,
      "learning_rate": 2.3933808261189026e-06,
      "loss": 0.6261,
      "step": 28526
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0123908519744873,
      "learning_rate": 2.3927419097211133e-06,
      "loss": 1.0395,
      "step": 28527
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8651365041732788,
      "learning_rate": 2.392102993323324e-06,
      "loss": 0.808,
      "step": 28528
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6764259338378906,
      "learning_rate": 2.3914640769255347e-06,
      "loss": 0.8188,
      "step": 28529
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3993983268737793,
      "learning_rate": 2.3908251605277454e-06,
      "loss": 0.4596,
      "step": 28530
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.0221071243286133,
      "learning_rate": 2.3901862441299557e-06,
      "loss": 0.9808,
      "step": 28531
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7720099687576294,
      "learning_rate": 2.3895473277321664e-06,
      "loss": 0.8891,
      "step": 28532
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0209131240844727,
      "learning_rate": 2.388908411334377e-06,
      "loss": 0.8186,
      "step": 28533
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.807969093322754,
      "learning_rate": 2.3882694949365878e-06,
      "loss": 1.5836,
      "step": 28534
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7667438983917236,
      "learning_rate": 2.3876305785387985e-06,
      "loss": 0.6365,
      "step": 28535
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6806384325027466,
      "learning_rate": 2.386991662141009e-06,
      "loss": 0.9058,
      "step": 28536
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1857144832611084,
      "learning_rate": 2.38635274574322e-06,
      "loss": 0.8457,
      "step": 28537
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.479655146598816,
      "learning_rate": 2.3857138293454306e-06,
      "loss": 0.6148,
      "step": 28538
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.548406720161438,
      "learning_rate": 2.3850749129476413e-06,
      "loss": 0.8618,
      "step": 28539
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.308682918548584,
      "learning_rate": 2.3844359965498516e-06,
      "loss": 0.3159,
      "step": 28540
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.841002106666565,
      "learning_rate": 2.3837970801520623e-06,
      "loss": 0.6656,
      "step": 28541
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7788033485412598,
      "learning_rate": 2.383158163754273e-06,
      "loss": 0.8295,
      "step": 28542
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.106147050857544,
      "learning_rate": 2.3825192473564837e-06,
      "loss": 1.0898,
      "step": 28543
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7663906812667847,
      "learning_rate": 2.3818803309586944e-06,
      "loss": 0.7479,
      "step": 28544
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.09591007232666,
      "learning_rate": 2.381241414560905e-06,
      "loss": 0.9614,
      "step": 28545
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7270138263702393,
      "learning_rate": 2.3806024981631158e-06,
      "loss": 0.8381,
      "step": 28546
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9194402694702148,
      "learning_rate": 2.3799635817653265e-06,
      "loss": 1.0466,
      "step": 28547
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1825034618377686,
      "learning_rate": 2.379324665367537e-06,
      "loss": 1.0019,
      "step": 28548
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.117138385772705,
      "learning_rate": 2.3786857489697474e-06,
      "loss": 0.845,
      "step": 28549
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0023791790008545,
      "learning_rate": 2.378046832571958e-06,
      "loss": 0.9501,
      "step": 28550
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.207571268081665,
      "learning_rate": 2.377407916174169e-06,
      "loss": 1.0047,
      "step": 28551
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.7852399349212646,
      "learning_rate": 2.3767689997763795e-06,
      "loss": 0.8571,
      "step": 28552
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8963913917541504,
      "learning_rate": 2.37613008337859e-06,
      "loss": 0.8082,
      "step": 28553
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.532486915588379,
      "learning_rate": 2.3754911669808005e-06,
      "loss": 0.7193,
      "step": 28554
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9698686599731445,
      "learning_rate": 2.3748522505830112e-06,
      "loss": 1.0018,
      "step": 28555
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0898663997650146,
      "learning_rate": 2.374213334185222e-06,
      "loss": 0.9309,
      "step": 28556
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.06622052192688,
      "learning_rate": 2.3735744177874326e-06,
      "loss": 0.6441,
      "step": 28557
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9330343008041382,
      "learning_rate": 2.3729355013896433e-06,
      "loss": 1.0997,
      "step": 28558
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3276890516281128,
      "learning_rate": 2.372296584991854e-06,
      "loss": 0.4265,
      "step": 28559
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.512605667114258,
      "learning_rate": 2.3716576685940647e-06,
      "loss": 1.0112,
      "step": 28560
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.638022541999817,
      "learning_rate": 2.371018752196275e-06,
      "loss": 0.8059,
      "step": 28561
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5166566371917725,
      "learning_rate": 2.3703798357984857e-06,
      "loss": 0.7633,
      "step": 28562
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2608875036239624,
      "learning_rate": 2.3697409194006964e-06,
      "loss": 0.3979,
      "step": 28563
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6043782234191895,
      "learning_rate": 2.369102003002907e-06,
      "loss": 0.4669,
      "step": 28564
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9696600437164307,
      "learning_rate": 2.368463086605118e-06,
      "loss": 0.8,
      "step": 28565
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6750965118408203,
      "learning_rate": 2.3678241702073285e-06,
      "loss": 0.7543,
      "step": 28566
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0433435440063477,
      "learning_rate": 2.367185253809539e-06,
      "loss": 0.8565,
      "step": 28567
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2611008882522583,
      "learning_rate": 2.36654633741175e-06,
      "loss": 0.3426,
      "step": 28568
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.718608856201172,
      "learning_rate": 2.3659074210139606e-06,
      "loss": 1.1769,
      "step": 28569
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9613943099975586,
      "learning_rate": 2.3652685046161713e-06,
      "loss": 0.8065,
      "step": 28570
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7795332670211792,
      "learning_rate": 2.3646295882183816e-06,
      "loss": 0.794,
      "step": 28571
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5002321004867554,
      "learning_rate": 2.3639906718205923e-06,
      "loss": 0.3353,
      "step": 28572
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7785991430282593,
      "learning_rate": 2.363351755422803e-06,
      "loss": 0.8588,
      "step": 28573
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.078873872756958,
      "learning_rate": 2.3627128390250137e-06,
      "loss": 0.8979,
      "step": 28574
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.958695888519287,
      "learning_rate": 2.3620739226272244e-06,
      "loss": 0.923,
      "step": 28575
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.658812403678894,
      "learning_rate": 2.361435006229435e-06,
      "loss": 0.8557,
      "step": 28576
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7620954513549805,
      "learning_rate": 2.3607960898316458e-06,
      "loss": 0.7918,
      "step": 28577
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.252788782119751,
      "learning_rate": 2.3601571734338565e-06,
      "loss": 0.7664,
      "step": 28578
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.918302059173584,
      "learning_rate": 2.359518257036067e-06,
      "loss": 0.8023,
      "step": 28579
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4905585050582886,
      "learning_rate": 2.3588793406382774e-06,
      "loss": 0.8319,
      "step": 28580
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.475183367729187,
      "learning_rate": 2.358240424240488e-06,
      "loss": 0.8086,
      "step": 28581
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1492645740509033,
      "learning_rate": 2.357601507842699e-06,
      "loss": 0.9521,
      "step": 28582
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8845093250274658,
      "learning_rate": 2.3569625914449095e-06,
      "loss": 0.8882,
      "step": 28583
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0426151752471924,
      "learning_rate": 2.3563236750471202e-06,
      "loss": 0.9584,
      "step": 28584
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9067628383636475,
      "learning_rate": 2.355684758649331e-06,
      "loss": 0.8175,
      "step": 28585
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5999531745910645,
      "learning_rate": 2.3550458422515416e-06,
      "loss": 0.8026,
      "step": 28586
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8019152879714966,
      "learning_rate": 2.3544069258537523e-06,
      "loss": 0.8948,
      "step": 28587
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2109484672546387,
      "learning_rate": 2.353768009455963e-06,
      "loss": 0.3285,
      "step": 28588
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4767940044403076,
      "learning_rate": 2.3531290930581733e-06,
      "loss": 0.7377,
      "step": 28589
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.480522871017456,
      "learning_rate": 2.352490176660384e-06,
      "loss": 0.6038,
      "step": 28590
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7597343921661377,
      "learning_rate": 2.3518512602625947e-06,
      "loss": 0.7812,
      "step": 28591
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5934767723083496,
      "learning_rate": 2.3512123438648054e-06,
      "loss": 0.7576,
      "step": 28592
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.7361207008361816,
      "learning_rate": 2.350573427467016e-06,
      "loss": 0.4878,
      "step": 28593
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7813156843185425,
      "learning_rate": 2.349934511069227e-06,
      "loss": 0.8122,
      "step": 28594
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0840625762939453,
      "learning_rate": 2.3492955946714375e-06,
      "loss": 0.8409,
      "step": 28595
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2836544513702393,
      "learning_rate": 2.3486566782736482e-06,
      "loss": 0.4295,
      "step": 28596
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2811191082000732,
      "learning_rate": 2.348017761875859e-06,
      "loss": 0.5015,
      "step": 28597
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.098135471343994,
      "learning_rate": 2.347378845478069e-06,
      "loss": 0.8515,
      "step": 28598
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.716107130050659,
      "learning_rate": 2.34673992908028e-06,
      "loss": 0.6067,
      "step": 28599
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.360670566558838,
      "learning_rate": 2.3461010126824906e-06,
      "loss": 0.5717,
      "step": 28600
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.380251407623291,
      "learning_rate": 2.3454620962847013e-06,
      "loss": 0.7672,
      "step": 28601
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0394303798675537,
      "learning_rate": 2.344823179886912e-06,
      "loss": 0.7426,
      "step": 28602
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.795048713684082,
      "learning_rate": 2.3441842634891227e-06,
      "loss": 0.6562,
      "step": 28603
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.624229907989502,
      "learning_rate": 2.3435453470913334e-06,
      "loss": 0.7809,
      "step": 28604
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.023503065109253,
      "learning_rate": 2.342906430693544e-06,
      "loss": 1.1294,
      "step": 28605
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0389325618743896,
      "learning_rate": 2.342267514295755e-06,
      "loss": 1.103,
      "step": 28606
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.623410940170288,
      "learning_rate": 2.341628597897965e-06,
      "loss": 0.7657,
      "step": 28607
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.810148000717163,
      "learning_rate": 2.3409896815001758e-06,
      "loss": 0.857,
      "step": 28608
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7505396604537964,
      "learning_rate": 2.3403507651023865e-06,
      "loss": 0.8642,
      "step": 28609
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.5670723915100098,
      "learning_rate": 2.339711848704597e-06,
      "loss": 0.7509,
      "step": 28610
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6150420904159546,
      "learning_rate": 2.339072932306808e-06,
      "loss": 0.8674,
      "step": 28611
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7255868911743164,
      "learning_rate": 2.3384340159090186e-06,
      "loss": 0.9171,
      "step": 28612
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2182457447052,
      "learning_rate": 2.3377950995112293e-06,
      "loss": 0.9433,
      "step": 28613
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.030289173126221,
      "learning_rate": 2.33715618311344e-06,
      "loss": 1.0457,
      "step": 28614
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.75432288646698,
      "learning_rate": 2.3365172667156507e-06,
      "loss": 0.8188,
      "step": 28615
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2036173343658447,
      "learning_rate": 2.335878350317861e-06,
      "loss": 0.784,
      "step": 28616
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9481152296066284,
      "learning_rate": 2.3352394339200716e-06,
      "loss": 0.7658,
      "step": 28617
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9154006242752075,
      "learning_rate": 2.3346005175222823e-06,
      "loss": 0.7736,
      "step": 28618
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.706069231033325,
      "learning_rate": 2.333961601124493e-06,
      "loss": 0.5006,
      "step": 28619
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8349109888076782,
      "learning_rate": 2.3333226847267037e-06,
      "loss": 0.909,
      "step": 28620
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4688308238983154,
      "learning_rate": 2.3326837683289144e-06,
      "loss": 0.7768,
      "step": 28621
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.81260085105896,
      "learning_rate": 2.332044851931125e-06,
      "loss": 0.8248,
      "step": 28622
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.131168842315674,
      "learning_rate": 2.331405935533336e-06,
      "loss": 0.841,
      "step": 28623
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5651055574417114,
      "learning_rate": 2.3307670191355465e-06,
      "loss": 0.7709,
      "step": 28624
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.425426483154297,
      "learning_rate": 2.330128102737757e-06,
      "loss": 1.0172,
      "step": 28625
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6647474765777588,
      "learning_rate": 2.3294891863399675e-06,
      "loss": 0.7392,
      "step": 28626
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9746413230895996,
      "learning_rate": 2.3288502699421782e-06,
      "loss": 0.7849,
      "step": 28627
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.324420928955078,
      "learning_rate": 2.328211353544389e-06,
      "loss": 0.9503,
      "step": 28628
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9189509153366089,
      "learning_rate": 2.3275724371465996e-06,
      "loss": 0.8319,
      "step": 28629
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1603469848632812,
      "learning_rate": 2.3269335207488103e-06,
      "loss": 1.1154,
      "step": 28630
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.3254377841949463,
      "learning_rate": 2.326294604351021e-06,
      "loss": 0.8288,
      "step": 28631
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.02805757522583,
      "learning_rate": 2.3256556879532317e-06,
      "loss": 0.9907,
      "step": 28632
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8358486890792847,
      "learning_rate": 2.3250167715554424e-06,
      "loss": 0.8467,
      "step": 28633
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5529221296310425,
      "learning_rate": 2.3243778551576527e-06,
      "loss": 0.4159,
      "step": 28634
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.6993703842163086,
      "learning_rate": 2.3237389387598634e-06,
      "loss": 1.3275,
      "step": 28635
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9056107997894287,
      "learning_rate": 2.323100022362074e-06,
      "loss": 0.7867,
      "step": 28636
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7648887634277344,
      "learning_rate": 2.322461105964285e-06,
      "loss": 0.4947,
      "step": 28637
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9004807472229004,
      "learning_rate": 2.3218221895664955e-06,
      "loss": 0.8672,
      "step": 28638
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.611907482147217,
      "learning_rate": 2.321183273168706e-06,
      "loss": 0.9343,
      "step": 28639
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9073199033737183,
      "learning_rate": 2.320544356770917e-06,
      "loss": 0.909,
      "step": 28640
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.068073034286499,
      "learning_rate": 2.3199054403731276e-06,
      "loss": 1.0459,
      "step": 28641
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4284162521362305,
      "learning_rate": 2.3192665239753383e-06,
      "loss": 0.4609,
      "step": 28642
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.780251383781433,
      "learning_rate": 2.318627607577549e-06,
      "loss": 0.8864,
      "step": 28643
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1981544494628906,
      "learning_rate": 2.3179886911797593e-06,
      "loss": 1.1853,
      "step": 28644
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8158457279205322,
      "learning_rate": 2.31734977478197e-06,
      "loss": 0.848,
      "step": 28645
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.503757119178772,
      "learning_rate": 2.3167108583841807e-06,
      "loss": 0.8527,
      "step": 28646
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.6114501953125,
      "learning_rate": 2.3160719419863914e-06,
      "loss": 1.6121,
      "step": 28647
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.965592384338379,
      "learning_rate": 2.315433025588602e-06,
      "loss": 1.0584,
      "step": 28648
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.644164800643921,
      "learning_rate": 2.3147941091908123e-06,
      "loss": 0.8881,
      "step": 28649
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1459243297576904,
      "learning_rate": 2.314155192793023e-06,
      "loss": 0.4633,
      "step": 28650
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2467286586761475,
      "learning_rate": 2.3135162763952337e-06,
      "loss": 0.8981,
      "step": 28651
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8376305103302002,
      "learning_rate": 2.3128773599974444e-06,
      "loss": 1.0315,
      "step": 28652
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8944298028945923,
      "learning_rate": 2.312238443599655e-06,
      "loss": 0.7785,
      "step": 28653
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.823409557342529,
      "learning_rate": 2.311599527201866e-06,
      "loss": 0.8282,
      "step": 28654
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0118274688720703,
      "learning_rate": 2.3109606108040765e-06,
      "loss": 0.5257,
      "step": 28655
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5193076133728027,
      "learning_rate": 2.310321694406287e-06,
      "loss": 0.751,
      "step": 28656
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4316571950912476,
      "learning_rate": 2.3096827780084975e-06,
      "loss": 0.7863,
      "step": 28657
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7875105142593384,
      "learning_rate": 2.3090438616107082e-06,
      "loss": 0.8513,
      "step": 28658
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9177024364471436,
      "learning_rate": 2.308404945212919e-06,
      "loss": 0.8513,
      "step": 28659
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.329006314277649,
      "learning_rate": 2.3077660288151296e-06,
      "loss": 0.4195,
      "step": 28660
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9805225133895874,
      "learning_rate": 2.3071271124173403e-06,
      "loss": 0.8803,
      "step": 28661
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.58708918094635,
      "learning_rate": 2.306488196019551e-06,
      "loss": 0.8004,
      "step": 28662
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.614281177520752,
      "learning_rate": 2.3058492796217617e-06,
      "loss": 1.5583,
      "step": 28663
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.979833722114563,
      "learning_rate": 2.3052103632239724e-06,
      "loss": 0.9705,
      "step": 28664
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.307372570037842,
      "learning_rate": 2.3045714468261827e-06,
      "loss": 0.8828,
      "step": 28665
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5562158823013306,
      "learning_rate": 2.3039325304283934e-06,
      "loss": 0.8516,
      "step": 28666
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.141205310821533,
      "learning_rate": 2.303293614030604e-06,
      "loss": 0.8442,
      "step": 28667
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6231915950775146,
      "learning_rate": 2.302654697632815e-06,
      "loss": 0.7364,
      "step": 28668
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6632765531539917,
      "learning_rate": 2.3020157812350255e-06,
      "loss": 0.5999,
      "step": 28669
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.418724536895752,
      "learning_rate": 2.301376864837236e-06,
      "loss": 0.3803,
      "step": 28670
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0147223472595215,
      "learning_rate": 2.300737948439447e-06,
      "loss": 0.4926,
      "step": 28671
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4085115194320679,
      "learning_rate": 2.3000990320416576e-06,
      "loss": 0.4875,
      "step": 28672
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9527485370635986,
      "learning_rate": 2.2994601156438683e-06,
      "loss": 0.8064,
      "step": 28673
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8682056665420532,
      "learning_rate": 2.2988211992460786e-06,
      "loss": 0.798,
      "step": 28674
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.197405815124512,
      "learning_rate": 2.2981822828482893e-06,
      "loss": 1.1863,
      "step": 28675
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4160733222961426,
      "learning_rate": 2.2975433664505e-06,
      "loss": 0.9703,
      "step": 28676
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0188238620758057,
      "learning_rate": 2.2969044500527107e-06,
      "loss": 0.7904,
      "step": 28677
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.7387571334838867,
      "learning_rate": 2.2962655336549214e-06,
      "loss": 0.8316,
      "step": 28678
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.3557703495025635,
      "learning_rate": 2.295626617257132e-06,
      "loss": 0.6375,
      "step": 28679
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2862846851348877,
      "learning_rate": 2.2949877008593428e-06,
      "loss": 0.9198,
      "step": 28680
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.035212993621826,
      "learning_rate": 2.2943487844615535e-06,
      "loss": 0.7845,
      "step": 28681
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8925986289978027,
      "learning_rate": 2.293709868063764e-06,
      "loss": 0.8915,
      "step": 28682
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.647155523300171,
      "learning_rate": 2.2930709516659745e-06,
      "loss": 0.5125,
      "step": 28683
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.633988618850708,
      "learning_rate": 2.292432035268185e-06,
      "loss": 0.8148,
      "step": 28684
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5228700637817383,
      "learning_rate": 2.291793118870396e-06,
      "loss": 0.8332,
      "step": 28685
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0458743572235107,
      "learning_rate": 2.2911542024726066e-06,
      "loss": 1.0248,
      "step": 28686
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6769999265670776,
      "learning_rate": 2.2905152860748173e-06,
      "loss": 0.76,
      "step": 28687
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5154480934143066,
      "learning_rate": 2.289876369677028e-06,
      "loss": 0.7495,
      "step": 28688
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5661122798919678,
      "learning_rate": 2.2892374532792387e-06,
      "loss": 0.6905,
      "step": 28689
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.3030176162719727,
      "learning_rate": 2.2885985368814494e-06,
      "loss": 0.9312,
      "step": 28690
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9436911344528198,
      "learning_rate": 2.28795962048366e-06,
      "loss": 0.8486,
      "step": 28691
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9099551439285278,
      "learning_rate": 2.2873207040858703e-06,
      "loss": 0.7987,
      "step": 28692
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.366392135620117,
      "learning_rate": 2.286681787688081e-06,
      "loss": 0.9281,
      "step": 28693
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.8171260356903076,
      "learning_rate": 2.2860428712902917e-06,
      "loss": 1.039,
      "step": 28694
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.207980155944824,
      "learning_rate": 2.2854039548925024e-06,
      "loss": 0.8228,
      "step": 28695
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.688388466835022,
      "learning_rate": 2.284765038494713e-06,
      "loss": 0.8402,
      "step": 28696
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8772048950195312,
      "learning_rate": 2.284126122096924e-06,
      "loss": 0.9169,
      "step": 28697
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.525045394897461,
      "learning_rate": 2.2834872056991345e-06,
      "loss": 0.9192,
      "step": 28698
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1129910945892334,
      "learning_rate": 2.2828482893013452e-06,
      "loss": 0.8487,
      "step": 28699
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.127171277999878,
      "learning_rate": 2.282209372903556e-06,
      "loss": 0.7458,
      "step": 28700
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.7931134700775146,
      "learning_rate": 2.281570456505766e-06,
      "loss": 0.7209,
      "step": 28701
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8231816291809082,
      "learning_rate": 2.280931540107977e-06,
      "loss": 0.785,
      "step": 28702
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.237849712371826,
      "learning_rate": 2.2802926237101876e-06,
      "loss": 1.0963,
      "step": 28703
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6958928108215332,
      "learning_rate": 2.2796537073123983e-06,
      "loss": 0.7845,
      "step": 28704
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.896040678024292,
      "learning_rate": 2.279014790914609e-06,
      "loss": 0.8245,
      "step": 28705
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0725884437561035,
      "learning_rate": 2.2783758745168197e-06,
      "loss": 0.7176,
      "step": 28706
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.551635503768921,
      "learning_rate": 2.2777369581190304e-06,
      "loss": 0.7153,
      "step": 28707
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1025233268737793,
      "learning_rate": 2.277098041721241e-06,
      "loss": 0.7804,
      "step": 28708
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.246230125427246,
      "learning_rate": 2.276459125323452e-06,
      "loss": 0.7008,
      "step": 28709
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5615053176879883,
      "learning_rate": 2.2758202089256625e-06,
      "loss": 0.8085,
      "step": 28710
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.314906597137451,
      "learning_rate": 2.2751812925278728e-06,
      "loss": 0.8348,
      "step": 28711
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6034884452819824,
      "learning_rate": 2.2745423761300835e-06,
      "loss": 0.788,
      "step": 28712
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8587137460708618,
      "learning_rate": 2.273903459732294e-06,
      "loss": 1.2168,
      "step": 28713
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.406660795211792,
      "learning_rate": 2.273264543334505e-06,
      "loss": 0.9023,
      "step": 28714
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.6543431282043457,
      "learning_rate": 2.2726256269367156e-06,
      "loss": 1.2364,
      "step": 28715
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.896726131439209,
      "learning_rate": 2.2719867105389263e-06,
      "loss": 0.778,
      "step": 28716
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.367843508720398,
      "learning_rate": 2.271347794141137e-06,
      "loss": 0.3251,
      "step": 28717
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.045997142791748,
      "learning_rate": 2.2707088777433477e-06,
      "loss": 1.4431,
      "step": 28718
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9121159315109253,
      "learning_rate": 2.2700699613455584e-06,
      "loss": 0.8771,
      "step": 28719
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7085825204849243,
      "learning_rate": 2.2694310449477687e-06,
      "loss": 0.7954,
      "step": 28720
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5632364749908447,
      "learning_rate": 2.2687921285499794e-06,
      "loss": 0.8072,
      "step": 28721
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.144895553588867,
      "learning_rate": 2.26815321215219e-06,
      "loss": 0.8074,
      "step": 28722
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.356684684753418,
      "learning_rate": 2.2675142957544008e-06,
      "loss": 0.8682,
      "step": 28723
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5925456285476685,
      "learning_rate": 2.2668753793566115e-06,
      "loss": 0.7719,
      "step": 28724
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5010454654693604,
      "learning_rate": 2.266236462958822e-06,
      "loss": 0.764,
      "step": 28725
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1879827976226807,
      "learning_rate": 2.265597546561033e-06,
      "loss": 0.8067,
      "step": 28726
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9326276779174805,
      "learning_rate": 2.2649586301632436e-06,
      "loss": 0.9126,
      "step": 28727
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9028273820877075,
      "learning_rate": 2.2643197137654543e-06,
      "loss": 0.9119,
      "step": 28728
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4714550971984863,
      "learning_rate": 2.2636807973676645e-06,
      "loss": 1.0429,
      "step": 28729
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7082722187042236,
      "learning_rate": 2.2630418809698752e-06,
      "loss": 0.8142,
      "step": 28730
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.789789915084839,
      "learning_rate": 2.262402964572086e-06,
      "loss": 1.1734,
      "step": 28731
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6046717166900635,
      "learning_rate": 2.2617640481742966e-06,
      "loss": 0.8355,
      "step": 28732
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.733588457107544,
      "learning_rate": 2.2611251317765073e-06,
      "loss": 0.8694,
      "step": 28733
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7342535257339478,
      "learning_rate": 2.260486215378718e-06,
      "loss": 0.9098,
      "step": 28734
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.143630027770996,
      "learning_rate": 2.2598472989809287e-06,
      "loss": 0.8083,
      "step": 28735
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.294935941696167,
      "learning_rate": 2.2592083825831394e-06,
      "loss": 1.1087,
      "step": 28736
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8728762865066528,
      "learning_rate": 2.25856946618535e-06,
      "loss": 0.7799,
      "step": 28737
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.662116527557373,
      "learning_rate": 2.2579305497875604e-06,
      "loss": 0.7487,
      "step": 28738
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8382431268692017,
      "learning_rate": 2.257291633389771e-06,
      "loss": 0.8975,
      "step": 28739
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.864199161529541,
      "learning_rate": 2.256652716991982e-06,
      "loss": 0.8628,
      "step": 28740
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.9745843410491943,
      "learning_rate": 2.2560138005941925e-06,
      "loss": 1.3558,
      "step": 28741
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.708939552307129,
      "learning_rate": 2.255374884196403e-06,
      "loss": 0.7657,
      "step": 28742
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.690700054168701,
      "learning_rate": 2.254735967798614e-06,
      "loss": 0.7198,
      "step": 28743
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.071028232574463,
      "learning_rate": 2.2540970514008246e-06,
      "loss": 0.796,
      "step": 28744
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.59993577003479,
      "learning_rate": 2.2534581350030353e-06,
      "loss": 0.7764,
      "step": 28745
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.497800350189209,
      "learning_rate": 2.2528192186052456e-06,
      "loss": 0.9209,
      "step": 28746
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7707465887069702,
      "learning_rate": 2.2521803022074563e-06,
      "loss": 0.8306,
      "step": 28747
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2634401321411133,
      "learning_rate": 2.251541385809667e-06,
      "loss": 0.9272,
      "step": 28748
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9995566606521606,
      "learning_rate": 2.2509024694118777e-06,
      "loss": 0.9356,
      "step": 28749
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7602729797363281,
      "learning_rate": 2.250263553014088e-06,
      "loss": 0.7747,
      "step": 28750
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9931364059448242,
      "learning_rate": 2.2496246366162987e-06,
      "loss": 1.1713,
      "step": 28751
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.058521032333374,
      "learning_rate": 2.2489857202185094e-06,
      "loss": 1.0594,
      "step": 28752
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.106105327606201,
      "learning_rate": 2.24834680382072e-06,
      "loss": 1.1183,
      "step": 28753
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7762988805770874,
      "learning_rate": 2.2477078874229308e-06,
      "loss": 0.8508,
      "step": 28754
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.3026976585388184,
      "learning_rate": 2.2470689710251415e-06,
      "loss": 1.2142,
      "step": 28755
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5923681259155273,
      "learning_rate": 2.246430054627352e-06,
      "loss": 0.8389,
      "step": 28756
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7715437412261963,
      "learning_rate": 2.245791138229563e-06,
      "loss": 0.8262,
      "step": 28757
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7893604040145874,
      "learning_rate": 2.2451522218317736e-06,
      "loss": 0.7724,
      "step": 28758
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9226235151290894,
      "learning_rate": 2.244513305433984e-06,
      "loss": 1.053,
      "step": 28759
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.978833794593811,
      "learning_rate": 2.2438743890361945e-06,
      "loss": 0.8159,
      "step": 28760
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8610038757324219,
      "learning_rate": 2.2432354726384052e-06,
      "loss": 0.9056,
      "step": 28761
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6015722751617432,
      "learning_rate": 2.242596556240616e-06,
      "loss": 0.7122,
      "step": 28762
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1549620628356934,
      "learning_rate": 2.2419576398428266e-06,
      "loss": 0.9526,
      "step": 28763
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9524956941604614,
      "learning_rate": 2.2413187234450373e-06,
      "loss": 0.8502,
      "step": 28764
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.984281301498413,
      "learning_rate": 2.240679807047248e-06,
      "loss": 1.1519,
      "step": 28765
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.5506234169006348,
      "learning_rate": 2.2400408906494587e-06,
      "loss": 0.8453,
      "step": 28766
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8538130521774292,
      "learning_rate": 2.2394019742516694e-06,
      "loss": 0.8632,
      "step": 28767
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6871201992034912,
      "learning_rate": 2.23876305785388e-06,
      "loss": 0.8269,
      "step": 28768
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.821144700050354,
      "learning_rate": 2.2381241414560904e-06,
      "loss": 0.789,
      "step": 28769
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.787999153137207,
      "learning_rate": 2.237485225058301e-06,
      "loss": 0.9967,
      "step": 28770
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.632550835609436,
      "learning_rate": 2.236846308660512e-06,
      "loss": 0.8189,
      "step": 28771
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.041815757751465,
      "learning_rate": 2.2362073922627225e-06,
      "loss": 0.6173,
      "step": 28772
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1055688858032227,
      "learning_rate": 2.2355684758649332e-06,
      "loss": 1.1585,
      "step": 28773
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7964879274368286,
      "learning_rate": 2.234929559467144e-06,
      "loss": 0.7954,
      "step": 28774
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5073585510253906,
      "learning_rate": 2.2342906430693546e-06,
      "loss": 0.7528,
      "step": 28775
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.7780613899230957,
      "learning_rate": 2.2336517266715653e-06,
      "loss": 0.4072,
      "step": 28776
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9046988487243652,
      "learning_rate": 2.233012810273776e-06,
      "loss": 0.9269,
      "step": 28777
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.276040554046631,
      "learning_rate": 2.2323738938759863e-06,
      "loss": 0.8491,
      "step": 28778
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7598735094070435,
      "learning_rate": 2.231734977478197e-06,
      "loss": 0.848,
      "step": 28779
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2867544889450073,
      "learning_rate": 2.2310960610804077e-06,
      "loss": 0.424,
      "step": 28780
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1427206993103027,
      "learning_rate": 2.2304571446826184e-06,
      "loss": 0.7918,
      "step": 28781
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5826530456542969,
      "learning_rate": 2.229818228284829e-06,
      "loss": 0.7689,
      "step": 28782
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7079260349273682,
      "learning_rate": 2.22917931188704e-06,
      "loss": 0.4663,
      "step": 28783
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0763096809387207,
      "learning_rate": 2.2285403954892505e-06,
      "loss": 0.8588,
      "step": 28784
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8840510845184326,
      "learning_rate": 2.227901479091461e-06,
      "loss": 0.886,
      "step": 28785
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.272344946861267,
      "learning_rate": 2.227262562693672e-06,
      "loss": 0.4027,
      "step": 28786
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.269572734832764,
      "learning_rate": 2.226623646295882e-06,
      "loss": 1.5928,
      "step": 28787
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.22536563873291,
      "learning_rate": 2.225984729898093e-06,
      "loss": 1.3669,
      "step": 28788
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.7899534702301025,
      "learning_rate": 2.2253458135003036e-06,
      "loss": 0.838,
      "step": 28789
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.4312562942504883,
      "learning_rate": 2.2247068971025143e-06,
      "loss": 0.7186,
      "step": 28790
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6421459913253784,
      "learning_rate": 2.224067980704725e-06,
      "loss": 0.8271,
      "step": 28791
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9228566884994507,
      "learning_rate": 2.2234290643069357e-06,
      "loss": 0.9233,
      "step": 28792
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.546597719192505,
      "learning_rate": 2.2227901479091464e-06,
      "loss": 0.7324,
      "step": 28793
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7525254487991333,
      "learning_rate": 2.222151231511357e-06,
      "loss": 0.8863,
      "step": 28794
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.586039423942566,
      "learning_rate": 2.2215123151135678e-06,
      "loss": 0.3845,
      "step": 28795
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.546119451522827,
      "learning_rate": 2.220873398715778e-06,
      "loss": 0.9314,
      "step": 28796
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.294760227203369,
      "learning_rate": 2.2202344823179887e-06,
      "loss": 0.9627,
      "step": 28797
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.576737403869629,
      "learning_rate": 2.2195955659201994e-06,
      "loss": 1.0589,
      "step": 28798
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2896995544433594,
      "learning_rate": 2.21895664952241e-06,
      "loss": 1.1487,
      "step": 28799
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.030477285385132,
      "learning_rate": 2.218317733124621e-06,
      "loss": 1.212,
      "step": 28800
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6565109491348267,
      "learning_rate": 2.2176788167268315e-06,
      "loss": 0.755,
      "step": 28801
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.16072940826416,
      "learning_rate": 2.2170399003290422e-06,
      "loss": 0.9008,
      "step": 28802
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7318530082702637,
      "learning_rate": 2.216400983931253e-06,
      "loss": 0.743,
      "step": 28803
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.3993749618530273,
      "learning_rate": 2.2157620675334636e-06,
      "loss": 1.0527,
      "step": 28804
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7650388479232788,
      "learning_rate": 2.215123151135674e-06,
      "loss": 0.9396,
      "step": 28805
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9568287134170532,
      "learning_rate": 2.2144842347378846e-06,
      "loss": 0.8866,
      "step": 28806
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7110015153884888,
      "learning_rate": 2.2138453183400953e-06,
      "loss": 0.8948,
      "step": 28807
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.125345468521118,
      "learning_rate": 2.213206401942306e-06,
      "loss": 0.8705,
      "step": 28808
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8326729536056519,
      "learning_rate": 2.2125674855445167e-06,
      "loss": 0.7728,
      "step": 28809
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2060391902923584,
      "learning_rate": 2.2119285691467274e-06,
      "loss": 0.937,
      "step": 28810
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2811591625213623,
      "learning_rate": 2.211289652748938e-06,
      "loss": 0.9332,
      "step": 28811
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7636518478393555,
      "learning_rate": 2.210650736351149e-06,
      "loss": 0.81,
      "step": 28812
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6115736961364746,
      "learning_rate": 2.2100118199533595e-06,
      "loss": 0.844,
      "step": 28813
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.062297344207764,
      "learning_rate": 2.20937290355557e-06,
      "loss": 1.0874,
      "step": 28814
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.973947048187256,
      "learning_rate": 2.2087339871577805e-06,
      "loss": 0.9607,
      "step": 28815
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2228939533233643,
      "learning_rate": 2.208095070759991e-06,
      "loss": 0.8574,
      "step": 28816
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.90962815284729,
      "learning_rate": 2.207456154362202e-06,
      "loss": 0.8429,
      "step": 28817
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.077753782272339,
      "learning_rate": 2.2068172379644126e-06,
      "loss": 1.2455,
      "step": 28818
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.262564182281494,
      "learning_rate": 2.2061783215666233e-06,
      "loss": 0.7695,
      "step": 28819
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4358859062194824,
      "learning_rate": 2.205539405168834e-06,
      "loss": 0.5867,
      "step": 28820
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1404449939727783,
      "learning_rate": 2.2049004887710447e-06,
      "loss": 0.7671,
      "step": 28821
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7437609434127808,
      "learning_rate": 2.2042615723732554e-06,
      "loss": 0.8479,
      "step": 28822
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4825507402420044,
      "learning_rate": 2.2036226559754657e-06,
      "loss": 0.3079,
      "step": 28823
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1247475147247314,
      "learning_rate": 2.2029837395776764e-06,
      "loss": 0.8648,
      "step": 28824
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8519290685653687,
      "learning_rate": 2.202344823179887e-06,
      "loss": 0.6402,
      "step": 28825
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8111528158187866,
      "learning_rate": 2.2017059067820978e-06,
      "loss": 0.7952,
      "step": 28826
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.04209041595459,
      "learning_rate": 2.2010669903843085e-06,
      "loss": 0.9639,
      "step": 28827
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.9723098278045654,
      "learning_rate": 2.200428073986519e-06,
      "loss": 1.0831,
      "step": 28828
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0263586044311523,
      "learning_rate": 2.19978915758873e-06,
      "loss": 0.935,
      "step": 28829
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.139019250869751,
      "learning_rate": 2.1991502411909406e-06,
      "loss": 1.0244,
      "step": 28830
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.8225977420806885,
      "learning_rate": 2.1985113247931513e-06,
      "loss": 0.4997,
      "step": 28831
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.6441476345062256,
      "learning_rate": 2.1978724083953615e-06,
      "loss": 1.1583,
      "step": 28832
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.998667597770691,
      "learning_rate": 2.1972334919975722e-06,
      "loss": 0.7743,
      "step": 28833
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5923691987991333,
      "learning_rate": 2.196594575599783e-06,
      "loss": 0.9231,
      "step": 28834
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1554229259490967,
      "learning_rate": 2.1959556592019936e-06,
      "loss": 0.7316,
      "step": 28835
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.678426742553711,
      "learning_rate": 2.1953167428042043e-06,
      "loss": 0.7669,
      "step": 28836
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8115105628967285,
      "learning_rate": 2.194677826406415e-06,
      "loss": 0.8273,
      "step": 28837
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.346103549003601,
      "learning_rate": 2.1940389100086257e-06,
      "loss": 0.8175,
      "step": 28838
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8553794622421265,
      "learning_rate": 2.1933999936108364e-06,
      "loss": 0.8126,
      "step": 28839
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9080730676651,
      "learning_rate": 2.192761077213047e-06,
      "loss": 0.9577,
      "step": 28840
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3879005908966064,
      "learning_rate": 2.192122160815258e-06,
      "loss": 0.3989,
      "step": 28841
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.6176202297210693,
      "learning_rate": 2.191483244417468e-06,
      "loss": 0.7916,
      "step": 28842
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.142362356185913,
      "learning_rate": 2.190844328019679e-06,
      "loss": 0.9034,
      "step": 28843
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.5446479320526123,
      "learning_rate": 2.1902054116218895e-06,
      "loss": 0.9961,
      "step": 28844
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.700206995010376,
      "learning_rate": 2.1895664952241e-06,
      "loss": 0.7791,
      "step": 28845
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2554738521575928,
      "learning_rate": 2.1889275788263105e-06,
      "loss": 0.7048,
      "step": 28846
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9370062351226807,
      "learning_rate": 2.188288662428521e-06,
      "loss": 0.8315,
      "step": 28847
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5685781240463257,
      "learning_rate": 2.187649746030732e-06,
      "loss": 0.8582,
      "step": 28848
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9634069204330444,
      "learning_rate": 2.1870108296329426e-06,
      "loss": 0.8282,
      "step": 28849
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6541539430618286,
      "learning_rate": 2.1863719132351533e-06,
      "loss": 0.7705,
      "step": 28850
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5738592147827148,
      "learning_rate": 2.185732996837364e-06,
      "loss": 0.7432,
      "step": 28851
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.45743989944458,
      "learning_rate": 2.1850940804395747e-06,
      "loss": 0.9419,
      "step": 28852
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6457843780517578,
      "learning_rate": 2.1844551640417854e-06,
      "loss": 0.8633,
      "step": 28853
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.134629726409912,
      "learning_rate": 2.1838162476439957e-06,
      "loss": 0.8165,
      "step": 28854
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0817434787750244,
      "learning_rate": 2.1831773312462064e-06,
      "loss": 0.8188,
      "step": 28855
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.419734477996826,
      "learning_rate": 2.182538414848417e-06,
      "loss": 0.8975,
      "step": 28856
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.673180341720581,
      "learning_rate": 2.1818994984506278e-06,
      "loss": 0.3816,
      "step": 28857
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9784687757492065,
      "learning_rate": 2.1812605820528385e-06,
      "loss": 0.8889,
      "step": 28858
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.107935667037964,
      "learning_rate": 2.180621665655049e-06,
      "loss": 0.8743,
      "step": 28859
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9375405311584473,
      "learning_rate": 2.17998274925726e-06,
      "loss": 1.0193,
      "step": 28860
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.043027639389038,
      "learning_rate": 2.1793438328594706e-06,
      "loss": 0.7203,
      "step": 28861
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5686126947402954,
      "learning_rate": 2.1787049164616813e-06,
      "loss": 0.8072,
      "step": 28862
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.99819278717041,
      "learning_rate": 2.1780660000638916e-06,
      "loss": 0.8808,
      "step": 28863
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.443737506866455,
      "learning_rate": 2.1774270836661023e-06,
      "loss": 1.0263,
      "step": 28864
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.482642412185669,
      "learning_rate": 2.176788167268313e-06,
      "loss": 0.9087,
      "step": 28865
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6948896646499634,
      "learning_rate": 2.1761492508705237e-06,
      "loss": 0.7361,
      "step": 28866
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.825045108795166,
      "learning_rate": 2.1755103344727344e-06,
      "loss": 0.9625,
      "step": 28867
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.882982611656189,
      "learning_rate": 2.174871418074945e-06,
      "loss": 0.8446,
      "step": 28868
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0685033798217773,
      "learning_rate": 2.1742325016771558e-06,
      "loss": 0.6151,
      "step": 28869
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9402965307235718,
      "learning_rate": 2.1735935852793665e-06,
      "loss": 0.8713,
      "step": 28870
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9604674577713013,
      "learning_rate": 2.172954668881577e-06,
      "loss": 0.7658,
      "step": 28871
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5674476623535156,
      "learning_rate": 2.1723157524837874e-06,
      "loss": 0.8414,
      "step": 28872
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.708138942718506,
      "learning_rate": 2.171676836085998e-06,
      "loss": 0.9315,
      "step": 28873
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.4761853218078613,
      "learning_rate": 2.171037919688209e-06,
      "loss": 0.6431,
      "step": 28874
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.467842936515808,
      "learning_rate": 2.1703990032904195e-06,
      "loss": 0.8355,
      "step": 28875
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6971956491470337,
      "learning_rate": 2.1697600868926302e-06,
      "loss": 0.79,
      "step": 28876
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7731807231903076,
      "learning_rate": 2.169121170494841e-06,
      "loss": 0.8909,
      "step": 28877
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.271728515625,
      "learning_rate": 2.1684822540970516e-06,
      "loss": 0.8275,
      "step": 28878
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.577985405921936,
      "learning_rate": 2.1678433376992623e-06,
      "loss": 0.8241,
      "step": 28879
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9949196577072144,
      "learning_rate": 2.167204421301473e-06,
      "loss": 0.7705,
      "step": 28880
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8650604486465454,
      "learning_rate": 2.1665655049036833e-06,
      "loss": 1.12,
      "step": 28881
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.403765082359314,
      "learning_rate": 2.165926588505894e-06,
      "loss": 0.8168,
      "step": 28882
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.9418795108795166,
      "learning_rate": 2.1652876721081047e-06,
      "loss": 0.5913,
      "step": 28883
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6676462888717651,
      "learning_rate": 2.1646487557103154e-06,
      "loss": 0.778,
      "step": 28884
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6359403133392334,
      "learning_rate": 2.164009839312526e-06,
      "loss": 0.7748,
      "step": 28885
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.870883584022522,
      "learning_rate": 2.163370922914737e-06,
      "loss": 0.7641,
      "step": 28886
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7870795726776123,
      "learning_rate": 2.1627320065169475e-06,
      "loss": 0.7927,
      "step": 28887
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.362220287322998,
      "learning_rate": 2.162093090119158e-06,
      "loss": 0.4143,
      "step": 28888
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7980494499206543,
      "learning_rate": 2.161454173721369e-06,
      "loss": 1.3112,
      "step": 28889
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8199118375778198,
      "learning_rate": 2.160815257323579e-06,
      "loss": 0.9215,
      "step": 28890
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1478753089904785,
      "learning_rate": 2.16017634092579e-06,
      "loss": 0.8204,
      "step": 28891
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8088897466659546,
      "learning_rate": 2.1595374245280006e-06,
      "loss": 0.7617,
      "step": 28892
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.348052978515625,
      "learning_rate": 2.1588985081302113e-06,
      "loss": 0.7722,
      "step": 28893
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8163440227508545,
      "learning_rate": 2.158259591732422e-06,
      "loss": 0.8414,
      "step": 28894
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0495500564575195,
      "learning_rate": 2.1576206753346327e-06,
      "loss": 0.7999,
      "step": 28895
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.9881317615509033,
      "learning_rate": 2.1569817589368434e-06,
      "loss": 0.7399,
      "step": 28896
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.57674241065979,
      "learning_rate": 2.156342842539054e-06,
      "loss": 0.7649,
      "step": 28897
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8541866540908813,
      "learning_rate": 2.1557039261412648e-06,
      "loss": 0.8341,
      "step": 28898
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.136967182159424,
      "learning_rate": 2.155065009743475e-06,
      "loss": 1.2808,
      "step": 28899
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.689015507698059,
      "learning_rate": 2.1544260933456858e-06,
      "loss": 0.792,
      "step": 28900
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8322813510894775,
      "learning_rate": 2.1537871769478965e-06,
      "loss": 0.8452,
      "step": 28901
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1755309104919434,
      "learning_rate": 2.153148260550107e-06,
      "loss": 0.7302,
      "step": 28902
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.078429698944092,
      "learning_rate": 2.152509344152318e-06,
      "loss": 0.8235,
      "step": 28903
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8516359329223633,
      "learning_rate": 2.1518704277545286e-06,
      "loss": 0.8224,
      "step": 28904
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.8888888359069824,
      "learning_rate": 2.1512315113567393e-06,
      "loss": 1.1077,
      "step": 28905
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8571767807006836,
      "learning_rate": 2.15059259495895e-06,
      "loss": 1.0741,
      "step": 28906
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.103279113769531,
      "learning_rate": 2.1499536785611607e-06,
      "loss": 1.3713,
      "step": 28907
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8921315670013428,
      "learning_rate": 2.1493147621633714e-06,
      "loss": 0.894,
      "step": 28908
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8551758527755737,
      "learning_rate": 2.1486758457655816e-06,
      "loss": 0.8242,
      "step": 28909
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8533133268356323,
      "learning_rate": 2.1480369293677923e-06,
      "loss": 0.754,
      "step": 28910
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3926030397415161,
      "learning_rate": 2.147398012970003e-06,
      "loss": 0.3373,
      "step": 28911
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1287577152252197,
      "learning_rate": 2.1467590965722137e-06,
      "loss": 0.8282,
      "step": 28912
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9164924621582031,
      "learning_rate": 2.1461201801744244e-06,
      "loss": 0.7595,
      "step": 28913
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.6057326793670654,
      "learning_rate": 2.145481263776635e-06,
      "loss": 0.6066,
      "step": 28914
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.429611325263977,
      "learning_rate": 2.144842347378846e-06,
      "loss": 0.3029,
      "step": 28915
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7737361192703247,
      "learning_rate": 2.1442034309810565e-06,
      "loss": 0.767,
      "step": 28916
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6187429428100586,
      "learning_rate": 2.1435645145832672e-06,
      "loss": 0.6895,
      "step": 28917
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2052125930786133,
      "learning_rate": 2.1429255981854775e-06,
      "loss": 0.6123,
      "step": 28918
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.9857735633850098,
      "learning_rate": 2.142286681787688e-06,
      "loss": 0.8861,
      "step": 28919
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.014270782470703,
      "learning_rate": 2.141647765389899e-06,
      "loss": 1.1182,
      "step": 28920
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9129869937896729,
      "learning_rate": 2.1410088489921096e-06,
      "loss": 0.7153,
      "step": 28921
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4886114597320557,
      "learning_rate": 2.1403699325943203e-06,
      "loss": 0.744,
      "step": 28922
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1158812046051025,
      "learning_rate": 2.139731016196531e-06,
      "loss": 0.8632,
      "step": 28923
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7906794548034668,
      "learning_rate": 2.1390920997987417e-06,
      "loss": 0.9113,
      "step": 28924
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9110097885131836,
      "learning_rate": 2.1384531834009524e-06,
      "loss": 0.8877,
      "step": 28925
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.466884136199951,
      "learning_rate": 2.137814267003163e-06,
      "loss": 0.8864,
      "step": 28926
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.310208797454834,
      "learning_rate": 2.1371753506053734e-06,
      "loss": 1.0099,
      "step": 28927
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7990078926086426,
      "learning_rate": 2.136536434207584e-06,
      "loss": 0.8095,
      "step": 28928
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9330247640609741,
      "learning_rate": 2.1358975178097948e-06,
      "loss": 0.8799,
      "step": 28929
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4713711738586426,
      "learning_rate": 2.1352586014120055e-06,
      "loss": 0.7628,
      "step": 28930
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.8712127208709717,
      "learning_rate": 2.134619685014216e-06,
      "loss": 0.5614,
      "step": 28931
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.563058376312256,
      "learning_rate": 2.133980768616427e-06,
      "loss": 0.5501,
      "step": 28932
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.624181866645813,
      "learning_rate": 2.1333418522186376e-06,
      "loss": 1.11,
      "step": 28933
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.43416166305542,
      "learning_rate": 2.1327029358208483e-06,
      "loss": 1.212,
      "step": 28934
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7123863697052,
      "learning_rate": 2.132064019423059e-06,
      "loss": 1.0659,
      "step": 28935
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9677318334579468,
      "learning_rate": 2.1314251030252693e-06,
      "loss": 0.7743,
      "step": 28936
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7277086973190308,
      "learning_rate": 2.13078618662748e-06,
      "loss": 0.8038,
      "step": 28937
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9992769956588745,
      "learning_rate": 2.1301472702296907e-06,
      "loss": 0.813,
      "step": 28938
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.817600250244141,
      "learning_rate": 2.129508353831901e-06,
      "loss": 1.3682,
      "step": 28939
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.700958013534546,
      "learning_rate": 2.1288694374341116e-06,
      "loss": 0.872,
      "step": 28940
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1807942390441895,
      "learning_rate": 2.1282305210363223e-06,
      "loss": 0.8325,
      "step": 28941
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7720096111297607,
      "learning_rate": 2.127591604638533e-06,
      "loss": 1.1334,
      "step": 28942
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.690960168838501,
      "learning_rate": 2.1269526882407437e-06,
      "loss": 0.8047,
      "step": 28943
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6251330375671387,
      "learning_rate": 2.1263137718429544e-06,
      "loss": 0.4479,
      "step": 28944
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.1928529739379883,
      "learning_rate": 2.125674855445165e-06,
      "loss": 0.7416,
      "step": 28945
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.8242340087890625,
      "learning_rate": 2.125035939047376e-06,
      "loss": 0.9918,
      "step": 28946
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8751726150512695,
      "learning_rate": 2.1243970226495865e-06,
      "loss": 0.4163,
      "step": 28947
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.5184216499328613,
      "learning_rate": 2.123758106251797e-06,
      "loss": 0.7001,
      "step": 28948
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.797484278678894,
      "learning_rate": 2.1231191898540075e-06,
      "loss": 0.5674,
      "step": 28949
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.2171287536621094,
      "learning_rate": 2.1224802734562182e-06,
      "loss": 0.7482,
      "step": 28950
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.028759717941284,
      "learning_rate": 2.121841357058429e-06,
      "loss": 1.0436,
      "step": 28951
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.04804801940918,
      "learning_rate": 2.1212024406606396e-06,
      "loss": 1.4611,
      "step": 28952
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.376248359680176,
      "learning_rate": 2.1205635242628503e-06,
      "loss": 1.1894,
      "step": 28953
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3628458976745605,
      "learning_rate": 2.119924607865061e-06,
      "loss": 0.3662,
      "step": 28954
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7001632452011108,
      "learning_rate": 2.1192856914672717e-06,
      "loss": 0.7977,
      "step": 28955
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.769093632698059,
      "learning_rate": 2.1186467750694824e-06,
      "loss": 0.8226,
      "step": 28956
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.707619071006775,
      "learning_rate": 2.1180078586716927e-06,
      "loss": 0.8633,
      "step": 28957
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4328105449676514,
      "learning_rate": 2.1173689422739034e-06,
      "loss": 0.9639,
      "step": 28958
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.237802505493164,
      "learning_rate": 2.116730025876114e-06,
      "loss": 0.8915,
      "step": 28959
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1127099990844727,
      "learning_rate": 2.116091109478325e-06,
      "loss": 0.801,
      "step": 28960
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5418533086776733,
      "learning_rate": 2.1154521930805355e-06,
      "loss": 0.8111,
      "step": 28961
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.0835554599761963,
      "learning_rate": 2.114813276682746e-06,
      "loss": 0.6425,
      "step": 28962
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6166054010391235,
      "learning_rate": 2.114174360284957e-06,
      "loss": 0.7885,
      "step": 28963
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6473207473754883,
      "learning_rate": 2.1135354438871676e-06,
      "loss": 0.8417,
      "step": 28964
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0464565753936768,
      "learning_rate": 2.1128965274893783e-06,
      "loss": 0.8338,
      "step": 28965
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6855053901672363,
      "learning_rate": 2.1122576110915886e-06,
      "loss": 0.7878,
      "step": 28966
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.022810459136963,
      "learning_rate": 2.1116186946937993e-06,
      "loss": 0.9005,
      "step": 28967
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4215493202209473,
      "learning_rate": 2.11097977829601e-06,
      "loss": 0.801,
      "step": 28968
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.8367843627929688,
      "learning_rate": 2.1103408618982207e-06,
      "loss": 1.0576,
      "step": 28969
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0136423110961914,
      "learning_rate": 2.1097019455004314e-06,
      "loss": 0.9398,
      "step": 28970
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2064990997314453,
      "learning_rate": 2.109063029102642e-06,
      "loss": 0.8811,
      "step": 28971
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.64731764793396,
      "learning_rate": 2.1084241127048528e-06,
      "loss": 0.8741,
      "step": 28972
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.701453447341919,
      "learning_rate": 2.1077851963070635e-06,
      "loss": 1.0781,
      "step": 28973
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9417682886123657,
      "learning_rate": 2.107146279909274e-06,
      "loss": 0.8526,
      "step": 28974
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.739898920059204,
      "learning_rate": 2.106507363511485e-06,
      "loss": 1.1157,
      "step": 28975
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.462085723876953,
      "learning_rate": 2.105868447113695e-06,
      "loss": 1.1801,
      "step": 28976
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9465539455413818,
      "learning_rate": 2.105229530715906e-06,
      "loss": 0.9038,
      "step": 28977
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.799065113067627,
      "learning_rate": 2.1045906143181165e-06,
      "loss": 0.888,
      "step": 28978
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6456249952316284,
      "learning_rate": 2.1039516979203272e-06,
      "loss": 0.8442,
      "step": 28979
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.533531904220581,
      "learning_rate": 2.103312781522538e-06,
      "loss": 0.7996,
      "step": 28980
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.209911823272705,
      "learning_rate": 2.1026738651247486e-06,
      "loss": 0.9716,
      "step": 28981
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.080968141555786,
      "learning_rate": 2.1020349487269593e-06,
      "loss": 0.8941,
      "step": 28982
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.865080714225769,
      "learning_rate": 2.10139603232917e-06,
      "loss": 0.9656,
      "step": 28983
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6534703969955444,
      "learning_rate": 2.1007571159313807e-06,
      "loss": 0.8224,
      "step": 28984
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.164788007736206,
      "learning_rate": 2.100118199533591e-06,
      "loss": 0.9396,
      "step": 28985
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.417726755142212,
      "learning_rate": 2.0994792831358017e-06,
      "loss": 0.9886,
      "step": 28986
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9811400175094604,
      "learning_rate": 2.0988403667380124e-06,
      "loss": 0.7312,
      "step": 28987
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.752403974533081,
      "learning_rate": 2.098201450340223e-06,
      "loss": 0.8364,
      "step": 28988
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.590810537338257,
      "learning_rate": 2.097562533942434e-06,
      "loss": 1.1709,
      "step": 28989
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7745509147644043,
      "learning_rate": 2.0969236175446445e-06,
      "loss": 0.8509,
      "step": 28990
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.5272367000579834,
      "learning_rate": 2.0962847011468552e-06,
      "loss": 0.7324,
      "step": 28991
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.400153398513794,
      "learning_rate": 2.095645784749066e-06,
      "loss": 0.8278,
      "step": 28992
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4153348207473755,
      "learning_rate": 2.0950068683512766e-06,
      "loss": 0.4262,
      "step": 28993
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7233185768127441,
      "learning_rate": 2.094367951953487e-06,
      "loss": 0.8818,
      "step": 28994
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.561613917350769,
      "learning_rate": 2.0937290355556976e-06,
      "loss": 0.8018,
      "step": 28995
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.055532932281494,
      "learning_rate": 2.0930901191579083e-06,
      "loss": 1.0735,
      "step": 28996
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.701695203781128,
      "learning_rate": 2.092451202760119e-06,
      "loss": 0.8536,
      "step": 28997
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7935466766357422,
      "learning_rate": 2.0918122863623297e-06,
      "loss": 0.9785,
      "step": 28998
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7953665256500244,
      "learning_rate": 2.0911733699645404e-06,
      "loss": 0.8193,
      "step": 28999
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.926120638847351,
      "learning_rate": 2.090534453566751e-06,
      "loss": 0.831,
      "step": 29000
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.612204670906067,
      "learning_rate": 2.089895537168962e-06,
      "loss": 0.774,
      "step": 29001
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.889962911605835,
      "learning_rate": 2.0892566207711725e-06,
      "loss": 0.8343,
      "step": 29002
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1712846755981445,
      "learning_rate": 2.0886177043733828e-06,
      "loss": 0.9071,
      "step": 29003
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9342526197433472,
      "learning_rate": 2.0879787879755935e-06,
      "loss": 0.8722,
      "step": 29004
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.690056562423706,
      "learning_rate": 2.087339871577804e-06,
      "loss": 1.0261,
      "step": 29005
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9186166524887085,
      "learning_rate": 2.086700955180015e-06,
      "loss": 0.8528,
      "step": 29006
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.847365140914917,
      "learning_rate": 2.0860620387822256e-06,
      "loss": 0.7444,
      "step": 29007
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.3134377002716064,
      "learning_rate": 2.0854231223844363e-06,
      "loss": 1.0577,
      "step": 29008
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8891444206237793,
      "learning_rate": 2.084784205986647e-06,
      "loss": 0.8047,
      "step": 29009
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.672839164733887,
      "learning_rate": 2.0841452895888577e-06,
      "loss": 1.5493,
      "step": 29010
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8551973104476929,
      "learning_rate": 2.0835063731910684e-06,
      "loss": 0.8949,
      "step": 29011
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.752682089805603,
      "learning_rate": 2.0828674567932786e-06,
      "loss": 0.8187,
      "step": 29012
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7884505987167358,
      "learning_rate": 2.0822285403954893e-06,
      "loss": 0.7895,
      "step": 29013
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.577589750289917,
      "learning_rate": 2.0815896239977e-06,
      "loss": 0.7889,
      "step": 29014
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.524231433868408,
      "learning_rate": 2.0809507075999107e-06,
      "loss": 1.1417,
      "step": 29015
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6637561321258545,
      "learning_rate": 2.0803117912021214e-06,
      "loss": 0.815,
      "step": 29016
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.660752058029175,
      "learning_rate": 2.079672874804332e-06,
      "loss": 1.1096,
      "step": 29017
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.869676947593689,
      "learning_rate": 2.079033958406543e-06,
      "loss": 0.7679,
      "step": 29018
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.544110894203186,
      "learning_rate": 2.0783950420087535e-06,
      "loss": 0.766,
      "step": 29019
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6312217712402344,
      "learning_rate": 2.0777561256109642e-06,
      "loss": 0.8254,
      "step": 29020
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.9586286544799805,
      "learning_rate": 2.0771172092131745e-06,
      "loss": 1.5538,
      "step": 29021
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9938870668411255,
      "learning_rate": 2.0764782928153852e-06,
      "loss": 0.8589,
      "step": 29022
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.003455400466919,
      "learning_rate": 2.075839376417596e-06,
      "loss": 0.8535,
      "step": 29023
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9848304986953735,
      "learning_rate": 2.0752004600198066e-06,
      "loss": 0.8018,
      "step": 29024
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.080855369567871,
      "learning_rate": 2.0745615436220173e-06,
      "loss": 1.0093,
      "step": 29025
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6632227897644043,
      "learning_rate": 2.073922627224228e-06,
      "loss": 0.5792,
      "step": 29026
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3643717765808105,
      "learning_rate": 2.0732837108264387e-06,
      "loss": 0.6698,
      "step": 29027
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6961535215377808,
      "learning_rate": 2.0726447944286494e-06,
      "loss": 0.9591,
      "step": 29028
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.882159233093262,
      "learning_rate": 2.07200587803086e-06,
      "loss": 1.191,
      "step": 29029
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.24001407623291,
      "learning_rate": 2.0713669616330704e-06,
      "loss": 0.7978,
      "step": 29030
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4845672845840454,
      "learning_rate": 2.070728045235281e-06,
      "loss": 0.7822,
      "step": 29031
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8332264423370361,
      "learning_rate": 2.070089128837492e-06,
      "loss": 0.7983,
      "step": 29032
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4017086029052734,
      "learning_rate": 2.0694502124397025e-06,
      "loss": 0.3842,
      "step": 29033
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7155978679656982,
      "learning_rate": 2.068811296041913e-06,
      "loss": 0.7951,
      "step": 29034
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.730497121810913,
      "learning_rate": 2.068172379644124e-06,
      "loss": 0.8463,
      "step": 29035
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.147836208343506,
      "learning_rate": 2.067533463246334e-06,
      "loss": 0.854,
      "step": 29036
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8755003213882446,
      "learning_rate": 2.066894546848545e-06,
      "loss": 0.8491,
      "step": 29037
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2461843490600586,
      "learning_rate": 2.0662556304507556e-06,
      "loss": 0.9136,
      "step": 29038
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.293597936630249,
      "learning_rate": 2.0656167140529663e-06,
      "loss": 0.346,
      "step": 29039
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.648979425430298,
      "learning_rate": 2.064977797655177e-06,
      "loss": 1.0717,
      "step": 29040
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.9066879749298096,
      "learning_rate": 2.0643388812573877e-06,
      "loss": 1.2343,
      "step": 29041
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7452372312545776,
      "learning_rate": 2.0636999648595984e-06,
      "loss": 0.8476,
      "step": 29042
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3690173625946045,
      "learning_rate": 2.0630610484618087e-06,
      "loss": 1.0622,
      "step": 29043
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5476677417755127,
      "learning_rate": 2.0624221320640194e-06,
      "loss": 0.7341,
      "step": 29044
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8330278396606445,
      "learning_rate": 2.06178321566623e-06,
      "loss": 0.9595,
      "step": 29045
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.835803985595703,
      "learning_rate": 2.0611442992684408e-06,
      "loss": 0.7698,
      "step": 29046
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8034095764160156,
      "learning_rate": 2.0605053828706515e-06,
      "loss": 0.9431,
      "step": 29047
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8115119934082031,
      "learning_rate": 2.059866466472862e-06,
      "loss": 0.8299,
      "step": 29048
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.543198823928833,
      "learning_rate": 2.059227550075073e-06,
      "loss": 0.9387,
      "step": 29049
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.815281867980957,
      "learning_rate": 2.0585886336772836e-06,
      "loss": 1.332,
      "step": 29050
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9455636739730835,
      "learning_rate": 2.0579497172794943e-06,
      "loss": 0.9726,
      "step": 29051
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4821691513061523,
      "learning_rate": 2.0573108008817045e-06,
      "loss": 0.7115,
      "step": 29052
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9726890325546265,
      "learning_rate": 2.0566718844839152e-06,
      "loss": 0.7314,
      "step": 29053
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.853049397468567,
      "learning_rate": 2.056032968086126e-06,
      "loss": 0.8184,
      "step": 29054
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.149310827255249,
      "learning_rate": 2.0553940516883366e-06,
      "loss": 0.993,
      "step": 29055
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.563757061958313,
      "learning_rate": 2.0547551352905473e-06,
      "loss": 0.8362,
      "step": 29056
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8211570978164673,
      "learning_rate": 2.054116218892758e-06,
      "loss": 0.7973,
      "step": 29057
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6302006244659424,
      "learning_rate": 2.0534773024949687e-06,
      "loss": 1.0735,
      "step": 29058
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.589511752128601,
      "learning_rate": 2.0528383860971794e-06,
      "loss": 0.7706,
      "step": 29059
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6230353116989136,
      "learning_rate": 2.05219946969939e-06,
      "loss": 0.8398,
      "step": 29060
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.841774582862854,
      "learning_rate": 2.0515605533016004e-06,
      "loss": 0.4749,
      "step": 29061
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5997209548950195,
      "learning_rate": 2.050921636903811e-06,
      "loss": 0.8092,
      "step": 29062
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7841579914093018,
      "learning_rate": 2.050282720506022e-06,
      "loss": 0.7504,
      "step": 29063
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7256957292556763,
      "learning_rate": 2.0496438041082325e-06,
      "loss": 0.8035,
      "step": 29064
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6985663175582886,
      "learning_rate": 2.049004887710443e-06,
      "loss": 0.9162,
      "step": 29065
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6850826740264893,
      "learning_rate": 2.048365971312654e-06,
      "loss": 1.0835,
      "step": 29066
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.0333027839660645,
      "learning_rate": 2.0477270549148646e-06,
      "loss": 1.1048,
      "step": 29067
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0644450187683105,
      "learning_rate": 2.0470881385170753e-06,
      "loss": 0.7361,
      "step": 29068
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.433529853820801,
      "learning_rate": 2.046449222119286e-06,
      "loss": 1.4465,
      "step": 29069
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.325406074523926,
      "learning_rate": 2.0458103057214963e-06,
      "loss": 1.1186,
      "step": 29070
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5179500579833984,
      "learning_rate": 2.045171389323707e-06,
      "loss": 0.8727,
      "step": 29071
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6876145601272583,
      "learning_rate": 2.0445324729259177e-06,
      "loss": 0.7972,
      "step": 29072
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.950176239013672,
      "learning_rate": 2.0438935565281284e-06,
      "loss": 1.0191,
      "step": 29073
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5077348947525024,
      "learning_rate": 2.043254640130339e-06,
      "loss": 0.831,
      "step": 29074
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0684988498687744,
      "learning_rate": 2.0426157237325498e-06,
      "loss": 0.492,
      "step": 29075
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.331882476806641,
      "learning_rate": 2.0419768073347605e-06,
      "loss": 1.0639,
      "step": 29076
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7165035009384155,
      "learning_rate": 2.041337890936971e-06,
      "loss": 0.837,
      "step": 29077
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7687366008758545,
      "learning_rate": 2.040698974539182e-06,
      "loss": 0.7802,
      "step": 29078
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4909502267837524,
      "learning_rate": 2.040060058141392e-06,
      "loss": 0.7443,
      "step": 29079
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.973937749862671,
      "learning_rate": 2.039421141743603e-06,
      "loss": 1.0186,
      "step": 29080
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6840101480484009,
      "learning_rate": 2.0387822253458136e-06,
      "loss": 0.8628,
      "step": 29081
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.991205215454102,
      "learning_rate": 2.0381433089480243e-06,
      "loss": 1.2147,
      "step": 29082
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.480635166168213,
      "learning_rate": 2.037504392550235e-06,
      "loss": 0.7359,
      "step": 29083
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.074467897415161,
      "learning_rate": 2.0368654761524457e-06,
      "loss": 0.818,
      "step": 29084
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.35814368724823,
      "learning_rate": 2.0362265597546564e-06,
      "loss": 0.3712,
      "step": 29085
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.779355525970459,
      "learning_rate": 2.035587643356867e-06,
      "loss": 0.8909,
      "step": 29086
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.676499843597412,
      "learning_rate": 2.0349487269590778e-06,
      "loss": 1.0337,
      "step": 29087
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8670878410339355,
      "learning_rate": 2.034309810561288e-06,
      "loss": 0.3686,
      "step": 29088
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9071608781814575,
      "learning_rate": 2.0336708941634987e-06,
      "loss": 0.7514,
      "step": 29089
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7474102973937988,
      "learning_rate": 2.0330319777657094e-06,
      "loss": 0.8005,
      "step": 29090
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.1065175533294678,
      "learning_rate": 2.03239306136792e-06,
      "loss": 0.8866,
      "step": 29091
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.5858445167541504,
      "learning_rate": 2.031754144970131e-06,
      "loss": 0.8212,
      "step": 29092
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8599258661270142,
      "learning_rate": 2.0311152285723415e-06,
      "loss": 0.8332,
      "step": 29093
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.947405457496643,
      "learning_rate": 2.0304763121745522e-06,
      "loss": 0.9254,
      "step": 29094
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.89499032497406,
      "learning_rate": 2.029837395776763e-06,
      "loss": 1.1441,
      "step": 29095
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.03605580329895,
      "learning_rate": 2.0291984793789736e-06,
      "loss": 1.006,
      "step": 29096
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8303054571151733,
      "learning_rate": 2.028559562981184e-06,
      "loss": 0.9235,
      "step": 29097
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9805774688720703,
      "learning_rate": 2.0279206465833946e-06,
      "loss": 0.8477,
      "step": 29098
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.148237705230713,
      "learning_rate": 2.0272817301856053e-06,
      "loss": 0.9269,
      "step": 29099
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.004826784133911,
      "learning_rate": 2.026642813787816e-06,
      "loss": 1.0541,
      "step": 29100
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8169183731079102,
      "learning_rate": 2.0260038973900267e-06,
      "loss": 0.8458,
      "step": 29101
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5668079853057861,
      "learning_rate": 2.0253649809922374e-06,
      "loss": 0.8192,
      "step": 29102
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2031443119049072,
      "learning_rate": 2.024726064594448e-06,
      "loss": 0.7515,
      "step": 29103
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.0833520889282227,
      "learning_rate": 2.024087148196659e-06,
      "loss": 0.8797,
      "step": 29104
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7756643295288086,
      "learning_rate": 2.0234482317988695e-06,
      "loss": 1.0815,
      "step": 29105
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6173545122146606,
      "learning_rate": 2.02280931540108e-06,
      "loss": 0.8332,
      "step": 29106
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.827707290649414,
      "learning_rate": 2.0221703990032905e-06,
      "loss": 0.8027,
      "step": 29107
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.917180299758911,
      "learning_rate": 2.021531482605501e-06,
      "loss": 1.0107,
      "step": 29108
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.714984178543091,
      "learning_rate": 2.020892566207712e-06,
      "loss": 0.8664,
      "step": 29109
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5519717931747437,
      "learning_rate": 2.0202536498099226e-06,
      "loss": 0.817,
      "step": 29110
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7360119819641113,
      "learning_rate": 2.0196147334121333e-06,
      "loss": 0.6112,
      "step": 29111
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9317959547042847,
      "learning_rate": 2.018975817014344e-06,
      "loss": 0.9227,
      "step": 29112
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.330526828765869,
      "learning_rate": 2.0183369006165547e-06,
      "loss": 0.6181,
      "step": 29113
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.968957543373108,
      "learning_rate": 2.0176979842187654e-06,
      "loss": 0.7278,
      "step": 29114
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8379764556884766,
      "learning_rate": 2.017059067820976e-06,
      "loss": 0.9812,
      "step": 29115
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.528904914855957,
      "learning_rate": 2.0164201514231864e-06,
      "loss": 0.8143,
      "step": 29116
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7719218730926514,
      "learning_rate": 2.015781235025397e-06,
      "loss": 1.0101,
      "step": 29117
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.1499364376068115,
      "learning_rate": 2.0151423186276078e-06,
      "loss": 0.7245,
      "step": 29118
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.8009185791015625,
      "learning_rate": 2.0145034022298185e-06,
      "loss": 1.7034,
      "step": 29119
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.10181999206543,
      "learning_rate": 2.013864485832029e-06,
      "loss": 1.7552,
      "step": 29120
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0143489837646484,
      "learning_rate": 2.01322556943424e-06,
      "loss": 0.9053,
      "step": 29121
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.1031434535980225,
      "learning_rate": 2.0125866530364506e-06,
      "loss": 0.4561,
      "step": 29122
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.8525190353393555,
      "learning_rate": 2.0119477366386613e-06,
      "loss": 0.8044,
      "step": 29123
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.448474884033203,
      "learning_rate": 2.011308820240872e-06,
      "loss": 0.5565,
      "step": 29124
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.599050760269165,
      "learning_rate": 2.0106699038430822e-06,
      "loss": 0.7899,
      "step": 29125
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8748894929885864,
      "learning_rate": 2.010030987445293e-06,
      "loss": 0.7225,
      "step": 29126
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.119845867156982,
      "learning_rate": 2.0093920710475036e-06,
      "loss": 1.4333,
      "step": 29127
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8919587135314941,
      "learning_rate": 2.0087531546497143e-06,
      "loss": 0.7784,
      "step": 29128
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2225725650787354,
      "learning_rate": 2.008114238251925e-06,
      "loss": 0.7943,
      "step": 29129
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4752726554870605,
      "learning_rate": 2.0074753218541357e-06,
      "loss": 0.8057,
      "step": 29130
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.6472811698913574,
      "learning_rate": 2.0068364054563464e-06,
      "loss": 1.1598,
      "step": 29131
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9698175191879272,
      "learning_rate": 2.006197489058557e-06,
      "loss": 0.7545,
      "step": 29132
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8609888553619385,
      "learning_rate": 2.0055585726607674e-06,
      "loss": 1.0672,
      "step": 29133
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9208754301071167,
      "learning_rate": 2.004919656262978e-06,
      "loss": 0.6982,
      "step": 29134
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5061365365982056,
      "learning_rate": 2.004280739865189e-06,
      "loss": 0.4501,
      "step": 29135
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0357165336608887,
      "learning_rate": 2.0036418234673995e-06,
      "loss": 0.7342,
      "step": 29136
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7211533784866333,
      "learning_rate": 2.00300290706961e-06,
      "loss": 0.7661,
      "step": 29137
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.954265594482422,
      "learning_rate": 2.0023639906718205e-06,
      "loss": 0.9863,
      "step": 29138
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0825202465057373,
      "learning_rate": 2.001725074274031e-06,
      "loss": 0.9198,
      "step": 29139
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7408182621002197,
      "learning_rate": 2.001086157876242e-06,
      "loss": 0.7824,
      "step": 29140
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.8631436824798584,
      "learning_rate": 2.0004472414784526e-06,
      "loss": 0.9906,
      "step": 29141
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.24482536315918,
      "learning_rate": 1.9998083250806633e-06,
      "loss": 1.3833,
      "step": 29142
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8219736814498901,
      "learning_rate": 1.999169408682874e-06,
      "loss": 1.0538,
      "step": 29143
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.0236659049987793,
      "learning_rate": 1.9985304922850847e-06,
      "loss": 0.7499,
      "step": 29144
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6526426076889038,
      "learning_rate": 1.9978915758872954e-06,
      "loss": 0.8079,
      "step": 29145
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0525894165039062,
      "learning_rate": 1.9972526594895057e-06,
      "loss": 0.8157,
      "step": 29146
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3853750228881836,
      "learning_rate": 1.9966137430917164e-06,
      "loss": 0.6993,
      "step": 29147
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.9027459621429443,
      "learning_rate": 1.995974826693927e-06,
      "loss": 0.8104,
      "step": 29148
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8593549728393555,
      "learning_rate": 1.9953359102961378e-06,
      "loss": 0.8541,
      "step": 29149
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6660797595977783,
      "learning_rate": 1.9946969938983485e-06,
      "loss": 0.8029,
      "step": 29150
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9748387336730957,
      "learning_rate": 1.994058077500559e-06,
      "loss": 0.7833,
      "step": 29151
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.072643280029297,
      "learning_rate": 1.99341916110277e-06,
      "loss": 1.1645,
      "step": 29152
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8431422710418701,
      "learning_rate": 1.9927802447049806e-06,
      "loss": 1.0488,
      "step": 29153
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9310272932052612,
      "learning_rate": 1.9921413283071913e-06,
      "loss": 1.0659,
      "step": 29154
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.193847179412842,
      "learning_rate": 1.9915024119094015e-06,
      "loss": 1.0133,
      "step": 29155
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.574652194976807,
      "learning_rate": 1.9908634955116122e-06,
      "loss": 1.1559,
      "step": 29156
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.8622093200683594,
      "learning_rate": 1.990224579113823e-06,
      "loss": 0.8088,
      "step": 29157
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6354475021362305,
      "learning_rate": 1.9895856627160336e-06,
      "loss": 0.8932,
      "step": 29158
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6770739555358887,
      "learning_rate": 1.9889467463182443e-06,
      "loss": 0.8393,
      "step": 29159
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.69184410572052,
      "learning_rate": 1.988307829920455e-06,
      "loss": 0.697,
      "step": 29160
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7186840772628784,
      "learning_rate": 1.9876689135226657e-06,
      "loss": 0.7287,
      "step": 29161
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.5357463359832764,
      "learning_rate": 1.9870299971248764e-06,
      "loss": 0.9922,
      "step": 29162
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.537320375442505,
      "learning_rate": 1.986391080727087e-06,
      "loss": 0.8412,
      "step": 29163
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3305788040161133,
      "learning_rate": 1.9857521643292974e-06,
      "loss": 0.3305,
      "step": 29164
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5932780504226685,
      "learning_rate": 1.985113247931508e-06,
      "loss": 0.7949,
      "step": 29165
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.48579740524292,
      "learning_rate": 1.984474331533719e-06,
      "loss": 0.6407,
      "step": 29166
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.767774224281311,
      "learning_rate": 1.9838354151359295e-06,
      "loss": 0.8496,
      "step": 29167
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.135104179382324,
      "learning_rate": 1.9831964987381402e-06,
      "loss": 0.7848,
      "step": 29168
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.907081127166748,
      "learning_rate": 1.982557582340351e-06,
      "loss": 0.8475,
      "step": 29169
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5609517097473145,
      "learning_rate": 1.9819186659425616e-06,
      "loss": 1.1719,
      "step": 29170
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.184035539627075,
      "learning_rate": 1.9812797495447723e-06,
      "loss": 0.8838,
      "step": 29171
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.349416732788086,
      "learning_rate": 1.980640833146983e-06,
      "loss": 0.8322,
      "step": 29172
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.094172477722168,
      "learning_rate": 1.9800019167491937e-06,
      "loss": 0.6892,
      "step": 29173
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6804720163345337,
      "learning_rate": 1.979363000351404e-06,
      "loss": 0.7207,
      "step": 29174
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2250072956085205,
      "learning_rate": 1.9787240839536147e-06,
      "loss": 0.8957,
      "step": 29175
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6218584775924683,
      "learning_rate": 1.9780851675558254e-06,
      "loss": 0.7072,
      "step": 29176
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4162763357162476,
      "learning_rate": 1.977446251158036e-06,
      "loss": 0.8594,
      "step": 29177
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.733896255493164,
      "learning_rate": 1.976807334760247e-06,
      "loss": 0.7787,
      "step": 29178
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.0397214889526367,
      "learning_rate": 1.9761684183624575e-06,
      "loss": 0.7216,
      "step": 29179
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.6437811851501465,
      "learning_rate": 1.975529501964668e-06,
      "loss": 1.1995,
      "step": 29180
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5503827333450317,
      "learning_rate": 1.974890585566879e-06,
      "loss": 0.854,
      "step": 29181
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7566473484039307,
      "learning_rate": 1.9742516691690896e-06,
      "loss": 0.6413,
      "step": 29182
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.212355852127075,
      "learning_rate": 1.9736127527713e-06,
      "loss": 0.9559,
      "step": 29183
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6779676675796509,
      "learning_rate": 1.9729738363735106e-06,
      "loss": 0.815,
      "step": 29184
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.340830683708191,
      "learning_rate": 1.9723349199757213e-06,
      "loss": 0.3549,
      "step": 29185
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.783210039138794,
      "learning_rate": 1.971696003577932e-06,
      "loss": 0.815,
      "step": 29186
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.6967673301696777,
      "learning_rate": 1.9710570871801427e-06,
      "loss": 1.1918,
      "step": 29187
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1555635929107666,
      "learning_rate": 1.9704181707823534e-06,
      "loss": 0.8548,
      "step": 29188
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.113319158554077,
      "learning_rate": 1.969779254384564e-06,
      "loss": 0.9369,
      "step": 29189
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6972655057907104,
      "learning_rate": 1.9691403379867748e-06,
      "loss": 0.7888,
      "step": 29190
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.8119897842407227,
      "learning_rate": 1.9685014215889855e-06,
      "loss": 1.2952,
      "step": 29191
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4178617000579834,
      "learning_rate": 1.9678625051911957e-06,
      "loss": 0.6436,
      "step": 29192
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.624004602432251,
      "learning_rate": 1.9672235887934064e-06,
      "loss": 0.8957,
      "step": 29193
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.030024766921997,
      "learning_rate": 1.966584672395617e-06,
      "loss": 0.8395,
      "step": 29194
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8366122245788574,
      "learning_rate": 1.965945755997828e-06,
      "loss": 0.8074,
      "step": 29195
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.797135353088379,
      "learning_rate": 1.9653068396000385e-06,
      "loss": 0.9503,
      "step": 29196
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7720263004302979,
      "learning_rate": 1.9646679232022492e-06,
      "loss": 0.8438,
      "step": 29197
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.056995153427124,
      "learning_rate": 1.96402900680446e-06,
      "loss": 1.1083,
      "step": 29198
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5835233926773071,
      "learning_rate": 1.9633900904066706e-06,
      "loss": 0.8683,
      "step": 29199
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.822311282157898,
      "learning_rate": 1.9627511740088813e-06,
      "loss": 0.8442,
      "step": 29200
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.260354518890381,
      "learning_rate": 1.9621122576110916e-06,
      "loss": 0.7429,
      "step": 29201
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.5345330238342285,
      "learning_rate": 1.9614733412133023e-06,
      "loss": 1.5526,
      "step": 29202
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.951802968978882,
      "learning_rate": 1.960834424815513e-06,
      "loss": 0.9009,
      "step": 29203
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8410437107086182,
      "learning_rate": 1.9601955084177237e-06,
      "loss": 0.8509,
      "step": 29204
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1799662113189697,
      "learning_rate": 1.9595565920199344e-06,
      "loss": 0.7437,
      "step": 29205
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.848547458648682,
      "learning_rate": 1.958917675622145e-06,
      "loss": 0.8858,
      "step": 29206
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.244788646697998,
      "learning_rate": 1.958278759224356e-06,
      "loss": 0.3397,
      "step": 29207
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7101836204528809,
      "learning_rate": 1.9576398428265665e-06,
      "loss": 0.8232,
      "step": 29208
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5680654048919678,
      "learning_rate": 1.9570009264287772e-06,
      "loss": 0.7341,
      "step": 29209
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.297471523284912,
      "learning_rate": 1.9563620100309875e-06,
      "loss": 0.7388,
      "step": 29210
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.749788522720337,
      "learning_rate": 1.955723093633198e-06,
      "loss": 0.8257,
      "step": 29211
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4830446243286133,
      "learning_rate": 1.955084177235409e-06,
      "loss": 0.7475,
      "step": 29212
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7182466983795166,
      "learning_rate": 1.9544452608376196e-06,
      "loss": 0.8529,
      "step": 29213
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3459415435791016,
      "learning_rate": 1.9538063444398303e-06,
      "loss": 0.929,
      "step": 29214
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.996775507926941,
      "learning_rate": 1.953167428042041e-06,
      "loss": 0.8969,
      "step": 29215
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.387747049331665,
      "learning_rate": 1.9525285116442517e-06,
      "loss": 0.8566,
      "step": 29216
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.412062168121338,
      "learning_rate": 1.9518895952464624e-06,
      "loss": 1.3412,
      "step": 29217
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.128019332885742,
      "learning_rate": 1.951250678848673e-06,
      "loss": 1.1618,
      "step": 29218
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9830172061920166,
      "learning_rate": 1.9506117624508834e-06,
      "loss": 0.9094,
      "step": 29219
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.911699652671814,
      "learning_rate": 1.949972846053094e-06,
      "loss": 0.7341,
      "step": 29220
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.119882106781006,
      "learning_rate": 1.9493339296553048e-06,
      "loss": 0.6541,
      "step": 29221
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.687746524810791,
      "learning_rate": 1.9486950132575155e-06,
      "loss": 0.7852,
      "step": 29222
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.633028268814087,
      "learning_rate": 1.948056096859726e-06,
      "loss": 0.8978,
      "step": 29223
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.7783966064453125,
      "learning_rate": 1.947417180461937e-06,
      "loss": 1.4503,
      "step": 29224
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.083836317062378,
      "learning_rate": 1.9467782640641476e-06,
      "loss": 0.7812,
      "step": 29225
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3296101093292236,
      "learning_rate": 1.9461393476663583e-06,
      "loss": 0.8256,
      "step": 29226
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7908722162246704,
      "learning_rate": 1.945500431268569e-06,
      "loss": 0.9175,
      "step": 29227
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.775292158126831,
      "learning_rate": 1.9448615148707793e-06,
      "loss": 0.9202,
      "step": 29228
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.9048309326171875,
      "learning_rate": 1.94422259847299e-06,
      "loss": 0.4679,
      "step": 29229
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.020087480545044,
      "learning_rate": 1.9435836820752007e-06,
      "loss": 1.0304,
      "step": 29230
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.655408501625061,
      "learning_rate": 1.942944765677411e-06,
      "loss": 0.7935,
      "step": 29231
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9818671941757202,
      "learning_rate": 1.9423058492796216e-06,
      "loss": 0.8168,
      "step": 29232
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.810550570487976,
      "learning_rate": 1.9416669328818323e-06,
      "loss": 0.879,
      "step": 29233
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7386983633041382,
      "learning_rate": 1.941028016484043e-06,
      "loss": 0.7995,
      "step": 29234
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0111472606658936,
      "learning_rate": 1.9403891000862537e-06,
      "loss": 0.8989,
      "step": 29235
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.120934963226318,
      "learning_rate": 1.9397501836884644e-06,
      "loss": 1.1215,
      "step": 29236
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.686900019645691,
      "learning_rate": 1.939111267290675e-06,
      "loss": 0.7677,
      "step": 29237
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.086958408355713,
      "learning_rate": 1.938472350892886e-06,
      "loss": 0.9933,
      "step": 29238
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.726971983909607,
      "learning_rate": 1.9378334344950965e-06,
      "loss": 0.8817,
      "step": 29239
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.227970600128174,
      "learning_rate": 1.9371945180973072e-06,
      "loss": 1.3429,
      "step": 29240
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0533924102783203,
      "learning_rate": 1.9365556016995175e-06,
      "loss": 0.6455,
      "step": 29241
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8755415678024292,
      "learning_rate": 1.935916685301728e-06,
      "loss": 1.0145,
      "step": 29242
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.996135711669922,
      "learning_rate": 1.935277768903939e-06,
      "loss": 0.6059,
      "step": 29243
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.190031051635742,
      "learning_rate": 1.9346388525061496e-06,
      "loss": 1.1247,
      "step": 29244
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2559521198272705,
      "learning_rate": 1.9339999361083603e-06,
      "loss": 0.878,
      "step": 29245
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3131113052368164,
      "learning_rate": 1.933361019710571e-06,
      "loss": 0.4002,
      "step": 29246
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7618614435195923,
      "learning_rate": 1.9327221033127817e-06,
      "loss": 0.8936,
      "step": 29247
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.036630868911743,
      "learning_rate": 1.9320831869149924e-06,
      "loss": 0.8279,
      "step": 29248
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6180225610733032,
      "learning_rate": 1.931444270517203e-06,
      "loss": 0.8071,
      "step": 29249
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4652035236358643,
      "learning_rate": 1.9308053541194134e-06,
      "loss": 0.7856,
      "step": 29250
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.1063127517700195,
      "learning_rate": 1.930166437721624e-06,
      "loss": 1.1951,
      "step": 29251
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8783355951309204,
      "learning_rate": 1.9295275213238348e-06,
      "loss": 0.7092,
      "step": 29252
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7347034215927124,
      "learning_rate": 1.9288886049260455e-06,
      "loss": 0.9259,
      "step": 29253
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.00242280960083,
      "learning_rate": 1.928249688528256e-06,
      "loss": 0.7132,
      "step": 29254
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6880396604537964,
      "learning_rate": 1.927610772130467e-06,
      "loss": 0.7965,
      "step": 29255
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6162601709365845,
      "learning_rate": 1.9269718557326776e-06,
      "loss": 0.7919,
      "step": 29256
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6116007566452026,
      "learning_rate": 1.9263329393348883e-06,
      "loss": 0.8475,
      "step": 29257
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.009370803833008,
      "learning_rate": 1.925694022937099e-06,
      "loss": 0.8572,
      "step": 29258
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.812556743621826,
      "learning_rate": 1.9250551065393093e-06,
      "loss": 1.1017,
      "step": 29259
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.712170958518982,
      "learning_rate": 1.92441619014152e-06,
      "loss": 0.8641,
      "step": 29260
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6675008535385132,
      "learning_rate": 1.9237772737437307e-06,
      "loss": 0.8508,
      "step": 29261
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.3531877994537354,
      "learning_rate": 1.9231383573459414e-06,
      "loss": 0.9035,
      "step": 29262
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0227746963500977,
      "learning_rate": 1.922499440948152e-06,
      "loss": 0.8673,
      "step": 29263
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.886791706085205,
      "learning_rate": 1.9218605245503628e-06,
      "loss": 0.8987,
      "step": 29264
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.107525587081909,
      "learning_rate": 1.9212216081525735e-06,
      "loss": 1.0726,
      "step": 29265
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3016629219055176,
      "learning_rate": 1.920582691754784e-06,
      "loss": 0.6335,
      "step": 29266
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.229159355163574,
      "learning_rate": 1.919943775356995e-06,
      "loss": 1.0628,
      "step": 29267
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5287023782730103,
      "learning_rate": 1.919304858959205e-06,
      "loss": 0.4027,
      "step": 29268
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6410140991210938,
      "learning_rate": 1.918665942561416e-06,
      "loss": 0.8215,
      "step": 29269
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6655032634735107,
      "learning_rate": 1.9180270261636265e-06,
      "loss": 0.8065,
      "step": 29270
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8169609308242798,
      "learning_rate": 1.9173881097658372e-06,
      "loss": 0.8232,
      "step": 29271
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6204661130905151,
      "learning_rate": 1.916749193368048e-06,
      "loss": 0.8589,
      "step": 29272
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0752551555633545,
      "learning_rate": 1.9161102769702586e-06,
      "loss": 0.7549,
      "step": 29273
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6497602462768555,
      "learning_rate": 1.9154713605724693e-06,
      "loss": 0.8636,
      "step": 29274
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6663891077041626,
      "learning_rate": 1.91483244417468e-06,
      "loss": 0.7618,
      "step": 29275
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5551272630691528,
      "learning_rate": 1.9141935277768907e-06,
      "loss": 0.8028,
      "step": 29276
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6689387559890747,
      "learning_rate": 1.913554611379101e-06,
      "loss": 0.8318,
      "step": 29277
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2384204864501953,
      "learning_rate": 1.9129156949813117e-06,
      "loss": 0.4719,
      "step": 29278
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.591893434524536,
      "learning_rate": 1.9122767785835224e-06,
      "loss": 0.9467,
      "step": 29279
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8445216417312622,
      "learning_rate": 1.911637862185733e-06,
      "loss": 0.7809,
      "step": 29280
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4654433727264404,
      "learning_rate": 1.910998945787944e-06,
      "loss": 0.7296,
      "step": 29281
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9356718063354492,
      "learning_rate": 1.9103600293901545e-06,
      "loss": 0.9154,
      "step": 29282
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.355139970779419,
      "learning_rate": 1.909721112992365e-06,
      "loss": 1.0091,
      "step": 29283
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2753443717956543,
      "learning_rate": 1.909082196594576e-06,
      "loss": 0.8917,
      "step": 29284
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0454018115997314,
      "learning_rate": 1.9084432801967866e-06,
      "loss": 0.7843,
      "step": 29285
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6872591972351074,
      "learning_rate": 1.907804363798997e-06,
      "loss": 0.8102,
      "step": 29286
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6355477571487427,
      "learning_rate": 1.9071654474012078e-06,
      "loss": 0.6831,
      "step": 29287
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5395727157592773,
      "learning_rate": 1.9065265310034185e-06,
      "loss": 0.7691,
      "step": 29288
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.208038806915283,
      "learning_rate": 1.905887614605629e-06,
      "loss": 1.0059,
      "step": 29289
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9647884368896484,
      "learning_rate": 1.9052486982078397e-06,
      "loss": 0.8058,
      "step": 29290
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.092053174972534,
      "learning_rate": 1.9046097818100504e-06,
      "loss": 0.7939,
      "step": 29291
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5994083881378174,
      "learning_rate": 1.903970865412261e-06,
      "loss": 0.7346,
      "step": 29292
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.100343704223633,
      "learning_rate": 1.9033319490144716e-06,
      "loss": 0.8656,
      "step": 29293
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.04677414894104,
      "learning_rate": 1.9026930326166823e-06,
      "loss": 0.7867,
      "step": 29294
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.760781168937683,
      "learning_rate": 1.902054116218893e-06,
      "loss": 0.8674,
      "step": 29295
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7428423166275024,
      "learning_rate": 1.9014151998211037e-06,
      "loss": 0.9755,
      "step": 29296
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9008499383926392,
      "learning_rate": 1.9007762834233144e-06,
      "loss": 0.884,
      "step": 29297
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7385839223861694,
      "learning_rate": 1.9001373670255249e-06,
      "loss": 0.968,
      "step": 29298
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5610860586166382,
      "learning_rate": 1.8994984506277356e-06,
      "loss": 0.8068,
      "step": 29299
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3060684204101562,
      "learning_rate": 1.8988595342299463e-06,
      "loss": 0.8143,
      "step": 29300
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.792486310005188,
      "learning_rate": 1.898220617832157e-06,
      "loss": 0.8192,
      "step": 29301
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6922249794006348,
      "learning_rate": 1.8975817014343674e-06,
      "loss": 0.6371,
      "step": 29302
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8524984121322632,
      "learning_rate": 1.8969427850365781e-06,
      "loss": 0.7651,
      "step": 29303
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.447856068611145,
      "learning_rate": 1.8963038686387888e-06,
      "loss": 0.7664,
      "step": 29304
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7472820281982422,
      "learning_rate": 1.8956649522409995e-06,
      "loss": 0.7718,
      "step": 29305
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.6016244888305664,
      "learning_rate": 1.8950260358432102e-06,
      "loss": 1.1087,
      "step": 29306
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2006866931915283,
      "learning_rate": 1.8943871194454207e-06,
      "loss": 0.9789,
      "step": 29307
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1304240226745605,
      "learning_rate": 1.8937482030476314e-06,
      "loss": 0.7707,
      "step": 29308
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0969910621643066,
      "learning_rate": 1.8931092866498421e-06,
      "loss": 0.8534,
      "step": 29309
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.802782654762268,
      "learning_rate": 1.8924703702520528e-06,
      "loss": 0.8527,
      "step": 29310
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.289332866668701,
      "learning_rate": 1.8918314538542633e-06,
      "loss": 0.6383,
      "step": 29311
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7606626749038696,
      "learning_rate": 1.891192537456474e-06,
      "loss": 0.7454,
      "step": 29312
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.2292256355285645,
      "learning_rate": 1.8905536210586847e-06,
      "loss": 1.1105,
      "step": 29313
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.502873182296753,
      "learning_rate": 1.8899147046608954e-06,
      "loss": 0.7703,
      "step": 29314
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.603763461112976,
      "learning_rate": 1.8892757882631061e-06,
      "loss": 0.6656,
      "step": 29315
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8187772035598755,
      "learning_rate": 1.8886368718653166e-06,
      "loss": 0.9322,
      "step": 29316
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.8076179027557373,
      "learning_rate": 1.8879979554675273e-06,
      "loss": 1.0979,
      "step": 29317
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.638796806335449,
      "learning_rate": 1.887359039069738e-06,
      "loss": 0.822,
      "step": 29318
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.1168532371521,
      "learning_rate": 1.8867201226719487e-06,
      "loss": 1.0139,
      "step": 29319
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.009432792663574,
      "learning_rate": 1.8860812062741592e-06,
      "loss": 0.9436,
      "step": 29320
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.851014614105225,
      "learning_rate": 1.88544228987637e-06,
      "loss": 1.5327,
      "step": 29321
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7002626657485962,
      "learning_rate": 1.8848033734785806e-06,
      "loss": 0.7509,
      "step": 29322
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.6522722244262695,
      "learning_rate": 1.8841644570807913e-06,
      "loss": 0.7425,
      "step": 29323
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9409215450286865,
      "learning_rate": 1.883525540683002e-06,
      "loss": 0.7983,
      "step": 29324
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6843664646148682,
      "learning_rate": 1.8828866242852125e-06,
      "loss": 0.6847,
      "step": 29325
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9996716976165771,
      "learning_rate": 1.882247707887423e-06,
      "loss": 0.8733,
      "step": 29326
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.003338098526001,
      "learning_rate": 1.8816087914896337e-06,
      "loss": 0.8914,
      "step": 29327
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6784849166870117,
      "learning_rate": 1.8809698750918442e-06,
      "loss": 0.8393,
      "step": 29328
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6989766359329224,
      "learning_rate": 1.8803309586940549e-06,
      "loss": 0.7733,
      "step": 29329
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8083648681640625,
      "learning_rate": 1.8796920422962656e-06,
      "loss": 0.8342,
      "step": 29330
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7001670598983765,
      "learning_rate": 1.8790531258984763e-06,
      "loss": 0.8988,
      "step": 29331
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.6320793628692627,
      "learning_rate": 1.878414209500687e-06,
      "loss": 0.7904,
      "step": 29332
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8422114849090576,
      "learning_rate": 1.8777752931028975e-06,
      "loss": 0.8959,
      "step": 29333
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4047659635543823,
      "learning_rate": 1.8771363767051082e-06,
      "loss": 0.3255,
      "step": 29334
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.711859703063965,
      "learning_rate": 1.8764974603073189e-06,
      "loss": 1.0478,
      "step": 29335
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.460393190383911,
      "learning_rate": 1.8758585439095296e-06,
      "loss": 0.7276,
      "step": 29336
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9262723922729492,
      "learning_rate": 1.87521962751174e-06,
      "loss": 0.8166,
      "step": 29337
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.5513763427734375,
      "learning_rate": 1.8745807111139507e-06,
      "loss": 0.9219,
      "step": 29338
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7556354999542236,
      "learning_rate": 1.8739417947161614e-06,
      "loss": 0.8025,
      "step": 29339
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.487508773803711,
      "learning_rate": 1.8733028783183721e-06,
      "loss": 0.4452,
      "step": 29340
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8060879707336426,
      "learning_rate": 1.8726639619205828e-06,
      "loss": 0.57,
      "step": 29341
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.281805992126465,
      "learning_rate": 1.8720250455227933e-06,
      "loss": 1.4245,
      "step": 29342
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8148549795150757,
      "learning_rate": 1.871386129125004e-06,
      "loss": 0.82,
      "step": 29343
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7216910123825073,
      "learning_rate": 1.8707472127272147e-06,
      "loss": 0.8265,
      "step": 29344
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4945876598358154,
      "learning_rate": 1.8701082963294254e-06,
      "loss": 0.8603,
      "step": 29345
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.062511444091797,
      "learning_rate": 1.869469379931636e-06,
      "loss": 0.7038,
      "step": 29346
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9738423824310303,
      "learning_rate": 1.8688304635338466e-06,
      "loss": 0.776,
      "step": 29347
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.068901300430298,
      "learning_rate": 1.8681915471360573e-06,
      "loss": 0.8575,
      "step": 29348
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3724961280822754,
      "learning_rate": 1.867552630738268e-06,
      "loss": 0.8675,
      "step": 29349
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2649903297424316,
      "learning_rate": 1.8669137143404787e-06,
      "loss": 0.7188,
      "step": 29350
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.043385028839111,
      "learning_rate": 1.8662747979426892e-06,
      "loss": 1.1625,
      "step": 29351
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.477372646331787,
      "learning_rate": 1.8656358815449e-06,
      "loss": 0.8269,
      "step": 29352
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.895242691040039,
      "learning_rate": 1.8649969651471106e-06,
      "loss": 0.8001,
      "step": 29353
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6478079557418823,
      "learning_rate": 1.8643580487493213e-06,
      "loss": 0.7142,
      "step": 29354
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9288758039474487,
      "learning_rate": 1.863719132351532e-06,
      "loss": 0.7776,
      "step": 29355
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8356021642684937,
      "learning_rate": 1.8630802159537425e-06,
      "loss": 0.8491,
      "step": 29356
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3441243171691895,
      "learning_rate": 1.8624412995559532e-06,
      "loss": 0.7016,
      "step": 29357
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5272369384765625,
      "learning_rate": 1.8618023831581639e-06,
      "loss": 0.7286,
      "step": 29358
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.533734679222107,
      "learning_rate": 1.8611634667603746e-06,
      "loss": 0.7291,
      "step": 29359
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8115613460540771,
      "learning_rate": 1.860524550362585e-06,
      "loss": 1.1085,
      "step": 29360
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.0304718017578125,
      "learning_rate": 1.8598856339647958e-06,
      "loss": 0.8526,
      "step": 29361
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.0199103355407715,
      "learning_rate": 1.8592467175670065e-06,
      "loss": 1.5176,
      "step": 29362
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.5934431552886963,
      "learning_rate": 1.8586078011692172e-06,
      "loss": 0.6233,
      "step": 29363
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.020321846008301,
      "learning_rate": 1.8579688847714279e-06,
      "loss": 0.882,
      "step": 29364
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.304002285003662,
      "learning_rate": 1.8573299683736384e-06,
      "loss": 1.0276,
      "step": 29365
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8116737604141235,
      "learning_rate": 1.856691051975849e-06,
      "loss": 0.8332,
      "step": 29366
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.03890323638916,
      "learning_rate": 1.8560521355780598e-06,
      "loss": 1.3958,
      "step": 29367
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.8495330810546875,
      "learning_rate": 1.8554132191802705e-06,
      "loss": 0.6812,
      "step": 29368
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.276888370513916,
      "learning_rate": 1.854774302782481e-06,
      "loss": 0.9111,
      "step": 29369
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6673626899719238,
      "learning_rate": 1.8541353863846917e-06,
      "loss": 0.8878,
      "step": 29370
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.342153072357178,
      "learning_rate": 1.8534964699869024e-06,
      "loss": 1.3127,
      "step": 29371
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8996707201004028,
      "learning_rate": 1.852857553589113e-06,
      "loss": 0.77,
      "step": 29372
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0119314193725586,
      "learning_rate": 1.8522186371913238e-06,
      "loss": 0.9878,
      "step": 29373
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8183181285858154,
      "learning_rate": 1.8515797207935342e-06,
      "loss": 0.8398,
      "step": 29374
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.355203151702881,
      "learning_rate": 1.850940804395745e-06,
      "loss": 0.8774,
      "step": 29375
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4974229335784912,
      "learning_rate": 1.8503018879979556e-06,
      "loss": 0.7077,
      "step": 29376
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9532318115234375,
      "learning_rate": 1.8496629716001663e-06,
      "loss": 0.7964,
      "step": 29377
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6814318895339966,
      "learning_rate": 1.8490240552023768e-06,
      "loss": 0.8451,
      "step": 29378
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1859519481658936,
      "learning_rate": 1.8483851388045875e-06,
      "loss": 0.7975,
      "step": 29379
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6821362972259521,
      "learning_rate": 1.8477462224067982e-06,
      "loss": 0.7292,
      "step": 29380
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5220600366592407,
      "learning_rate": 1.847107306009009e-06,
      "loss": 0.8597,
      "step": 29381
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4897031784057617,
      "learning_rate": 1.8464683896112196e-06,
      "loss": 0.7749,
      "step": 29382
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.8991317749023438,
      "learning_rate": 1.8458294732134301e-06,
      "loss": 0.8942,
      "step": 29383
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.995252013206482,
      "learning_rate": 1.8451905568156408e-06,
      "loss": 0.8912,
      "step": 29384
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9098774194717407,
      "learning_rate": 1.8445516404178515e-06,
      "loss": 0.8593,
      "step": 29385
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0345044136047363,
      "learning_rate": 1.8439127240200622e-06,
      "loss": 0.8619,
      "step": 29386
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.086871862411499,
      "learning_rate": 1.843273807622273e-06,
      "loss": 0.8136,
      "step": 29387
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7131099700927734,
      "learning_rate": 1.8426348912244834e-06,
      "loss": 0.8764,
      "step": 29388
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.551697015762329,
      "learning_rate": 1.8419959748266941e-06,
      "loss": 0.7275,
      "step": 29389
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6267869472503662,
      "learning_rate": 1.8413570584289048e-06,
      "loss": 0.7121,
      "step": 29390
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7362395524978638,
      "learning_rate": 1.8407181420311155e-06,
      "loss": 0.8576,
      "step": 29391
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6646887063980103,
      "learning_rate": 1.840079225633326e-06,
      "loss": 0.8675,
      "step": 29392
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9161561727523804,
      "learning_rate": 1.8394403092355367e-06,
      "loss": 0.8052,
      "step": 29393
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.498600006103516,
      "learning_rate": 1.8388013928377474e-06,
      "loss": 1.1383,
      "step": 29394
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.074831247329712,
      "learning_rate": 1.838162476439958e-06,
      "loss": 0.8974,
      "step": 29395
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8388961553573608,
      "learning_rate": 1.8375235600421688e-06,
      "loss": 0.8288,
      "step": 29396
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.5499439239501953,
      "learning_rate": 1.8368846436443793e-06,
      "loss": 0.8472,
      "step": 29397
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1697919368743896,
      "learning_rate": 1.83624572724659e-06,
      "loss": 0.8087,
      "step": 29398
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.9405934810638428,
      "learning_rate": 1.8356068108488007e-06,
      "loss": 0.873,
      "step": 29399
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6310193538665771,
      "learning_rate": 1.8349678944510114e-06,
      "loss": 0.7855,
      "step": 29400
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0658576488494873,
      "learning_rate": 1.8343289780532219e-06,
      "loss": 0.8465,
      "step": 29401
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7733081579208374,
      "learning_rate": 1.8336900616554326e-06,
      "loss": 0.7438,
      "step": 29402
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7987033128738403,
      "learning_rate": 1.8330511452576433e-06,
      "loss": 0.8875,
      "step": 29403
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9833590984344482,
      "learning_rate": 1.832412228859854e-06,
      "loss": 0.979,
      "step": 29404
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.552732229232788,
      "learning_rate": 1.8317733124620647e-06,
      "loss": 0.8631,
      "step": 29405
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.420548439025879,
      "learning_rate": 1.8311343960642752e-06,
      "loss": 0.6097,
      "step": 29406
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1883437633514404,
      "learning_rate": 1.8304954796664859e-06,
      "loss": 1.026,
      "step": 29407
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.421510219573975,
      "learning_rate": 1.8298565632686966e-06,
      "loss": 1.205,
      "step": 29408
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8950196504592896,
      "learning_rate": 1.8292176468709073e-06,
      "loss": 0.8376,
      "step": 29409
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.087834119796753,
      "learning_rate": 1.8285787304731177e-06,
      "loss": 0.8988,
      "step": 29410
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7538295984268188,
      "learning_rate": 1.8279398140753284e-06,
      "loss": 0.927,
      "step": 29411
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9101660251617432,
      "learning_rate": 1.8273008976775391e-06,
      "loss": 0.9022,
      "step": 29412
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1449248790740967,
      "learning_rate": 1.8266619812797498e-06,
      "loss": 0.7922,
      "step": 29413
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.699673175811768,
      "learning_rate": 1.8260230648819605e-06,
      "loss": 1.2814,
      "step": 29414
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.101022720336914,
      "learning_rate": 1.825384148484171e-06,
      "loss": 1.2032,
      "step": 29415
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3553979396820068,
      "learning_rate": 1.8247452320863817e-06,
      "loss": 0.3596,
      "step": 29416
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.791419267654419,
      "learning_rate": 1.8241063156885924e-06,
      "loss": 0.8694,
      "step": 29417
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4332799911499023,
      "learning_rate": 1.8234673992908031e-06,
      "loss": 0.5261,
      "step": 29418
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9283101558685303,
      "learning_rate": 1.8228284828930136e-06,
      "loss": 0.9544,
      "step": 29419
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.892113208770752,
      "learning_rate": 1.8221895664952243e-06,
      "loss": 1.3726,
      "step": 29420
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6475517749786377,
      "learning_rate": 1.821550650097435e-06,
      "loss": 0.8009,
      "step": 29421
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2089545726776123,
      "learning_rate": 1.8209117336996455e-06,
      "loss": 0.9549,
      "step": 29422
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7964587211608887,
      "learning_rate": 1.820272817301856e-06,
      "loss": 0.8244,
      "step": 29423
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6411937475204468,
      "learning_rate": 1.8196339009040667e-06,
      "loss": 0.8234,
      "step": 29424
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0978944301605225,
      "learning_rate": 1.8189949845062774e-06,
      "loss": 0.8933,
      "step": 29425
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.521619439125061,
      "learning_rate": 1.818356068108488e-06,
      "loss": 0.8319,
      "step": 29426
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8472617864608765,
      "learning_rate": 1.8177171517106986e-06,
      "loss": 0.9613,
      "step": 29427
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2500054836273193,
      "learning_rate": 1.8170782353129093e-06,
      "loss": 0.8135,
      "step": 29428
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.766615867614746,
      "learning_rate": 1.81643931891512e-06,
      "loss": 0.7961,
      "step": 29429
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5379934310913086,
      "learning_rate": 1.8158004025173307e-06,
      "loss": 0.3716,
      "step": 29430
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.4972615242004395,
      "learning_rate": 1.8151614861195414e-06,
      "loss": 0.9859,
      "step": 29431
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9943723678588867,
      "learning_rate": 1.8145225697217519e-06,
      "loss": 0.9077,
      "step": 29432
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7987874746322632,
      "learning_rate": 1.8138836533239626e-06,
      "loss": 1.1362,
      "step": 29433
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4013407230377197,
      "learning_rate": 1.8132447369261733e-06,
      "loss": 0.4421,
      "step": 29434
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.861891746520996,
      "learning_rate": 1.812605820528384e-06,
      "loss": 0.9383,
      "step": 29435
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.979148507118225,
      "learning_rate": 1.8119669041305945e-06,
      "loss": 1.0812,
      "step": 29436
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5667462348937988,
      "learning_rate": 1.8113279877328052e-06,
      "loss": 0.7536,
      "step": 29437
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.066629409790039,
      "learning_rate": 1.8106890713350159e-06,
      "loss": 0.7018,
      "step": 29438
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0492494106292725,
      "learning_rate": 1.8100501549372266e-06,
      "loss": 0.8367,
      "step": 29439
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0285017490386963,
      "learning_rate": 1.8094112385394373e-06,
      "loss": 0.7858,
      "step": 29440
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.980833888053894,
      "learning_rate": 1.8087723221416478e-06,
      "loss": 0.7496,
      "step": 29441
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7296797037124634,
      "learning_rate": 1.8081334057438585e-06,
      "loss": 0.8944,
      "step": 29442
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8822709321975708,
      "learning_rate": 1.8074944893460692e-06,
      "loss": 0.7748,
      "step": 29443
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3168888092041016,
      "learning_rate": 1.8068555729482799e-06,
      "loss": 0.8418,
      "step": 29444
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.536081314086914,
      "learning_rate": 1.8062166565504903e-06,
      "loss": 0.9493,
      "step": 29445
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3697123527526855,
      "learning_rate": 1.805577740152701e-06,
      "loss": 0.8181,
      "step": 29446
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8067307472229004,
      "learning_rate": 1.8049388237549117e-06,
      "loss": 0.7787,
      "step": 29447
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9075883626937866,
      "learning_rate": 1.8042999073571224e-06,
      "loss": 0.7613,
      "step": 29448
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4905967712402344,
      "learning_rate": 1.8036609909593331e-06,
      "loss": 0.9016,
      "step": 29449
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.767469048500061,
      "learning_rate": 1.8030220745615436e-06,
      "loss": 0.9934,
      "step": 29450
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.2236833572387695,
      "learning_rate": 1.8023831581637543e-06,
      "loss": 1.1097,
      "step": 29451
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.693626403808594,
      "learning_rate": 1.801744241765965e-06,
      "loss": 1.2558,
      "step": 29452
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9885631799697876,
      "learning_rate": 1.8011053253681757e-06,
      "loss": 1.0092,
      "step": 29453
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.4152324199676514,
      "learning_rate": 1.8004664089703864e-06,
      "loss": 1.0574,
      "step": 29454
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4152010679244995,
      "learning_rate": 1.799827492572597e-06,
      "loss": 0.7866,
      "step": 29455
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2453103065490723,
      "learning_rate": 1.7991885761748076e-06,
      "loss": 1.0099,
      "step": 29456
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5099717378616333,
      "learning_rate": 1.7985496597770183e-06,
      "loss": 0.7175,
      "step": 29457
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.876778244972229,
      "learning_rate": 1.797910743379229e-06,
      "loss": 0.8335,
      "step": 29458
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6946772336959839,
      "learning_rate": 1.7972718269814395e-06,
      "loss": 0.7809,
      "step": 29459
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.882135272026062,
      "learning_rate": 1.7966329105836502e-06,
      "loss": 0.9196,
      "step": 29460
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.860684633255005,
      "learning_rate": 1.795993994185861e-06,
      "loss": 0.8145,
      "step": 29461
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9009506702423096,
      "learning_rate": 1.7953550777880716e-06,
      "loss": 0.8544,
      "step": 29462
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.298496246337891,
      "learning_rate": 1.7947161613902823e-06,
      "loss": 1.024,
      "step": 29463
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8044006824493408,
      "learning_rate": 1.7940772449924928e-06,
      "loss": 0.8092,
      "step": 29464
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.01271653175354,
      "learning_rate": 1.7934383285947035e-06,
      "loss": 0.7983,
      "step": 29465
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1381053924560547,
      "learning_rate": 1.7927994121969142e-06,
      "loss": 0.8504,
      "step": 29466
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9597669839859009,
      "learning_rate": 1.7921604957991249e-06,
      "loss": 0.8142,
      "step": 29467
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.427712917327881,
      "learning_rate": 1.7915215794013354e-06,
      "loss": 1.0995,
      "step": 29468
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5072957277297974,
      "learning_rate": 1.790882663003546e-06,
      "loss": 0.7714,
      "step": 29469
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1848092079162598,
      "learning_rate": 1.7902437466057568e-06,
      "loss": 0.9806,
      "step": 29470
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.998889446258545,
      "learning_rate": 1.7896048302079675e-06,
      "loss": 1.128,
      "step": 29471
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4838597774505615,
      "learning_rate": 1.7889659138101782e-06,
      "loss": 0.7756,
      "step": 29472
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6309205293655396,
      "learning_rate": 1.7883269974123887e-06,
      "loss": 0.8472,
      "step": 29473
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.101407527923584,
      "learning_rate": 1.7876880810145994e-06,
      "loss": 1.0072,
      "step": 29474
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0323495864868164,
      "learning_rate": 1.78704916461681e-06,
      "loss": 0.8873,
      "step": 29475
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.101343870162964,
      "learning_rate": 1.7864102482190208e-06,
      "loss": 0.8817,
      "step": 29476
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2306318283081055,
      "learning_rate": 1.7857713318212313e-06,
      "loss": 0.9367,
      "step": 29477
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.7002758979797363,
      "learning_rate": 1.785132415423442e-06,
      "loss": 0.9933,
      "step": 29478
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.378704786300659,
      "learning_rate": 1.7844934990256527e-06,
      "loss": 1.1856,
      "step": 29479
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.540339469909668,
      "learning_rate": 1.7838545826278634e-06,
      "loss": 0.731,
      "step": 29480
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.024783134460449,
      "learning_rate": 1.783215666230074e-06,
      "loss": 1.0342,
      "step": 29481
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0634453296661377,
      "learning_rate": 1.7825767498322845e-06,
      "loss": 0.8327,
      "step": 29482
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6084685325622559,
      "learning_rate": 1.7819378334344952e-06,
      "loss": 0.7699,
      "step": 29483
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.552389144897461,
      "learning_rate": 1.781298917036706e-06,
      "loss": 0.9666,
      "step": 29484
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.652117371559143,
      "learning_rate": 1.7806600006389166e-06,
      "loss": 0.9226,
      "step": 29485
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7586511373519897,
      "learning_rate": 1.7800210842411271e-06,
      "loss": 0.732,
      "step": 29486
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8035987615585327,
      "learning_rate": 1.7793821678433378e-06,
      "loss": 0.8371,
      "step": 29487
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.594748497009277,
      "learning_rate": 1.7787432514455485e-06,
      "loss": 1.2326,
      "step": 29488
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.116832971572876,
      "learning_rate": 1.7781043350477592e-06,
      "loss": 0.9149,
      "step": 29489
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.40043830871582,
      "learning_rate": 1.77746541864997e-06,
      "loss": 1.1261,
      "step": 29490
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.96074640750885,
      "learning_rate": 1.7768265022521804e-06,
      "loss": 1.0493,
      "step": 29491
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.210146903991699,
      "learning_rate": 1.7761875858543911e-06,
      "loss": 0.7849,
      "step": 29492
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.3234612941741943,
      "learning_rate": 1.7755486694566018e-06,
      "loss": 0.7511,
      "step": 29493
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.65667200088501,
      "learning_rate": 1.7749097530588125e-06,
      "loss": 1.2237,
      "step": 29494
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5833123922348022,
      "learning_rate": 1.7742708366610232e-06,
      "loss": 0.7435,
      "step": 29495
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.5022664070129395,
      "learning_rate": 1.7736319202632337e-06,
      "loss": 1.3621,
      "step": 29496
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6972161531448364,
      "learning_rate": 1.7729930038654444e-06,
      "loss": 0.8765,
      "step": 29497
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6600321531295776,
      "learning_rate": 1.7723540874676551e-06,
      "loss": 1.1025,
      "step": 29498
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4522581100463867,
      "learning_rate": 1.7717151710698658e-06,
      "loss": 0.5524,
      "step": 29499
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2096869945526123,
      "learning_rate": 1.7710762546720763e-06,
      "loss": 0.8584,
      "step": 29500
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0437254905700684,
      "learning_rate": 1.770437338274287e-06,
      "loss": 0.7394,
      "step": 29501
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.14202618598938,
      "learning_rate": 1.7697984218764977e-06,
      "loss": 0.8311,
      "step": 29502
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7707358598709106,
      "learning_rate": 1.7691595054787084e-06,
      "loss": 0.8448,
      "step": 29503
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.3771417140960693,
      "learning_rate": 1.768520589080919e-06,
      "loss": 0.9004,
      "step": 29504
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.48932945728302,
      "learning_rate": 1.7678816726831296e-06,
      "loss": 0.8428,
      "step": 29505
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.2239933013916016,
      "learning_rate": 1.7672427562853403e-06,
      "loss": 1.0792,
      "step": 29506
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.7987520694732666,
      "learning_rate": 1.766603839887551e-06,
      "loss": 0.8225,
      "step": 29507
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.096776008605957,
      "learning_rate": 1.7659649234897617e-06,
      "loss": 1.7057,
      "step": 29508
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7488813400268555,
      "learning_rate": 1.7653260070919722e-06,
      "loss": 0.7711,
      "step": 29509
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6810402870178223,
      "learning_rate": 1.7646870906941829e-06,
      "loss": 0.8805,
      "step": 29510
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.399043083190918,
      "learning_rate": 1.7640481742963936e-06,
      "loss": 1.2665,
      "step": 29511
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.6470165252685547,
      "learning_rate": 1.7634092578986043e-06,
      "loss": 0.7816,
      "step": 29512
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3143243789672852,
      "learning_rate": 1.762770341500815e-06,
      "loss": 0.3965,
      "step": 29513
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.409630298614502,
      "learning_rate": 1.7621314251030255e-06,
      "loss": 0.4339,
      "step": 29514
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.029995918273926,
      "learning_rate": 1.7614925087052362e-06,
      "loss": 0.962,
      "step": 29515
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.164647340774536,
      "learning_rate": 1.7608535923074469e-06,
      "loss": 0.8918,
      "step": 29516
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.292039394378662,
      "learning_rate": 1.7602146759096576e-06,
      "loss": 0.9186,
      "step": 29517
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.629112958908081,
      "learning_rate": 1.759575759511868e-06,
      "loss": 0.7374,
      "step": 29518
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8088003396987915,
      "learning_rate": 1.7589368431140785e-06,
      "loss": 0.8365,
      "step": 29519
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5932531356811523,
      "learning_rate": 1.7582979267162892e-06,
      "loss": 0.7267,
      "step": 29520
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.802096128463745,
      "learning_rate": 1.7576590103185e-06,
      "loss": 0.6101,
      "step": 29521
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.10945200920105,
      "learning_rate": 1.7570200939207104e-06,
      "loss": 0.8273,
      "step": 29522
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.16133975982666,
      "learning_rate": 1.7563811775229211e-06,
      "loss": 0.6966,
      "step": 29523
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.747332215309143,
      "learning_rate": 1.7557422611251318e-06,
      "loss": 0.861,
      "step": 29524
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3522151708602905,
      "learning_rate": 1.7551033447273425e-06,
      "loss": 0.383,
      "step": 29525
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7943967580795288,
      "learning_rate": 1.754464428329553e-06,
      "loss": 0.7296,
      "step": 29526
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9364317655563354,
      "learning_rate": 1.7538255119317637e-06,
      "loss": 0.9316,
      "step": 29527
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9750280380249023,
      "learning_rate": 1.7531865955339744e-06,
      "loss": 0.8377,
      "step": 29528
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.964484214782715,
      "learning_rate": 1.7525476791361851e-06,
      "loss": 0.9853,
      "step": 29529
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7470015287399292,
      "learning_rate": 1.7519087627383958e-06,
      "loss": 0.7048,
      "step": 29530
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.2701189517974854,
      "learning_rate": 1.7512698463406063e-06,
      "loss": 0.9076,
      "step": 29531
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.008310317993164,
      "learning_rate": 1.750630929942817e-06,
      "loss": 1.0818,
      "step": 29532
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5581355094909668,
      "learning_rate": 1.7499920135450277e-06,
      "loss": 1.0588,
      "step": 29533
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.4099490642547607,
      "learning_rate": 1.7493530971472384e-06,
      "loss": 0.9917,
      "step": 29534
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8751935958862305,
      "learning_rate": 1.7487141807494489e-06,
      "loss": 0.8652,
      "step": 29535
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.2786264419555664,
      "learning_rate": 1.7480752643516596e-06,
      "loss": 0.8047,
      "step": 29536
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5477951765060425,
      "learning_rate": 1.7474363479538703e-06,
      "loss": 0.8223,
      "step": 29537
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7671056985855103,
      "learning_rate": 1.746797431556081e-06,
      "loss": 1.1098,
      "step": 29538
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.820798635482788,
      "learning_rate": 1.7461585151582917e-06,
      "loss": 1.3294,
      "step": 29539
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9004242420196533,
      "learning_rate": 1.7455195987605022e-06,
      "loss": 0.8428,
      "step": 29540
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6980063915252686,
      "learning_rate": 1.7448806823627129e-06,
      "loss": 0.7474,
      "step": 29541
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7181941270828247,
      "learning_rate": 1.7442417659649236e-06,
      "loss": 0.8659,
      "step": 29542
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.7307868003845215,
      "learning_rate": 1.7436028495671343e-06,
      "loss": 1.1376,
      "step": 29543
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.3971848487854,
      "learning_rate": 1.7429639331693448e-06,
      "loss": 1.3952,
      "step": 29544
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8443855047225952,
      "learning_rate": 1.7423250167715555e-06,
      "loss": 0.8822,
      "step": 29545
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.524242877960205,
      "learning_rate": 1.7416861003737662e-06,
      "loss": 1.3429,
      "step": 29546
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6575851440429688,
      "learning_rate": 1.7410471839759769e-06,
      "loss": 0.8082,
      "step": 29547
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.8487465381622314,
      "learning_rate": 1.7404082675781876e-06,
      "loss": 1.0269,
      "step": 29548
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6513659954071045,
      "learning_rate": 1.739769351180398e-06,
      "loss": 0.8076,
      "step": 29549
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.2329695224761963,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 1.2487,
      "step": 29550
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5224637985229492,
      "learning_rate": 1.7384915183848195e-06,
      "loss": 0.8063,
      "step": 29551
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.257507562637329,
      "learning_rate": 1.7378526019870302e-06,
      "loss": 0.9546,
      "step": 29552
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6025071144104004,
      "learning_rate": 1.7372136855892409e-06,
      "loss": 0.7402,
      "step": 29553
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.5590591430664062,
      "learning_rate": 1.7365747691914513e-06,
      "loss": 1.0184,
      "step": 29554
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7279192209243774,
      "learning_rate": 1.735935852793662e-06,
      "loss": 0.8641,
      "step": 29555
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.802578330039978,
      "learning_rate": 1.7352969363958727e-06,
      "loss": 0.7979,
      "step": 29556
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0358173847198486,
      "learning_rate": 1.7346580199980834e-06,
      "loss": 0.8157,
      "step": 29557
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0500619411468506,
      "learning_rate": 1.734019103600294e-06,
      "loss": 0.8664,
      "step": 29558
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6309229135513306,
      "learning_rate": 1.7333801872025046e-06,
      "loss": 0.7029,
      "step": 29559
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.035714864730835,
      "learning_rate": 1.7327412708047153e-06,
      "loss": 0.8978,
      "step": 29560
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.695929527282715,
      "learning_rate": 1.732102354406926e-06,
      "loss": 1.0133,
      "step": 29561
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.410885810852051,
      "learning_rate": 1.7314634380091367e-06,
      "loss": 1.3855,
      "step": 29562
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9382025003433228,
      "learning_rate": 1.7308245216113472e-06,
      "loss": 0.7556,
      "step": 29563
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0059874057769775,
      "learning_rate": 1.730185605213558e-06,
      "loss": 0.8403,
      "step": 29564
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.577343225479126,
      "learning_rate": 1.7295466888157686e-06,
      "loss": 0.8122,
      "step": 29565
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.523216962814331,
      "learning_rate": 1.7289077724179793e-06,
      "loss": 0.7637,
      "step": 29566
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5550183057785034,
      "learning_rate": 1.7282688560201898e-06,
      "loss": 0.8273,
      "step": 29567
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8390756845474243,
      "learning_rate": 1.7276299396224005e-06,
      "loss": 0.8759,
      "step": 29568
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7791428565979004,
      "learning_rate": 1.7269910232246112e-06,
      "loss": 0.9114,
      "step": 29569
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.279094934463501,
      "learning_rate": 1.726352106826822e-06,
      "loss": 0.3272,
      "step": 29570
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.108804941177368,
      "learning_rate": 1.7257131904290326e-06,
      "loss": 0.9942,
      "step": 29571
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1538407802581787,
      "learning_rate": 1.725074274031243e-06,
      "loss": 0.8768,
      "step": 29572
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.2446656227111816,
      "learning_rate": 1.7244353576334538e-06,
      "loss": 0.7463,
      "step": 29573
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.5234529972076416,
      "learning_rate": 1.7237964412356645e-06,
      "loss": 1.0225,
      "step": 29574
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8823624849319458,
      "learning_rate": 1.7231575248378752e-06,
      "loss": 0.7789,
      "step": 29575
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.395716905593872,
      "learning_rate": 1.7225186084400857e-06,
      "loss": 0.8343,
      "step": 29576
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.291456699371338,
      "learning_rate": 1.7218796920422964e-06,
      "loss": 0.8907,
      "step": 29577
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0455515384674072,
      "learning_rate": 1.721240775644507e-06,
      "loss": 0.7513,
      "step": 29578
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.035161018371582,
      "learning_rate": 1.7206018592467178e-06,
      "loss": 0.8457,
      "step": 29579
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.5841903686523438,
      "learning_rate": 1.7199629428489285e-06,
      "loss": 0.9016,
      "step": 29580
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7038978338241577,
      "learning_rate": 1.719324026451139e-06,
      "loss": 0.88,
      "step": 29581
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0057623386383057,
      "learning_rate": 1.7186851100533497e-06,
      "loss": 0.7006,
      "step": 29582
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1957907676696777,
      "learning_rate": 1.7180461936555604e-06,
      "loss": 0.7555,
      "step": 29583
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.5616700649261475,
      "learning_rate": 1.717407277257771e-06,
      "loss": 0.8868,
      "step": 29584
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9621167182922363,
      "learning_rate": 1.7167683608599816e-06,
      "loss": 0.8828,
      "step": 29585
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7721796035766602,
      "learning_rate": 1.7161294444621923e-06,
      "loss": 0.933,
      "step": 29586
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8641666173934937,
      "learning_rate": 1.715490528064403e-06,
      "loss": 0.7726,
      "step": 29587
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.137037515640259,
      "learning_rate": 1.7148516116666137e-06,
      "loss": 0.7096,
      "step": 29588
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0240864753723145,
      "learning_rate": 1.7142126952688244e-06,
      "loss": 0.8145,
      "step": 29589
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.574873685836792,
      "learning_rate": 1.7135737788710348e-06,
      "loss": 0.7491,
      "step": 29590
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6953999996185303,
      "learning_rate": 1.7129348624732455e-06,
      "loss": 0.8448,
      "step": 29591
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.6814045906066895,
      "learning_rate": 1.7122959460754562e-06,
      "loss": 1.0461,
      "step": 29592
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9955674409866333,
      "learning_rate": 1.711657029677667e-06,
      "loss": 0.4307,
      "step": 29593
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6599494218826294,
      "learning_rate": 1.7110181132798776e-06,
      "loss": 0.8218,
      "step": 29594
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6602909564971924,
      "learning_rate": 1.7103791968820881e-06,
      "loss": 0.5745,
      "step": 29595
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.1159064769744873,
      "learning_rate": 1.7097402804842988e-06,
      "loss": 0.7539,
      "step": 29596
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.160018444061279,
      "learning_rate": 1.7091013640865095e-06,
      "loss": 1.0533,
      "step": 29597
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5528671741485596,
      "learning_rate": 1.7084624476887202e-06,
      "loss": 0.7443,
      "step": 29598
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4129719734191895,
      "learning_rate": 1.7078235312909307e-06,
      "loss": 0.3591,
      "step": 29599
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6594853401184082,
      "learning_rate": 1.7071846148931414e-06,
      "loss": 0.831,
      "step": 29600
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.8331003189086914,
      "learning_rate": 1.7065456984953521e-06,
      "loss": 0.7808,
      "step": 29601
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.514151692390442,
      "learning_rate": 1.7059067820975628e-06,
      "loss": 0.3533,
      "step": 29602
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8561375141143799,
      "learning_rate": 1.7052678656997735e-06,
      "loss": 0.814,
      "step": 29603
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8801994323730469,
      "learning_rate": 1.704628949301984e-06,
      "loss": 0.9025,
      "step": 29604
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7733126878738403,
      "learning_rate": 1.7039900329041947e-06,
      "loss": 0.8259,
      "step": 29605
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.9478330612182617,
      "learning_rate": 1.7033511165064054e-06,
      "loss": 0.4983,
      "step": 29606
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0480284690856934,
      "learning_rate": 1.7027122001086161e-06,
      "loss": 1.2376,
      "step": 29607
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7892471551895142,
      "learning_rate": 1.7020732837108266e-06,
      "loss": 0.8778,
      "step": 29608
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5229780673980713,
      "learning_rate": 1.7014343673130373e-06,
      "loss": 0.7599,
      "step": 29609
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.145453691482544,
      "learning_rate": 1.700795450915248e-06,
      "loss": 0.8403,
      "step": 29610
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0574634075164795,
      "learning_rate": 1.7001565345174587e-06,
      "loss": 0.8627,
      "step": 29611
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3982326984405518,
      "learning_rate": 1.6995176181196694e-06,
      "loss": 0.8617,
      "step": 29612
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2685487270355225,
      "learning_rate": 1.6988787017218799e-06,
      "loss": 0.7064,
      "step": 29613
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5912531614303589,
      "learning_rate": 1.6982397853240906e-06,
      "loss": 0.7836,
      "step": 29614
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5567377805709839,
      "learning_rate": 1.6976008689263013e-06,
      "loss": 0.7815,
      "step": 29615
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3417046070098877,
      "learning_rate": 1.6969619525285116e-06,
      "loss": 0.7454,
      "step": 29616
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8386964797973633,
      "learning_rate": 1.6963230361307223e-06,
      "loss": 0.8376,
      "step": 29617
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.046779155731201,
      "learning_rate": 1.695684119732933e-06,
      "loss": 1.2248,
      "step": 29618
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5841546058654785,
      "learning_rate": 1.6950452033351437e-06,
      "loss": 0.7528,
      "step": 29619
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.558621883392334,
      "learning_rate": 1.6944062869373544e-06,
      "loss": 0.9443,
      "step": 29620
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1326308250427246,
      "learning_rate": 1.6937673705395649e-06,
      "loss": 0.8867,
      "step": 29621
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3987765312194824,
      "learning_rate": 1.6931284541417756e-06,
      "loss": 0.9136,
      "step": 29622
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.544458389282227,
      "learning_rate": 1.6924895377439863e-06,
      "loss": 0.9183,
      "step": 29623
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4700899124145508,
      "learning_rate": 1.691850621346197e-06,
      "loss": 0.7938,
      "step": 29624
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6232688426971436,
      "learning_rate": 1.6912117049484074e-06,
      "loss": 0.7291,
      "step": 29625
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.396373987197876,
      "learning_rate": 1.6905727885506181e-06,
      "loss": 0.8005,
      "step": 29626
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9060661792755127,
      "learning_rate": 1.6899338721528288e-06,
      "loss": 0.831,
      "step": 29627
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6252353191375732,
      "learning_rate": 1.6892949557550395e-06,
      "loss": 0.8478,
      "step": 29628
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1951100826263428,
      "learning_rate": 1.6886560393572502e-06,
      "loss": 0.7836,
      "step": 29629
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.246826171875,
      "learning_rate": 1.6880171229594607e-06,
      "loss": 0.7859,
      "step": 29630
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6703298091888428,
      "learning_rate": 1.6873782065616714e-06,
      "loss": 0.9521,
      "step": 29631
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.815800189971924,
      "learning_rate": 1.6867392901638821e-06,
      "loss": 0.7699,
      "step": 29632
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5194175243377686,
      "learning_rate": 1.6861003737660928e-06,
      "loss": 0.8301,
      "step": 29633
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.254650592803955,
      "learning_rate": 1.6854614573683033e-06,
      "loss": 0.8767,
      "step": 29634
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.7528223991394043,
      "learning_rate": 1.684822540970514e-06,
      "loss": 0.7041,
      "step": 29635
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.003777027130127,
      "learning_rate": 1.6841836245727247e-06,
      "loss": 1.0787,
      "step": 29636
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7167834043502808,
      "learning_rate": 1.6835447081749354e-06,
      "loss": 0.7452,
      "step": 29637
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.710926055908203,
      "learning_rate": 1.6829057917771461e-06,
      "loss": 1.032,
      "step": 29638
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.753201723098755,
      "learning_rate": 1.6822668753793566e-06,
      "loss": 0.8419,
      "step": 29639
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2826777696609497,
      "learning_rate": 1.6816279589815673e-06,
      "loss": 0.3378,
      "step": 29640
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8728069067001343,
      "learning_rate": 1.680989042583778e-06,
      "loss": 0.795,
      "step": 29641
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9095134735107422,
      "learning_rate": 1.6803501261859887e-06,
      "loss": 0.7704,
      "step": 29642
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.031508207321167,
      "learning_rate": 1.6797112097881992e-06,
      "loss": 0.9531,
      "step": 29643
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.202889919281006,
      "learning_rate": 1.6790722933904099e-06,
      "loss": 1.0613,
      "step": 29644
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9791656732559204,
      "learning_rate": 1.6784333769926206e-06,
      "loss": 0.886,
      "step": 29645
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5148056745529175,
      "learning_rate": 1.6777944605948313e-06,
      "loss": 0.3748,
      "step": 29646
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6749855279922485,
      "learning_rate": 1.677155544197042e-06,
      "loss": 0.794,
      "step": 29647
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.06807541847229,
      "learning_rate": 1.6765166277992525e-06,
      "loss": 0.629,
      "step": 29648
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.663745403289795,
      "learning_rate": 1.6758777114014632e-06,
      "loss": 1.3062,
      "step": 29649
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6316014528274536,
      "learning_rate": 1.6752387950036739e-06,
      "loss": 0.7855,
      "step": 29650
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.230634927749634,
      "learning_rate": 1.6745998786058846e-06,
      "loss": 0.8729,
      "step": 29651
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6875625848770142,
      "learning_rate": 1.6739609622080953e-06,
      "loss": 0.8006,
      "step": 29652
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.2866692543029785,
      "learning_rate": 1.6733220458103058e-06,
      "loss": 0.8111,
      "step": 29653
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.914714813232422,
      "learning_rate": 1.6726831294125165e-06,
      "loss": 0.9951,
      "step": 29654
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.745062232017517,
      "learning_rate": 1.6720442130147272e-06,
      "loss": 0.8971,
      "step": 29655
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.2335968017578125,
      "learning_rate": 1.6714052966169379e-06,
      "loss": 1.4108,
      "step": 29656
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.567999839782715,
      "learning_rate": 1.6707663802191484e-06,
      "loss": 1.0781,
      "step": 29657
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.433783769607544,
      "learning_rate": 1.670127463821359e-06,
      "loss": 0.3728,
      "step": 29658
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.122330665588379,
      "learning_rate": 1.6694885474235698e-06,
      "loss": 0.9001,
      "step": 29659
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.290728807449341,
      "learning_rate": 1.6688496310257805e-06,
      "loss": 0.8694,
      "step": 29660
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.650928020477295,
      "learning_rate": 1.6682107146279912e-06,
      "loss": 0.8491,
      "step": 29661
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.539940357208252,
      "learning_rate": 1.6675717982302016e-06,
      "loss": 1.4326,
      "step": 29662
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.112128496170044,
      "learning_rate": 1.6669328818324123e-06,
      "loss": 0.8943,
      "step": 29663
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.598227620124817,
      "learning_rate": 1.666293965434623e-06,
      "loss": 0.7871,
      "step": 29664
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.349538564682007,
      "learning_rate": 1.6656550490368337e-06,
      "loss": 0.8576,
      "step": 29665
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.621739387512207,
      "learning_rate": 1.6650161326390442e-06,
      "loss": 0.7667,
      "step": 29666
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8753094673156738,
      "learning_rate": 1.664377216241255e-06,
      "loss": 0.9294,
      "step": 29667
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.796488881111145,
      "learning_rate": 1.6637382998434656e-06,
      "loss": 0.897,
      "step": 29668
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8686538934707642,
      "learning_rate": 1.6630993834456763e-06,
      "loss": 0.8686,
      "step": 29669
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.657761335372925,
      "learning_rate": 1.662460467047887e-06,
      "loss": 0.832,
      "step": 29670
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.030080556869507,
      "learning_rate": 1.6618215506500975e-06,
      "loss": 0.7989,
      "step": 29671
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5460870265960693,
      "learning_rate": 1.6611826342523082e-06,
      "loss": 0.7657,
      "step": 29672
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7272659540176392,
      "learning_rate": 1.660543717854519e-06,
      "loss": 0.8505,
      "step": 29673
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.910485744476318,
      "learning_rate": 1.6599048014567296e-06,
      "loss": 1.2203,
      "step": 29674
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.2554802894592285,
      "learning_rate": 1.6592658850589401e-06,
      "loss": 0.986,
      "step": 29675
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6936439275741577,
      "learning_rate": 1.6586269686611508e-06,
      "loss": 0.8614,
      "step": 29676
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.25014591217041,
      "learning_rate": 1.6579880522633615e-06,
      "loss": 0.8224,
      "step": 29677
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.155846118927002,
      "learning_rate": 1.6573491358655722e-06,
      "loss": 1.0321,
      "step": 29678
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.481911540031433,
      "learning_rate": 1.656710219467783e-06,
      "loss": 0.7813,
      "step": 29679
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.732781171798706,
      "learning_rate": 1.6560713030699934e-06,
      "loss": 0.5286,
      "step": 29680
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5400980710983276,
      "learning_rate": 1.655432386672204e-06,
      "loss": 0.7507,
      "step": 29681
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.551497220993042,
      "learning_rate": 1.6547934702744148e-06,
      "loss": 0.7258,
      "step": 29682
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.645480155944824,
      "learning_rate": 1.6541545538766255e-06,
      "loss": 0.6041,
      "step": 29683
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6369832754135132,
      "learning_rate": 1.653515637478836e-06,
      "loss": 0.7098,
      "step": 29684
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.8355937004089355,
      "learning_rate": 1.6528767210810467e-06,
      "loss": 1.2282,
      "step": 29685
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.2606520652771,
      "learning_rate": 1.6522378046832574e-06,
      "loss": 1.4125,
      "step": 29686
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.733839750289917,
      "learning_rate": 1.651598888285468e-06,
      "loss": 0.731,
      "step": 29687
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.4558985233306885,
      "learning_rate": 1.6509599718876788e-06,
      "loss": 0.5957,
      "step": 29688
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5934553146362305,
      "learning_rate": 1.6503210554898893e-06,
      "loss": 0.788,
      "step": 29689
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.842282772064209,
      "learning_rate": 1.6496821390921e-06,
      "loss": 0.9133,
      "step": 29690
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.513182282447815,
      "learning_rate": 1.6490432226943107e-06,
      "loss": 0.7482,
      "step": 29691
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.511888861656189,
      "learning_rate": 1.6484043062965214e-06,
      "loss": 0.6854,
      "step": 29692
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.645672559738159,
      "learning_rate": 1.647765389898732e-06,
      "loss": 0.8083,
      "step": 29693
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9879405498504639,
      "learning_rate": 1.6471264735009426e-06,
      "loss": 0.7914,
      "step": 29694
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6542519330978394,
      "learning_rate": 1.6464875571031533e-06,
      "loss": 0.7893,
      "step": 29695
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.24314546585083,
      "learning_rate": 1.645848640705364e-06,
      "loss": 0.7157,
      "step": 29696
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.5144312381744385,
      "learning_rate": 1.6452097243075747e-06,
      "loss": 0.7618,
      "step": 29697
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0034146308898926,
      "learning_rate": 1.6445708079097851e-06,
      "loss": 0.9086,
      "step": 29698
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.617656946182251,
      "learning_rate": 1.6439318915119958e-06,
      "loss": 0.7866,
      "step": 29699
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9342844486236572,
      "learning_rate": 1.6432929751142065e-06,
      "loss": 0.7695,
      "step": 29700
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.42179536819458,
      "learning_rate": 1.6426540587164172e-06,
      "loss": 1.2721,
      "step": 29701
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0788352489471436,
      "learning_rate": 1.642015142318628e-06,
      "loss": 0.7268,
      "step": 29702
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6374622583389282,
      "learning_rate": 1.6413762259208384e-06,
      "loss": 0.831,
      "step": 29703
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.8491196632385254,
      "learning_rate": 1.6407373095230491e-06,
      "loss": 0.791,
      "step": 29704
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.703514575958252,
      "learning_rate": 1.6400983931252598e-06,
      "loss": 0.7936,
      "step": 29705
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.017287254333496,
      "learning_rate": 1.6394594767274705e-06,
      "loss": 0.766,
      "step": 29706
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2877657413482666,
      "learning_rate": 1.638820560329681e-06,
      "loss": 0.3289,
      "step": 29707
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5106375217437744,
      "learning_rate": 1.6381816439318917e-06,
      "loss": 0.7301,
      "step": 29708
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.542734146118164,
      "learning_rate": 1.6375427275341024e-06,
      "loss": 0.8419,
      "step": 29709
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.844865083694458,
      "learning_rate": 1.6369038111363131e-06,
      "loss": 0.7947,
      "step": 29710
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9079524278640747,
      "learning_rate": 1.6362648947385238e-06,
      "loss": 1.0401,
      "step": 29711
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6203534603118896,
      "learning_rate": 1.635625978340734e-06,
      "loss": 0.765,
      "step": 29712
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7376227378845215,
      "learning_rate": 1.6349870619429448e-06,
      "loss": 0.6855,
      "step": 29713
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7658377885818481,
      "learning_rate": 1.6343481455451555e-06,
      "loss": 0.8303,
      "step": 29714
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.5135786533355713,
      "learning_rate": 1.633709229147366e-06,
      "loss": 0.8539,
      "step": 29715
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9111794233322144,
      "learning_rate": 1.6330703127495767e-06,
      "loss": 0.8022,
      "step": 29716
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.948238730430603,
      "learning_rate": 1.6324313963517874e-06,
      "loss": 0.8789,
      "step": 29717
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.2545759677886963,
      "learning_rate": 1.631792479953998e-06,
      "loss": 0.818,
      "step": 29718
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.318133592605591,
      "learning_rate": 1.6311535635562088e-06,
      "loss": 0.8466,
      "step": 29719
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.388567566871643,
      "learning_rate": 1.6305146471584193e-06,
      "loss": 0.3132,
      "step": 29720
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.65054452419281,
      "learning_rate": 1.62987573076063e-06,
      "loss": 0.7549,
      "step": 29721
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.8395750522613525,
      "learning_rate": 1.6292368143628407e-06,
      "loss": 1.0675,
      "step": 29722
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.8972461223602295,
      "learning_rate": 1.6285978979650514e-06,
      "loss": 0.9161,
      "step": 29723
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.770465850830078,
      "learning_rate": 1.6279589815672619e-06,
      "loss": 1.1371,
      "step": 29724
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.4776828289031982,
      "learning_rate": 1.6273200651694726e-06,
      "loss": 1.0469,
      "step": 29725
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.473888635635376,
      "learning_rate": 1.6266811487716833e-06,
      "loss": 0.9697,
      "step": 29726
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.932729482650757,
      "learning_rate": 1.626042232373894e-06,
      "loss": 0.7595,
      "step": 29727
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.575777769088745,
      "learning_rate": 1.6254033159761047e-06,
      "loss": 0.8099,
      "step": 29728
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0103397369384766,
      "learning_rate": 1.6247643995783152e-06,
      "loss": 0.881,
      "step": 29729
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4746323823928833,
      "learning_rate": 1.6241254831805259e-06,
      "loss": 0.7851,
      "step": 29730
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3504332304000854,
      "learning_rate": 1.6234865667827366e-06,
      "loss": 0.3393,
      "step": 29731
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6044398546218872,
      "learning_rate": 1.6228476503849473e-06,
      "loss": 0.8205,
      "step": 29732
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.063401460647583,
      "learning_rate": 1.6222087339871577e-06,
      "loss": 0.6928,
      "step": 29733
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.2196500301361084,
      "learning_rate": 1.6215698175893684e-06,
      "loss": 0.9237,
      "step": 29734
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6305252313613892,
      "learning_rate": 1.6209309011915791e-06,
      "loss": 0.8406,
      "step": 29735
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5316402912139893,
      "learning_rate": 1.6202919847937898e-06,
      "loss": 0.7382,
      "step": 29736
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3360939025878906,
      "learning_rate": 1.6196530683960005e-06,
      "loss": 0.3911,
      "step": 29737
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0286433696746826,
      "learning_rate": 1.619014151998211e-06,
      "loss": 1.3034,
      "step": 29738
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9194248914718628,
      "learning_rate": 1.6183752356004217e-06,
      "loss": 0.9845,
      "step": 29739
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3906340599060059,
      "learning_rate": 1.6177363192026324e-06,
      "loss": 0.4081,
      "step": 29740
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6153712272644043,
      "learning_rate": 1.6170974028048431e-06,
      "loss": 0.8353,
      "step": 29741
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.983788251876831,
      "learning_rate": 1.6164584864070536e-06,
      "loss": 0.9247,
      "step": 29742
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8242971897125244,
      "learning_rate": 1.6158195700092643e-06,
      "loss": 0.86,
      "step": 29743
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8387717008590698,
      "learning_rate": 1.615180653611475e-06,
      "loss": 0.9443,
      "step": 29744
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.050481081008911,
      "learning_rate": 1.6145417372136857e-06,
      "loss": 0.6987,
      "step": 29745
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9169018268585205,
      "learning_rate": 1.6139028208158964e-06,
      "loss": 0.8724,
      "step": 29746
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6167311668395996,
      "learning_rate": 1.613263904418107e-06,
      "loss": 0.8115,
      "step": 29747
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8604012727737427,
      "learning_rate": 1.6126249880203176e-06,
      "loss": 0.9839,
      "step": 29748
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5553650856018066,
      "learning_rate": 1.6119860716225283e-06,
      "loss": 0.7893,
      "step": 29749
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6516458988189697,
      "learning_rate": 1.611347155224739e-06,
      "loss": 0.6849,
      "step": 29750
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.761688232421875,
      "learning_rate": 1.6107082388269495e-06,
      "loss": 0.8296,
      "step": 29751
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5709893703460693,
      "learning_rate": 1.6100693224291602e-06,
      "loss": 0.8127,
      "step": 29752
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8950080871582031,
      "learning_rate": 1.6094304060313709e-06,
      "loss": 0.7996,
      "step": 29753
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.387080669403076,
      "learning_rate": 1.6087914896335816e-06,
      "loss": 0.9596,
      "step": 29754
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.060587167739868,
      "learning_rate": 1.6081525732357923e-06,
      "loss": 0.8721,
      "step": 29755
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.189288139343262,
      "learning_rate": 1.6075136568380028e-06,
      "loss": 1.0031,
      "step": 29756
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8399639129638672,
      "learning_rate": 1.6068747404402135e-06,
      "loss": 0.7928,
      "step": 29757
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.658469557762146,
      "learning_rate": 1.6062358240424242e-06,
      "loss": 0.8252,
      "step": 29758
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7341132164001465,
      "learning_rate": 1.6055969076446349e-06,
      "loss": 0.7554,
      "step": 29759
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.8774023056030273,
      "learning_rate": 1.6049579912468456e-06,
      "loss": 0.9724,
      "step": 29760
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0115058422088623,
      "learning_rate": 1.604319074849056e-06,
      "loss": 1.1853,
      "step": 29761
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5964455604553223,
      "learning_rate": 1.6036801584512668e-06,
      "loss": 0.7263,
      "step": 29762
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5129269361495972,
      "learning_rate": 1.6030412420534775e-06,
      "loss": 0.8401,
      "step": 29763
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.810469627380371,
      "learning_rate": 1.6024023256556882e-06,
      "loss": 0.7885,
      "step": 29764
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.553220748901367,
      "learning_rate": 1.6017634092578987e-06,
      "loss": 1.2231,
      "step": 29765
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.030578851699829,
      "learning_rate": 1.6011244928601094e-06,
      "loss": 0.775,
      "step": 29766
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.773078680038452,
      "learning_rate": 1.60048557646232e-06,
      "loss": 0.8423,
      "step": 29767
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9900298118591309,
      "learning_rate": 1.5998466600645308e-06,
      "loss": 0.7378,
      "step": 29768
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.491934895515442,
      "learning_rate": 1.5992077436667415e-06,
      "loss": 0.4017,
      "step": 29769
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7391468286514282,
      "learning_rate": 1.598568827268952e-06,
      "loss": 0.8863,
      "step": 29770
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8470216989517212,
      "learning_rate": 1.5979299108711626e-06,
      "loss": 0.7706,
      "step": 29771
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.941207766532898,
      "learning_rate": 1.5972909944733733e-06,
      "loss": 0.8266,
      "step": 29772
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.739320755004883,
      "learning_rate": 1.596652078075584e-06,
      "loss": 1.499,
      "step": 29773
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8321831226348877,
      "learning_rate": 1.5960131616777945e-06,
      "loss": 0.7415,
      "step": 29774
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7570698261260986,
      "learning_rate": 1.5953742452800052e-06,
      "loss": 0.805,
      "step": 29775
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.8616881370544434,
      "learning_rate": 1.594735328882216e-06,
      "loss": 0.9373,
      "step": 29776
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.406665563583374,
      "learning_rate": 1.5940964124844266e-06,
      "loss": 0.8034,
      "step": 29777
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9674376249313354,
      "learning_rate": 1.5934574960866373e-06,
      "loss": 0.8077,
      "step": 29778
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.3823113441467285,
      "learning_rate": 1.5928185796888478e-06,
      "loss": 0.975,
      "step": 29779
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.4397735595703125,
      "learning_rate": 1.5921796632910585e-06,
      "loss": 0.5996,
      "step": 29780
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.7628018856048584,
      "learning_rate": 1.5915407468932692e-06,
      "loss": 0.833,
      "step": 29781
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.5088069438934326,
      "learning_rate": 1.59090183049548e-06,
      "loss": 0.7989,
      "step": 29782
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9130321741104126,
      "learning_rate": 1.5902629140976904e-06,
      "loss": 0.865,
      "step": 29783
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.65986967086792,
      "learning_rate": 1.5896239976999011e-06,
      "loss": 0.9845,
      "step": 29784
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.5189030170440674,
      "learning_rate": 1.5889850813021118e-06,
      "loss": 0.9129,
      "step": 29785
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.2532942295074463,
      "learning_rate": 1.5883461649043225e-06,
      "loss": 0.8331,
      "step": 29786
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7860307693481445,
      "learning_rate": 1.5877072485065332e-06,
      "loss": 0.846,
      "step": 29787
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6738561391830444,
      "learning_rate": 1.5870683321087437e-06,
      "loss": 0.7767,
      "step": 29788
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2884624004364014,
      "learning_rate": 1.5864294157109544e-06,
      "loss": 0.3688,
      "step": 29789
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1857707500457764,
      "learning_rate": 1.585790499313165e-06,
      "loss": 0.9459,
      "step": 29790
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.390854835510254,
      "learning_rate": 1.5851515829153758e-06,
      "loss": 0.8729,
      "step": 29791
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.230137586593628,
      "learning_rate": 1.5845126665175865e-06,
      "loss": 0.8152,
      "step": 29792
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.624449610710144,
      "learning_rate": 1.583873750119797e-06,
      "loss": 0.9395,
      "step": 29793
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3936502933502197,
      "learning_rate": 1.5832348337220077e-06,
      "loss": 0.8474,
      "step": 29794
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6503305435180664,
      "learning_rate": 1.5825959173242184e-06,
      "loss": 0.7927,
      "step": 29795
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.7611918449401855,
      "learning_rate": 1.581957000926429e-06,
      "loss": 1.0768,
      "step": 29796
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9928395748138428,
      "learning_rate": 1.5813180845286396e-06,
      "loss": 0.8194,
      "step": 29797
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.850892424583435,
      "learning_rate": 1.5806791681308503e-06,
      "loss": 0.8757,
      "step": 29798
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.076383352279663,
      "learning_rate": 1.580040251733061e-06,
      "loss": 0.7498,
      "step": 29799
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.952398657798767,
      "learning_rate": 1.5794013353352717e-06,
      "loss": 0.8311,
      "step": 29800
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.965039849281311,
      "learning_rate": 1.5787624189374824e-06,
      "loss": 0.7433,
      "step": 29801
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.7647478580474854,
      "learning_rate": 1.5781235025396929e-06,
      "loss": 1.1633,
      "step": 29802
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.707616925239563,
      "learning_rate": 1.5774845861419036e-06,
      "loss": 0.7433,
      "step": 29803
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.343963146209717,
      "learning_rate": 1.5768456697441143e-06,
      "loss": 0.7696,
      "step": 29804
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6813395023345947,
      "learning_rate": 1.576206753346325e-06,
      "loss": 0.5389,
      "step": 29805
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.5339114665985107,
      "learning_rate": 1.5755678369485355e-06,
      "loss": 0.5958,
      "step": 29806
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.072352647781372,
      "learning_rate": 1.5749289205507462e-06,
      "loss": 0.8844,
      "step": 29807
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7975611686706543,
      "learning_rate": 1.5742900041529569e-06,
      "loss": 0.7982,
      "step": 29808
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9879796504974365,
      "learning_rate": 1.5736510877551671e-06,
      "loss": 1.0583,
      "step": 29809
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0077037811279297,
      "learning_rate": 1.5730121713573778e-06,
      "loss": 0.9371,
      "step": 29810
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.7671706676483154,
      "learning_rate": 1.5723732549595885e-06,
      "loss": 0.8674,
      "step": 29811
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0300769805908203,
      "learning_rate": 1.5717343385617992e-06,
      "loss": 0.7673,
      "step": 29812
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.868441104888916,
      "learning_rate": 1.57109542216401e-06,
      "loss": 0.8967,
      "step": 29813
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.8604111671447754,
      "learning_rate": 1.5704565057662204e-06,
      "loss": 0.8181,
      "step": 29814
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5146424770355225,
      "learning_rate": 1.5698175893684311e-06,
      "loss": 0.8003,
      "step": 29815
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.591883897781372,
      "learning_rate": 1.5691786729706418e-06,
      "loss": 0.7368,
      "step": 29816
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.1387264728546143,
      "learning_rate": 1.5685397565728525e-06,
      "loss": 0.8877,
      "step": 29817
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.5539183616638184,
      "learning_rate": 1.5679008401750632e-06,
      "loss": 1.2671,
      "step": 29818
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5922964811325073,
      "learning_rate": 1.5672619237772737e-06,
      "loss": 0.7829,
      "step": 29819
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5732975006103516,
      "learning_rate": 1.5666230073794844e-06,
      "loss": 0.7755,
      "step": 29820
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1843602657318115,
      "learning_rate": 1.565984090981695e-06,
      "loss": 0.5097,
      "step": 29821
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.6182861328125,
      "learning_rate": 1.5653451745839058e-06,
      "loss": 1.2334,
      "step": 29822
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1314280033111572,
      "learning_rate": 1.5647062581861163e-06,
      "loss": 0.8953,
      "step": 29823
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0450966358184814,
      "learning_rate": 1.564067341788327e-06,
      "loss": 0.4832,
      "step": 29824
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0687384605407715,
      "learning_rate": 1.5634284253905377e-06,
      "loss": 0.8377,
      "step": 29825
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.207893133163452,
      "learning_rate": 1.5627895089927484e-06,
      "loss": 0.9712,
      "step": 29826
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.952497959136963,
      "learning_rate": 1.562150592594959e-06,
      "loss": 0.9257,
      "step": 29827
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5931198596954346,
      "learning_rate": 1.5615116761971696e-06,
      "loss": 1.096,
      "step": 29828
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.412383794784546,
      "learning_rate": 1.5608727597993803e-06,
      "loss": 0.4996,
      "step": 29829
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8696999549865723,
      "learning_rate": 1.560233843401591e-06,
      "loss": 0.7462,
      "step": 29830
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.6421051025390625,
      "learning_rate": 1.5595949270038017e-06,
      "loss": 1.1833,
      "step": 29831
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.9646854400634766,
      "learning_rate": 1.5589560106060122e-06,
      "loss": 0.7807,
      "step": 29832
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.566150426864624,
      "learning_rate": 1.5583170942082229e-06,
      "loss": 0.8335,
      "step": 29833
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9629656076431274,
      "learning_rate": 1.5576781778104336e-06,
      "loss": 0.751,
      "step": 29834
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.967956781387329,
      "learning_rate": 1.5570392614126443e-06,
      "loss": 0.9172,
      "step": 29835
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6768951416015625,
      "learning_rate": 1.556400345014855e-06,
      "loss": 0.8328,
      "step": 29836
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.767781972885132,
      "learning_rate": 1.5557614286170655e-06,
      "loss": 0.5625,
      "step": 29837
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3881795406341553,
      "learning_rate": 1.5551225122192762e-06,
      "loss": 0.8762,
      "step": 29838
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.157938480377197,
      "learning_rate": 1.5544835958214869e-06,
      "loss": 0.9733,
      "step": 29839
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.833632230758667,
      "learning_rate": 1.5538446794236976e-06,
      "loss": 0.9321,
      "step": 29840
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7157334089279175,
      "learning_rate": 1.553205763025908e-06,
      "loss": 0.7982,
      "step": 29841
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.743521809577942,
      "learning_rate": 1.5525668466281187e-06,
      "loss": 0.7753,
      "step": 29842
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.929645299911499,
      "learning_rate": 1.5519279302303294e-06,
      "loss": 0.4825,
      "step": 29843
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5386252403259277,
      "learning_rate": 1.5512890138325401e-06,
      "loss": 0.7862,
      "step": 29844
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8193849325180054,
      "learning_rate": 1.5506500974347508e-06,
      "loss": 0.7853,
      "step": 29845
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3852527141571045,
      "learning_rate": 1.5500111810369613e-06,
      "loss": 0.4459,
      "step": 29846
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.05792498588562,
      "learning_rate": 1.549372264639172e-06,
      "loss": 0.9114,
      "step": 29847
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5382484197616577,
      "learning_rate": 1.5487333482413827e-06,
      "loss": 0.7557,
      "step": 29848
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8812302350997925,
      "learning_rate": 1.5480944318435934e-06,
      "loss": 1.0646,
      "step": 29849
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.560274362564087,
      "learning_rate": 1.547455515445804e-06,
      "loss": 0.5583,
      "step": 29850
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2997379302978516,
      "learning_rate": 1.5468165990480146e-06,
      "loss": 0.3215,
      "step": 29851
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.674767017364502,
      "learning_rate": 1.5461776826502253e-06,
      "loss": 0.8313,
      "step": 29852
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4672436714172363,
      "learning_rate": 1.545538766252436e-06,
      "loss": 0.664,
      "step": 29853
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4058926105499268,
      "learning_rate": 1.5448998498546467e-06,
      "loss": 0.7962,
      "step": 29854
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.6269595623016357,
      "learning_rate": 1.5442609334568572e-06,
      "loss": 0.785,
      "step": 29855
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.091061592102051,
      "learning_rate": 1.543622017059068e-06,
      "loss": 0.8593,
      "step": 29856
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4915117025375366,
      "learning_rate": 1.5429831006612786e-06,
      "loss": 0.7498,
      "step": 29857
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.227608323097229,
      "learning_rate": 1.5423441842634893e-06,
      "loss": 0.4383,
      "step": 29858
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.432953119277954,
      "learning_rate": 1.5417052678657e-06,
      "loss": 1.0029,
      "step": 29859
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9047802686691284,
      "learning_rate": 1.5410663514679105e-06,
      "loss": 0.8853,
      "step": 29860
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5881654024124146,
      "learning_rate": 1.5404274350701212e-06,
      "loss": 0.8075,
      "step": 29861
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.0019078254699707,
      "learning_rate": 1.539788518672332e-06,
      "loss": 0.8585,
      "step": 29862
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.109225034713745,
      "learning_rate": 1.5391496022745426e-06,
      "loss": 0.9005,
      "step": 29863
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1411612033843994,
      "learning_rate": 1.538510685876753e-06,
      "loss": 0.9583,
      "step": 29864
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.651908040046692,
      "learning_rate": 1.5378717694789638e-06,
      "loss": 0.7865,
      "step": 29865
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2317025661468506,
      "learning_rate": 1.5372328530811745e-06,
      "loss": 1.1548,
      "step": 29866
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7078886032104492,
      "learning_rate": 1.5365939366833852e-06,
      "loss": 0.9611,
      "step": 29867
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.274721622467041,
      "learning_rate": 1.5359550202855959e-06,
      "loss": 0.7862,
      "step": 29868
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.484182834625244,
      "learning_rate": 1.5353161038878064e-06,
      "loss": 0.9866,
      "step": 29869
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.972108840942383,
      "learning_rate": 1.534677187490017e-06,
      "loss": 1.1939,
      "step": 29870
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6490356922149658,
      "learning_rate": 1.5340382710922278e-06,
      "loss": 0.7947,
      "step": 29871
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8848689794540405,
      "learning_rate": 1.5333993546944385e-06,
      "loss": 0.8082,
      "step": 29872
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2922677993774414,
      "learning_rate": 1.532760438296649e-06,
      "loss": 0.9751,
      "step": 29873
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8081796169281006,
      "learning_rate": 1.5321215218988597e-06,
      "loss": 0.8555,
      "step": 29874
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6809394359588623,
      "learning_rate": 1.5314826055010704e-06,
      "loss": 0.8344,
      "step": 29875
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.3678297996521,
      "learning_rate": 1.530843689103281e-06,
      "loss": 1.1609,
      "step": 29876
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9795887470245361,
      "learning_rate": 1.5302047727054918e-06,
      "loss": 0.9065,
      "step": 29877
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.7647738456726074,
      "learning_rate": 1.5295658563077022e-06,
      "loss": 1.0559,
      "step": 29878
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.7082154750823975,
      "learning_rate": 1.528926939909913e-06,
      "loss": 1.1012,
      "step": 29879
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3965113162994385,
      "learning_rate": 1.5282880235121236e-06,
      "loss": 1.0246,
      "step": 29880
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.206576347351074,
      "learning_rate": 1.5276491071143343e-06,
      "loss": 0.8913,
      "step": 29881
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.257014513015747,
      "learning_rate": 1.5270101907165448e-06,
      "loss": 0.8419,
      "step": 29882
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.425447940826416,
      "learning_rate": 1.5263712743187555e-06,
      "loss": 0.7984,
      "step": 29883
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6497620344161987,
      "learning_rate": 1.5257323579209662e-06,
      "loss": 0.8442,
      "step": 29884
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.829559564590454,
      "learning_rate": 1.525093441523177e-06,
      "loss": 0.6371,
      "step": 29885
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6226599216461182,
      "learning_rate": 1.5244545251253876e-06,
      "loss": 0.7129,
      "step": 29886
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.7933545112609863,
      "learning_rate": 1.5238156087275981e-06,
      "loss": 0.5725,
      "step": 29887
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7480655908584595,
      "learning_rate": 1.5231766923298088e-06,
      "loss": 0.8204,
      "step": 29888
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.835226058959961,
      "learning_rate": 1.5225377759320195e-06,
      "loss": 0.7406,
      "step": 29889
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.691710114479065,
      "learning_rate": 1.5218988595342302e-06,
      "loss": 0.7416,
      "step": 29890
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.940050482749939,
      "learning_rate": 1.521259943136441e-06,
      "loss": 0.9475,
      "step": 29891
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1980717182159424,
      "learning_rate": 1.5206210267386514e-06,
      "loss": 0.8705,
      "step": 29892
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.846387267112732,
      "learning_rate": 1.5199821103408621e-06,
      "loss": 0.8587,
      "step": 29893
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.244340896606445,
      "learning_rate": 1.5193431939430728e-06,
      "loss": 1.3508,
      "step": 29894
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5781731605529785,
      "learning_rate": 1.5187042775452835e-06,
      "loss": 0.8393,
      "step": 29895
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9550305604934692,
      "learning_rate": 1.518065361147494e-06,
      "loss": 0.7269,
      "step": 29896
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.648491859436035,
      "learning_rate": 1.5174264447497047e-06,
      "loss": 0.7837,
      "step": 29897
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.675021171569824,
      "learning_rate": 1.5167875283519154e-06,
      "loss": 1.0304,
      "step": 29898
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0118319988250732,
      "learning_rate": 1.516148611954126e-06,
      "loss": 1.1006,
      "step": 29899
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.9597277641296387,
      "learning_rate": 1.5155096955563368e-06,
      "loss": 0.8407,
      "step": 29900
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.929136276245117,
      "learning_rate": 1.5148707791585473e-06,
      "loss": 0.8277,
      "step": 29901
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5789011716842651,
      "learning_rate": 1.514231862760758e-06,
      "loss": 0.7574,
      "step": 29902
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.684342622756958,
      "learning_rate": 1.5135929463629687e-06,
      "loss": 0.925,
      "step": 29903
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4542049169540405,
      "learning_rate": 1.5129540299651794e-06,
      "loss": 0.8098,
      "step": 29904
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5474674701690674,
      "learning_rate": 1.5123151135673897e-06,
      "loss": 0.7289,
      "step": 29905
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6465405225753784,
      "learning_rate": 1.5116761971696004e-06,
      "loss": 0.9282,
      "step": 29906
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.912960410118103,
      "learning_rate": 1.511037280771811e-06,
      "loss": 0.8143,
      "step": 29907
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6323373317718506,
      "learning_rate": 1.5103983643740216e-06,
      "loss": 0.7513,
      "step": 29908
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4356788396835327,
      "learning_rate": 1.5097594479762323e-06,
      "loss": 0.349,
      "step": 29909
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.3073174953460693,
      "learning_rate": 1.509120531578443e-06,
      "loss": 0.6322,
      "step": 29910
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2433829307556152,
      "learning_rate": 1.5084816151806537e-06,
      "loss": 0.7754,
      "step": 29911
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5112128257751465,
      "learning_rate": 1.5078426987828644e-06,
      "loss": 0.9149,
      "step": 29912
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8448108434677124,
      "learning_rate": 1.5072037823850748e-06,
      "loss": 0.7068,
      "step": 29913
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.062152147293091,
      "learning_rate": 1.5065648659872855e-06,
      "loss": 0.9551,
      "step": 29914
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.728467583656311,
      "learning_rate": 1.5059259495894962e-06,
      "loss": 0.7606,
      "step": 29915
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5271899700164795,
      "learning_rate": 1.505287033191707e-06,
      "loss": 0.7576,
      "step": 29916
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.6892952919006348,
      "learning_rate": 1.5046481167939176e-06,
      "loss": 0.8092,
      "step": 29917
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.399076223373413,
      "learning_rate": 1.5040092003961281e-06,
      "loss": 0.6925,
      "step": 29918
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5915303230285645,
      "learning_rate": 1.5033702839983388e-06,
      "loss": 0.9575,
      "step": 29919
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.4619436264038086,
      "learning_rate": 1.5027313676005495e-06,
      "loss": 1.1144,
      "step": 29920
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0222537517547607,
      "learning_rate": 1.5020924512027602e-06,
      "loss": 0.8901,
      "step": 29921
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1584370136260986,
      "learning_rate": 1.5014535348049707e-06,
      "loss": 0.8446,
      "step": 29922
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8778159618377686,
      "learning_rate": 1.5008146184071814e-06,
      "loss": 0.8701,
      "step": 29923
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.7552554607391357,
      "learning_rate": 1.5001757020093921e-06,
      "loss": 1.1901,
      "step": 29924
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6108814477920532,
      "learning_rate": 1.4995367856116028e-06,
      "loss": 0.8167,
      "step": 29925
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7868043184280396,
      "learning_rate": 1.4988978692138135e-06,
      "loss": 0.8374,
      "step": 29926
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.165288209915161,
      "learning_rate": 1.498258952816024e-06,
      "loss": 0.9475,
      "step": 29927
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.574022889137268,
      "learning_rate": 1.4976200364182347e-06,
      "loss": 0.8446,
      "step": 29928
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.098938465118408,
      "learning_rate": 1.4969811200204454e-06,
      "loss": 0.8924,
      "step": 29929
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8084419965744019,
      "learning_rate": 1.496342203622656e-06,
      "loss": 0.8722,
      "step": 29930
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6601651906967163,
      "learning_rate": 1.4957032872248666e-06,
      "loss": 0.7382,
      "step": 29931
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6105793714523315,
      "learning_rate": 1.4950643708270773e-06,
      "loss": 0.8513,
      "step": 29932
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7218962907791138,
      "learning_rate": 1.494425454429288e-06,
      "loss": 0.815,
      "step": 29933
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.601110816001892,
      "learning_rate": 1.4937865380314987e-06,
      "loss": 0.7935,
      "step": 29934
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5034406185150146,
      "learning_rate": 1.4931476216337094e-06,
      "loss": 0.792,
      "step": 29935
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.846886396408081,
      "learning_rate": 1.4925087052359199e-06,
      "loss": 0.7685,
      "step": 29936
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7992949485778809,
      "learning_rate": 1.4918697888381306e-06,
      "loss": 1.2421,
      "step": 29937
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7891424894332886,
      "learning_rate": 1.4912308724403413e-06,
      "loss": 0.9232,
      "step": 29938
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4229350090026855,
      "learning_rate": 1.490591956042552e-06,
      "loss": 0.6199,
      "step": 29939
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5430898666381836,
      "learning_rate": 1.4899530396447625e-06,
      "loss": 0.6812,
      "step": 29940
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.077914237976074,
      "learning_rate": 1.4893141232469732e-06,
      "loss": 0.8416,
      "step": 29941
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.318920612335205,
      "learning_rate": 1.4886752068491839e-06,
      "loss": 0.8966,
      "step": 29942
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2177503108978271,
      "learning_rate": 1.4880362904513946e-06,
      "loss": 0.3332,
      "step": 29943
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6467570066452026,
      "learning_rate": 1.4873973740536053e-06,
      "loss": 0.8461,
      "step": 29944
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9365819692611694,
      "learning_rate": 1.4867584576558158e-06,
      "loss": 1.1589,
      "step": 29945
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.135201930999756,
      "learning_rate": 1.4861195412580265e-06,
      "loss": 0.7055,
      "step": 29946
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9572350978851318,
      "learning_rate": 1.4854806248602372e-06,
      "loss": 0.8456,
      "step": 29947
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.638401746749878,
      "learning_rate": 1.4848417084624479e-06,
      "loss": 0.8927,
      "step": 29948
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7953568696975708,
      "learning_rate": 1.4842027920646583e-06,
      "loss": 0.7712,
      "step": 29949
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7948431968688965,
      "learning_rate": 1.483563875666869e-06,
      "loss": 0.8753,
      "step": 29950
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.57865047454834,
      "learning_rate": 1.4829249592690797e-06,
      "loss": 0.9965,
      "step": 29951
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.055384874343872,
      "learning_rate": 1.4822860428712904e-06,
      "loss": 0.8142,
      "step": 29952
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.796468496322632,
      "learning_rate": 1.4816471264735011e-06,
      "loss": 0.8363,
      "step": 29953
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.181718587875366,
      "learning_rate": 1.4810082100757116e-06,
      "loss": 0.9878,
      "step": 29954
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.969955563545227,
      "learning_rate": 1.4803692936779223e-06,
      "loss": 0.8746,
      "step": 29955
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.938009262084961,
      "learning_rate": 1.479730377280133e-06,
      "loss": 0.8104,
      "step": 29956
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5320892333984375,
      "learning_rate": 1.4790914608823437e-06,
      "loss": 1.0507,
      "step": 29957
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1290156841278076,
      "learning_rate": 1.4784525444845544e-06,
      "loss": 1.1801,
      "step": 29958
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.748668909072876,
      "learning_rate": 1.477813628086765e-06,
      "loss": 0.9292,
      "step": 29959
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7349988222122192,
      "learning_rate": 1.4771747116889756e-06,
      "loss": 0.828,
      "step": 29960
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.7738165855407715,
      "learning_rate": 1.4765357952911863e-06,
      "loss": 0.7892,
      "step": 29961
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.84038782119751,
      "learning_rate": 1.475896878893397e-06,
      "loss": 1.271,
      "step": 29962
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7956846952438354,
      "learning_rate": 1.4752579624956075e-06,
      "loss": 0.7115,
      "step": 29963
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6184730529785156,
      "learning_rate": 1.4746190460978182e-06,
      "loss": 0.7157,
      "step": 29964
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3429455757141113,
      "learning_rate": 1.473980129700029e-06,
      "loss": 0.7388,
      "step": 29965
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1980273723602295,
      "learning_rate": 1.4733412133022396e-06,
      "loss": 0.8313,
      "step": 29966
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7709723711013794,
      "learning_rate": 1.4727022969044503e-06,
      "loss": 0.7295,
      "step": 29967
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.074955940246582,
      "learning_rate": 1.4720633805066608e-06,
      "loss": 0.879,
      "step": 29968
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.12168288230896,
      "learning_rate": 1.4714244641088715e-06,
      "loss": 0.887,
      "step": 29969
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2674012184143066,
      "learning_rate": 1.4707855477110822e-06,
      "loss": 1.2602,
      "step": 29970
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7924091815948486,
      "learning_rate": 1.470146631313293e-06,
      "loss": 0.7741,
      "step": 29971
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9145976305007935,
      "learning_rate": 1.4695077149155034e-06,
      "loss": 0.8961,
      "step": 29972
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.620788335800171,
      "learning_rate": 1.468868798517714e-06,
      "loss": 0.7886,
      "step": 29973
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5740748643875122,
      "learning_rate": 1.4682298821199248e-06,
      "loss": 0.7798,
      "step": 29974
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.417395830154419,
      "learning_rate": 1.4675909657221355e-06,
      "loss": 0.7768,
      "step": 29975
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.7224996089935303,
      "learning_rate": 1.4669520493243462e-06,
      "loss": 0.7193,
      "step": 29976
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4263577461242676,
      "learning_rate": 1.4663131329265567e-06,
      "loss": 0.854,
      "step": 29977
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5739953517913818,
      "learning_rate": 1.4656742165287674e-06,
      "loss": 0.8174,
      "step": 29978
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5381590127944946,
      "learning_rate": 1.465035300130978e-06,
      "loss": 0.7343,
      "step": 29979
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.542578101158142,
      "learning_rate": 1.4643963837331888e-06,
      "loss": 0.7483,
      "step": 29980
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6182773113250732,
      "learning_rate": 1.4637574673353993e-06,
      "loss": 0.696,
      "step": 29981
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7554830312728882,
      "learning_rate": 1.46311855093761e-06,
      "loss": 0.8883,
      "step": 29982
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6937298774719238,
      "learning_rate": 1.4624796345398207e-06,
      "loss": 0.825,
      "step": 29983
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7687081098556519,
      "learning_rate": 1.4618407181420314e-06,
      "loss": 0.901,
      "step": 29984
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0170302391052246,
      "learning_rate": 1.461201801744242e-06,
      "loss": 0.8388,
      "step": 29985
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8098795413970947,
      "learning_rate": 1.4605628853464525e-06,
      "loss": 0.8817,
      "step": 29986
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6474045515060425,
      "learning_rate": 1.4599239689486632e-06,
      "loss": 0.7664,
      "step": 29987
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9874547719955444,
      "learning_rate": 1.459285052550874e-06,
      "loss": 0.8696,
      "step": 29988
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2289130687713623,
      "learning_rate": 1.4586461361530846e-06,
      "loss": 0.8958,
      "step": 29989
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.194753170013428,
      "learning_rate": 1.4580072197552953e-06,
      "loss": 1.2112,
      "step": 29990
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2988369464874268,
      "learning_rate": 1.4573683033575058e-06,
      "loss": 1.011,
      "step": 29991
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.94949471950531,
      "learning_rate": 1.4567293869597165e-06,
      "loss": 0.764,
      "step": 29992
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.089169979095459,
      "learning_rate": 1.4560904705619272e-06,
      "loss": 0.9109,
      "step": 29993
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5934717655181885,
      "learning_rate": 1.455451554164138e-06,
      "loss": 0.8106,
      "step": 29994
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.841338872909546,
      "learning_rate": 1.4548126377663484e-06,
      "loss": 0.8775,
      "step": 29995
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.655856966972351,
      "learning_rate": 1.4541737213685591e-06,
      "loss": 0.8131,
      "step": 29996
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.547137498855591,
      "learning_rate": 1.4535348049707698e-06,
      "loss": 0.6411,
      "step": 29997
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7762290239334106,
      "learning_rate": 1.4528958885729805e-06,
      "loss": 0.8396,
      "step": 29998
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8855655193328857,
      "learning_rate": 1.4522569721751912e-06,
      "loss": 0.7106,
      "step": 29999
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.349170207977295,
      "learning_rate": 1.4516180557774017e-06,
      "loss": 0.691,
      "step": 30000
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8464370965957642,
      "learning_rate": 1.4509791393796124e-06,
      "loss": 0.8656,
      "step": 30001
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0128283500671387,
      "learning_rate": 1.450340222981823e-06,
      "loss": 0.9249,
      "step": 30002
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.770452618598938,
      "learning_rate": 1.4497013065840334e-06,
      "loss": 0.8457,
      "step": 30003
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7967816591262817,
      "learning_rate": 1.449062390186244e-06,
      "loss": 0.8368,
      "step": 30004
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7071977853775024,
      "learning_rate": 1.4484234737884548e-06,
      "loss": 0.9049,
      "step": 30005
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.8997721672058105,
      "learning_rate": 1.4477845573906655e-06,
      "loss": 0.9913,
      "step": 30006
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.84083890914917,
      "learning_rate": 1.447145640992876e-06,
      "loss": 0.8297,
      "step": 30007
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.71725594997406,
      "learning_rate": 1.4465067245950867e-06,
      "loss": 0.8222,
      "step": 30008
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.4821982383728027,
      "learning_rate": 1.4458678081972974e-06,
      "loss": 0.8038,
      "step": 30009
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6443486213684082,
      "learning_rate": 1.445228891799508e-06,
      "loss": 0.7053,
      "step": 30010
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.506581783294678,
      "learning_rate": 1.4445899754017188e-06,
      "loss": 1.128,
      "step": 30011
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.954460620880127,
      "learning_rate": 1.4439510590039293e-06,
      "loss": 1.1531,
      "step": 30012
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2265102863311768,
      "learning_rate": 1.44331214260614e-06,
      "loss": 0.8262,
      "step": 30013
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.9776978492736816,
      "learning_rate": 1.4426732262083507e-06,
      "loss": 0.8018,
      "step": 30014
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7257059812545776,
      "learning_rate": 1.4420343098105614e-06,
      "loss": 0.8039,
      "step": 30015
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.214418411254883,
      "learning_rate": 1.4413953934127719e-06,
      "loss": 0.6894,
      "step": 30016
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4212400913238525,
      "learning_rate": 1.4407564770149826e-06,
      "loss": 0.8577,
      "step": 30017
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.125014305114746,
      "learning_rate": 1.4401175606171933e-06,
      "loss": 0.9719,
      "step": 30018
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.1161370277404785,
      "learning_rate": 1.439478644219404e-06,
      "loss": 1.0051,
      "step": 30019
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5051040649414062,
      "learning_rate": 1.4388397278216147e-06,
      "loss": 0.7782,
      "step": 30020
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.945363998413086,
      "learning_rate": 1.4382008114238251e-06,
      "loss": 0.9415,
      "step": 30021
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.809238076210022,
      "learning_rate": 1.4375618950260358e-06,
      "loss": 0.7772,
      "step": 30022
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.7169318199157715,
      "learning_rate": 1.4369229786282465e-06,
      "loss": 0.8299,
      "step": 30023
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2054100036621094,
      "learning_rate": 1.4362840622304572e-06,
      "loss": 0.8375,
      "step": 30024
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4636058807373047,
      "learning_rate": 1.435645145832668e-06,
      "loss": 0.7852,
      "step": 30025
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.7149367332458496,
      "learning_rate": 1.4350062294348784e-06,
      "loss": 0.8769,
      "step": 30026
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.27944278717041,
      "learning_rate": 1.4343673130370891e-06,
      "loss": 0.8809,
      "step": 30027
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.660207748413086,
      "learning_rate": 1.4337283966392998e-06,
      "loss": 0.778,
      "step": 30028
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9738376140594482,
      "learning_rate": 1.4330894802415105e-06,
      "loss": 0.757,
      "step": 30029
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.6414785385131836,
      "learning_rate": 1.432450563843721e-06,
      "loss": 0.7139,
      "step": 30030
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.262739419937134,
      "learning_rate": 1.4318116474459317e-06,
      "loss": 0.9809,
      "step": 30031
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5997297763824463,
      "learning_rate": 1.4311727310481424e-06,
      "loss": 0.7104,
      "step": 30032
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.344839096069336,
      "learning_rate": 1.4305338146503531e-06,
      "loss": 0.9515,
      "step": 30033
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.5596649646759033,
      "learning_rate": 1.4298948982525638e-06,
      "loss": 0.8244,
      "step": 30034
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.278005838394165,
      "learning_rate": 1.4292559818547743e-06,
      "loss": 1.0885,
      "step": 30035
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5014028549194336,
      "learning_rate": 1.428617065456985e-06,
      "loss": 0.5082,
      "step": 30036
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8150273561477661,
      "learning_rate": 1.4279781490591957e-06,
      "loss": 0.7247,
      "step": 30037
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.299485921859741,
      "learning_rate": 1.4273392326614064e-06,
      "loss": 0.7308,
      "step": 30038
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.019904851913452,
      "learning_rate": 1.426700316263617e-06,
      "loss": 1.1403,
      "step": 30039
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6379144191741943,
      "learning_rate": 1.4260613998658276e-06,
      "loss": 0.777,
      "step": 30040
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5753369331359863,
      "learning_rate": 1.4254224834680383e-06,
      "loss": 0.6247,
      "step": 30041
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8973616361618042,
      "learning_rate": 1.424783567070249e-06,
      "loss": 0.7764,
      "step": 30042
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5826084613800049,
      "learning_rate": 1.4241446506724597e-06,
      "loss": 0.8691,
      "step": 30043
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.947313666343689,
      "learning_rate": 1.4235057342746702e-06,
      "loss": 1.0144,
      "step": 30044
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0918900966644287,
      "learning_rate": 1.4228668178768809e-06,
      "loss": 0.8796,
      "step": 30045
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3881186246871948,
      "learning_rate": 1.4222279014790916e-06,
      "loss": 0.417,
      "step": 30046
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8270398378372192,
      "learning_rate": 1.4215889850813023e-06,
      "loss": 0.8886,
      "step": 30047
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.647371768951416,
      "learning_rate": 1.4209500686835128e-06,
      "loss": 0.8235,
      "step": 30048
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4402763843536377,
      "learning_rate": 1.4203111522857235e-06,
      "loss": 0.8205,
      "step": 30049
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.149688243865967,
      "learning_rate": 1.4196722358879342e-06,
      "loss": 0.8621,
      "step": 30050
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8376911878585815,
      "learning_rate": 1.4190333194901449e-06,
      "loss": 0.7292,
      "step": 30051
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8921469449996948,
      "learning_rate": 1.4183944030923556e-06,
      "loss": 0.804,
      "step": 30052
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5315666198730469,
      "learning_rate": 1.417755486694566e-06,
      "loss": 0.8042,
      "step": 30053
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.093141555786133,
      "learning_rate": 1.4171165702967768e-06,
      "loss": 1.0172,
      "step": 30054
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0874743461608887,
      "learning_rate": 1.4164776538989875e-06,
      "loss": 0.8832,
      "step": 30055
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.857411503791809,
      "learning_rate": 1.4158387375011982e-06,
      "loss": 0.8088,
      "step": 30056
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.524873971939087,
      "learning_rate": 1.4151998211034089e-06,
      "loss": 0.7653,
      "step": 30057
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8703352212905884,
      "learning_rate": 1.4145609047056193e-06,
      "loss": 0.7824,
      "step": 30058
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5728302001953125,
      "learning_rate": 1.41392198830783e-06,
      "loss": 0.9357,
      "step": 30059
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6024646759033203,
      "learning_rate": 1.4132830719100407e-06,
      "loss": 0.768,
      "step": 30060
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.5138494968414307,
      "learning_rate": 1.4126441555122514e-06,
      "loss": 1.0616,
      "step": 30061
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1145145893096924,
      "learning_rate": 1.412005239114462e-06,
      "loss": 0.8561,
      "step": 30062
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6356745958328247,
      "learning_rate": 1.4113663227166726e-06,
      "loss": 0.8531,
      "step": 30063
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.808815836906433,
      "learning_rate": 1.4107274063188833e-06,
      "loss": 0.8704,
      "step": 30064
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.767451047897339,
      "learning_rate": 1.410088489921094e-06,
      "loss": 0.8627,
      "step": 30065
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3275203704833984,
      "learning_rate": 1.4094495735233047e-06,
      "loss": 0.6849,
      "step": 30066
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3979538679122925,
      "learning_rate": 1.4088106571255152e-06,
      "loss": 0.3826,
      "step": 30067
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0331807136535645,
      "learning_rate": 1.408171740727726e-06,
      "loss": 0.952,
      "step": 30068
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9187638759613037,
      "learning_rate": 1.4075328243299366e-06,
      "loss": 0.8413,
      "step": 30069
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7497384548187256,
      "learning_rate": 1.4068939079321473e-06,
      "loss": 0.8037,
      "step": 30070
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.7009716033935547,
      "learning_rate": 1.4062549915343578e-06,
      "loss": 0.662,
      "step": 30071
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7272392511367798,
      "learning_rate": 1.4056160751365685e-06,
      "loss": 0.8434,
      "step": 30072
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.7532434463500977,
      "learning_rate": 1.4049771587387792e-06,
      "loss": 0.9696,
      "step": 30073
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0346744060516357,
      "learning_rate": 1.40433824234099e-06,
      "loss": 0.8001,
      "step": 30074
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1492459774017334,
      "learning_rate": 1.4036993259432006e-06,
      "loss": 1.2953,
      "step": 30075
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5071356296539307,
      "learning_rate": 1.403060409545411e-06,
      "loss": 0.9894,
      "step": 30076
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9347501993179321,
      "learning_rate": 1.4024214931476218e-06,
      "loss": 0.8762,
      "step": 30077
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3422051668167114,
      "learning_rate": 1.4017825767498325e-06,
      "loss": 0.386,
      "step": 30078
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8382763862609863,
      "learning_rate": 1.4011436603520432e-06,
      "loss": 0.8738,
      "step": 30079
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2761389017105103,
      "learning_rate": 1.4005047439542537e-06,
      "loss": 0.3088,
      "step": 30080
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.0528368949890137,
      "learning_rate": 1.3998658275564644e-06,
      "loss": 1.288,
      "step": 30081
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.127784013748169,
      "learning_rate": 1.399226911158675e-06,
      "loss": 0.7809,
      "step": 30082
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6861222982406616,
      "learning_rate": 1.3985879947608858e-06,
      "loss": 0.8977,
      "step": 30083
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.9110519886016846,
      "learning_rate": 1.3979490783630965e-06,
      "loss": 1.203,
      "step": 30084
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8582122325897217,
      "learning_rate": 1.397310161965307e-06,
      "loss": 0.9116,
      "step": 30085
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.14687180519104,
      "learning_rate": 1.3966712455675177e-06,
      "loss": 1.021,
      "step": 30086
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9356447458267212,
      "learning_rate": 1.3960323291697284e-06,
      "loss": 0.7172,
      "step": 30087
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6677690744400024,
      "learning_rate": 1.395393412771939e-06,
      "loss": 0.7946,
      "step": 30088
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.103285312652588,
      "learning_rate": 1.3947544963741498e-06,
      "loss": 0.9094,
      "step": 30089
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.778413772583008,
      "learning_rate": 1.3941155799763603e-06,
      "loss": 0.7342,
      "step": 30090
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.122765064239502,
      "learning_rate": 1.393476663578571e-06,
      "loss": 0.9015,
      "step": 30091
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9732131958007812,
      "learning_rate": 1.3928377471807817e-06,
      "loss": 0.7752,
      "step": 30092
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5372560024261475,
      "learning_rate": 1.3921988307829924e-06,
      "loss": 0.9216,
      "step": 30093
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6023733615875244,
      "learning_rate": 1.3915599143852029e-06,
      "loss": 0.8342,
      "step": 30094
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5622546672821045,
      "learning_rate": 1.3909209979874136e-06,
      "loss": 0.7725,
      "step": 30095
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9937405586242676,
      "learning_rate": 1.3902820815896243e-06,
      "loss": 1.2709,
      "step": 30096
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.9413883686065674,
      "learning_rate": 1.389643165191835e-06,
      "loss": 0.5497,
      "step": 30097
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6835732460021973,
      "learning_rate": 1.3890042487940457e-06,
      "loss": 0.8353,
      "step": 30098
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.249204158782959,
      "learning_rate": 1.388365332396256e-06,
      "loss": 0.8824,
      "step": 30099
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8087403774261475,
      "learning_rate": 1.3877264159984666e-06,
      "loss": 0.8869,
      "step": 30100
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.436268925666809,
      "learning_rate": 1.3870874996006773e-06,
      "loss": 0.7583,
      "step": 30101
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8511642217636108,
      "learning_rate": 1.3864485832028878e-06,
      "loss": 0.816,
      "step": 30102
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5382411479949951,
      "learning_rate": 1.3858096668050985e-06,
      "loss": 0.8314,
      "step": 30103
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6481839418411255,
      "learning_rate": 1.3851707504073092e-06,
      "loss": 0.8292,
      "step": 30104
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.997395396232605,
      "learning_rate": 1.38453183400952e-06,
      "loss": 0.8201,
      "step": 30105
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2285122871398926,
      "learning_rate": 1.3838929176117304e-06,
      "loss": 0.6874,
      "step": 30106
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7426552772521973,
      "learning_rate": 1.383254001213941e-06,
      "loss": 0.6916,
      "step": 30107
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8076342344284058,
      "learning_rate": 1.3826150848161518e-06,
      "loss": 0.7776,
      "step": 30108
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3576111793518066,
      "learning_rate": 1.3819761684183625e-06,
      "loss": 0.8282,
      "step": 30109
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.4853882789611816,
      "learning_rate": 1.3813372520205732e-06,
      "loss": 0.797,
      "step": 30110
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.47934889793396,
      "learning_rate": 1.3806983356227837e-06,
      "loss": 0.7394,
      "step": 30111
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.769075632095337,
      "learning_rate": 1.3800594192249944e-06,
      "loss": 0.8054,
      "step": 30112
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7914268970489502,
      "learning_rate": 1.379420502827205e-06,
      "loss": 0.7658,
      "step": 30113
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7524104118347168,
      "learning_rate": 1.3787815864294158e-06,
      "loss": 0.7808,
      "step": 30114
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7738734483718872,
      "learning_rate": 1.3781426700316263e-06,
      "loss": 0.9564,
      "step": 30115
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3944849967956543,
      "learning_rate": 1.377503753633837e-06,
      "loss": 0.8596,
      "step": 30116
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8153468370437622,
      "learning_rate": 1.3768648372360477e-06,
      "loss": 0.723,
      "step": 30117
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6640379428863525,
      "learning_rate": 1.3762259208382584e-06,
      "loss": 0.8615,
      "step": 30118
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8136038780212402,
      "learning_rate": 1.375587004440469e-06,
      "loss": 0.812,
      "step": 30119
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5578693151474,
      "learning_rate": 1.3749480880426796e-06,
      "loss": 0.7783,
      "step": 30120
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.583985447883606,
      "learning_rate": 1.3743091716448903e-06,
      "loss": 0.7846,
      "step": 30121
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7993583679199219,
      "learning_rate": 1.373670255247101e-06,
      "loss": 0.83,
      "step": 30122
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6612666845321655,
      "learning_rate": 1.3730313388493117e-06,
      "loss": 0.8011,
      "step": 30123
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.745827078819275,
      "learning_rate": 1.3723924224515224e-06,
      "loss": 0.8648,
      "step": 30124
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9541503190994263,
      "learning_rate": 1.3717535060537329e-06,
      "loss": 0.863,
      "step": 30125
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6950035095214844,
      "learning_rate": 1.3711145896559436e-06,
      "loss": 0.7412,
      "step": 30126
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9602292776107788,
      "learning_rate": 1.3704756732581543e-06,
      "loss": 1.1376,
      "step": 30127
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5574873685836792,
      "learning_rate": 1.369836756860365e-06,
      "loss": 0.4237,
      "step": 30128
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.285616636276245,
      "learning_rate": 1.3691978404625754e-06,
      "loss": 0.7949,
      "step": 30129
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.419997215270996,
      "learning_rate": 1.3685589240647861e-06,
      "loss": 1.007,
      "step": 30130
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8739299774169922,
      "learning_rate": 1.3679200076669968e-06,
      "loss": 0.8347,
      "step": 30131
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.003551721572876,
      "learning_rate": 1.3672810912692075e-06,
      "loss": 0.8303,
      "step": 30132
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0411014556884766,
      "learning_rate": 1.3666421748714182e-06,
      "loss": 0.7698,
      "step": 30133
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7706589698791504,
      "learning_rate": 1.3660032584736287e-06,
      "loss": 0.7915,
      "step": 30134
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7309770584106445,
      "learning_rate": 1.3653643420758394e-06,
      "loss": 0.8342,
      "step": 30135
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4481253623962402,
      "learning_rate": 1.3647254256780501e-06,
      "loss": 0.7789,
      "step": 30136
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.031114101409912,
      "learning_rate": 1.3640865092802608e-06,
      "loss": 1.01,
      "step": 30137
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.874173641204834,
      "learning_rate": 1.3634475928824713e-06,
      "loss": 1.095,
      "step": 30138
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6023025512695312,
      "learning_rate": 1.362808676484682e-06,
      "loss": 0.7259,
      "step": 30139
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3468537330627441,
      "learning_rate": 1.3621697600868927e-06,
      "loss": 0.8442,
      "step": 30140
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.442288398742676,
      "learning_rate": 1.3615308436891034e-06,
      "loss": 0.7837,
      "step": 30141
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7001407146453857,
      "learning_rate": 1.3608919272913141e-06,
      "loss": 0.7506,
      "step": 30142
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7388213872909546,
      "learning_rate": 1.3602530108935246e-06,
      "loss": 0.7282,
      "step": 30143
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.017638683319092,
      "learning_rate": 1.3596140944957353e-06,
      "loss": 0.9992,
      "step": 30144
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.816097617149353,
      "learning_rate": 1.358975178097946e-06,
      "loss": 0.9378,
      "step": 30145
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8452216386795044,
      "learning_rate": 1.3583362617001567e-06,
      "loss": 0.7894,
      "step": 30146
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.580910325050354,
      "learning_rate": 1.3576973453023672e-06,
      "loss": 0.7692,
      "step": 30147
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8228495121002197,
      "learning_rate": 1.357058428904578e-06,
      "loss": 0.8559,
      "step": 30148
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.654893636703491,
      "learning_rate": 1.3564195125067886e-06,
      "loss": 0.481,
      "step": 30149
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9813010692596436,
      "learning_rate": 1.3557805961089993e-06,
      "loss": 0.8056,
      "step": 30150
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1429896354675293,
      "learning_rate": 1.35514167971121e-06,
      "loss": 0.8401,
      "step": 30151
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.279685974121094,
      "learning_rate": 1.3545027633134205e-06,
      "loss": 0.9751,
      "step": 30152
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.013740062713623,
      "learning_rate": 1.3538638469156312e-06,
      "loss": 0.9266,
      "step": 30153
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5872033834457397,
      "learning_rate": 1.3532249305178419e-06,
      "loss": 0.8061,
      "step": 30154
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0002732276916504,
      "learning_rate": 1.3525860141200526e-06,
      "loss": 1.1798,
      "step": 30155
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9261237382888794,
      "learning_rate": 1.3519470977222633e-06,
      "loss": 0.861,
      "step": 30156
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.398392915725708,
      "learning_rate": 1.3513081813244738e-06,
      "loss": 0.7326,
      "step": 30157
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.180219888687134,
      "learning_rate": 1.3506692649266845e-06,
      "loss": 0.8153,
      "step": 30158
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.773120641708374,
      "learning_rate": 1.3500303485288952e-06,
      "loss": 0.8377,
      "step": 30159
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8180127143859863,
      "learning_rate": 1.3493914321311059e-06,
      "loss": 0.8954,
      "step": 30160
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8234580755233765,
      "learning_rate": 1.3487525157333164e-06,
      "loss": 0.83,
      "step": 30161
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8243581056594849,
      "learning_rate": 1.348113599335527e-06,
      "loss": 0.9083,
      "step": 30162
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.875382661819458,
      "learning_rate": 1.3474746829377378e-06,
      "loss": 0.8032,
      "step": 30163
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5488059520721436,
      "learning_rate": 1.3468357665399485e-06,
      "loss": 0.8731,
      "step": 30164
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.230686664581299,
      "learning_rate": 1.3461968501421592e-06,
      "loss": 0.8519,
      "step": 30165
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.916431188583374,
      "learning_rate": 1.3455579337443696e-06,
      "loss": 0.8727,
      "step": 30166
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.373360514640808,
      "learning_rate": 1.3449190173465803e-06,
      "loss": 0.7082,
      "step": 30167
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.9475128650665283,
      "learning_rate": 1.344280100948791e-06,
      "loss": 1.2279,
      "step": 30168
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8514577150344849,
      "learning_rate": 1.3436411845510017e-06,
      "loss": 0.8403,
      "step": 30169
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.617825984954834,
      "learning_rate": 1.3430022681532122e-06,
      "loss": 0.8688,
      "step": 30170
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.333425521850586,
      "learning_rate": 1.342363351755423e-06,
      "loss": 1.0692,
      "step": 30171
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.97259521484375,
      "learning_rate": 1.3417244353576336e-06,
      "loss": 0.7612,
      "step": 30172
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.163562536239624,
      "learning_rate": 1.3410855189598443e-06,
      "loss": 0.9845,
      "step": 30173
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5582669973373413,
      "learning_rate": 1.340446602562055e-06,
      "loss": 0.8378,
      "step": 30174
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9608763456344604,
      "learning_rate": 1.3398076861642655e-06,
      "loss": 0.8557,
      "step": 30175
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.997093915939331,
      "learning_rate": 1.3391687697664762e-06,
      "loss": 1.0247,
      "step": 30176
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2348124980926514,
      "learning_rate": 1.338529853368687e-06,
      "loss": 0.8378,
      "step": 30177
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6783050298690796,
      "learning_rate": 1.3378909369708976e-06,
      "loss": 0.8711,
      "step": 30178
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5569196939468384,
      "learning_rate": 1.3372520205731081e-06,
      "loss": 0.7927,
      "step": 30179
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5114901065826416,
      "learning_rate": 1.3366131041753188e-06,
      "loss": 0.7368,
      "step": 30180
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5762379169464111,
      "learning_rate": 1.3359741877775295e-06,
      "loss": 0.8096,
      "step": 30181
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0038862228393555,
      "learning_rate": 1.3353352713797402e-06,
      "loss": 0.9798,
      "step": 30182
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.716064453125,
      "learning_rate": 1.334696354981951e-06,
      "loss": 0.8267,
      "step": 30183
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.770041584968567,
      "learning_rate": 1.3340574385841614e-06,
      "loss": 0.8365,
      "step": 30184
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6863881349563599,
      "learning_rate": 1.333418522186372e-06,
      "loss": 0.7258,
      "step": 30185
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.046750783920288,
      "learning_rate": 1.3327796057885828e-06,
      "loss": 1.0671,
      "step": 30186
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.964080810546875,
      "learning_rate": 1.3321406893907935e-06,
      "loss": 0.729,
      "step": 30187
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.9401872158050537,
      "learning_rate": 1.331501772993004e-06,
      "loss": 1.1679,
      "step": 30188
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.214083433151245,
      "learning_rate": 1.3308628565952147e-06,
      "loss": 0.8806,
      "step": 30189
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4720672369003296,
      "learning_rate": 1.3302239401974254e-06,
      "loss": 0.8534,
      "step": 30190
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8289413452148438,
      "learning_rate": 1.329585023799636e-06,
      "loss": 0.782,
      "step": 30191
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8041363954544067,
      "learning_rate": 1.3289461074018468e-06,
      "loss": 0.7673,
      "step": 30192
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9791887998580933,
      "learning_rate": 1.3283071910040573e-06,
      "loss": 0.7821,
      "step": 30193
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.688073754310608,
      "learning_rate": 1.327668274606268e-06,
      "loss": 0.8429,
      "step": 30194
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9828213453292847,
      "learning_rate": 1.3270293582084785e-06,
      "loss": 0.7205,
      "step": 30195
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.354299068450928,
      "learning_rate": 1.326390441810689e-06,
      "loss": 1.1176,
      "step": 30196
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9670954942703247,
      "learning_rate": 1.3257515254128997e-06,
      "loss": 0.6919,
      "step": 30197
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9414373636245728,
      "learning_rate": 1.3251126090151104e-06,
      "loss": 0.8119,
      "step": 30198
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6424822807312012,
      "learning_rate": 1.324473692617321e-06,
      "loss": 0.8232,
      "step": 30199
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.789402961730957,
      "learning_rate": 1.3238347762195318e-06,
      "loss": 1.2693,
      "step": 30200
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.3944432735443115,
      "learning_rate": 1.3231958598217422e-06,
      "loss": 1.0341,
      "step": 30201
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.007507562637329,
      "learning_rate": 1.322556943423953e-06,
      "loss": 0.8762,
      "step": 30202
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.322505950927734,
      "learning_rate": 1.3219180270261636e-06,
      "loss": 1.2045,
      "step": 30203
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7334017753601074,
      "learning_rate": 1.3212791106283743e-06,
      "loss": 0.7754,
      "step": 30204
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7780207395553589,
      "learning_rate": 1.3206401942305848e-06,
      "loss": 0.7512,
      "step": 30205
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8015172481536865,
      "learning_rate": 1.3200012778327955e-06,
      "loss": 0.7494,
      "step": 30206
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5214381217956543,
      "learning_rate": 1.3193623614350062e-06,
      "loss": 0.8164,
      "step": 30207
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.068258285522461,
      "learning_rate": 1.318723445037217e-06,
      "loss": 0.8324,
      "step": 30208
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.416563868522644,
      "learning_rate": 1.3180845286394276e-06,
      "loss": 0.8865,
      "step": 30209
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8460829257965088,
      "learning_rate": 1.3174456122416381e-06,
      "loss": 0.8293,
      "step": 30210
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9122682809829712,
      "learning_rate": 1.3168066958438488e-06,
      "loss": 0.8441,
      "step": 30211
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8278112411499023,
      "learning_rate": 1.3161677794460595e-06,
      "loss": 0.7658,
      "step": 30212
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4370102882385254,
      "learning_rate": 1.3155288630482702e-06,
      "loss": 0.5409,
      "step": 30213
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.820348858833313,
      "learning_rate": 1.3148899466504807e-06,
      "loss": 0.8303,
      "step": 30214
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.695908546447754,
      "learning_rate": 1.3142510302526914e-06,
      "loss": 0.7585,
      "step": 30215
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7706843614578247,
      "learning_rate": 1.313612113854902e-06,
      "loss": 0.7601,
      "step": 30216
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3798702955245972,
      "learning_rate": 1.3129731974571128e-06,
      "loss": 0.3432,
      "step": 30217
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9978605508804321,
      "learning_rate": 1.3123342810593235e-06,
      "loss": 0.9143,
      "step": 30218
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.829947829246521,
      "learning_rate": 1.311695364661534e-06,
      "loss": 0.7749,
      "step": 30219
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6413161754608154,
      "learning_rate": 1.3110564482637447e-06,
      "loss": 0.7968,
      "step": 30220
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6191843748092651,
      "learning_rate": 1.3104175318659554e-06,
      "loss": 0.7229,
      "step": 30221
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.614130973815918,
      "learning_rate": 1.309778615468166e-06,
      "loss": 1.1468,
      "step": 30222
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9432697296142578,
      "learning_rate": 1.3091396990703768e-06,
      "loss": 0.8402,
      "step": 30223
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8169491291046143,
      "learning_rate": 1.3085007826725873e-06,
      "loss": 0.8216,
      "step": 30224
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.541329264640808,
      "learning_rate": 1.307861866274798e-06,
      "loss": 0.8147,
      "step": 30225
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1828114986419678,
      "learning_rate": 1.3072229498770087e-06,
      "loss": 0.856,
      "step": 30226
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.7170748710632324,
      "learning_rate": 1.3065840334792194e-06,
      "loss": 1.0424,
      "step": 30227
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5964879989624023,
      "learning_rate": 1.3059451170814299e-06,
      "loss": 0.8658,
      "step": 30228
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.5370755195617676,
      "learning_rate": 1.3053062006836406e-06,
      "loss": 0.9142,
      "step": 30229
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.2806224822998047,
      "learning_rate": 1.3046672842858513e-06,
      "loss": 0.7913,
      "step": 30230
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7939857244491577,
      "learning_rate": 1.304028367888062e-06,
      "loss": 0.7598,
      "step": 30231
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3951342105865479,
      "learning_rate": 1.3033894514902727e-06,
      "loss": 0.6808,
      "step": 30232
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8143203258514404,
      "learning_rate": 1.3027505350924832e-06,
      "loss": 0.8996,
      "step": 30233
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.352548837661743,
      "learning_rate": 1.3021116186946939e-06,
      "loss": 0.8993,
      "step": 30234
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6676342487335205,
      "learning_rate": 1.3014727022969046e-06,
      "loss": 0.77,
      "step": 30235
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.4453444480896,
      "learning_rate": 1.3008337858991153e-06,
      "loss": 0.9604,
      "step": 30236
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0155792236328125,
      "learning_rate": 1.3001948695013257e-06,
      "loss": 0.8066,
      "step": 30237
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.396714210510254,
      "learning_rate": 1.2995559531035364e-06,
      "loss": 0.8699,
      "step": 30238
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6671416759490967,
      "learning_rate": 1.2989170367057471e-06,
      "loss": 0.9165,
      "step": 30239
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.791768550872803,
      "learning_rate": 1.2982781203079578e-06,
      "loss": 1.2395,
      "step": 30240
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8992063999176025,
      "learning_rate": 1.2976392039101685e-06,
      "loss": 0.8555,
      "step": 30241
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.086399555206299,
      "learning_rate": 1.297000287512379e-06,
      "loss": 0.8641,
      "step": 30242
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9398289918899536,
      "learning_rate": 1.2963613711145897e-06,
      "loss": 0.9807,
      "step": 30243
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7533924579620361,
      "learning_rate": 1.2957224547168004e-06,
      "loss": 0.8065,
      "step": 30244
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8270018100738525,
      "learning_rate": 1.2950835383190111e-06,
      "loss": 0.8776,
      "step": 30245
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.9588875770568848,
      "learning_rate": 1.2944446219212216e-06,
      "loss": 0.5457,
      "step": 30246
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9775856733322144,
      "learning_rate": 1.2938057055234323e-06,
      "loss": 0.8489,
      "step": 30247
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9571863412857056,
      "learning_rate": 1.293166789125643e-06,
      "loss": 0.775,
      "step": 30248
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4108712673187256,
      "learning_rate": 1.2925278727278537e-06,
      "loss": 0.6305,
      "step": 30249
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.5791361331939697,
      "learning_rate": 1.2918889563300644e-06,
      "loss": 0.7257,
      "step": 30250
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.0739645957946777,
      "learning_rate": 1.291250039932275e-06,
      "loss": 0.904,
      "step": 30251
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0053584575653076,
      "learning_rate": 1.2906111235344856e-06,
      "loss": 1.3481,
      "step": 30252
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.180468559265137,
      "learning_rate": 1.2899722071366963e-06,
      "loss": 0.8266,
      "step": 30253
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.796037673950195,
      "learning_rate": 1.289333290738907e-06,
      "loss": 1.6853,
      "step": 30254
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.665475368499756,
      "learning_rate": 1.2886943743411177e-06,
      "loss": 1.0344,
      "step": 30255
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.5522565841674805,
      "learning_rate": 1.2880554579433282e-06,
      "loss": 1.2708,
      "step": 30256
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5376441478729248,
      "learning_rate": 1.287416541545539e-06,
      "loss": 0.7776,
      "step": 30257
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8353431224823,
      "learning_rate": 1.2867776251477496e-06,
      "loss": 0.6683,
      "step": 30258
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7332981824874878,
      "learning_rate": 1.2861387087499603e-06,
      "loss": 0.7597,
      "step": 30259
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5411714315414429,
      "learning_rate": 1.2854997923521708e-06,
      "loss": 0.8345,
      "step": 30260
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6300077438354492,
      "learning_rate": 1.2848608759543815e-06,
      "loss": 0.8156,
      "step": 30261
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.3726704120635986,
      "learning_rate": 1.2842219595565922e-06,
      "loss": 0.869,
      "step": 30262
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8730876445770264,
      "learning_rate": 1.2835830431588029e-06,
      "loss": 0.8127,
      "step": 30263
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3679649829864502,
      "learning_rate": 1.2829441267610136e-06,
      "loss": 0.3997,
      "step": 30264
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.902742385864258,
      "learning_rate": 1.282305210363224e-06,
      "loss": 1.2167,
      "step": 30265
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4621169567108154,
      "learning_rate": 1.2816662939654348e-06,
      "loss": 0.9215,
      "step": 30266
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0243799686431885,
      "learning_rate": 1.2810273775676455e-06,
      "loss": 0.8546,
      "step": 30267
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.136058330535889,
      "learning_rate": 1.2803884611698562e-06,
      "loss": 1.4523,
      "step": 30268
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7895305156707764,
      "learning_rate": 1.2797495447720667e-06,
      "loss": 0.8488,
      "step": 30269
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.911118745803833,
      "learning_rate": 1.2791106283742774e-06,
      "loss": 1.2141,
      "step": 30270
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7746315002441406,
      "learning_rate": 1.278471711976488e-06,
      "loss": 0.8801,
      "step": 30271
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6965930461883545,
      "learning_rate": 1.2778327955786988e-06,
      "loss": 0.7386,
      "step": 30272
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8858641386032104,
      "learning_rate": 1.2771938791809095e-06,
      "loss": 0.8191,
      "step": 30273
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.545562744140625,
      "learning_rate": 1.27655496278312e-06,
      "loss": 0.654,
      "step": 30274
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6440941095352173,
      "learning_rate": 1.2759160463853306e-06,
      "loss": 0.7662,
      "step": 30275
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.638343572616577,
      "learning_rate": 1.2752771299875413e-06,
      "loss": 0.8801,
      "step": 30276
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9399915933609009,
      "learning_rate": 1.274638213589752e-06,
      "loss": 0.9463,
      "step": 30277
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.904296398162842,
      "learning_rate": 1.2739992971919625e-06,
      "loss": 1.4117,
      "step": 30278
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.220562934875488,
      "learning_rate": 1.2733603807941732e-06,
      "loss": 0.9896,
      "step": 30279
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9004155397415161,
      "learning_rate": 1.272721464396384e-06,
      "loss": 1.2179,
      "step": 30280
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4684224128723145,
      "learning_rate": 1.2720825479985946e-06,
      "loss": 0.8319,
      "step": 30281
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5365403890609741,
      "learning_rate": 1.2714436316008053e-06,
      "loss": 0.7674,
      "step": 30282
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6210366487503052,
      "learning_rate": 1.2708047152030158e-06,
      "loss": 0.8053,
      "step": 30283
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.2004501819610596,
      "learning_rate": 1.2701657988052265e-06,
      "loss": 0.5759,
      "step": 30284
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.199601650238037,
      "learning_rate": 1.2695268824074372e-06,
      "loss": 1.3525,
      "step": 30285
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8755651712417603,
      "learning_rate": 1.268887966009648e-06,
      "loss": 0.8636,
      "step": 30286
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.71329665184021,
      "learning_rate": 1.2682490496118584e-06,
      "loss": 0.7847,
      "step": 30287
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5821599960327148,
      "learning_rate": 1.2676101332140691e-06,
      "loss": 0.8681,
      "step": 30288
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0174522399902344,
      "learning_rate": 1.2669712168162798e-06,
      "loss": 1.1525,
      "step": 30289
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.438058853149414,
      "learning_rate": 1.2663323004184905e-06,
      "loss": 0.8277,
      "step": 30290
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.269970417022705,
      "learning_rate": 1.2656933840207012e-06,
      "loss": 0.3433,
      "step": 30291
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.782467246055603,
      "learning_rate": 1.2650544676229115e-06,
      "loss": 0.7857,
      "step": 30292
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9586130380630493,
      "learning_rate": 1.2644155512251222e-06,
      "loss": 1.0939,
      "step": 30293
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0495452880859375,
      "learning_rate": 1.2637766348273329e-06,
      "loss": 0.6933,
      "step": 30294
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9625011682510376,
      "learning_rate": 1.2631377184295434e-06,
      "loss": 0.7989,
      "step": 30295
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7629953622817993,
      "learning_rate": 1.262498802031754e-06,
      "loss": 0.7576,
      "step": 30296
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3449710607528687,
      "learning_rate": 1.2618598856339648e-06,
      "loss": 0.3513,
      "step": 30297
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.844019412994385,
      "learning_rate": 1.2612209692361755e-06,
      "loss": 1.3579,
      "step": 30298
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.758382797241211,
      "learning_rate": 1.2605820528383862e-06,
      "loss": 0.9296,
      "step": 30299
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.840773105621338,
      "learning_rate": 1.2599431364405967e-06,
      "loss": 1.8268,
      "step": 30300
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.484173059463501,
      "learning_rate": 1.2593042200428074e-06,
      "loss": 0.7937,
      "step": 30301
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.260099411010742,
      "learning_rate": 1.258665303645018e-06,
      "loss": 0.9053,
      "step": 30302
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.064969062805176,
      "learning_rate": 1.2580263872472288e-06,
      "loss": 1.3649,
      "step": 30303
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6645244359970093,
      "learning_rate": 1.2573874708494393e-06,
      "loss": 0.8832,
      "step": 30304
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.83335542678833,
      "learning_rate": 1.25674855445165e-06,
      "loss": 1.095,
      "step": 30305
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.150732517242432,
      "learning_rate": 1.2561096380538607e-06,
      "loss": 0.995,
      "step": 30306
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8431437015533447,
      "learning_rate": 1.2554707216560714e-06,
      "loss": 0.8373,
      "step": 30307
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6387009620666504,
      "learning_rate": 1.254831805258282e-06,
      "loss": 0.8429,
      "step": 30308
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.825153112411499,
      "learning_rate": 1.2541928888604925e-06,
      "loss": 0.7702,
      "step": 30309
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6627390384674072,
      "learning_rate": 1.2535539724627032e-06,
      "loss": 0.8486,
      "step": 30310
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0963668823242188,
      "learning_rate": 1.252915056064914e-06,
      "loss": 0.9298,
      "step": 30311
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8363369703292847,
      "learning_rate": 1.2522761396671246e-06,
      "loss": 0.892,
      "step": 30312
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6039366722106934,
      "learning_rate": 1.2516372232693351e-06,
      "loss": 0.6673,
      "step": 30313
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8055856227874756,
      "learning_rate": 1.2509983068715458e-06,
      "loss": 0.8185,
      "step": 30314
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6327311992645264,
      "learning_rate": 1.2503593904737565e-06,
      "loss": 0.9242,
      "step": 30315
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.558369755744934,
      "learning_rate": 1.2497204740759672e-06,
      "loss": 0.6913,
      "step": 30316
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3561408519744873,
      "learning_rate": 1.249081557678178e-06,
      "loss": 0.8967,
      "step": 30317
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.213097333908081,
      "learning_rate": 1.2484426412803884e-06,
      "loss": 0.8614,
      "step": 30318
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8971692323684692,
      "learning_rate": 1.2478037248825991e-06,
      "loss": 0.8446,
      "step": 30319
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0868468284606934,
      "learning_rate": 1.2471648084848098e-06,
      "loss": 1.1413,
      "step": 30320
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.7010436058044434,
      "learning_rate": 1.2465258920870205e-06,
      "loss": 0.9563,
      "step": 30321
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.5261974334716797,
      "learning_rate": 1.2458869756892312e-06,
      "loss": 0.528,
      "step": 30322
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9314801692962646,
      "learning_rate": 1.2452480592914417e-06,
      "loss": 0.9487,
      "step": 30323
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6425929069519043,
      "learning_rate": 1.2446091428936524e-06,
      "loss": 0.6551,
      "step": 30324
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8508939743041992,
      "learning_rate": 1.2439702264958631e-06,
      "loss": 0.8982,
      "step": 30325
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9962860345840454,
      "learning_rate": 1.2433313100980738e-06,
      "loss": 0.8067,
      "step": 30326
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5868141651153564,
      "learning_rate": 1.2426923937002843e-06,
      "loss": 0.6863,
      "step": 30327
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8525727987289429,
      "learning_rate": 1.242053477302495e-06,
      "loss": 0.7766,
      "step": 30328
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8484452962875366,
      "learning_rate": 1.2414145609047057e-06,
      "loss": 0.6812,
      "step": 30329
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.163214921951294,
      "learning_rate": 1.2407756445069164e-06,
      "loss": 0.9625,
      "step": 30330
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.7338736057281494,
      "learning_rate": 1.240136728109127e-06,
      "loss": 0.964,
      "step": 30331
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.936062216758728,
      "learning_rate": 1.2394978117113376e-06,
      "loss": 0.7496,
      "step": 30332
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9413076639175415,
      "learning_rate": 1.2388588953135483e-06,
      "loss": 0.8658,
      "step": 30333
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.412638187408447,
      "learning_rate": 1.238219978915759e-06,
      "loss": 0.9978,
      "step": 30334
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.3499929904937744,
      "learning_rate": 1.2375810625179697e-06,
      "loss": 0.9296,
      "step": 30335
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9821468591690063,
      "learning_rate": 1.2369421461201802e-06,
      "loss": 0.9273,
      "step": 30336
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9196398258209229,
      "learning_rate": 1.2363032297223909e-06,
      "loss": 1.0074,
      "step": 30337
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2775840759277344,
      "learning_rate": 1.2356643133246016e-06,
      "loss": 0.6931,
      "step": 30338
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3108959197998047,
      "learning_rate": 1.2350253969268123e-06,
      "loss": 0.786,
      "step": 30339
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.979040503501892,
      "learning_rate": 1.234386480529023e-06,
      "loss": 0.7084,
      "step": 30340
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5191799402236938,
      "learning_rate": 1.2337475641312335e-06,
      "loss": 0.7485,
      "step": 30341
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9570099115371704,
      "learning_rate": 1.2331086477334442e-06,
      "loss": 0.7579,
      "step": 30342
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.479449987411499,
      "learning_rate": 1.2324697313356549e-06,
      "loss": 0.3638,
      "step": 30343
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9217952489852905,
      "learning_rate": 1.2318308149378656e-06,
      "loss": 0.7642,
      "step": 30344
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.521944284439087,
      "learning_rate": 1.231191898540076e-06,
      "loss": 0.9375,
      "step": 30345
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.99892258644104,
      "learning_rate": 1.2305529821422867e-06,
      "loss": 0.8556,
      "step": 30346
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.244457244873047,
      "learning_rate": 1.2299140657444974e-06,
      "loss": 0.8702,
      "step": 30347
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.696925163269043,
      "learning_rate": 1.2292751493467081e-06,
      "loss": 1.3133,
      "step": 30348
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3144195079803467,
      "learning_rate": 1.2286362329489188e-06,
      "loss": 0.5274,
      "step": 30349
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5806026458740234,
      "learning_rate": 1.2279973165511293e-06,
      "loss": 0.7758,
      "step": 30350
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0557475090026855,
      "learning_rate": 1.22735840015334e-06,
      "loss": 0.4288,
      "step": 30351
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.996537208557129,
      "learning_rate": 1.2267194837555507e-06,
      "loss": 0.8556,
      "step": 30352
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7079143524169922,
      "learning_rate": 1.2260805673577614e-06,
      "loss": 0.8018,
      "step": 30353
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0048484802246094,
      "learning_rate": 1.225441650959972e-06,
      "loss": 0.7975,
      "step": 30354
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6226298809051514,
      "learning_rate": 1.2248027345621826e-06,
      "loss": 0.723,
      "step": 30355
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2770824432373047,
      "learning_rate": 1.2241638181643933e-06,
      "loss": 0.3339,
      "step": 30356
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0400233268737793,
      "learning_rate": 1.223524901766604e-06,
      "loss": 0.5645,
      "step": 30357
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.583108425140381,
      "learning_rate": 1.2228859853688147e-06,
      "loss": 1.1201,
      "step": 30358
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8867502212524414,
      "learning_rate": 1.2222470689710252e-06,
      "loss": 0.6081,
      "step": 30359
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2348546981811523,
      "learning_rate": 1.221608152573236e-06,
      "loss": 0.5512,
      "step": 30360
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.1951403617858887,
      "learning_rate": 1.2209692361754466e-06,
      "loss": 0.7957,
      "step": 30361
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6445362567901611,
      "learning_rate": 1.2203303197776573e-06,
      "loss": 1.0847,
      "step": 30362
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6626614332199097,
      "learning_rate": 1.219691403379868e-06,
      "loss": 0.6801,
      "step": 30363
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.678335428237915,
      "learning_rate": 1.2190524869820785e-06,
      "loss": 0.8138,
      "step": 30364
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6341537237167358,
      "learning_rate": 1.218413570584289e-06,
      "loss": 0.7477,
      "step": 30365
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.05143404006958,
      "learning_rate": 1.2177746541864997e-06,
      "loss": 0.9165,
      "step": 30366
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6574890613555908,
      "learning_rate": 1.2171357377887104e-06,
      "loss": 0.8372,
      "step": 30367
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8266454935073853,
      "learning_rate": 1.216496821390921e-06,
      "loss": 0.8504,
      "step": 30368
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.836959958076477,
      "learning_rate": 1.2158579049931318e-06,
      "loss": 0.85,
      "step": 30369
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0285918712615967,
      "learning_rate": 1.2152189885953423e-06,
      "loss": 0.7377,
      "step": 30370
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.63392972946167,
      "learning_rate": 1.214580072197553e-06,
      "loss": 1.2276,
      "step": 30371
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2914106845855713,
      "learning_rate": 1.2139411557997637e-06,
      "loss": 0.6968,
      "step": 30372
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2020111083984375,
      "learning_rate": 1.2133022394019744e-06,
      "loss": 0.8168,
      "step": 30373
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4592533111572266,
      "learning_rate": 1.2126633230041849e-06,
      "loss": 0.7302,
      "step": 30374
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.408799171447754,
      "learning_rate": 1.2120244066063956e-06,
      "loss": 0.9073,
      "step": 30375
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.565554618835449,
      "learning_rate": 1.2113854902086063e-06,
      "loss": 1.0568,
      "step": 30376
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8723875284194946,
      "learning_rate": 1.210746573810817e-06,
      "loss": 1.1839,
      "step": 30377
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9884297847747803,
      "learning_rate": 1.2101076574130277e-06,
      "loss": 0.9477,
      "step": 30378
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.121917247772217,
      "learning_rate": 1.2094687410152382e-06,
      "loss": 0.8713,
      "step": 30379
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1411244869232178,
      "learning_rate": 1.2088298246174489e-06,
      "loss": 1.2033,
      "step": 30380
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.28192138671875,
      "learning_rate": 1.2081909082196596e-06,
      "loss": 1.3077,
      "step": 30381
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7114137411117554,
      "learning_rate": 1.2075519918218703e-06,
      "loss": 0.8381,
      "step": 30382
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3773629665374756,
      "learning_rate": 1.2069130754240807e-06,
      "loss": 0.88,
      "step": 30383
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3744919300079346,
      "learning_rate": 1.2062741590262914e-06,
      "loss": 1.0906,
      "step": 30384
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8547452688217163,
      "learning_rate": 1.2056352426285021e-06,
      "loss": 0.7971,
      "step": 30385
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8181158304214478,
      "learning_rate": 1.2049963262307128e-06,
      "loss": 0.9263,
      "step": 30386
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8573509454727173,
      "learning_rate": 1.2043574098329235e-06,
      "loss": 0.8324,
      "step": 30387
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.605055093765259,
      "learning_rate": 1.203718493435134e-06,
      "loss": 0.9395,
      "step": 30388
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1089181900024414,
      "learning_rate": 1.2030795770373447e-06,
      "loss": 0.8401,
      "step": 30389
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.250652313232422,
      "learning_rate": 1.2024406606395554e-06,
      "loss": 0.7161,
      "step": 30390
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.127494812011719,
      "learning_rate": 1.2018017442417661e-06,
      "loss": 1.0289,
      "step": 30391
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.013439178466797,
      "learning_rate": 1.2011628278439768e-06,
      "loss": 0.7655,
      "step": 30392
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.362966775894165,
      "learning_rate": 1.2005239114461873e-06,
      "loss": 0.7404,
      "step": 30393
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.112809181213379,
      "learning_rate": 1.199884995048398e-06,
      "loss": 0.8071,
      "step": 30394
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.670689344406128,
      "learning_rate": 1.1992460786506087e-06,
      "loss": 1.0285,
      "step": 30395
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7009823322296143,
      "learning_rate": 1.1986071622528194e-06,
      "loss": 0.9067,
      "step": 30396
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8672970533370972,
      "learning_rate": 1.19796824585503e-06,
      "loss": 0.7282,
      "step": 30397
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8263412714004517,
      "learning_rate": 1.1973293294572406e-06,
      "loss": 0.8622,
      "step": 30398
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4687178134918213,
      "learning_rate": 1.1966904130594513e-06,
      "loss": 0.3354,
      "step": 30399
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.85706627368927,
      "learning_rate": 1.196051496661662e-06,
      "loss": 0.7789,
      "step": 30400
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.968878149986267,
      "learning_rate": 1.1954125802638727e-06,
      "loss": 0.8508,
      "step": 30401
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4637060165405273,
      "learning_rate": 1.1947736638660832e-06,
      "loss": 1.1847,
      "step": 30402
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.283132553100586,
      "learning_rate": 1.1941347474682939e-06,
      "loss": 1.2929,
      "step": 30403
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4588884115219116,
      "learning_rate": 1.1934958310705046e-06,
      "loss": 0.7356,
      "step": 30404
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8835386037826538,
      "learning_rate": 1.1928569146727153e-06,
      "loss": 1.0713,
      "step": 30405
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4790730476379395,
      "learning_rate": 1.1922179982749258e-06,
      "loss": 0.8155,
      "step": 30406
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8607821464538574,
      "learning_rate": 1.1915790818771365e-06,
      "loss": 0.6603,
      "step": 30407
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.8026559352874756,
      "learning_rate": 1.1909401654793472e-06,
      "loss": 0.6229,
      "step": 30408
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9207860231399536,
      "learning_rate": 1.1903012490815579e-06,
      "loss": 0.7453,
      "step": 30409
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6950199604034424,
      "learning_rate": 1.1896623326837686e-06,
      "loss": 0.7357,
      "step": 30410
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.0469043254852295,
      "learning_rate": 1.189023416285979e-06,
      "loss": 0.9431,
      "step": 30411
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2399909496307373,
      "learning_rate": 1.1883844998881898e-06,
      "loss": 0.7532,
      "step": 30412
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1028881072998047,
      "learning_rate": 1.1877455834904003e-06,
      "loss": 0.8325,
      "step": 30413
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1654703617095947,
      "learning_rate": 1.187106667092611e-06,
      "loss": 1.0654,
      "step": 30414
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.623894214630127,
      "learning_rate": 1.1864677506948217e-06,
      "loss": 0.7221,
      "step": 30415
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0441973209381104,
      "learning_rate": 1.1858288342970324e-06,
      "loss": 0.8167,
      "step": 30416
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.08345365524292,
      "learning_rate": 1.1851899178992428e-06,
      "loss": 0.8016,
      "step": 30417
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7724974155426025,
      "learning_rate": 1.1845510015014535e-06,
      "loss": 0.9728,
      "step": 30418
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4132821559906006,
      "learning_rate": 1.1839120851036642e-06,
      "loss": 0.3321,
      "step": 30419
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.700256109237671,
      "learning_rate": 1.183273168705875e-06,
      "loss": 0.9763,
      "step": 30420
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8348948955535889,
      "learning_rate": 1.1826342523080856e-06,
      "loss": 0.8737,
      "step": 30421
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.759202718734741,
      "learning_rate": 1.1819953359102961e-06,
      "loss": 0.8745,
      "step": 30422
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.299661636352539,
      "learning_rate": 1.1813564195125068e-06,
      "loss": 0.8638,
      "step": 30423
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8945845365524292,
      "learning_rate": 1.1807175031147175e-06,
      "loss": 1.0378,
      "step": 30424
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.803915023803711,
      "learning_rate": 1.1800785867169282e-06,
      "loss": 0.5569,
      "step": 30425
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.5132031440734863,
      "learning_rate": 1.1794396703191387e-06,
      "loss": 0.5379,
      "step": 30426
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9009848833084106,
      "learning_rate": 1.1788007539213494e-06,
      "loss": 0.7475,
      "step": 30427
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8104221820831299,
      "learning_rate": 1.1781618375235601e-06,
      "loss": 0.7065,
      "step": 30428
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5426814556121826,
      "learning_rate": 1.1775229211257708e-06,
      "loss": 0.7692,
      "step": 30429
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8619685173034668,
      "learning_rate": 1.1768840047279815e-06,
      "loss": 0.9064,
      "step": 30430
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8510289192199707,
      "learning_rate": 1.176245088330192e-06,
      "loss": 0.805,
      "step": 30431
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1726431846618652,
      "learning_rate": 1.1756061719324027e-06,
      "loss": 1.1694,
      "step": 30432
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8167479038238525,
      "learning_rate": 1.1749672555346134e-06,
      "loss": 0.818,
      "step": 30433
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.465275526046753,
      "learning_rate": 1.1743283391368241e-06,
      "loss": 0.7622,
      "step": 30434
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.6093804836273193,
      "learning_rate": 1.1736894227390346e-06,
      "loss": 1.0454,
      "step": 30435
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.999118447303772,
      "learning_rate": 1.1730505063412453e-06,
      "loss": 0.4754,
      "step": 30436
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.187516212463379,
      "learning_rate": 1.172411589943456e-06,
      "loss": 0.9282,
      "step": 30437
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.139801025390625,
      "learning_rate": 1.1717726735456667e-06,
      "loss": 1.1661,
      "step": 30438
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8891518115997314,
      "learning_rate": 1.1711337571478774e-06,
      "loss": 0.8436,
      "step": 30439
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.49784779548645,
      "learning_rate": 1.1704948407500879e-06,
      "loss": 0.9308,
      "step": 30440
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.5547678470611572,
      "learning_rate": 1.1698559243522986e-06,
      "loss": 0.9528,
      "step": 30441
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0055370330810547,
      "learning_rate": 1.1692170079545093e-06,
      "loss": 0.9997,
      "step": 30442
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6997699737548828,
      "learning_rate": 1.16857809155672e-06,
      "loss": 0.9024,
      "step": 30443
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2898354530334473,
      "learning_rate": 1.1679391751589305e-06,
      "loss": 0.9042,
      "step": 30444
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.815710425376892,
      "learning_rate": 1.1673002587611412e-06,
      "loss": 0.9247,
      "step": 30445
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5999600887298584,
      "learning_rate": 1.1666613423633519e-06,
      "loss": 0.788,
      "step": 30446
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6003600358963013,
      "learning_rate": 1.1660224259655626e-06,
      "loss": 0.7843,
      "step": 30447
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4822492599487305,
      "learning_rate": 1.1653835095677733e-06,
      "loss": 0.5917,
      "step": 30448
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6031540632247925,
      "learning_rate": 1.1647445931699838e-06,
      "loss": 0.7972,
      "step": 30449
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7738392353057861,
      "learning_rate": 1.1641056767721945e-06,
      "loss": 0.7849,
      "step": 30450
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1649258136749268,
      "learning_rate": 1.1634667603744052e-06,
      "loss": 0.9185,
      "step": 30451
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.9971909523010254,
      "learning_rate": 1.1628278439766159e-06,
      "loss": 1.3439,
      "step": 30452
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2571969032287598,
      "learning_rate": 1.1621889275788263e-06,
      "loss": 0.8361,
      "step": 30453
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.0055339336395264,
      "learning_rate": 1.161550011181037e-06,
      "loss": 1.0533,
      "step": 30454
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.624764084815979,
      "learning_rate": 1.1609110947832477e-06,
      "loss": 0.5913,
      "step": 30455
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.097405195236206,
      "learning_rate": 1.1602721783854584e-06,
      "loss": 1.1663,
      "step": 30456
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8571865558624268,
      "learning_rate": 1.1596332619876691e-06,
      "loss": 0.8549,
      "step": 30457
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8805174827575684,
      "learning_rate": 1.1589943455898796e-06,
      "loss": 0.827,
      "step": 30458
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.6827290058135986,
      "learning_rate": 1.1583554291920903e-06,
      "loss": 1.2329,
      "step": 30459
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.006923198699951,
      "learning_rate": 1.157716512794301e-06,
      "loss": 0.8913,
      "step": 30460
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1217355728149414,
      "learning_rate": 1.1570775963965115e-06,
      "loss": 0.7996,
      "step": 30461
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9806374311447144,
      "learning_rate": 1.1564386799987222e-06,
      "loss": 0.8545,
      "step": 30462
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0004751682281494,
      "learning_rate": 1.155799763600933e-06,
      "loss": 1.187,
      "step": 30463
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7179865837097168,
      "learning_rate": 1.1551608472031434e-06,
      "loss": 0.8001,
      "step": 30464
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.9888904094696045,
      "learning_rate": 1.1545219308053541e-06,
      "loss": 0.9779,
      "step": 30465
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6927815675735474,
      "learning_rate": 1.1538830144075648e-06,
      "loss": 0.8942,
      "step": 30466
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.6503796577453613,
      "learning_rate": 1.1532440980097755e-06,
      "loss": 1.0627,
      "step": 30467
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.7250607013702393,
      "learning_rate": 1.1526051816119862e-06,
      "loss": 0.7608,
      "step": 30468
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.3373098373413086,
      "learning_rate": 1.1519662652141967e-06,
      "loss": 0.7493,
      "step": 30469
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.229686975479126,
      "learning_rate": 1.1513273488164074e-06,
      "loss": 0.767,
      "step": 30470
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.700260043144226,
      "learning_rate": 1.150688432418618e-06,
      "loss": 0.8649,
      "step": 30471
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.356199026107788,
      "learning_rate": 1.1500495160208288e-06,
      "loss": 0.7949,
      "step": 30472
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.135643720626831,
      "learning_rate": 1.1494105996230393e-06,
      "loss": 1.0598,
      "step": 30473
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.4358441829681396,
      "learning_rate": 1.14877168322525e-06,
      "loss": 0.7131,
      "step": 30474
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4039602279663086,
      "learning_rate": 1.1481327668274607e-06,
      "loss": 0.3579,
      "step": 30475
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7765284776687622,
      "learning_rate": 1.1474938504296714e-06,
      "loss": 0.9181,
      "step": 30476
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6966434717178345,
      "learning_rate": 1.146854934031882e-06,
      "loss": 0.7747,
      "step": 30477
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9783748388290405,
      "learning_rate": 1.1462160176340926e-06,
      "loss": 0.8236,
      "step": 30478
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.152613639831543,
      "learning_rate": 1.1455771012363033e-06,
      "loss": 0.9451,
      "step": 30479
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.770724892616272,
      "learning_rate": 1.144938184838514e-06,
      "loss": 0.7324,
      "step": 30480
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.8050873279571533,
      "learning_rate": 1.1442992684407247e-06,
      "loss": 0.8292,
      "step": 30481
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.722655177116394,
      "learning_rate": 1.1436603520429352e-06,
      "loss": 0.823,
      "step": 30482
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.002126932144165,
      "learning_rate": 1.1430214356451459e-06,
      "loss": 0.8082,
      "step": 30483
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5552270412445068,
      "learning_rate": 1.1423825192473566e-06,
      "loss": 0.7502,
      "step": 30484
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2436418533325195,
      "learning_rate": 1.1417436028495673e-06,
      "loss": 0.5037,
      "step": 30485
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7412978410720825,
      "learning_rate": 1.141104686451778e-06,
      "loss": 0.8628,
      "step": 30486
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.238767147064209,
      "learning_rate": 1.1404657700539885e-06,
      "loss": 0.8677,
      "step": 30487
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.637096405029297,
      "learning_rate": 1.1398268536561992e-06,
      "loss": 0.8793,
      "step": 30488
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2571566104888916,
      "learning_rate": 1.1391879372584099e-06,
      "loss": 0.9155,
      "step": 30489
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.7208197116851807,
      "learning_rate": 1.1385490208606206e-06,
      "loss": 1.232,
      "step": 30490
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.132742404937744,
      "learning_rate": 1.1379101044628313e-06,
      "loss": 0.8591,
      "step": 30491
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6909347772598267,
      "learning_rate": 1.1372711880650417e-06,
      "loss": 0.8876,
      "step": 30492
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.975749969482422,
      "learning_rate": 1.1366322716672524e-06,
      "loss": 0.8859,
      "step": 30493
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0496575832366943,
      "learning_rate": 1.1359933552694631e-06,
      "loss": 0.7482,
      "step": 30494
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2485733032226562,
      "learning_rate": 1.1353544388716738e-06,
      "loss": 0.783,
      "step": 30495
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.497370958328247,
      "learning_rate": 1.1347155224738843e-06,
      "loss": 0.9944,
      "step": 30496
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.52498197555542,
      "learning_rate": 1.134076606076095e-06,
      "loss": 0.8001,
      "step": 30497
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9579969644546509,
      "learning_rate": 1.1334376896783057e-06,
      "loss": 0.9463,
      "step": 30498
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4140052795410156,
      "learning_rate": 1.1327987732805164e-06,
      "loss": 0.7807,
      "step": 30499
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4595375061035156,
      "learning_rate": 1.1321598568827271e-06,
      "loss": 0.923,
      "step": 30500
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.320279359817505,
      "learning_rate": 1.1315209404849376e-06,
      "loss": 0.9915,
      "step": 30501
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.170348882675171,
      "learning_rate": 1.1308820240871483e-06,
      "loss": 0.69,
      "step": 30502
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7782188653945923,
      "learning_rate": 1.130243107689359e-06,
      "loss": 0.906,
      "step": 30503
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6651800870895386,
      "learning_rate": 1.1296041912915697e-06,
      "loss": 0.6718,
      "step": 30504
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5096590518951416,
      "learning_rate": 1.1289652748937802e-06,
      "loss": 0.6653,
      "step": 30505
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6598143577575684,
      "learning_rate": 1.128326358495991e-06,
      "loss": 0.3521,
      "step": 30506
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.856532335281372,
      "learning_rate": 1.1276874420982016e-06,
      "loss": 0.8147,
      "step": 30507
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1204724311828613,
      "learning_rate": 1.1270485257004123e-06,
      "loss": 0.9457,
      "step": 30508
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.303310871124268,
      "learning_rate": 1.1264096093026228e-06,
      "loss": 0.9555,
      "step": 30509
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8509618043899536,
      "learning_rate": 1.1257706929048335e-06,
      "loss": 0.87,
      "step": 30510
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3896170854568481,
      "learning_rate": 1.125131776507044e-06,
      "loss": 0.4044,
      "step": 30511
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.632826328277588,
      "learning_rate": 1.1244928601092547e-06,
      "loss": 0.8386,
      "step": 30512
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.585343360900879,
      "learning_rate": 1.1238539437114654e-06,
      "loss": 1.0977,
      "step": 30513
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.088850736618042,
      "learning_rate": 1.123215027313676e-06,
      "loss": 0.8992,
      "step": 30514
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3461601734161377,
      "learning_rate": 1.1225761109158868e-06,
      "loss": 0.344,
      "step": 30515
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.8992247581481934,
      "learning_rate": 1.1219371945180973e-06,
      "loss": 0.6446,
      "step": 30516
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8443711996078491,
      "learning_rate": 1.121298278120308e-06,
      "loss": 0.8566,
      "step": 30517
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0586514472961426,
      "learning_rate": 1.1206593617225187e-06,
      "loss": 0.8098,
      "step": 30518
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.632379531860352,
      "learning_rate": 1.1200204453247294e-06,
      "loss": 1.1817,
      "step": 30519
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.8269004821777344,
      "learning_rate": 1.11938152892694e-06,
      "loss": 1.1416,
      "step": 30520
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3202903270721436,
      "learning_rate": 1.1187426125291506e-06,
      "loss": 0.8498,
      "step": 30521
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.090362548828125,
      "learning_rate": 1.1181036961313613e-06,
      "loss": 0.7715,
      "step": 30522
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7719800472259521,
      "learning_rate": 1.117464779733572e-06,
      "loss": 1.0702,
      "step": 30523
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.82218599319458,
      "learning_rate": 1.1168258633357827e-06,
      "loss": 0.7723,
      "step": 30524
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.822059988975525,
      "learning_rate": 1.1161869469379931e-06,
      "loss": 1.1255,
      "step": 30525
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0042500495910645,
      "learning_rate": 1.1155480305402038e-06,
      "loss": 1.1071,
      "step": 30526
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8018555641174316,
      "learning_rate": 1.1149091141424145e-06,
      "loss": 0.8967,
      "step": 30527
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9502683877944946,
      "learning_rate": 1.1142701977446252e-06,
      "loss": 0.9478,
      "step": 30528
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7087618112564087,
      "learning_rate": 1.113631281346836e-06,
      "loss": 0.8348,
      "step": 30529
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.168094635009766,
      "learning_rate": 1.1129923649490464e-06,
      "loss": 1.6461,
      "step": 30530
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7063101530075073,
      "learning_rate": 1.1123534485512571e-06,
      "loss": 0.7639,
      "step": 30531
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.522783041000366,
      "learning_rate": 1.1117145321534678e-06,
      "loss": 0.8933,
      "step": 30532
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.8033106327056885,
      "learning_rate": 1.1110756157556785e-06,
      "loss": 0.6032,
      "step": 30533
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3293559551239014,
      "learning_rate": 1.110436699357889e-06,
      "loss": 0.8278,
      "step": 30534
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7859972715377808,
      "learning_rate": 1.1097977829600997e-06,
      "loss": 0.7882,
      "step": 30535
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8869011402130127,
      "learning_rate": 1.1091588665623104e-06,
      "loss": 0.8167,
      "step": 30536
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9245967864990234,
      "learning_rate": 1.1085199501645211e-06,
      "loss": 0.87,
      "step": 30537
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2112767696380615,
      "learning_rate": 1.1078810337667318e-06,
      "loss": 1.0279,
      "step": 30538
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.137174606323242,
      "learning_rate": 1.1072421173689423e-06,
      "loss": 0.8834,
      "step": 30539
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.832443356513977,
      "learning_rate": 1.106603200971153e-06,
      "loss": 0.6422,
      "step": 30540
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7140569686889648,
      "learning_rate": 1.1059642845733637e-06,
      "loss": 0.7501,
      "step": 30541
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.8991332054138184,
      "learning_rate": 1.1053253681755744e-06,
      "loss": 1.1637,
      "step": 30542
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7574868202209473,
      "learning_rate": 1.104686451777785e-06,
      "loss": 0.7836,
      "step": 30543
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.090049743652344,
      "learning_rate": 1.1040475353799956e-06,
      "loss": 1.2205,
      "step": 30544
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.846284866333008,
      "learning_rate": 1.1034086189822063e-06,
      "loss": 0.9676,
      "step": 30545
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6652092933654785,
      "learning_rate": 1.102769702584417e-06,
      "loss": 0.8672,
      "step": 30546
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.603463649749756,
      "learning_rate": 1.1021307861866277e-06,
      "loss": 1.4862,
      "step": 30547
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9173473119735718,
      "learning_rate": 1.1014918697888382e-06,
      "loss": 0.7776,
      "step": 30548
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.175521373748779,
      "learning_rate": 1.1008529533910489e-06,
      "loss": 1.3009,
      "step": 30549
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0585992336273193,
      "learning_rate": 1.1002140369932596e-06,
      "loss": 0.7959,
      "step": 30550
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6698448657989502,
      "learning_rate": 1.0995751205954703e-06,
      "loss": 0.8497,
      "step": 30551
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.8591666221618652,
      "learning_rate": 1.0989362041976808e-06,
      "loss": 0.8842,
      "step": 30552
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.846513271331787,
      "learning_rate": 1.0982972877998915e-06,
      "loss": 0.9303,
      "step": 30553
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.8272705078125,
      "learning_rate": 1.0976583714021022e-06,
      "loss": 1.0261,
      "step": 30554
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7457767724990845,
      "learning_rate": 1.0970194550043129e-06,
      "loss": 0.7724,
      "step": 30555
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1341488361358643,
      "learning_rate": 1.0963805386065236e-06,
      "loss": 1.1829,
      "step": 30556
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.483150959014893,
      "learning_rate": 1.095741622208734e-06,
      "loss": 0.9076,
      "step": 30557
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.5619237422943115,
      "learning_rate": 1.0951027058109448e-06,
      "loss": 0.7401,
      "step": 30558
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6633168458938599,
      "learning_rate": 1.0944637894131553e-06,
      "loss": 0.7246,
      "step": 30559
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4256389141082764,
      "learning_rate": 1.093824873015366e-06,
      "loss": 0.8241,
      "step": 30560
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.764342188835144,
      "learning_rate": 1.0931859566175767e-06,
      "loss": 0.7832,
      "step": 30561
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6415038108825684,
      "learning_rate": 1.0925470402197874e-06,
      "loss": 0.7702,
      "step": 30562
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7423349618911743,
      "learning_rate": 1.0919081238219978e-06,
      "loss": 0.8954,
      "step": 30563
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7370872497558594,
      "learning_rate": 1.0912692074242085e-06,
      "loss": 0.925,
      "step": 30564
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9159934520721436,
      "learning_rate": 1.0906302910264192e-06,
      "loss": 0.9514,
      "step": 30565
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7967485189437866,
      "learning_rate": 1.08999137462863e-06,
      "loss": 0.8316,
      "step": 30566
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1143393516540527,
      "learning_rate": 1.0893524582308406e-06,
      "loss": 1.0778,
      "step": 30567
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0998215675354004,
      "learning_rate": 1.0887135418330511e-06,
      "loss": 0.8641,
      "step": 30568
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.892636775970459,
      "learning_rate": 1.0880746254352618e-06,
      "loss": 0.9137,
      "step": 30569
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.007940769195557,
      "learning_rate": 1.0874357090374725e-06,
      "loss": 1.3447,
      "step": 30570
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.08737325668335,
      "learning_rate": 1.0867967926396832e-06,
      "loss": 0.9243,
      "step": 30571
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8015377521514893,
      "learning_rate": 1.0861578762418937e-06,
      "loss": 0.7643,
      "step": 30572
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.380340814590454,
      "learning_rate": 1.0855189598441044e-06,
      "loss": 0.4475,
      "step": 30573
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6317602396011353,
      "learning_rate": 1.0848800434463151e-06,
      "loss": 0.7285,
      "step": 30574
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.044553756713867,
      "learning_rate": 1.0842411270485258e-06,
      "loss": 0.8423,
      "step": 30575
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7477916479110718,
      "learning_rate": 1.0836022106507365e-06,
      "loss": 0.8203,
      "step": 30576
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7751814126968384,
      "learning_rate": 1.082963294252947e-06,
      "loss": 0.749,
      "step": 30577
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.8613595962524414,
      "learning_rate": 1.0823243778551577e-06,
      "loss": 0.7531,
      "step": 30578
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6779851913452148,
      "learning_rate": 1.0816854614573684e-06,
      "loss": 0.8127,
      "step": 30579
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.215172290802002,
      "learning_rate": 1.081046545059579e-06,
      "loss": 0.7163,
      "step": 30580
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3368650674819946,
      "learning_rate": 1.0804076286617896e-06,
      "loss": 0.3211,
      "step": 30581
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.834270477294922,
      "learning_rate": 1.0797687122640003e-06,
      "loss": 1.4446,
      "step": 30582
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0457146167755127,
      "learning_rate": 1.079129795866211e-06,
      "loss": 0.8782,
      "step": 30583
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.622168779373169,
      "learning_rate": 1.0784908794684217e-06,
      "loss": 0.9244,
      "step": 30584
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6011813879013062,
      "learning_rate": 1.0778519630706324e-06,
      "loss": 0.8393,
      "step": 30585
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6701009273529053,
      "learning_rate": 1.0772130466728429e-06,
      "loss": 0.7931,
      "step": 30586
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6560193300247192,
      "learning_rate": 1.0765741302750536e-06,
      "loss": 0.7838,
      "step": 30587
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.842488408088684,
      "learning_rate": 1.0759352138772643e-06,
      "loss": 0.7767,
      "step": 30588
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.7576894760131836,
      "learning_rate": 1.075296297479475e-06,
      "loss": 1.1326,
      "step": 30589
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.904815435409546,
      "learning_rate": 1.0746573810816857e-06,
      "loss": 0.745,
      "step": 30590
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5727958679199219,
      "learning_rate": 1.0740184646838962e-06,
      "loss": 0.7949,
      "step": 30591
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.7384848594665527,
      "learning_rate": 1.0733795482861069e-06,
      "loss": 0.8551,
      "step": 30592
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.269617557525635,
      "learning_rate": 1.0727406318883176e-06,
      "loss": 0.9846,
      "step": 30593
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6974947452545166,
      "learning_rate": 1.0721017154905283e-06,
      "loss": 0.7793,
      "step": 30594
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6904244422912598,
      "learning_rate": 1.0714627990927388e-06,
      "loss": 0.795,
      "step": 30595
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.6603920459747314,
      "learning_rate": 1.0708238826949495e-06,
      "loss": 0.743,
      "step": 30596
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8186246156692505,
      "learning_rate": 1.0701849662971602e-06,
      "loss": 0.7602,
      "step": 30597
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.519017457962036,
      "learning_rate": 1.0695460498993709e-06,
      "loss": 0.7178,
      "step": 30598
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.359959363937378,
      "learning_rate": 1.0689071335015816e-06,
      "loss": 0.4716,
      "step": 30599
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7262256145477295,
      "learning_rate": 1.068268217103792e-06,
      "loss": 0.7686,
      "step": 30600
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.787327766418457,
      "learning_rate": 1.0676293007060027e-06,
      "loss": 0.9705,
      "step": 30601
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6221809387207031,
      "learning_rate": 1.0669903843082134e-06,
      "loss": 0.6994,
      "step": 30602
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5716959238052368,
      "learning_rate": 1.0663514679104241e-06,
      "loss": 0.4671,
      "step": 30603
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.8181979656219482,
      "learning_rate": 1.0657125515126346e-06,
      "loss": 0.9094,
      "step": 30604
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1288557052612305,
      "learning_rate": 1.0650736351148453e-06,
      "loss": 0.9037,
      "step": 30605
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0807628631591797,
      "learning_rate": 1.0644347187170558e-06,
      "loss": 0.8664,
      "step": 30606
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0968644618988037,
      "learning_rate": 1.0637958023192665e-06,
      "loss": 1.043,
      "step": 30607
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.5363144874572754,
      "learning_rate": 1.0631568859214772e-06,
      "loss": 1.2641,
      "step": 30608
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.192553997039795,
      "learning_rate": 1.062517969523688e-06,
      "loss": 0.8224,
      "step": 30609
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.37110710144043,
      "learning_rate": 1.0618790531258984e-06,
      "loss": 0.9271,
      "step": 30610
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.707057476043701,
      "learning_rate": 1.0612401367281091e-06,
      "loss": 0.9149,
      "step": 30611
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9824295043945312,
      "learning_rate": 1.0606012203303198e-06,
      "loss": 0.7636,
      "step": 30612
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.843961238861084,
      "learning_rate": 1.0599623039325305e-06,
      "loss": 0.8577,
      "step": 30613
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.779817581176758,
      "learning_rate": 1.0593233875347412e-06,
      "loss": 0.7147,
      "step": 30614
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6833934783935547,
      "learning_rate": 1.0586844711369517e-06,
      "loss": 0.8305,
      "step": 30615
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2397122383117676,
      "learning_rate": 1.0580455547391624e-06,
      "loss": 0.8917,
      "step": 30616
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.498725891113281,
      "learning_rate": 1.057406638341373e-06,
      "loss": 0.9926,
      "step": 30617
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.6244027614593506,
      "learning_rate": 1.0567677219435838e-06,
      "loss": 0.8346,
      "step": 30618
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8065024614334106,
      "learning_rate": 1.0561288055457943e-06,
      "loss": 0.745,
      "step": 30619
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8057218790054321,
      "learning_rate": 1.055489889148005e-06,
      "loss": 0.7862,
      "step": 30620
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.137768268585205,
      "learning_rate": 1.0548509727502157e-06,
      "loss": 0.9525,
      "step": 30621
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.1069176197052,
      "learning_rate": 1.0542120563524264e-06,
      "loss": 0.7597,
      "step": 30622
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8836344480514526,
      "learning_rate": 1.053573139954637e-06,
      "loss": 0.8453,
      "step": 30623
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.296273708343506,
      "learning_rate": 1.0529342235568476e-06,
      "loss": 1.0698,
      "step": 30624
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.359849452972412,
      "learning_rate": 1.0522953071590583e-06,
      "loss": 0.8794,
      "step": 30625
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0089802742004395,
      "learning_rate": 1.051656390761269e-06,
      "loss": 0.8287,
      "step": 30626
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.055689573287964,
      "learning_rate": 1.0510174743634797e-06,
      "loss": 1.0733,
      "step": 30627
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9268678426742554,
      "learning_rate": 1.0503785579656904e-06,
      "loss": 0.849,
      "step": 30628
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8071262836456299,
      "learning_rate": 1.0497396415679009e-06,
      "loss": 0.7888,
      "step": 30629
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7429778575897217,
      "learning_rate": 1.0491007251701116e-06,
      "loss": 0.8859,
      "step": 30630
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.3223061561584473,
      "learning_rate": 1.0484618087723223e-06,
      "loss": 1.0336,
      "step": 30631
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8541069030761719,
      "learning_rate": 1.047822892374533e-06,
      "loss": 0.7562,
      "step": 30632
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.961914539337158,
      "learning_rate": 1.0471839759767434e-06,
      "loss": 1.0403,
      "step": 30633
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.363596439361572,
      "learning_rate": 1.0465450595789541e-06,
      "loss": 1.2041,
      "step": 30634
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2565114498138428,
      "learning_rate": 1.0459061431811648e-06,
      "loss": 0.7835,
      "step": 30635
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0370774269104004,
      "learning_rate": 1.0452672267833755e-06,
      "loss": 0.8128,
      "step": 30636
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.844241738319397,
      "learning_rate": 1.0446283103855862e-06,
      "loss": 0.8672,
      "step": 30637
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.113046646118164,
      "learning_rate": 1.0439893939877967e-06,
      "loss": 0.9076,
      "step": 30638
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0212275981903076,
      "learning_rate": 1.0433504775900074e-06,
      "loss": 0.9116,
      "step": 30639
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9508951902389526,
      "learning_rate": 1.0427115611922181e-06,
      "loss": 0.7446,
      "step": 30640
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.137594699859619,
      "learning_rate": 1.0420726447944288e-06,
      "loss": 0.7359,
      "step": 30641
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8613276481628418,
      "learning_rate": 1.0414337283966393e-06,
      "loss": 0.846,
      "step": 30642
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.593149423599243,
      "learning_rate": 1.04079481199885e-06,
      "loss": 0.7435,
      "step": 30643
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.06855583190918,
      "learning_rate": 1.0401558956010607e-06,
      "loss": 1.1008,
      "step": 30644
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5130867958068848,
      "learning_rate": 1.0395169792032714e-06,
      "loss": 0.4508,
      "step": 30645
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7331591844558716,
      "learning_rate": 1.0388780628054821e-06,
      "loss": 0.8123,
      "step": 30646
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.8940048217773438,
      "learning_rate": 1.0382391464076926e-06,
      "loss": 0.9778,
      "step": 30647
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7402088642120361,
      "learning_rate": 1.0376002300099033e-06,
      "loss": 0.8356,
      "step": 30648
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6349623203277588,
      "learning_rate": 1.036961313612114e-06,
      "loss": 0.7738,
      "step": 30649
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8014299869537354,
      "learning_rate": 1.0363223972143247e-06,
      "loss": 0.8093,
      "step": 30650
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.179581880569458,
      "learning_rate": 1.0356834808165352e-06,
      "loss": 0.7756,
      "step": 30651
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5071282386779785,
      "learning_rate": 1.035044564418746e-06,
      "loss": 0.7798,
      "step": 30652
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.884493112564087,
      "learning_rate": 1.0344056480209566e-06,
      "loss": 0.8828,
      "step": 30653
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.6782772541046143,
      "learning_rate": 1.033766731623167e-06,
      "loss": 0.7719,
      "step": 30654
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3789780139923096,
      "learning_rate": 1.0331278152253778e-06,
      "loss": 0.8155,
      "step": 30655
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6729187965393066,
      "learning_rate": 1.0324888988275885e-06,
      "loss": 0.8648,
      "step": 30656
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7991362810134888,
      "learning_rate": 1.0318499824297992e-06,
      "loss": 0.8836,
      "step": 30657
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.0288479328155518,
      "learning_rate": 1.0312110660320097e-06,
      "loss": 0.5701,
      "step": 30658
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.830556869506836,
      "learning_rate": 1.0305721496342204e-06,
      "loss": 0.716,
      "step": 30659
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.640009880065918,
      "learning_rate": 1.029933233236431e-06,
      "loss": 0.8221,
      "step": 30660
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6398285627365112,
      "learning_rate": 1.0292943168386418e-06,
      "loss": 0.7999,
      "step": 30661
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5756646394729614,
      "learning_rate": 1.0286554004408523e-06,
      "loss": 0.7827,
      "step": 30662
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5112110376358032,
      "learning_rate": 1.028016484043063e-06,
      "loss": 0.747,
      "step": 30663
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9749794006347656,
      "learning_rate": 1.0273775676452737e-06,
      "loss": 0.8672,
      "step": 30664
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.681504487991333,
      "learning_rate": 1.0267386512474844e-06,
      "loss": 0.7325,
      "step": 30665
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4363362789154053,
      "learning_rate": 1.026099734849695e-06,
      "loss": 1.0438,
      "step": 30666
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.199503183364868,
      "learning_rate": 1.0254608184519056e-06,
      "loss": 0.748,
      "step": 30667
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3245829343795776,
      "learning_rate": 1.0248219020541163e-06,
      "loss": 0.4417,
      "step": 30668
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8988215923309326,
      "learning_rate": 1.024182985656327e-06,
      "loss": 0.7006,
      "step": 30669
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.750339508056641,
      "learning_rate": 1.0235440692585377e-06,
      "loss": 0.8487,
      "step": 30670
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.024684190750122,
      "learning_rate": 1.0229051528607481e-06,
      "loss": 0.9488,
      "step": 30671
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.29189395904541,
      "learning_rate": 1.0222662364629588e-06,
      "loss": 0.574,
      "step": 30672
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.036710500717163,
      "learning_rate": 1.0216273200651695e-06,
      "loss": 0.9426,
      "step": 30673
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.619929075241089,
      "learning_rate": 1.0209884036673802e-06,
      "loss": 0.6671,
      "step": 30674
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4750491380691528,
      "learning_rate": 1.020349487269591e-06,
      "loss": 0.3751,
      "step": 30675
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.729512095451355,
      "learning_rate": 1.0197105708718014e-06,
      "loss": 0.7403,
      "step": 30676
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9646064043045044,
      "learning_rate": 1.0190716544740121e-06,
      "loss": 0.8503,
      "step": 30677
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6345818042755127,
      "learning_rate": 1.0184327380762228e-06,
      "loss": 0.8003,
      "step": 30678
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0328686237335205,
      "learning_rate": 1.0177938216784335e-06,
      "loss": 1.0328,
      "step": 30679
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3545541763305664,
      "learning_rate": 1.017154905280644e-06,
      "loss": 0.9061,
      "step": 30680
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.483630180358887,
      "learning_rate": 1.0165159888828547e-06,
      "loss": 1.2539,
      "step": 30681
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.97399640083313,
      "learning_rate": 1.0158770724850654e-06,
      "loss": 0.9879,
      "step": 30682
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5948576927185059,
      "learning_rate": 1.0152381560872761e-06,
      "loss": 0.6655,
      "step": 30683
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8595726490020752,
      "learning_rate": 1.0145992396894868e-06,
      "loss": 0.8595,
      "step": 30684
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.124337673187256,
      "learning_rate": 1.0139603232916973e-06,
      "loss": 1.0637,
      "step": 30685
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1298465728759766,
      "learning_rate": 1.013321406893908e-06,
      "loss": 0.8053,
      "step": 30686
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5601959228515625,
      "learning_rate": 1.0126824904961187e-06,
      "loss": 0.8108,
      "step": 30687
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.590101957321167,
      "learning_rate": 1.0120435740983294e-06,
      "loss": 0.776,
      "step": 30688
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0722146034240723,
      "learning_rate": 1.01140465770054e-06,
      "loss": 0.7427,
      "step": 30689
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.810800075531006,
      "learning_rate": 1.0107657413027506e-06,
      "loss": 0.8119,
      "step": 30690
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8133008480072021,
      "learning_rate": 1.0101268249049613e-06,
      "loss": 0.8283,
      "step": 30691
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3984122276306152,
      "learning_rate": 1.009487908507172e-06,
      "loss": 0.9743,
      "step": 30692
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2490592002868652,
      "learning_rate": 1.0088489921093827e-06,
      "loss": 0.8938,
      "step": 30693
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.384168028831482,
      "learning_rate": 1.0082100757115932e-06,
      "loss": 0.3514,
      "step": 30694
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9998433589935303,
      "learning_rate": 1.0075711593138039e-06,
      "loss": 0.7848,
      "step": 30695
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6495397090911865,
      "learning_rate": 1.0069322429160146e-06,
      "loss": 0.7482,
      "step": 30696
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.372981309890747,
      "learning_rate": 1.0062933265182253e-06,
      "loss": 0.6531,
      "step": 30697
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.786439299583435,
      "learning_rate": 1.005654410120436e-06,
      "loss": 0.7945,
      "step": 30698
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1212503910064697,
      "learning_rate": 1.0050154937226465e-06,
      "loss": 0.9218,
      "step": 30699
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.9795403480529785,
      "learning_rate": 1.0043765773248572e-06,
      "loss": 1.5072,
      "step": 30700
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7333375215530396,
      "learning_rate": 1.0037376609270679e-06,
      "loss": 0.8631,
      "step": 30701
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.763770341873169,
      "learning_rate": 1.0030987445292786e-06,
      "loss": 0.6056,
      "step": 30702
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.7344753742218018,
      "learning_rate": 1.002459828131489e-06,
      "loss": 0.761,
      "step": 30703
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5028353929519653,
      "learning_rate": 1.0018209117336998e-06,
      "loss": 0.8026,
      "step": 30704
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.7252533435821533,
      "learning_rate": 1.0011819953359102e-06,
      "loss": 0.9042,
      "step": 30705
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4442620277404785,
      "learning_rate": 1.000543078938121e-06,
      "loss": 0.7927,
      "step": 30706
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4423444271087646,
      "learning_rate": 9.999041625403316e-07,
      "loss": 0.8151,
      "step": 30707
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7153908014297485,
      "learning_rate": 9.992652461425423e-07,
      "loss": 0.7734,
      "step": 30708
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.723732829093933,
      "learning_rate": 9.986263297447528e-07,
      "loss": 0.8909,
      "step": 30709
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.3985280990600586,
      "learning_rate": 9.979874133469635e-07,
      "loss": 0.9809,
      "step": 30710
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4924184083938599,
      "learning_rate": 9.973484969491742e-07,
      "loss": 0.804,
      "step": 30711
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.037717580795288,
      "learning_rate": 9.96709580551385e-07,
      "loss": 1.007,
      "step": 30712
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1062605381011963,
      "learning_rate": 9.960706641535956e-07,
      "loss": 0.9258,
      "step": 30713
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.710769772529602,
      "learning_rate": 9.954317477558061e-07,
      "loss": 0.8599,
      "step": 30714
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7558412551879883,
      "learning_rate": 9.947928313580168e-07,
      "loss": 0.795,
      "step": 30715
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2028989791870117,
      "learning_rate": 9.941539149602275e-07,
      "loss": 0.7726,
      "step": 30716
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2564456462860107,
      "learning_rate": 9.935149985624382e-07,
      "loss": 0.7559,
      "step": 30717
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.782346248626709,
      "learning_rate": 9.928760821646487e-07,
      "loss": 0.7494,
      "step": 30718
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.141507148742676,
      "learning_rate": 9.922371657668594e-07,
      "loss": 0.8638,
      "step": 30719
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.427854061126709,
      "learning_rate": 9.915982493690701e-07,
      "loss": 0.8329,
      "step": 30720
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.222410202026367,
      "learning_rate": 9.909593329712808e-07,
      "loss": 0.5629,
      "step": 30721
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4237767457962036,
      "learning_rate": 9.903204165734915e-07,
      "loss": 0.3538,
      "step": 30722
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.686981678009033,
      "learning_rate": 9.89681500175702e-07,
      "loss": 1.1094,
      "step": 30723
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.9959611892700195,
      "learning_rate": 9.890425837779127e-07,
      "loss": 0.9424,
      "step": 30724
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.8857405185699463,
      "learning_rate": 9.884036673801234e-07,
      "loss": 1.0614,
      "step": 30725
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.584477186203003,
      "learning_rate": 9.87764750982334e-07,
      "loss": 0.807,
      "step": 30726
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.640244483947754,
      "learning_rate": 9.871258345845448e-07,
      "loss": 0.8927,
      "step": 30727
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.030030250549316,
      "learning_rate": 9.864869181867553e-07,
      "loss": 1.4093,
      "step": 30728
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8731127977371216,
      "learning_rate": 9.85848001788966e-07,
      "loss": 0.881,
      "step": 30729
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0998337268829346,
      "learning_rate": 9.852090853911767e-07,
      "loss": 0.8695,
      "step": 30730
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6799591779708862,
      "learning_rate": 9.845701689933874e-07,
      "loss": 0.755,
      "step": 30731
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.786971926689148,
      "learning_rate": 9.839312525955979e-07,
      "loss": 0.8229,
      "step": 30732
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5824052095413208,
      "learning_rate": 9.832923361978086e-07,
      "loss": 0.7892,
      "step": 30733
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.3284223079681396,
      "learning_rate": 9.826534198000193e-07,
      "loss": 0.8452,
      "step": 30734
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.1373865604400635,
      "learning_rate": 9.8201450340223e-07,
      "loss": 0.8243,
      "step": 30735
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.5914483070373535,
      "learning_rate": 9.813755870044407e-07,
      "loss": 1.5972,
      "step": 30736
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9560422897338867,
      "learning_rate": 9.807366706066512e-07,
      "loss": 0.8134,
      "step": 30737
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.3109660148620605,
      "learning_rate": 9.800977542088619e-07,
      "loss": 1.0704,
      "step": 30738
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.5863022804260254,
      "learning_rate": 9.794588378110726e-07,
      "loss": 0.8795,
      "step": 30739
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1467719078063965,
      "learning_rate": 9.788199214132833e-07,
      "loss": 0.9412,
      "step": 30740
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.651571750640869,
      "learning_rate": 9.781810050154937e-07,
      "loss": 0.8781,
      "step": 30741
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.325268268585205,
      "learning_rate": 9.775420886177044e-07,
      "loss": 0.785,
      "step": 30742
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.042832374572754,
      "learning_rate": 9.769031722199151e-07,
      "loss": 0.7543,
      "step": 30743
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.398235321044922,
      "learning_rate": 9.762642558221258e-07,
      "loss": 0.9125,
      "step": 30744
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8330700397491455,
      "learning_rate": 9.756253394243365e-07,
      "loss": 0.8826,
      "step": 30745
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9165544509887695,
      "learning_rate": 9.74986423026547e-07,
      "loss": 0.7231,
      "step": 30746
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.6971817016601562,
      "learning_rate": 9.743475066287577e-07,
      "loss": 0.6265,
      "step": 30747
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.990346908569336,
      "learning_rate": 9.737085902309684e-07,
      "loss": 0.816,
      "step": 30748
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.478578567504883,
      "learning_rate": 9.730696738331791e-07,
      "loss": 1.2368,
      "step": 30749
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.724331855773926,
      "learning_rate": 9.724307574353896e-07,
      "loss": 1.2864,
      "step": 30750
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6912895441055298,
      "learning_rate": 9.717918410376003e-07,
      "loss": 0.8049,
      "step": 30751
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8140830993652344,
      "learning_rate": 9.711529246398108e-07,
      "loss": 0.8631,
      "step": 30752
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.380315065383911,
      "learning_rate": 9.705140082420215e-07,
      "loss": 1.0978,
      "step": 30753
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5053497552871704,
      "learning_rate": 9.698750918442322e-07,
      "loss": 0.7566,
      "step": 30754
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6851716041564941,
      "learning_rate": 9.69236175446443e-07,
      "loss": 0.7222,
      "step": 30755
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.457435131072998,
      "learning_rate": 9.685972590486536e-07,
      "loss": 0.6651,
      "step": 30756
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8826338052749634,
      "learning_rate": 9.67958342650864e-07,
      "loss": 0.8292,
      "step": 30757
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.833187580108643,
      "learning_rate": 9.673194262530748e-07,
      "loss": 1.2646,
      "step": 30758
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.921604156494141,
      "learning_rate": 9.666805098552855e-07,
      "loss": 1.1163,
      "step": 30759
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7163032293319702,
      "learning_rate": 9.660415934574962e-07,
      "loss": 0.7313,
      "step": 30760
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.0852766036987305,
      "learning_rate": 9.654026770597067e-07,
      "loss": 1.1592,
      "step": 30761
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9090038537979126,
      "learning_rate": 9.647637606619174e-07,
      "loss": 0.8712,
      "step": 30762
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.786423683166504,
      "learning_rate": 9.64124844264128e-07,
      "loss": 0.6152,
      "step": 30763
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4697414636611938,
      "learning_rate": 9.634859278663388e-07,
      "loss": 0.3574,
      "step": 30764
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.3211469650268555,
      "learning_rate": 9.628470114685495e-07,
      "loss": 1.3485,
      "step": 30765
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2434839010238647,
      "learning_rate": 9.6220809507076e-07,
      "loss": 0.3669,
      "step": 30766
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.14304780960083,
      "learning_rate": 9.615691786729707e-07,
      "loss": 1.2012,
      "step": 30767
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6806033849716187,
      "learning_rate": 9.609302622751814e-07,
      "loss": 0.3553,
      "step": 30768
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8442491292953491,
      "learning_rate": 9.60291345877392e-07,
      "loss": 0.7933,
      "step": 30769
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.624245643615723,
      "learning_rate": 9.596524294796026e-07,
      "loss": 1.5973,
      "step": 30770
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8754591941833496,
      "learning_rate": 9.590135130818133e-07,
      "loss": 0.7566,
      "step": 30771
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4973903894424438,
      "learning_rate": 9.58374596684024e-07,
      "loss": 0.797,
      "step": 30772
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0548739433288574,
      "learning_rate": 9.577356802862347e-07,
      "loss": 0.9122,
      "step": 30773
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.357778549194336,
      "learning_rate": 9.570967638884454e-07,
      "loss": 1.4279,
      "step": 30774
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5208200216293335,
      "learning_rate": 9.564578474906559e-07,
      "loss": 0.7829,
      "step": 30775
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.117013454437256,
      "learning_rate": 9.558189310928666e-07,
      "loss": 0.9569,
      "step": 30776
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.3142616748809814,
      "learning_rate": 9.551800146950773e-07,
      "loss": 0.7,
      "step": 30777
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7658324241638184,
      "learning_rate": 9.54541098297288e-07,
      "loss": 0.8293,
      "step": 30778
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.037060260772705,
      "learning_rate": 9.539021818994984e-07,
      "loss": 1.1424,
      "step": 30779
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7936055660247803,
      "learning_rate": 9.532632655017092e-07,
      "loss": 0.8057,
      "step": 30780
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.557258129119873,
      "learning_rate": 9.526243491039198e-07,
      "loss": 1.155,
      "step": 30781
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6916855573654175,
      "learning_rate": 9.519854327061305e-07,
      "loss": 0.8366,
      "step": 30782
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7482984066009521,
      "learning_rate": 9.513465163083411e-07,
      "loss": 0.841,
      "step": 30783
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.9864203929901123,
      "learning_rate": 9.507075999105518e-07,
      "loss": 0.7049,
      "step": 30784
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.6899497509002686,
      "learning_rate": 9.500686835127624e-07,
      "loss": 0.9194,
      "step": 30785
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.023486852645874,
      "learning_rate": 9.494297671149731e-07,
      "loss": 0.7858,
      "step": 30786
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5363041162490845,
      "learning_rate": 9.487908507171837e-07,
      "loss": 0.7772,
      "step": 30787
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8185288906097412,
      "learning_rate": 9.481519343193944e-07,
      "loss": 0.8789,
      "step": 30788
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8486355543136597,
      "learning_rate": 9.475130179216051e-07,
      "loss": 0.792,
      "step": 30789
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6708348989486694,
      "learning_rate": 9.468741015238157e-07,
      "loss": 0.7663,
      "step": 30790
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.336045742034912,
      "learning_rate": 9.462351851260264e-07,
      "loss": 0.8693,
      "step": 30791
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9973338842391968,
      "learning_rate": 9.45596268728237e-07,
      "loss": 0.7845,
      "step": 30792
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6182295083999634,
      "learning_rate": 9.449573523304477e-07,
      "loss": 0.7498,
      "step": 30793
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2560362815856934,
      "learning_rate": 9.443184359326583e-07,
      "loss": 1.2276,
      "step": 30794
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9632833003997803,
      "learning_rate": 9.43679519534869e-07,
      "loss": 0.8499,
      "step": 30795
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9652073383331299,
      "learning_rate": 9.430406031370796e-07,
      "loss": 0.8095,
      "step": 30796
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.884407877922058,
      "learning_rate": 9.424016867392903e-07,
      "loss": 1.2457,
      "step": 30797
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9261196851730347,
      "learning_rate": 9.41762770341501e-07,
      "loss": 1.2779,
      "step": 30798
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.474215507507324,
      "learning_rate": 9.411238539437115e-07,
      "loss": 0.8479,
      "step": 30799
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7845624685287476,
      "learning_rate": 9.404849375459221e-07,
      "loss": 0.8108,
      "step": 30800
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8580501079559326,
      "learning_rate": 9.398460211481328e-07,
      "loss": 0.8133,
      "step": 30801
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.012636661529541,
      "learning_rate": 9.392071047503435e-07,
      "loss": 0.8943,
      "step": 30802
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.26255202293396,
      "learning_rate": 9.385681883525541e-07,
      "loss": 1.1575,
      "step": 30803
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.831897497177124,
      "learning_rate": 9.379292719547648e-07,
      "loss": 0.9203,
      "step": 30804
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.2882256507873535,
      "learning_rate": 9.372903555569754e-07,
      "loss": 0.951,
      "step": 30805
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5309948921203613,
      "learning_rate": 9.366514391591861e-07,
      "loss": 0.7415,
      "step": 30806
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.62200665473938,
      "learning_rate": 9.360125227613967e-07,
      "loss": 0.8281,
      "step": 30807
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0175223350524902,
      "learning_rate": 9.353736063636074e-07,
      "loss": 0.8658,
      "step": 30808
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.994032859802246,
      "learning_rate": 9.34734689965818e-07,
      "loss": 0.7658,
      "step": 30809
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.1307692527771,
      "learning_rate": 9.340957735680287e-07,
      "loss": 1.3728,
      "step": 30810
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9010082483291626,
      "learning_rate": 9.334568571702394e-07,
      "loss": 0.9012,
      "step": 30811
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.896483302116394,
      "learning_rate": 9.3281794077245e-07,
      "loss": 0.8656,
      "step": 30812
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0575878620147705,
      "learning_rate": 9.321790243746607e-07,
      "loss": 1.2239,
      "step": 30813
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7843884229660034,
      "learning_rate": 9.315401079768712e-07,
      "loss": 0.8778,
      "step": 30814
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5414047241210938,
      "learning_rate": 9.309011915790819e-07,
      "loss": 0.8266,
      "step": 30815
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6052751541137695,
      "learning_rate": 9.302622751812925e-07,
      "loss": 0.6894,
      "step": 30816
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8716132640838623,
      "learning_rate": 9.296233587835032e-07,
      "loss": 0.8324,
      "step": 30817
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1328482627868652,
      "learning_rate": 9.289844423857139e-07,
      "loss": 1.1786,
      "step": 30818
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.095266819000244,
      "learning_rate": 9.283455259879245e-07,
      "loss": 0.7857,
      "step": 30819
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.473611831665039,
      "learning_rate": 9.277066095901352e-07,
      "loss": 0.727,
      "step": 30820
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9193789958953857,
      "learning_rate": 9.270676931923458e-07,
      "loss": 0.6902,
      "step": 30821
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9884730577468872,
      "learning_rate": 9.264287767945565e-07,
      "loss": 0.8335,
      "step": 30822
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7080631256103516,
      "learning_rate": 9.257898603967671e-07,
      "loss": 0.8883,
      "step": 30823
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6644262075424194,
      "learning_rate": 9.251509439989778e-07,
      "loss": 0.7869,
      "step": 30824
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0877761840820312,
      "learning_rate": 9.245120276011884e-07,
      "loss": 0.7941,
      "step": 30825
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.390470027923584,
      "learning_rate": 9.238731112033991e-07,
      "loss": 1.0073,
      "step": 30826
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5777705907821655,
      "learning_rate": 9.232341948056098e-07,
      "loss": 0.7662,
      "step": 30827
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5474244356155396,
      "learning_rate": 9.225952784078204e-07,
      "loss": 0.7283,
      "step": 30828
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6969505548477173,
      "learning_rate": 9.219563620100311e-07,
      "loss": 0.8283,
      "step": 30829
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6984643936157227,
      "learning_rate": 9.213174456122417e-07,
      "loss": 0.7676,
      "step": 30830
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0026447772979736,
      "learning_rate": 9.206785292144524e-07,
      "loss": 0.9153,
      "step": 30831
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9361480474472046,
      "learning_rate": 9.20039612816663e-07,
      "loss": 0.9737,
      "step": 30832
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.416058540344238,
      "learning_rate": 9.194006964188737e-07,
      "loss": 0.8404,
      "step": 30833
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7797844409942627,
      "learning_rate": 9.187617800210844e-07,
      "loss": 0.7439,
      "step": 30834
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.018995761871338,
      "learning_rate": 9.18122863623295e-07,
      "loss": 0.7903,
      "step": 30835
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.509904384613037,
      "learning_rate": 9.174839472255057e-07,
      "loss": 0.7925,
      "step": 30836
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7241584062576294,
      "learning_rate": 9.168450308277163e-07,
      "loss": 0.8673,
      "step": 30837
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.418137550354004,
      "learning_rate": 9.16206114429927e-07,
      "loss": 0.9556,
      "step": 30838
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8076186180114746,
      "learning_rate": 9.155671980321376e-07,
      "loss": 0.7735,
      "step": 30839
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5261226892471313,
      "learning_rate": 9.149282816343483e-07,
      "loss": 0.3373,
      "step": 30840
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4931728839874268,
      "learning_rate": 9.142893652365589e-07,
      "loss": 0.7461,
      "step": 30841
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.671783924102783,
      "learning_rate": 9.136504488387696e-07,
      "loss": 1.0575,
      "step": 30842
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2412495613098145,
      "learning_rate": 9.130115324409803e-07,
      "loss": 0.7984,
      "step": 30843
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.359095573425293,
      "learning_rate": 9.123726160431909e-07,
      "loss": 0.5289,
      "step": 30844
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8835848569869995,
      "learning_rate": 9.117336996454016e-07,
      "loss": 0.7597,
      "step": 30845
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9485794305801392,
      "learning_rate": 9.110947832476122e-07,
      "loss": 0.6982,
      "step": 30846
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6583222150802612,
      "learning_rate": 9.104558668498228e-07,
      "loss": 0.7236,
      "step": 30847
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7130358219146729,
      "learning_rate": 9.098169504520334e-07,
      "loss": 0.8472,
      "step": 30848
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.638427734375,
      "learning_rate": 9.09178034054244e-07,
      "loss": 0.7333,
      "step": 30849
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3626654148101807,
      "learning_rate": 9.085391176564546e-07,
      "loss": 0.7993,
      "step": 30850
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.325256586074829,
      "learning_rate": 9.079002012586653e-07,
      "loss": 1.2079,
      "step": 30851
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.033360719680786,
      "learning_rate": 9.072612848608759e-07,
      "loss": 0.6533,
      "step": 30852
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.468126893043518,
      "learning_rate": 9.066223684630866e-07,
      "loss": 0.4656,
      "step": 30853
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.701393723487854,
      "learning_rate": 9.059834520652972e-07,
      "loss": 0.8177,
      "step": 30854
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.269103527069092,
      "learning_rate": 9.053445356675079e-07,
      "loss": 0.9286,
      "step": 30855
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.265505075454712,
      "learning_rate": 9.047056192697186e-07,
      "loss": 0.9779,
      "step": 30856
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1580326557159424,
      "learning_rate": 9.040667028719292e-07,
      "loss": 0.8225,
      "step": 30857
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.570619106292725,
      "learning_rate": 9.034277864741399e-07,
      "loss": 1.2175,
      "step": 30858
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.366276979446411,
      "learning_rate": 9.027888700763505e-07,
      "loss": 0.9168,
      "step": 30859
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7439881563186646,
      "learning_rate": 9.021499536785612e-07,
      "loss": 0.6669,
      "step": 30860
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.8703529834747314,
      "learning_rate": 9.015110372807718e-07,
      "loss": 0.9401,
      "step": 30861
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0160627365112305,
      "learning_rate": 9.008721208829825e-07,
      "loss": 0.8571,
      "step": 30862
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8359928131103516,
      "learning_rate": 9.002332044851932e-07,
      "loss": 0.8617,
      "step": 30863
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.1835551261901855,
      "learning_rate": 8.995942880874038e-07,
      "loss": 0.9861,
      "step": 30864
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.264639139175415,
      "learning_rate": 8.989553716896145e-07,
      "loss": 0.7881,
      "step": 30865
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.785819411277771,
      "learning_rate": 8.983164552918251e-07,
      "loss": 0.8072,
      "step": 30866
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.855502963066101,
      "learning_rate": 8.976775388940358e-07,
      "loss": 0.8372,
      "step": 30867
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2115674018859863,
      "learning_rate": 8.970386224962464e-07,
      "loss": 0.8773,
      "step": 30868
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.5468533039093018,
      "learning_rate": 8.963997060984571e-07,
      "loss": 0.8717,
      "step": 30869
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8574414253234863,
      "learning_rate": 8.957607897006677e-07,
      "loss": 0.8622,
      "step": 30870
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.4753880500793457,
      "learning_rate": 8.951218733028784e-07,
      "loss": 0.745,
      "step": 30871
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8345465660095215,
      "learning_rate": 8.944829569050891e-07,
      "loss": 0.7701,
      "step": 30872
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4957690238952637,
      "learning_rate": 8.938440405072997e-07,
      "loss": 0.8351,
      "step": 30873
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6890367269515991,
      "learning_rate": 8.932051241095104e-07,
      "loss": 0.9146,
      "step": 30874
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.321019649505615,
      "learning_rate": 8.92566207711721e-07,
      "loss": 0.9755,
      "step": 30875
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.122627258300781,
      "learning_rate": 8.919272913139317e-07,
      "loss": 0.715,
      "step": 30876
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.123138904571533,
      "learning_rate": 8.912883749161423e-07,
      "loss": 0.9176,
      "step": 30877
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6747345924377441,
      "learning_rate": 8.90649458518353e-07,
      "loss": 0.8792,
      "step": 30878
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.922548294067383,
      "learning_rate": 8.900105421205636e-07,
      "loss": 0.9148,
      "step": 30879
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4910465478897095,
      "learning_rate": 8.893716257227743e-07,
      "loss": 0.77,
      "step": 30880
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.192958116531372,
      "learning_rate": 8.88732709324985e-07,
      "loss": 0.7708,
      "step": 30881
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.7341854572296143,
      "learning_rate": 8.880937929271956e-07,
      "loss": 0.7887,
      "step": 30882
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.5811338424682617,
      "learning_rate": 8.874548765294063e-07,
      "loss": 0.8524,
      "step": 30883
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6506046056747437,
      "learning_rate": 8.868159601316169e-07,
      "loss": 0.8112,
      "step": 30884
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.870793342590332,
      "learning_rate": 8.861770437338276e-07,
      "loss": 1.0819,
      "step": 30885
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.798045635223389,
      "learning_rate": 8.855381273360381e-07,
      "loss": 1.134,
      "step": 30886
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.816913604736328,
      "learning_rate": 8.848992109382488e-07,
      "loss": 1.1036,
      "step": 30887
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.273224830627441,
      "learning_rate": 8.842602945404595e-07,
      "loss": 1.1396,
      "step": 30888
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.517789363861084,
      "learning_rate": 8.836213781426701e-07,
      "loss": 0.4811,
      "step": 30889
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.563779354095459,
      "learning_rate": 8.829824617448808e-07,
      "loss": 1.4899,
      "step": 30890
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.032618522644043,
      "learning_rate": 8.823435453470914e-07,
      "loss": 1.1619,
      "step": 30891
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8113346099853516,
      "learning_rate": 8.817046289493021e-07,
      "loss": 0.9705,
      "step": 30892
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3879261016845703,
      "learning_rate": 8.810657125515127e-07,
      "loss": 0.7165,
      "step": 30893
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.752753973007202,
      "learning_rate": 8.804267961537234e-07,
      "loss": 1.0146,
      "step": 30894
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.195062637329102,
      "learning_rate": 8.79787879755934e-07,
      "loss": 1.2433,
      "step": 30895
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.975305199623108,
      "learning_rate": 8.791489633581446e-07,
      "loss": 0.8873,
      "step": 30896
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7948110103607178,
      "learning_rate": 8.785100469603552e-07,
      "loss": 0.8145,
      "step": 30897
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.881470799446106,
      "learning_rate": 8.778711305625659e-07,
      "loss": 0.7604,
      "step": 30898
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7229775190353394,
      "learning_rate": 8.772322141647765e-07,
      "loss": 1.1946,
      "step": 30899
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7072937488555908,
      "learning_rate": 8.765932977669872e-07,
      "loss": 0.8735,
      "step": 30900
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.307193756103516,
      "learning_rate": 8.759543813691979e-07,
      "loss": 0.6487,
      "step": 30901
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.8869097232818604,
      "learning_rate": 8.753154649714085e-07,
      "loss": 1.1622,
      "step": 30902
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9298127889633179,
      "learning_rate": 8.746765485736192e-07,
      "loss": 0.7972,
      "step": 30903
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4965403079986572,
      "learning_rate": 8.740376321758298e-07,
      "loss": 0.8993,
      "step": 30904
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.767502784729004,
      "learning_rate": 8.733987157780405e-07,
      "loss": 0.6992,
      "step": 30905
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8294172286987305,
      "learning_rate": 8.727597993802511e-07,
      "loss": 0.797,
      "step": 30906
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5231605768203735,
      "learning_rate": 8.721208829824618e-07,
      "loss": 0.6587,
      "step": 30907
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.06601619720459,
      "learning_rate": 8.714819665846724e-07,
      "loss": 1.0052,
      "step": 30908
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9245209693908691,
      "learning_rate": 8.708430501868831e-07,
      "loss": 0.9359,
      "step": 30909
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.037579536437988,
      "learning_rate": 8.702041337890938e-07,
      "loss": 0.8052,
      "step": 30910
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.235238552093506,
      "learning_rate": 8.695652173913044e-07,
      "loss": 0.9003,
      "step": 30911
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5479129552841187,
      "learning_rate": 8.689263009935151e-07,
      "loss": 0.7934,
      "step": 30912
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7634053230285645,
      "learning_rate": 8.682873845957257e-07,
      "loss": 0.7605,
      "step": 30913
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8220773935317993,
      "learning_rate": 8.676484681979364e-07,
      "loss": 0.9088,
      "step": 30914
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.0976619720458984,
      "learning_rate": 8.67009551800147e-07,
      "loss": 1.1937,
      "step": 30915
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7149267196655273,
      "learning_rate": 8.663706354023577e-07,
      "loss": 0.8916,
      "step": 30916
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7755413055419922,
      "learning_rate": 8.657317190045684e-07,
      "loss": 0.7754,
      "step": 30917
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.284170389175415,
      "learning_rate": 8.65092802606779e-07,
      "loss": 0.5949,
      "step": 30918
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.787325143814087,
      "learning_rate": 8.644538862089897e-07,
      "loss": 1.2943,
      "step": 30919
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.968822956085205,
      "learning_rate": 8.638149698112003e-07,
      "loss": 1.186,
      "step": 30920
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.825368881225586,
      "learning_rate": 8.63176053413411e-07,
      "loss": 0.75,
      "step": 30921
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9414572715759277,
      "learning_rate": 8.625371370156215e-07,
      "loss": 1.1805,
      "step": 30922
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6252336502075195,
      "learning_rate": 8.618982206178322e-07,
      "loss": 0.8479,
      "step": 30923
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.274018287658691,
      "learning_rate": 8.612593042200428e-07,
      "loss": 1.0461,
      "step": 30924
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8528896570205688,
      "learning_rate": 8.606203878222535e-07,
      "loss": 0.8644,
      "step": 30925
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4754066467285156,
      "learning_rate": 8.599814714244642e-07,
      "loss": 0.7626,
      "step": 30926
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.089301586151123,
      "learning_rate": 8.593425550266748e-07,
      "loss": 0.7643,
      "step": 30927
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8895715475082397,
      "learning_rate": 8.587036386288855e-07,
      "loss": 0.8759,
      "step": 30928
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8088706731796265,
      "learning_rate": 8.580647222310961e-07,
      "loss": 0.8327,
      "step": 30929
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.115110874176025,
      "learning_rate": 8.574258058333068e-07,
      "loss": 1.0618,
      "step": 30930
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.078416347503662,
      "learning_rate": 8.567868894355174e-07,
      "loss": 1.1339,
      "step": 30931
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.625643014907837,
      "learning_rate": 8.561479730377281e-07,
      "loss": 0.7578,
      "step": 30932
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7129231691360474,
      "learning_rate": 8.555090566399388e-07,
      "loss": 0.7235,
      "step": 30933
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9050661325454712,
      "learning_rate": 8.548701402421494e-07,
      "loss": 0.8205,
      "step": 30934
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.153592109680176,
      "learning_rate": 8.542312238443601e-07,
      "loss": 0.8602,
      "step": 30935
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.923514723777771,
      "learning_rate": 8.535923074465707e-07,
      "loss": 0.9068,
      "step": 30936
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0405561923980713,
      "learning_rate": 8.529533910487814e-07,
      "loss": 0.7959,
      "step": 30937
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8544585704803467,
      "learning_rate": 8.52314474650992e-07,
      "loss": 0.7085,
      "step": 30938
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.475053071975708,
      "learning_rate": 8.516755582532027e-07,
      "loss": 0.9029,
      "step": 30939
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.840787649154663,
      "learning_rate": 8.510366418554133e-07,
      "loss": 0.95,
      "step": 30940
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9068982601165771,
      "learning_rate": 8.50397725457624e-07,
      "loss": 0.8554,
      "step": 30941
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8314061164855957,
      "learning_rate": 8.497588090598347e-07,
      "loss": 1.0608,
      "step": 30942
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6657181978225708,
      "learning_rate": 8.491198926620453e-07,
      "loss": 0.7938,
      "step": 30943
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8320529460906982,
      "learning_rate": 8.484809762642558e-07,
      "loss": 0.7506,
      "step": 30944
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9713176488876343,
      "learning_rate": 8.478420598664665e-07,
      "loss": 0.9041,
      "step": 30945
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.427085518836975,
      "learning_rate": 8.472031434686772e-07,
      "loss": 0.7754,
      "step": 30946
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5739086866378784,
      "learning_rate": 8.465642270708878e-07,
      "loss": 0.7718,
      "step": 30947
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.075479030609131,
      "learning_rate": 8.459253106730985e-07,
      "loss": 0.9025,
      "step": 30948
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.9100871086120605,
      "learning_rate": 8.452863942753091e-07,
      "loss": 1.5,
      "step": 30949
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.990343451499939,
      "learning_rate": 8.446474778775198e-07,
      "loss": 0.831,
      "step": 30950
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.6999013423919678,
      "learning_rate": 8.440085614797304e-07,
      "loss": 0.5843,
      "step": 30951
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.700730323791504,
      "learning_rate": 8.433696450819411e-07,
      "loss": 1.1144,
      "step": 30952
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.792373538017273,
      "learning_rate": 8.427307286841517e-07,
      "loss": 0.8854,
      "step": 30953
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.711024284362793,
      "learning_rate": 8.420918122863624e-07,
      "loss": 0.7866,
      "step": 30954
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6037825345993042,
      "learning_rate": 8.414528958885731e-07,
      "loss": 0.6851,
      "step": 30955
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8669984340667725,
      "learning_rate": 8.408139794907837e-07,
      "loss": 0.8063,
      "step": 30956
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8024526834487915,
      "learning_rate": 8.401750630929944e-07,
      "loss": 0.5748,
      "step": 30957
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1614878177642822,
      "learning_rate": 8.395361466952049e-07,
      "loss": 0.7976,
      "step": 30958
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.080883741378784,
      "learning_rate": 8.388972302974156e-07,
      "loss": 0.7395,
      "step": 30959
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4391305446624756,
      "learning_rate": 8.382583138996262e-07,
      "loss": 0.4336,
      "step": 30960
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2709169387817383,
      "learning_rate": 8.376193975018369e-07,
      "loss": 0.9768,
      "step": 30961
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7638168334960938,
      "learning_rate": 8.369804811040476e-07,
      "loss": 0.8197,
      "step": 30962
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9422966241836548,
      "learning_rate": 8.363415647062582e-07,
      "loss": 0.9226,
      "step": 30963
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8261641263961792,
      "learning_rate": 8.357026483084689e-07,
      "loss": 0.8674,
      "step": 30964
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3622145652770996,
      "learning_rate": 8.350637319106795e-07,
      "loss": 0.8193,
      "step": 30965
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.101625680923462,
      "learning_rate": 8.344248155128902e-07,
      "loss": 0.7358,
      "step": 30966
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7509156465530396,
      "learning_rate": 8.337858991151008e-07,
      "loss": 1.0064,
      "step": 30967
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0773508548736572,
      "learning_rate": 8.331469827173115e-07,
      "loss": 0.9091,
      "step": 30968
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7026625871658325,
      "learning_rate": 8.325080663195221e-07,
      "loss": 0.7463,
      "step": 30969
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.337442874908447,
      "learning_rate": 8.318691499217328e-07,
      "loss": 0.8985,
      "step": 30970
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4372761249542236,
      "learning_rate": 8.312302335239435e-07,
      "loss": 0.8419,
      "step": 30971
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.624648332595825,
      "learning_rate": 8.305913171261541e-07,
      "loss": 0.8566,
      "step": 30972
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.100637674331665,
      "learning_rate": 8.299524007283648e-07,
      "loss": 0.6758,
      "step": 30973
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7612602710723877,
      "learning_rate": 8.293134843305754e-07,
      "loss": 0.8232,
      "step": 30974
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0684375762939453,
      "learning_rate": 8.286745679327861e-07,
      "loss": 1.1427,
      "step": 30975
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.112175941467285,
      "learning_rate": 8.280356515349967e-07,
      "loss": 0.6817,
      "step": 30976
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7884482145309448,
      "learning_rate": 8.273967351372074e-07,
      "loss": 0.8547,
      "step": 30977
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.826292514801025,
      "learning_rate": 8.26757818739418e-07,
      "loss": 1.6631,
      "step": 30978
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.589916467666626,
      "learning_rate": 8.261189023416287e-07,
      "loss": 0.7267,
      "step": 30979
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6813249588012695,
      "learning_rate": 8.254799859438394e-07,
      "loss": 0.8351,
      "step": 30980
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7425180673599243,
      "learning_rate": 8.2484106954605e-07,
      "loss": 0.9028,
      "step": 30981
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.146592140197754,
      "learning_rate": 8.242021531482607e-07,
      "loss": 0.3816,
      "step": 30982
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9674444198608398,
      "learning_rate": 8.235632367504713e-07,
      "loss": 0.8083,
      "step": 30983
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.85384202003479,
      "learning_rate": 8.22924320352682e-07,
      "loss": 1.5086,
      "step": 30984
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4505562782287598,
      "learning_rate": 8.222854039548926e-07,
      "loss": 0.8629,
      "step": 30985
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.761976718902588,
      "learning_rate": 8.216464875571033e-07,
      "loss": 0.8631,
      "step": 30986
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.105034828186035,
      "learning_rate": 8.21007571159314e-07,
      "loss": 0.9353,
      "step": 30987
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4518282413482666,
      "learning_rate": 8.203686547615246e-07,
      "loss": 1.0356,
      "step": 30988
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7036758661270142,
      "learning_rate": 8.197297383637353e-07,
      "loss": 0.8424,
      "step": 30989
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2060046195983887,
      "learning_rate": 8.190908219659459e-07,
      "loss": 0.8165,
      "step": 30990
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7016035318374634,
      "learning_rate": 8.184519055681566e-07,
      "loss": 0.8665,
      "step": 30991
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8728928565979004,
      "learning_rate": 8.17812989170367e-07,
      "loss": 0.7731,
      "step": 30992
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1993560791015625,
      "learning_rate": 8.171740727725778e-07,
      "loss": 0.5361,
      "step": 30993
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.173724889755249,
      "learning_rate": 8.165351563747883e-07,
      "loss": 0.957,
      "step": 30994
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8656980991363525,
      "learning_rate": 8.15896239976999e-07,
      "loss": 0.7844,
      "step": 30995
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.113929033279419,
      "learning_rate": 8.152573235792096e-07,
      "loss": 0.9121,
      "step": 30996
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7013397216796875,
      "learning_rate": 8.146184071814203e-07,
      "loss": 0.7672,
      "step": 30997
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.465055465698242,
      "learning_rate": 8.139794907836309e-07,
      "loss": 0.6942,
      "step": 30998
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8498793840408325,
      "learning_rate": 8.133405743858416e-07,
      "loss": 0.772,
      "step": 30999
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1867308616638184,
      "learning_rate": 8.127016579880523e-07,
      "loss": 0.7453,
      "step": 31000
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0557618141174316,
      "learning_rate": 8.120627415902629e-07,
      "loss": 0.7872,
      "step": 31001
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7954589128494263,
      "learning_rate": 8.114238251924736e-07,
      "loss": 0.7247,
      "step": 31002
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6697256565093994,
      "learning_rate": 8.107849087946842e-07,
      "loss": 0.8613,
      "step": 31003
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.728541612625122,
      "learning_rate": 8.101459923968949e-07,
      "loss": 0.8685,
      "step": 31004
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5042860507965088,
      "learning_rate": 8.095070759991055e-07,
      "loss": 0.3994,
      "step": 31005
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.3081841468811035,
      "learning_rate": 8.088681596013162e-07,
      "loss": 0.6947,
      "step": 31006
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6220017671585083,
      "learning_rate": 8.082292432035268e-07,
      "loss": 0.7119,
      "step": 31007
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5872572660446167,
      "learning_rate": 8.075903268057375e-07,
      "loss": 0.8714,
      "step": 31008
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1572229862213135,
      "learning_rate": 8.069514104079482e-07,
      "loss": 0.8494,
      "step": 31009
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7704325914382935,
      "learning_rate": 8.063124940101588e-07,
      "loss": 0.7496,
      "step": 31010
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.744567632675171,
      "learning_rate": 8.056735776123695e-07,
      "loss": 1.122,
      "step": 31011
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6366418600082397,
      "learning_rate": 8.050346612145801e-07,
      "loss": 0.7808,
      "step": 31012
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.141486167907715,
      "learning_rate": 8.043957448167908e-07,
      "loss": 0.8844,
      "step": 31013
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0797781944274902,
      "learning_rate": 8.037568284190014e-07,
      "loss": 0.8986,
      "step": 31014
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7077676057815552,
      "learning_rate": 8.031179120212121e-07,
      "loss": 0.8024,
      "step": 31015
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.838989019393921,
      "learning_rate": 8.024789956234228e-07,
      "loss": 0.8612,
      "step": 31016
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.756356954574585,
      "learning_rate": 8.018400792256334e-07,
      "loss": 0.922,
      "step": 31017
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6335797309875488,
      "learning_rate": 8.012011628278441e-07,
      "loss": 0.7604,
      "step": 31018
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8859360218048096,
      "learning_rate": 8.005622464300547e-07,
      "loss": 0.7797,
      "step": 31019
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9187434911727905,
      "learning_rate": 7.999233300322654e-07,
      "loss": 0.7761,
      "step": 31020
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.41007137298584,
      "learning_rate": 7.99284413634476e-07,
      "loss": 1.0334,
      "step": 31021
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8028689622879028,
      "learning_rate": 7.986454972366867e-07,
      "loss": 0.963,
      "step": 31022
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9230338335037231,
      "learning_rate": 7.980065808388973e-07,
      "loss": 0.9536,
      "step": 31023
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.8520753383636475,
      "learning_rate": 7.97367664441108e-07,
      "loss": 0.6105,
      "step": 31024
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9472912549972534,
      "learning_rate": 7.967287480433187e-07,
      "loss": 0.827,
      "step": 31025
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9796884059906006,
      "learning_rate": 7.960898316455293e-07,
      "loss": 1.0878,
      "step": 31026
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6311556100845337,
      "learning_rate": 7.9545091524774e-07,
      "loss": 0.7924,
      "step": 31027
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9229741096496582,
      "learning_rate": 7.948119988499506e-07,
      "loss": 0.8061,
      "step": 31028
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.483687400817871,
      "learning_rate": 7.941730824521613e-07,
      "loss": 0.7846,
      "step": 31029
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.681287169456482,
      "learning_rate": 7.935341660543718e-07,
      "loss": 0.8076,
      "step": 31030
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.124671697616577,
      "learning_rate": 7.928952496565825e-07,
      "loss": 0.735,
      "step": 31031
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6862696409225464,
      "learning_rate": 7.922563332587932e-07,
      "loss": 0.7529,
      "step": 31032
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1582753658294678,
      "learning_rate": 7.916174168610038e-07,
      "loss": 0.8112,
      "step": 31033
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.402484178543091,
      "learning_rate": 7.909785004632145e-07,
      "loss": 0.7924,
      "step": 31034
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0339982509613037,
      "learning_rate": 7.903395840654251e-07,
      "loss": 0.7552,
      "step": 31035
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.855044960975647,
      "learning_rate": 7.897006676676358e-07,
      "loss": 0.9031,
      "step": 31036
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1159632205963135,
      "learning_rate": 7.890617512698464e-07,
      "loss": 0.8235,
      "step": 31037
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8818999528884888,
      "learning_rate": 7.884228348720571e-07,
      "loss": 0.7792,
      "step": 31038
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6524567604064941,
      "learning_rate": 7.877839184742677e-07,
      "loss": 0.8842,
      "step": 31039
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.0174291133880615,
      "learning_rate": 7.871450020764784e-07,
      "loss": 0.9357,
      "step": 31040
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1678411960601807,
      "learning_rate": 7.865060856786889e-07,
      "loss": 0.8361,
      "step": 31041
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8176831007003784,
      "learning_rate": 7.858671692808996e-07,
      "loss": 0.8388,
      "step": 31042
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6496857404708862,
      "learning_rate": 7.852282528831102e-07,
      "loss": 0.783,
      "step": 31043
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.782226800918579,
      "learning_rate": 7.845893364853209e-07,
      "loss": 0.8716,
      "step": 31044
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7288676500320435,
      "learning_rate": 7.839504200875316e-07,
      "loss": 0.7303,
      "step": 31045
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4796321392059326,
      "learning_rate": 7.833115036897422e-07,
      "loss": 0.6043,
      "step": 31046
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.429011344909668,
      "learning_rate": 7.826725872919529e-07,
      "loss": 0.9074,
      "step": 31047
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4806232452392578,
      "learning_rate": 7.820336708941635e-07,
      "loss": 0.399,
      "step": 31048
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.1905455589294434,
      "learning_rate": 7.813947544963742e-07,
      "loss": 0.8773,
      "step": 31049
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.162127494812012,
      "learning_rate": 7.807558380985848e-07,
      "loss": 1.397,
      "step": 31050
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.179643392562866,
      "learning_rate": 7.801169217007955e-07,
      "loss": 0.686,
      "step": 31051
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5858126878738403,
      "learning_rate": 7.794780053030061e-07,
      "loss": 0.8783,
      "step": 31052
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1104772090911865,
      "learning_rate": 7.788390889052168e-07,
      "loss": 0.8093,
      "step": 31053
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9659520387649536,
      "learning_rate": 7.782001725074275e-07,
      "loss": 0.6786,
      "step": 31054
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6904113292694092,
      "learning_rate": 7.775612561096381e-07,
      "loss": 0.82,
      "step": 31055
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.693640112876892,
      "learning_rate": 7.769223397118488e-07,
      "loss": 0.8606,
      "step": 31056
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.608628273010254,
      "learning_rate": 7.762834233140594e-07,
      "loss": 0.8269,
      "step": 31057
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.3811733722686768,
      "learning_rate": 7.756445069162701e-07,
      "loss": 0.9912,
      "step": 31058
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5199493169784546,
      "learning_rate": 7.750055905184807e-07,
      "loss": 0.7688,
      "step": 31059
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.226041316986084,
      "learning_rate": 7.743666741206914e-07,
      "loss": 0.9709,
      "step": 31060
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.136826038360596,
      "learning_rate": 7.73727757722902e-07,
      "loss": 1.605,
      "step": 31061
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.979206919670105,
      "learning_rate": 7.730888413251127e-07,
      "loss": 0.792,
      "step": 31062
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7048593759536743,
      "learning_rate": 7.724499249273234e-07,
      "loss": 0.8825,
      "step": 31063
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.237697124481201,
      "learning_rate": 7.71811008529534e-07,
      "loss": 0.8742,
      "step": 31064
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7186036109924316,
      "learning_rate": 7.711720921317447e-07,
      "loss": 1.1313,
      "step": 31065
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.384920835494995,
      "learning_rate": 7.705331757339552e-07,
      "loss": 0.781,
      "step": 31066
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3241658210754395,
      "learning_rate": 7.69894259336166e-07,
      "loss": 0.8441,
      "step": 31067
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.14731502532959,
      "learning_rate": 7.692553429383765e-07,
      "loss": 0.9142,
      "step": 31068
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.062751531600952,
      "learning_rate": 7.686164265405872e-07,
      "loss": 0.87,
      "step": 31069
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.712218165397644,
      "learning_rate": 7.679775101427979e-07,
      "loss": 0.9069,
      "step": 31070
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7867462635040283,
      "learning_rate": 7.673385937450085e-07,
      "loss": 0.8651,
      "step": 31071
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6610291004180908,
      "learning_rate": 7.666996773472192e-07,
      "loss": 0.9206,
      "step": 31072
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6764836311340332,
      "learning_rate": 7.660607609494298e-07,
      "loss": 0.7526,
      "step": 31073
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5345909595489502,
      "learning_rate": 7.654218445516405e-07,
      "loss": 0.7473,
      "step": 31074
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.002444267272949,
      "learning_rate": 7.647829281538511e-07,
      "loss": 0.9381,
      "step": 31075
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.534230947494507,
      "learning_rate": 7.641440117560618e-07,
      "loss": 0.9334,
      "step": 31076
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.761965274810791,
      "learning_rate": 7.635050953582724e-07,
      "loss": 0.8638,
      "step": 31077
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.7204105854034424,
      "learning_rate": 7.628661789604831e-07,
      "loss": 0.9315,
      "step": 31078
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.767278790473938,
      "learning_rate": 7.622272625626938e-07,
      "loss": 0.7348,
      "step": 31079
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.860666275024414,
      "learning_rate": 7.615883461649044e-07,
      "loss": 1.2746,
      "step": 31080
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.699539065361023,
      "learning_rate": 7.609494297671151e-07,
      "loss": 0.79,
      "step": 31081
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.518092632293701,
      "learning_rate": 7.603105133693257e-07,
      "loss": 0.9135,
      "step": 31082
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.589875221252441,
      "learning_rate": 7.596715969715364e-07,
      "loss": 1.165,
      "step": 31083
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.65756893157959,
      "learning_rate": 7.59032680573747e-07,
      "loss": 0.9729,
      "step": 31084
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0501139163970947,
      "learning_rate": 7.583937641759577e-07,
      "loss": 1.2047,
      "step": 31085
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8187834024429321,
      "learning_rate": 7.577548477781684e-07,
      "loss": 0.8013,
      "step": 31086
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9881641864776611,
      "learning_rate": 7.57115931380379e-07,
      "loss": 0.846,
      "step": 31087
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5271661281585693,
      "learning_rate": 7.564770149825897e-07,
      "loss": 0.7797,
      "step": 31088
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7112802267074585,
      "learning_rate": 7.558380985848002e-07,
      "loss": 0.9023,
      "step": 31089
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6973639726638794,
      "learning_rate": 7.551991821870108e-07,
      "loss": 0.8351,
      "step": 31090
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8845500946044922,
      "learning_rate": 7.545602657892215e-07,
      "loss": 0.7935,
      "step": 31091
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.727782964706421,
      "learning_rate": 7.539213493914322e-07,
      "loss": 1.0979,
      "step": 31092
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5953409671783447,
      "learning_rate": 7.532824329936428e-07,
      "loss": 0.8716,
      "step": 31093
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4573169946670532,
      "learning_rate": 7.526435165958535e-07,
      "loss": 0.6551,
      "step": 31094
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6328063011169434,
      "learning_rate": 7.520046001980641e-07,
      "loss": 0.743,
      "step": 31095
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9263607263565063,
      "learning_rate": 7.513656838002748e-07,
      "loss": 0.7593,
      "step": 31096
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6275739669799805,
      "learning_rate": 7.507267674024854e-07,
      "loss": 0.8356,
      "step": 31097
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8557308912277222,
      "learning_rate": 7.500878510046961e-07,
      "loss": 0.9052,
      "step": 31098
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4238067865371704,
      "learning_rate": 7.494489346069068e-07,
      "loss": 0.749,
      "step": 31099
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.604596734046936,
      "learning_rate": 7.488100182091174e-07,
      "loss": 0.7598,
      "step": 31100
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7958790063858032,
      "learning_rate": 7.48171101811328e-07,
      "loss": 0.8213,
      "step": 31101
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.625307321548462,
      "learning_rate": 7.475321854135386e-07,
      "loss": 0.775,
      "step": 31102
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6710020303726196,
      "learning_rate": 7.468932690157493e-07,
      "loss": 0.7101,
      "step": 31103
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.7920312881469727,
      "learning_rate": 7.462543526179599e-07,
      "loss": 0.5107,
      "step": 31104
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6933499574661255,
      "learning_rate": 7.456154362201706e-07,
      "loss": 0.8556,
      "step": 31105
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.6633002758026123,
      "learning_rate": 7.449765198223812e-07,
      "loss": 1.0545,
      "step": 31106
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.8560736179351807,
      "learning_rate": 7.443376034245919e-07,
      "loss": 0.7833,
      "step": 31107
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.875885009765625,
      "learning_rate": 7.436986870268026e-07,
      "loss": 0.4189,
      "step": 31108
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.899546504020691,
      "learning_rate": 7.430597706290132e-07,
      "loss": 0.9118,
      "step": 31109
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.932376503944397,
      "learning_rate": 7.424208542312239e-07,
      "loss": 0.9482,
      "step": 31110
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0030100345611572,
      "learning_rate": 7.417819378334345e-07,
      "loss": 0.8703,
      "step": 31111
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9504987001419067,
      "learning_rate": 7.411430214356452e-07,
      "loss": 0.88,
      "step": 31112
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.243267059326172,
      "learning_rate": 7.405041050378558e-07,
      "loss": 0.731,
      "step": 31113
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.624068021774292,
      "learning_rate": 7.398651886400665e-07,
      "loss": 0.7164,
      "step": 31114
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.149524688720703,
      "learning_rate": 7.392262722422772e-07,
      "loss": 0.9681,
      "step": 31115
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2462315559387207,
      "learning_rate": 7.385873558444878e-07,
      "loss": 0.9933,
      "step": 31116
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.113241195678711,
      "learning_rate": 7.379484394466985e-07,
      "loss": 1.2104,
      "step": 31117
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6650128364562988,
      "learning_rate": 7.373095230489091e-07,
      "loss": 0.7935,
      "step": 31118
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.6845638751983643,
      "learning_rate": 7.366706066511198e-07,
      "loss": 0.8805,
      "step": 31119
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0318076610565186,
      "learning_rate": 7.360316902533304e-07,
      "loss": 0.8848,
      "step": 31120
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.298013687133789,
      "learning_rate": 7.353927738555411e-07,
      "loss": 0.7981,
      "step": 31121
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8149299621582031,
      "learning_rate": 7.347538574577517e-07,
      "loss": 0.9764,
      "step": 31122
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6244072914123535,
      "learning_rate": 7.341149410599624e-07,
      "loss": 0.824,
      "step": 31123
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.406671166419983,
      "learning_rate": 7.334760246621731e-07,
      "loss": 0.2673,
      "step": 31124
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7704753875732422,
      "learning_rate": 7.328371082643837e-07,
      "loss": 0.8448,
      "step": 31125
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8362209796905518,
      "learning_rate": 7.321981918665944e-07,
      "loss": 0.8175,
      "step": 31126
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.863032579421997,
      "learning_rate": 7.31559275468805e-07,
      "loss": 0.7116,
      "step": 31127
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.754965901374817,
      "learning_rate": 7.309203590710157e-07,
      "loss": 0.7881,
      "step": 31128
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1962473392486572,
      "learning_rate": 7.302814426732263e-07,
      "loss": 0.9227,
      "step": 31129
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.6320929527282715,
      "learning_rate": 7.29642526275437e-07,
      "loss": 1.0427,
      "step": 31130
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.578282117843628,
      "learning_rate": 7.290036098776477e-07,
      "loss": 0.8819,
      "step": 31131
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6241966485977173,
      "learning_rate": 7.283646934798583e-07,
      "loss": 0.8145,
      "step": 31132
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8674396276474,
      "learning_rate": 7.27725777082069e-07,
      "loss": 0.975,
      "step": 31133
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3990797996520996,
      "learning_rate": 7.270868606842796e-07,
      "loss": 0.7764,
      "step": 31134
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4393515586853027,
      "learning_rate": 7.264479442864903e-07,
      "loss": 0.6485,
      "step": 31135
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1867363452911377,
      "learning_rate": 7.258090278887009e-07,
      "loss": 0.8373,
      "step": 31136
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7796039581298828,
      "learning_rate": 7.251701114909115e-07,
      "loss": 0.3497,
      "step": 31137
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7907371520996094,
      "learning_rate": 7.24531195093122e-07,
      "loss": 0.7551,
      "step": 31138
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.055591344833374,
      "learning_rate": 7.238922786953327e-07,
      "loss": 0.9236,
      "step": 31139
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2602357864379883,
      "learning_rate": 7.232533622975433e-07,
      "loss": 0.608,
      "step": 31140
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1378862857818604,
      "learning_rate": 7.22614445899754e-07,
      "loss": 0.8662,
      "step": 31141
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6870940923690796,
      "learning_rate": 7.219755295019646e-07,
      "loss": 0.3574,
      "step": 31142
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.519783020019531,
      "learning_rate": 7.213366131041753e-07,
      "loss": 1.5738,
      "step": 31143
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5843701362609863,
      "learning_rate": 7.206976967063859e-07,
      "loss": 0.7937,
      "step": 31144
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.692506790161133,
      "learning_rate": 7.200587803085966e-07,
      "loss": 0.942,
      "step": 31145
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.409752368927002,
      "learning_rate": 7.194198639108073e-07,
      "loss": 1.0599,
      "step": 31146
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0860683917999268,
      "learning_rate": 7.187809475130179e-07,
      "loss": 0.8763,
      "step": 31147
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8568133115768433,
      "learning_rate": 7.181420311152286e-07,
      "loss": 0.8851,
      "step": 31148
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4332964420318604,
      "learning_rate": 7.175031147174392e-07,
      "loss": 0.4814,
      "step": 31149
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8544964790344238,
      "learning_rate": 7.168641983196499e-07,
      "loss": 0.8765,
      "step": 31150
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.3063838481903076,
      "learning_rate": 7.162252819218605e-07,
      "loss": 0.7647,
      "step": 31151
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2097041606903076,
      "learning_rate": 7.155863655240712e-07,
      "loss": 0.8109,
      "step": 31152
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9598345756530762,
      "learning_rate": 7.149474491262819e-07,
      "loss": 0.8364,
      "step": 31153
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0473270416259766,
      "learning_rate": 7.143085327284925e-07,
      "loss": 1.1168,
      "step": 31154
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.1393325328826904,
      "learning_rate": 7.136696163307032e-07,
      "loss": 0.8425,
      "step": 31155
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.5368497371673584,
      "learning_rate": 7.130306999329138e-07,
      "loss": 0.7398,
      "step": 31156
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.842437744140625,
      "learning_rate": 7.123917835351245e-07,
      "loss": 0.8891,
      "step": 31157
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6814327239990234,
      "learning_rate": 7.117528671373351e-07,
      "loss": 0.8266,
      "step": 31158
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.685049533843994,
      "learning_rate": 7.111139507395458e-07,
      "loss": 1.2421,
      "step": 31159
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.933852195739746,
      "learning_rate": 7.104750343417564e-07,
      "loss": 0.7951,
      "step": 31160
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.887060284614563,
      "learning_rate": 7.098361179439671e-07,
      "loss": 0.879,
      "step": 31161
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.218867778778076,
      "learning_rate": 7.091972015461778e-07,
      "loss": 0.7533,
      "step": 31162
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.96415376663208,
      "learning_rate": 7.085582851483884e-07,
      "loss": 0.8964,
      "step": 31163
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9654358625411987,
      "learning_rate": 7.079193687505991e-07,
      "loss": 0.8669,
      "step": 31164
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9833968877792358,
      "learning_rate": 7.072804523528097e-07,
      "loss": 0.8284,
      "step": 31165
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.53651762008667,
      "learning_rate": 7.066415359550204e-07,
      "loss": 0.8248,
      "step": 31166
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.8618955612182617,
      "learning_rate": 7.06002619557231e-07,
      "loss": 0.7302,
      "step": 31167
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.189944267272949,
      "learning_rate": 7.053637031594417e-07,
      "loss": 0.5468,
      "step": 31168
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.458028793334961,
      "learning_rate": 7.047247867616524e-07,
      "loss": 0.7554,
      "step": 31169
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8064453601837158,
      "learning_rate": 7.04085870363863e-07,
      "loss": 0.7641,
      "step": 31170
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.504452705383301,
      "learning_rate": 7.034469539660737e-07,
      "loss": 1.006,
      "step": 31171
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6797370910644531,
      "learning_rate": 7.028080375682843e-07,
      "loss": 0.69,
      "step": 31172
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.83793306350708,
      "learning_rate": 7.02169121170495e-07,
      "loss": 0.8758,
      "step": 31173
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8708120584487915,
      "learning_rate": 7.015302047727055e-07,
      "loss": 0.759,
      "step": 31174
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1361600160598755,
      "learning_rate": 7.008912883749162e-07,
      "loss": 0.3946,
      "step": 31175
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1879327297210693,
      "learning_rate": 7.002523719771268e-07,
      "loss": 0.8088,
      "step": 31176
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4053657054901123,
      "learning_rate": 6.996134555793375e-07,
      "loss": 0.964,
      "step": 31177
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8392971754074097,
      "learning_rate": 6.989745391815482e-07,
      "loss": 0.8702,
      "step": 31178
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.957327961921692,
      "learning_rate": 6.983356227837588e-07,
      "loss": 0.9154,
      "step": 31179
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1313273906707764,
      "learning_rate": 6.976967063859695e-07,
      "loss": 0.8922,
      "step": 31180
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4557342529296875,
      "learning_rate": 6.970577899881801e-07,
      "loss": 0.9151,
      "step": 31181
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9197958707809448,
      "learning_rate": 6.964188735903908e-07,
      "loss": 1.0213,
      "step": 31182
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5294806957244873,
      "learning_rate": 6.957799571926014e-07,
      "loss": 0.7356,
      "step": 31183
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6122002601623535,
      "learning_rate": 6.951410407948121e-07,
      "loss": 0.7265,
      "step": 31184
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6239601373672485,
      "learning_rate": 6.945021243970228e-07,
      "loss": 0.7429,
      "step": 31185
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.665693998336792,
      "learning_rate": 6.938632079992333e-07,
      "loss": 0.9188,
      "step": 31186
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.617988109588623,
      "learning_rate": 6.932242916014439e-07,
      "loss": 0.7944,
      "step": 31187
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.7695529460906982,
      "learning_rate": 6.925853752036546e-07,
      "loss": 1.0165,
      "step": 31188
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5934350490570068,
      "learning_rate": 6.919464588058652e-07,
      "loss": 0.833,
      "step": 31189
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8539667129516602,
      "learning_rate": 6.913075424080759e-07,
      "loss": 0.8292,
      "step": 31190
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7770843505859375,
      "learning_rate": 6.906686260102866e-07,
      "loss": 0.7639,
      "step": 31191
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9854108095169067,
      "learning_rate": 6.900297096124972e-07,
      "loss": 0.8567,
      "step": 31192
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.0025646686553955,
      "learning_rate": 6.893907932147079e-07,
      "loss": 0.8792,
      "step": 31193
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8942543268203735,
      "learning_rate": 6.887518768169185e-07,
      "loss": 0.8795,
      "step": 31194
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2347567081451416,
      "learning_rate": 6.881129604191292e-07,
      "loss": 0.9147,
      "step": 31195
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.6425485610961914,
      "learning_rate": 6.874740440213398e-07,
      "loss": 0.8844,
      "step": 31196
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.7490274906158447,
      "learning_rate": 6.868351276235505e-07,
      "loss": 0.854,
      "step": 31197
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.189347982406616,
      "learning_rate": 6.861962112257612e-07,
      "loss": 0.8088,
      "step": 31198
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9995801448822021,
      "learning_rate": 6.855572948279718e-07,
      "loss": 0.7578,
      "step": 31199
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.252333164215088,
      "learning_rate": 6.849183784301825e-07,
      "loss": 1.2775,
      "step": 31200
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7374820709228516,
      "learning_rate": 6.842794620323931e-07,
      "loss": 0.7829,
      "step": 31201
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.384978771209717,
      "learning_rate": 6.836405456346038e-07,
      "loss": 0.7461,
      "step": 31202
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.3180835247039795,
      "learning_rate": 6.830016292368144e-07,
      "loss": 0.6439,
      "step": 31203
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1916017532348633,
      "learning_rate": 6.823627128390251e-07,
      "loss": 0.6848,
      "step": 31204
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.014737129211426,
      "learning_rate": 6.817237964412357e-07,
      "loss": 1.1159,
      "step": 31205
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9443204402923584,
      "learning_rate": 6.810848800434464e-07,
      "loss": 0.8042,
      "step": 31206
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.029117107391357,
      "learning_rate": 6.804459636456571e-07,
      "loss": 1.2302,
      "step": 31207
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.188333511352539,
      "learning_rate": 6.798070472478677e-07,
      "loss": 0.8767,
      "step": 31208
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6626989841461182,
      "learning_rate": 6.791681308500784e-07,
      "loss": 0.6726,
      "step": 31209
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.468214988708496,
      "learning_rate": 6.78529214452289e-07,
      "loss": 0.3855,
      "step": 31210
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7830500602722168,
      "learning_rate": 6.778902980544996e-07,
      "loss": 0.8239,
      "step": 31211
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.049521446228027,
      "learning_rate": 6.772513816567102e-07,
      "loss": 1.2109,
      "step": 31212
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9647103548049927,
      "learning_rate": 6.766124652589209e-07,
      "loss": 0.7099,
      "step": 31213
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7221623659133911,
      "learning_rate": 6.759735488611316e-07,
      "loss": 0.8402,
      "step": 31214
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.9300146102905273,
      "learning_rate": 6.753346324633422e-07,
      "loss": 0.7682,
      "step": 31215
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1155519485473633,
      "learning_rate": 6.746957160655529e-07,
      "loss": 0.733,
      "step": 31216
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.2148752212524414,
      "learning_rate": 6.740567996677635e-07,
      "loss": 0.7346,
      "step": 31217
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7950266599655151,
      "learning_rate": 6.734178832699742e-07,
      "loss": 0.7553,
      "step": 31218
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8423534631729126,
      "learning_rate": 6.727789668721848e-07,
      "loss": 0.8932,
      "step": 31219
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.3112337589263916,
      "learning_rate": 6.721400504743955e-07,
      "loss": 0.8517,
      "step": 31220
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.5266618728637695,
      "learning_rate": 6.715011340766061e-07,
      "loss": 1.0411,
      "step": 31221
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8333146572113037,
      "learning_rate": 6.708622176788168e-07,
      "loss": 0.8334,
      "step": 31222
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.362516403198242,
      "learning_rate": 6.702233012810275e-07,
      "loss": 1.001,
      "step": 31223
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.089254856109619,
      "learning_rate": 6.695843848832381e-07,
      "loss": 1.3973,
      "step": 31224
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5414857864379883,
      "learning_rate": 6.689454684854488e-07,
      "loss": 0.8107,
      "step": 31225
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0669801235198975,
      "learning_rate": 6.683065520876594e-07,
      "loss": 0.7963,
      "step": 31226
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7325681447982788,
      "learning_rate": 6.676676356898701e-07,
      "loss": 0.7691,
      "step": 31227
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.525272846221924,
      "learning_rate": 6.670287192920807e-07,
      "loss": 0.9695,
      "step": 31228
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9961475133895874,
      "learning_rate": 6.663898028942914e-07,
      "loss": 0.8349,
      "step": 31229
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9044479131698608,
      "learning_rate": 6.65750886496502e-07,
      "loss": 0.7688,
      "step": 31230
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.7195606231689453,
      "learning_rate": 6.651119700987127e-07,
      "loss": 0.9555,
      "step": 31231
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.96207594871521,
      "learning_rate": 6.644730537009234e-07,
      "loss": 0.6183,
      "step": 31232
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8271392583847046,
      "learning_rate": 6.63834137303134e-07,
      "loss": 0.826,
      "step": 31233
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.032024621963501,
      "learning_rate": 6.631952209053445e-07,
      "loss": 0.8009,
      "step": 31234
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2640057802200317,
      "learning_rate": 6.625563045075552e-07,
      "loss": 0.4244,
      "step": 31235
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3189585208892822,
      "learning_rate": 6.619173881097659e-07,
      "loss": 0.8165,
      "step": 31236
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9538609981536865,
      "learning_rate": 6.612784717119765e-07,
      "loss": 0.888,
      "step": 31237
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.038180828094482,
      "learning_rate": 6.606395553141872e-07,
      "loss": 0.8199,
      "step": 31238
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.736171007156372,
      "learning_rate": 6.600006389163978e-07,
      "loss": 0.8297,
      "step": 31239
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.9785680770874023,
      "learning_rate": 6.593617225186085e-07,
      "loss": 0.9699,
      "step": 31240
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.391293525695801,
      "learning_rate": 6.587228061208191e-07,
      "loss": 1.0543,
      "step": 31241
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.063867568969727,
      "learning_rate": 6.580838897230298e-07,
      "loss": 1.1304,
      "step": 31242
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7959610223770142,
      "learning_rate": 6.574449733252404e-07,
      "loss": 0.8639,
      "step": 31243
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9775108098983765,
      "learning_rate": 6.56806056927451e-07,
      "loss": 0.7268,
      "step": 31244
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1940114498138428,
      "learning_rate": 6.561671405296618e-07,
      "loss": 0.8885,
      "step": 31245
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6838515996932983,
      "learning_rate": 6.555282241318723e-07,
      "loss": 0.7462,
      "step": 31246
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8939944505691528,
      "learning_rate": 6.54889307734083e-07,
      "loss": 0.7098,
      "step": 31247
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1406021118164062,
      "learning_rate": 6.542503913362936e-07,
      "loss": 0.9347,
      "step": 31248
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1473612785339355,
      "learning_rate": 6.536114749385043e-07,
      "loss": 0.7912,
      "step": 31249
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.810765027999878,
      "learning_rate": 6.529725585407149e-07,
      "loss": 0.7586,
      "step": 31250
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8185434341430664,
      "learning_rate": 6.523336421429256e-07,
      "loss": 0.8168,
      "step": 31251
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.849402904510498,
      "learning_rate": 6.516947257451363e-07,
      "loss": 0.8049,
      "step": 31252
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4342076778411865,
      "learning_rate": 6.510558093473469e-07,
      "loss": 1.1894,
      "step": 31253
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.083631753921509,
      "learning_rate": 6.504168929495576e-07,
      "loss": 1.0897,
      "step": 31254
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.589321255683899,
      "learning_rate": 6.497779765517682e-07,
      "loss": 0.7171,
      "step": 31255
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.114222764968872,
      "learning_rate": 6.491390601539789e-07,
      "loss": 0.786,
      "step": 31256
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.8634274005889893,
      "learning_rate": 6.485001437561895e-07,
      "loss": 0.7895,
      "step": 31257
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8509575128555298,
      "learning_rate": 6.478612273584002e-07,
      "loss": 0.7521,
      "step": 31258
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0542356967926025,
      "learning_rate": 6.472223109606108e-07,
      "loss": 0.866,
      "step": 31259
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.347504734992981,
      "learning_rate": 6.465833945628215e-07,
      "loss": 0.3671,
      "step": 31260
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8740726709365845,
      "learning_rate": 6.459444781650322e-07,
      "loss": 0.8105,
      "step": 31261
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.925765037536621,
      "learning_rate": 6.453055617672428e-07,
      "loss": 0.8877,
      "step": 31262
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9366129636764526,
      "learning_rate": 6.446666453694535e-07,
      "loss": 0.8763,
      "step": 31263
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9047315120697021,
      "learning_rate": 6.440277289716641e-07,
      "loss": 0.956,
      "step": 31264
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7505156993865967,
      "learning_rate": 6.433888125738748e-07,
      "loss": 0.6995,
      "step": 31265
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.953967809677124,
      "learning_rate": 6.427498961760854e-07,
      "loss": 1.0433,
      "step": 31266
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1814124584198,
      "learning_rate": 6.421109797782961e-07,
      "loss": 0.6438,
      "step": 31267
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.6447131633758545,
      "learning_rate": 6.414720633805068e-07,
      "loss": 1.179,
      "step": 31268
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.9597949981689453,
      "learning_rate": 6.408331469827174e-07,
      "loss": 0.8209,
      "step": 31269
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8339180946350098,
      "learning_rate": 6.401942305849281e-07,
      "loss": 0.8008,
      "step": 31270
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6514476537704468,
      "learning_rate": 6.395553141871387e-07,
      "loss": 0.8015,
      "step": 31271
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7298673391342163,
      "learning_rate": 6.389163977893494e-07,
      "loss": 0.8165,
      "step": 31272
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.688593864440918,
      "learning_rate": 6.3827748139156e-07,
      "loss": 0.8333,
      "step": 31273
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.760565996170044,
      "learning_rate": 6.376385649937707e-07,
      "loss": 0.8673,
      "step": 31274
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1256372928619385,
      "learning_rate": 6.369996485959813e-07,
      "loss": 0.8932,
      "step": 31275
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.490602731704712,
      "learning_rate": 6.36360732198192e-07,
      "loss": 0.8323,
      "step": 31276
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.108468770980835,
      "learning_rate": 6.357218158004027e-07,
      "loss": 1.1568,
      "step": 31277
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7289847135543823,
      "learning_rate": 6.350828994026133e-07,
      "loss": 0.4131,
      "step": 31278
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6179561614990234,
      "learning_rate": 6.34443983004824e-07,
      "loss": 0.9519,
      "step": 31279
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.470028400421143,
      "learning_rate": 6.338050666070346e-07,
      "loss": 0.9195,
      "step": 31280
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.0823025703430176,
      "learning_rate": 6.331661502092453e-07,
      "loss": 0.9166,
      "step": 31281
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8564457893371582,
      "learning_rate": 6.325272338114557e-07,
      "loss": 0.8669,
      "step": 31282
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.07287859916687,
      "learning_rate": 6.318883174136664e-07,
      "loss": 1.1188,
      "step": 31283
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.135218381881714,
      "learning_rate": 6.31249401015877e-07,
      "loss": 0.7141,
      "step": 31284
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.188390016555786,
      "learning_rate": 6.306104846180877e-07,
      "loss": 0.9193,
      "step": 31285
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8954238891601562,
      "learning_rate": 6.299715682202983e-07,
      "loss": 1.0464,
      "step": 31286
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.974435806274414,
      "learning_rate": 6.29332651822509e-07,
      "loss": 1.2005,
      "step": 31287
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.1797332763671875,
      "learning_rate": 6.286937354247196e-07,
      "loss": 1.1719,
      "step": 31288
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.5638325214385986,
      "learning_rate": 6.280548190269303e-07,
      "loss": 0.9081,
      "step": 31289
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2942020893096924,
      "learning_rate": 6.27415902629141e-07,
      "loss": 0.7157,
      "step": 31290
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.497039556503296,
      "learning_rate": 6.267769862313516e-07,
      "loss": 0.4355,
      "step": 31291
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1448581218719482,
      "learning_rate": 6.261380698335623e-07,
      "loss": 0.8142,
      "step": 31292
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.639078140258789,
      "learning_rate": 6.254991534357729e-07,
      "loss": 0.4753,
      "step": 31293
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.876697063446045,
      "learning_rate": 6.248602370379836e-07,
      "loss": 1.6184,
      "step": 31294
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.772863745689392,
      "learning_rate": 6.242213206401942e-07,
      "loss": 0.8476,
      "step": 31295
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.470952272415161,
      "learning_rate": 6.235824042424049e-07,
      "loss": 0.7396,
      "step": 31296
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0694501399993896,
      "learning_rate": 6.229434878446156e-07,
      "loss": 0.7707,
      "step": 31297
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7733314037322998,
      "learning_rate": 6.223045714468262e-07,
      "loss": 0.7793,
      "step": 31298
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4656524658203125,
      "learning_rate": 6.216656550490369e-07,
      "loss": 0.3901,
      "step": 31299
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6642193794250488,
      "learning_rate": 6.210267386512475e-07,
      "loss": 0.8705,
      "step": 31300
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.170719861984253,
      "learning_rate": 6.203878222534582e-07,
      "loss": 0.7623,
      "step": 31301
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.773095607757568,
      "learning_rate": 6.197489058556688e-07,
      "loss": 1.3842,
      "step": 31302
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.800869941711426,
      "learning_rate": 6.191099894578795e-07,
      "loss": 1.278,
      "step": 31303
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.385946035385132,
      "learning_rate": 6.184710730600901e-07,
      "loss": 0.9813,
      "step": 31304
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6112735271453857,
      "learning_rate": 6.178321566623008e-07,
      "loss": 0.8372,
      "step": 31305
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4814021587371826,
      "learning_rate": 6.171932402645115e-07,
      "loss": 0.7731,
      "step": 31306
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3140981197357178,
      "learning_rate": 6.165543238667221e-07,
      "loss": 0.8039,
      "step": 31307
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6571264266967773,
      "learning_rate": 6.159154074689328e-07,
      "loss": 0.7747,
      "step": 31308
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0533413887023926,
      "learning_rate": 6.152764910711434e-07,
      "loss": 0.762,
      "step": 31309
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1121134757995605,
      "learning_rate": 6.146375746733541e-07,
      "loss": 0.898,
      "step": 31310
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8504414558410645,
      "learning_rate": 6.139986582755647e-07,
      "loss": 0.9349,
      "step": 31311
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2913455963134766,
      "learning_rate": 6.133597418777754e-07,
      "loss": 0.8941,
      "step": 31312
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.502913236618042,
      "learning_rate": 6.12720825479986e-07,
      "loss": 0.8248,
      "step": 31313
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5833382606506348,
      "learning_rate": 6.120819090821967e-07,
      "loss": 0.784,
      "step": 31314
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4273598194122314,
      "learning_rate": 6.114429926844074e-07,
      "loss": 0.9962,
      "step": 31315
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7756015062332153,
      "learning_rate": 6.10804076286618e-07,
      "loss": 0.7707,
      "step": 31316
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.1207423210144043,
      "learning_rate": 6.101651598888287e-07,
      "loss": 0.8669,
      "step": 31317
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.345571517944336,
      "learning_rate": 6.095262434910392e-07,
      "loss": 0.3426,
      "step": 31318
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5574721097946167,
      "learning_rate": 6.088873270932498e-07,
      "loss": 0.7584,
      "step": 31319
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.994398832321167,
      "learning_rate": 6.082484106954605e-07,
      "loss": 0.8528,
      "step": 31320
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.229440927505493,
      "learning_rate": 6.076094942976711e-07,
      "loss": 0.6353,
      "step": 31321
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.028001070022583,
      "learning_rate": 6.069705778998818e-07,
      "loss": 0.866,
      "step": 31322
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5161288976669312,
      "learning_rate": 6.063316615020924e-07,
      "loss": 0.7258,
      "step": 31323
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9308425188064575,
      "learning_rate": 6.056927451043031e-07,
      "loss": 0.7758,
      "step": 31324
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.6186954975128174,
      "learning_rate": 6.050538287065138e-07,
      "loss": 1.0666,
      "step": 31325
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.26710844039917,
      "learning_rate": 6.044149123087244e-07,
      "loss": 0.8388,
      "step": 31326
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5884085893630981,
      "learning_rate": 6.037759959109351e-07,
      "loss": 0.4223,
      "step": 31327
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9039446115493774,
      "learning_rate": 6.031370795131457e-07,
      "loss": 0.89,
      "step": 31328
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.197481155395508,
      "learning_rate": 6.024981631153564e-07,
      "loss": 0.8718,
      "step": 31329
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.671142578125,
      "learning_rate": 6.01859246717567e-07,
      "loss": 1.1579,
      "step": 31330
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1257057189941406,
      "learning_rate": 6.012203303197777e-07,
      "loss": 0.8314,
      "step": 31331
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4251315593719482,
      "learning_rate": 6.005814139219884e-07,
      "loss": 0.3694,
      "step": 31332
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.200765371322632,
      "learning_rate": 5.99942497524199e-07,
      "loss": 0.937,
      "step": 31333
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7270234823226929,
      "learning_rate": 5.993035811264097e-07,
      "loss": 0.8573,
      "step": 31334
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.008125066757202,
      "learning_rate": 5.986646647286203e-07,
      "loss": 1.0078,
      "step": 31335
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8346288204193115,
      "learning_rate": 5.98025748330831e-07,
      "loss": 0.9173,
      "step": 31336
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.118819236755371,
      "learning_rate": 5.973868319330416e-07,
      "loss": 0.8874,
      "step": 31337
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7190479040145874,
      "learning_rate": 5.967479155352523e-07,
      "loss": 0.7834,
      "step": 31338
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4272520542144775,
      "learning_rate": 5.961089991374629e-07,
      "loss": 0.7172,
      "step": 31339
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8831456899642944,
      "learning_rate": 5.954700827396736e-07,
      "loss": 0.8133,
      "step": 31340
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.025620937347412,
      "learning_rate": 5.948311663418843e-07,
      "loss": 0.9093,
      "step": 31341
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4975532293319702,
      "learning_rate": 5.941922499440949e-07,
      "loss": 0.8094,
      "step": 31342
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7169008255004883,
      "learning_rate": 5.935533335463055e-07,
      "loss": 0.815,
      "step": 31343
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5790902376174927,
      "learning_rate": 5.929144171485162e-07,
      "loss": 0.7721,
      "step": 31344
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.8943514823913574,
      "learning_rate": 5.922755007507268e-07,
      "loss": 0.9199,
      "step": 31345
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.898484706878662,
      "learning_rate": 5.916365843529375e-07,
      "loss": 1.334,
      "step": 31346
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.9992637634277344,
      "learning_rate": 5.909976679551481e-07,
      "loss": 1.1635,
      "step": 31347
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.62120521068573,
      "learning_rate": 5.903587515573588e-07,
      "loss": 0.7845,
      "step": 31348
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.571323871612549,
      "learning_rate": 5.897198351595694e-07,
      "loss": 0.5484,
      "step": 31349
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3261301517486572,
      "learning_rate": 5.890809187617801e-07,
      "loss": 0.9759,
      "step": 31350
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7718745470046997,
      "learning_rate": 5.884420023639908e-07,
      "loss": 0.8096,
      "step": 31351
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.05441951751709,
      "learning_rate": 5.878030859662014e-07,
      "loss": 0.5289,
      "step": 31352
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.711786150932312,
      "learning_rate": 5.871641695684121e-07,
      "loss": 0.7019,
      "step": 31353
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1192691326141357,
      "learning_rate": 5.865252531706226e-07,
      "loss": 0.8759,
      "step": 31354
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2369325160980225,
      "learning_rate": 5.858863367728333e-07,
      "loss": 0.8574,
      "step": 31355
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1740968227386475,
      "learning_rate": 5.852474203750439e-07,
      "loss": 0.8947,
      "step": 31356
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5170105695724487,
      "learning_rate": 5.846085039772546e-07,
      "loss": 0.713,
      "step": 31357
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.936863660812378,
      "learning_rate": 5.839695875794652e-07,
      "loss": 0.7908,
      "step": 31358
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.495988845825195,
      "learning_rate": 5.833306711816759e-07,
      "loss": 1.0572,
      "step": 31359
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.291860342025757,
      "learning_rate": 5.826917547838866e-07,
      "loss": 0.9771,
      "step": 31360
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7400165796279907,
      "learning_rate": 5.820528383860972e-07,
      "loss": 0.729,
      "step": 31361
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.273048162460327,
      "learning_rate": 5.814139219883079e-07,
      "loss": 0.8154,
      "step": 31362
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6543684005737305,
      "learning_rate": 5.807750055905185e-07,
      "loss": 0.8104,
      "step": 31363
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.379478931427002,
      "learning_rate": 5.801360891927292e-07,
      "loss": 0.9401,
      "step": 31364
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.2959985733032227,
      "learning_rate": 5.794971727949398e-07,
      "loss": 0.871,
      "step": 31365
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.994820475578308,
      "learning_rate": 5.788582563971505e-07,
      "loss": 0.7473,
      "step": 31366
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4636175632476807,
      "learning_rate": 5.782193399993611e-07,
      "loss": 0.7885,
      "step": 31367
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7327784299850464,
      "learning_rate": 5.775804236015717e-07,
      "loss": 0.8566,
      "step": 31368
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6565769910812378,
      "learning_rate": 5.769415072037824e-07,
      "loss": 0.7643,
      "step": 31369
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9750300645828247,
      "learning_rate": 5.763025908059931e-07,
      "loss": 0.8356,
      "step": 31370
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5859168767929077,
      "learning_rate": 5.756636744082037e-07,
      "loss": 0.7714,
      "step": 31371
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5401455163955688,
      "learning_rate": 5.750247580104144e-07,
      "loss": 0.8407,
      "step": 31372
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.319181203842163,
      "learning_rate": 5.74385841612625e-07,
      "loss": 0.7921,
      "step": 31373
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.74342155456543,
      "learning_rate": 5.737469252148357e-07,
      "loss": 1.1277,
      "step": 31374
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.753124952316284,
      "learning_rate": 5.731080088170463e-07,
      "loss": 0.3973,
      "step": 31375
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.549149513244629,
      "learning_rate": 5.72469092419257e-07,
      "loss": 0.7846,
      "step": 31376
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.010754346847534,
      "learning_rate": 5.718301760214676e-07,
      "loss": 0.8599,
      "step": 31377
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8097513914108276,
      "learning_rate": 5.711912596236783e-07,
      "loss": 0.8717,
      "step": 31378
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.882107734680176,
      "learning_rate": 5.70552343225889e-07,
      "loss": 1.4309,
      "step": 31379
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.305548667907715,
      "learning_rate": 5.699134268280996e-07,
      "loss": 0.7009,
      "step": 31380
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3087146282196045,
      "learning_rate": 5.692745104303103e-07,
      "loss": 0.8354,
      "step": 31381
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.270850896835327,
      "learning_rate": 5.686355940325209e-07,
      "loss": 0.8484,
      "step": 31382
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4392880201339722,
      "learning_rate": 5.679966776347316e-07,
      "loss": 0.467,
      "step": 31383
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7821884155273438,
      "learning_rate": 5.673577612369422e-07,
      "loss": 0.864,
      "step": 31384
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.32755708694458,
      "learning_rate": 5.667188448391529e-07,
      "loss": 0.5957,
      "step": 31385
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.095901012420654,
      "learning_rate": 5.660799284413636e-07,
      "loss": 0.9988,
      "step": 31386
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.5853841304779053,
      "learning_rate": 5.654410120435742e-07,
      "loss": 0.8784,
      "step": 31387
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9763723611831665,
      "learning_rate": 5.648020956457849e-07,
      "loss": 1.1979,
      "step": 31388
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8366297483444214,
      "learning_rate": 5.641631792479955e-07,
      "loss": 0.8704,
      "step": 31389
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.286834239959717,
      "learning_rate": 5.635242628502062e-07,
      "loss": 0.8265,
      "step": 31390
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5710639953613281,
      "learning_rate": 5.628853464524167e-07,
      "loss": 0.8765,
      "step": 31391
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.03755521774292,
      "learning_rate": 5.622464300546273e-07,
      "loss": 0.7695,
      "step": 31392
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.098681926727295,
      "learning_rate": 5.61607513656838e-07,
      "loss": 1.4883,
      "step": 31393
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3270275592803955,
      "learning_rate": 5.609685972590486e-07,
      "loss": 0.9363,
      "step": 31394
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8805171251296997,
      "learning_rate": 5.603296808612593e-07,
      "loss": 0.8212,
      "step": 31395
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6842758655548096,
      "learning_rate": 5.5969076446347e-07,
      "loss": 0.8675,
      "step": 31396
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1657912731170654,
      "learning_rate": 5.590518480656806e-07,
      "loss": 0.951,
      "step": 31397
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9741500616073608,
      "learning_rate": 5.584129316678913e-07,
      "loss": 0.762,
      "step": 31398
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1543421745300293,
      "learning_rate": 5.577740152701019e-07,
      "loss": 0.9942,
      "step": 31399
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1077377796173096,
      "learning_rate": 5.571350988723126e-07,
      "loss": 1.1273,
      "step": 31400
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8800791501998901,
      "learning_rate": 5.564961824745232e-07,
      "loss": 0.7232,
      "step": 31401
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2142333984375,
      "learning_rate": 5.558572660767339e-07,
      "loss": 0.8261,
      "step": 31402
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2748360633850098,
      "learning_rate": 5.552183496789445e-07,
      "loss": 0.8129,
      "step": 31403
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7233459949493408,
      "learning_rate": 5.545794332811552e-07,
      "loss": 0.7367,
      "step": 31404
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5725276470184326,
      "learning_rate": 5.539405168833659e-07,
      "loss": 0.624,
      "step": 31405
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6962062120437622,
      "learning_rate": 5.533016004855765e-07,
      "loss": 0.8073,
      "step": 31406
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5863088369369507,
      "learning_rate": 5.526626840877872e-07,
      "loss": 0.6974,
      "step": 31407
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.246646881103516,
      "learning_rate": 5.520237676899978e-07,
      "loss": 1.105,
      "step": 31408
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3507871627807617,
      "learning_rate": 5.513848512922085e-07,
      "loss": 0.7608,
      "step": 31409
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6646721363067627,
      "learning_rate": 5.507459348944191e-07,
      "loss": 0.8317,
      "step": 31410
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.9097774028778076,
      "learning_rate": 5.501070184966298e-07,
      "loss": 0.6748,
      "step": 31411
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.032697916030884,
      "learning_rate": 5.494681020988404e-07,
      "loss": 0.9075,
      "step": 31412
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.183089017868042,
      "learning_rate": 5.488291857010511e-07,
      "loss": 0.8578,
      "step": 31413
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9648362398147583,
      "learning_rate": 5.481902693032618e-07,
      "loss": 0.8577,
      "step": 31414
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.19528865814209,
      "learning_rate": 5.475513529054724e-07,
      "loss": 0.9951,
      "step": 31415
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.252103090286255,
      "learning_rate": 5.46912436507683e-07,
      "loss": 1.085,
      "step": 31416
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6708874702453613,
      "learning_rate": 5.462735201098937e-07,
      "loss": 0.8316,
      "step": 31417
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7309045791625977,
      "learning_rate": 5.456346037121043e-07,
      "loss": 0.7262,
      "step": 31418
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8363280296325684,
      "learning_rate": 5.44995687314315e-07,
      "loss": 0.818,
      "step": 31419
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.6646199226379395,
      "learning_rate": 5.443567709165256e-07,
      "loss": 1.6209,
      "step": 31420
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.982590436935425,
      "learning_rate": 5.437178545187363e-07,
      "loss": 1.047,
      "step": 31421
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.986161231994629,
      "learning_rate": 5.430789381209469e-07,
      "loss": 0.8294,
      "step": 31422
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9352985620498657,
      "learning_rate": 5.424400217231576e-07,
      "loss": 0.835,
      "step": 31423
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6091796159744263,
      "learning_rate": 5.418011053253683e-07,
      "loss": 0.8142,
      "step": 31424
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1419949531555176,
      "learning_rate": 5.411621889275789e-07,
      "loss": 0.7811,
      "step": 31425
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.967961311340332,
      "learning_rate": 5.405232725297896e-07,
      "loss": 0.7694,
      "step": 31426
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6147288084030151,
      "learning_rate": 5.398843561320001e-07,
      "loss": 0.7603,
      "step": 31427
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.2228055000305176,
      "learning_rate": 5.392454397342108e-07,
      "loss": 1.0022,
      "step": 31428
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6959214210510254,
      "learning_rate": 5.386065233364214e-07,
      "loss": 0.9077,
      "step": 31429
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9271595478057861,
      "learning_rate": 5.379676069386321e-07,
      "loss": 0.8556,
      "step": 31430
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.733708143234253,
      "learning_rate": 5.373286905408428e-07,
      "loss": 0.9909,
      "step": 31431
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.03301739692688,
      "learning_rate": 5.366897741430534e-07,
      "loss": 0.7819,
      "step": 31432
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2490999698638916,
      "learning_rate": 5.360508577452641e-07,
      "loss": 0.7805,
      "step": 31433
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3587863445281982,
      "learning_rate": 5.354119413474747e-07,
      "loss": 0.8115,
      "step": 31434
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6775025129318237,
      "learning_rate": 5.347730249496854e-07,
      "loss": 0.6146,
      "step": 31435
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7764286994934082,
      "learning_rate": 5.34134108551896e-07,
      "loss": 0.8629,
      "step": 31436
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6358551979064941,
      "learning_rate": 5.334951921541067e-07,
      "loss": 1.1323,
      "step": 31437
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.285102367401123,
      "learning_rate": 5.328562757563173e-07,
      "loss": 0.8378,
      "step": 31438
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.9413983821868896,
      "learning_rate": 5.322173593585279e-07,
      "loss": 0.6306,
      "step": 31439
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8291194438934326,
      "learning_rate": 5.315784429607386e-07,
      "loss": 0.796,
      "step": 31440
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6953551769256592,
      "learning_rate": 5.309395265629492e-07,
      "loss": 1.0679,
      "step": 31441
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.747227668762207,
      "learning_rate": 5.303006101651599e-07,
      "loss": 0.8835,
      "step": 31442
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.906236410140991,
      "learning_rate": 5.296616937673706e-07,
      "loss": 0.5637,
      "step": 31443
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.168401002883911,
      "learning_rate": 5.290227773695812e-07,
      "loss": 0.7212,
      "step": 31444
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4770116806030273,
      "learning_rate": 5.283838609717919e-07,
      "loss": 1.0593,
      "step": 31445
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4857062101364136,
      "learning_rate": 5.277449445740025e-07,
      "loss": 0.3607,
      "step": 31446
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9667648077011108,
      "learning_rate": 5.271060281762132e-07,
      "loss": 0.7875,
      "step": 31447
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7482874393463135,
      "learning_rate": 5.264671117784238e-07,
      "loss": 0.8459,
      "step": 31448
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8588789701461792,
      "learning_rate": 5.258281953806345e-07,
      "loss": 0.7569,
      "step": 31449
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8640536069869995,
      "learning_rate": 5.251892789828452e-07,
      "loss": 0.6499,
      "step": 31450
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8176909685134888,
      "learning_rate": 5.245503625850558e-07,
      "loss": 0.8391,
      "step": 31451
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.902167320251465,
      "learning_rate": 5.239114461872665e-07,
      "loss": 1.354,
      "step": 31452
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2051234245300293,
      "learning_rate": 5.232725297894771e-07,
      "loss": 0.9891,
      "step": 31453
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6047661304473877,
      "learning_rate": 5.226336133916878e-07,
      "loss": 0.7801,
      "step": 31454
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9791063070297241,
      "learning_rate": 5.219946969938984e-07,
      "loss": 0.703,
      "step": 31455
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.818145990371704,
      "learning_rate": 5.213557805961091e-07,
      "loss": 0.9161,
      "step": 31456
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0201170444488525,
      "learning_rate": 5.207168641983197e-07,
      "loss": 0.8394,
      "step": 31457
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.774354100227356,
      "learning_rate": 5.200779478005304e-07,
      "loss": 0.8263,
      "step": 31458
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3840806484222412,
      "learning_rate": 5.194390314027411e-07,
      "loss": 0.7329,
      "step": 31459
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.7544267177581787,
      "learning_rate": 5.188001150049517e-07,
      "loss": 0.3459,
      "step": 31460
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.8268961906433105,
      "learning_rate": 5.181611986071624e-07,
      "loss": 1.1835,
      "step": 31461
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.016540765762329,
      "learning_rate": 5.17522282209373e-07,
      "loss": 0.8679,
      "step": 31462
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4646646976470947,
      "learning_rate": 5.168833658115835e-07,
      "loss": 0.935,
      "step": 31463
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6513954401016235,
      "learning_rate": 5.162444494137942e-07,
      "loss": 0.757,
      "step": 31464
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.4541211128234863,
      "learning_rate": 5.156055330160048e-07,
      "loss": 0.429,
      "step": 31465
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.567268133163452,
      "learning_rate": 5.149666166182155e-07,
      "loss": 0.8138,
      "step": 31466
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0685391426086426,
      "learning_rate": 5.143277002204261e-07,
      "loss": 0.7934,
      "step": 31467
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.18134880065918,
      "learning_rate": 5.136887838226368e-07,
      "loss": 0.8128,
      "step": 31468
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8475382328033447,
      "learning_rate": 5.130498674248475e-07,
      "loss": 1.0691,
      "step": 31469
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.836553931236267,
      "learning_rate": 5.124109510270581e-07,
      "loss": 0.9496,
      "step": 31470
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.243372678756714,
      "learning_rate": 5.117720346292688e-07,
      "loss": 0.7254,
      "step": 31471
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.200472354888916,
      "learning_rate": 5.111331182314794e-07,
      "loss": 0.924,
      "step": 31472
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.154297351837158,
      "learning_rate": 5.104942018336901e-07,
      "loss": 0.8978,
      "step": 31473
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.583304762840271,
      "learning_rate": 5.098552854359007e-07,
      "loss": 0.8122,
      "step": 31474
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.95341157913208,
      "learning_rate": 5.092163690381114e-07,
      "loss": 1.3003,
      "step": 31475
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.955048680305481,
      "learning_rate": 5.08577452640322e-07,
      "loss": 0.8817,
      "step": 31476
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7699717283248901,
      "learning_rate": 5.079385362425327e-07,
      "loss": 0.7988,
      "step": 31477
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5322694778442383,
      "learning_rate": 5.072996198447434e-07,
      "loss": 0.7031,
      "step": 31478
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.6564059257507324,
      "learning_rate": 5.06660703446954e-07,
      "loss": 0.8634,
      "step": 31479
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1441662311553955,
      "learning_rate": 5.060217870491647e-07,
      "loss": 0.9006,
      "step": 31480
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9062920808792114,
      "learning_rate": 5.053828706513753e-07,
      "loss": 0.9105,
      "step": 31481
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5278500318527222,
      "learning_rate": 5.04743954253586e-07,
      "loss": 0.7546,
      "step": 31482
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.6251959800720215,
      "learning_rate": 5.041050378557966e-07,
      "loss": 1.1651,
      "step": 31483
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6232614517211914,
      "learning_rate": 5.034661214580073e-07,
      "loss": 0.8084,
      "step": 31484
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2252416610717773,
      "learning_rate": 5.02827205060218e-07,
      "loss": 0.6566,
      "step": 31485
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8687479496002197,
      "learning_rate": 5.021882886624286e-07,
      "loss": 0.9241,
      "step": 31486
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.7410378456115723,
      "learning_rate": 5.015493722646393e-07,
      "loss": 0.7136,
      "step": 31487
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7847834825515747,
      "learning_rate": 5.009104558668499e-07,
      "loss": 0.8767,
      "step": 31488
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.4642367362976074,
      "learning_rate": 5.002715394690605e-07,
      "loss": 0.8183,
      "step": 31489
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8005595207214355,
      "learning_rate": 4.996326230712712e-07,
      "loss": 0.8298,
      "step": 31490
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.772123336791992,
      "learning_rate": 4.989937066734818e-07,
      "loss": 1.3665,
      "step": 31491
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.646854043006897,
      "learning_rate": 4.983547902756925e-07,
      "loss": 0.7046,
      "step": 31492
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.169307708740234,
      "learning_rate": 4.977158738779031e-07,
      "loss": 1.1126,
      "step": 31493
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6343656778335571,
      "learning_rate": 4.970769574801138e-07,
      "loss": 0.7998,
      "step": 31494
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2429251670837402,
      "learning_rate": 4.964380410823244e-07,
      "loss": 0.3036,
      "step": 31495
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3359663486480713,
      "learning_rate": 4.957991246845351e-07,
      "loss": 0.5701,
      "step": 31496
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.491726875305176,
      "learning_rate": 4.951602082867458e-07,
      "loss": 0.9499,
      "step": 31497
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3870036602020264,
      "learning_rate": 4.945212918889563e-07,
      "loss": 0.862,
      "step": 31498
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.72830069065094,
      "learning_rate": 4.93882375491167e-07,
      "loss": 0.8355,
      "step": 31499
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4785151481628418,
      "learning_rate": 4.932434590933776e-07,
      "loss": 0.6581,
      "step": 31500
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7262276411056519,
      "learning_rate": 4.926045426955883e-07,
      "loss": 0.7816,
      "step": 31501
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7863011360168457,
      "learning_rate": 4.919656262977989e-07,
      "loss": 0.963,
      "step": 31502
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3060595989227295,
      "learning_rate": 4.913267099000096e-07,
      "loss": 1.2259,
      "step": 31503
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5516518354415894,
      "learning_rate": 4.906877935022203e-07,
      "loss": 0.7259,
      "step": 31504
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.145292043685913,
      "learning_rate": 4.900488771044309e-07,
      "loss": 0.9516,
      "step": 31505
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.669996500015259,
      "learning_rate": 4.894099607066416e-07,
      "loss": 0.7912,
      "step": 31506
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8412996530532837,
      "learning_rate": 4.887710443088522e-07,
      "loss": 0.7183,
      "step": 31507
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.421624183654785,
      "learning_rate": 4.881321279110629e-07,
      "loss": 1.2416,
      "step": 31508
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.544928789138794,
      "learning_rate": 4.874932115132735e-07,
      "loss": 1.2065,
      "step": 31509
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.548986792564392,
      "learning_rate": 4.868542951154842e-07,
      "loss": 0.8098,
      "step": 31510
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.575657606124878,
      "learning_rate": 4.862153787176948e-07,
      "loss": 0.9158,
      "step": 31511
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.5028953552246094,
      "learning_rate": 4.855764623199054e-07,
      "loss": 1.2188,
      "step": 31512
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.4190938472747803,
      "learning_rate": 4.849375459221161e-07,
      "loss": 0.7571,
      "step": 31513
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8853178024291992,
      "learning_rate": 4.842986295243268e-07,
      "loss": 0.7852,
      "step": 31514
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5956649780273438,
      "learning_rate": 4.836597131265374e-07,
      "loss": 0.8472,
      "step": 31515
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9222679138183594,
      "learning_rate": 4.830207967287481e-07,
      "loss": 0.7723,
      "step": 31516
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.017784357070923,
      "learning_rate": 4.823818803309587e-07,
      "loss": 0.6604,
      "step": 31517
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.5484020709991455,
      "learning_rate": 4.817429639331694e-07,
      "loss": 0.8308,
      "step": 31518
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.179429531097412,
      "learning_rate": 4.8110404753538e-07,
      "loss": 0.7831,
      "step": 31519
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5875372886657715,
      "learning_rate": 4.804651311375907e-07,
      "loss": 0.7802,
      "step": 31520
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7923001050949097,
      "learning_rate": 4.798262147398013e-07,
      "loss": 0.8028,
      "step": 31521
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5389350652694702,
      "learning_rate": 4.79187298342012e-07,
      "loss": 0.7605,
      "step": 31522
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7988299131393433,
      "learning_rate": 4.785483819442227e-07,
      "loss": 0.8079,
      "step": 31523
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.91190242767334,
      "learning_rate": 4.779094655464333e-07,
      "loss": 0.8189,
      "step": 31524
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.7746782302856445,
      "learning_rate": 4.77270549148644e-07,
      "loss": 1.195,
      "step": 31525
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6371815204620361,
      "learning_rate": 4.766316327508546e-07,
      "loss": 0.77,
      "step": 31526
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7886165380477905,
      "learning_rate": 4.7599271635306527e-07,
      "loss": 0.815,
      "step": 31527
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.6662027835845947,
      "learning_rate": 4.753537999552759e-07,
      "loss": 0.7517,
      "step": 31528
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2620999813079834,
      "learning_rate": 4.7471488355748656e-07,
      "loss": 0.5968,
      "step": 31529
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.242461919784546,
      "learning_rate": 4.740759671596972e-07,
      "loss": 0.6886,
      "step": 31530
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6157829761505127,
      "learning_rate": 4.7343705076190786e-07,
      "loss": 0.6518,
      "step": 31531
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9290406703948975,
      "learning_rate": 4.727981343641185e-07,
      "loss": 0.8281,
      "step": 31532
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.199471950531006,
      "learning_rate": 4.7215921796632915e-07,
      "loss": 0.8061,
      "step": 31533
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9474996328353882,
      "learning_rate": 4.715203015685398e-07,
      "loss": 0.79,
      "step": 31534
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8045871257781982,
      "learning_rate": 4.708813851707505e-07,
      "loss": 0.9149,
      "step": 31535
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.947467565536499,
      "learning_rate": 4.7024246877296104e-07,
      "loss": 0.8535,
      "step": 31536
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1093080043792725,
      "learning_rate": 4.6960355237517174e-07,
      "loss": 0.8054,
      "step": 31537
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7581171989440918,
      "learning_rate": 4.689646359773824e-07,
      "loss": 0.7954,
      "step": 31538
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4684135913848877,
      "learning_rate": 4.6832571957959303e-07,
      "loss": 0.4644,
      "step": 31539
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1370837688446045,
      "learning_rate": 4.676868031818037e-07,
      "loss": 0.8583,
      "step": 31540
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5598201751708984,
      "learning_rate": 4.6704788678401433e-07,
      "loss": 0.4663,
      "step": 31541
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7279752492904663,
      "learning_rate": 4.66408970386225e-07,
      "loss": 0.781,
      "step": 31542
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.756820797920227,
      "learning_rate": 4.657700539884356e-07,
      "loss": 0.7685,
      "step": 31543
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2896337509155273,
      "learning_rate": 4.6513113759064627e-07,
      "loss": 0.8814,
      "step": 31544
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5119516849517822,
      "learning_rate": 4.6449222119285697e-07,
      "loss": 0.7337,
      "step": 31545
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9816492795944214,
      "learning_rate": 4.638533047950676e-07,
      "loss": 0.8173,
      "step": 31546
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6836475133895874,
      "learning_rate": 4.6321438839727826e-07,
      "loss": 0.8351,
      "step": 31547
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.324021100997925,
      "learning_rate": 4.625754719994889e-07,
      "loss": 0.7994,
      "step": 31548
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0660133361816406,
      "learning_rate": 4.6193655560169956e-07,
      "loss": 0.9996,
      "step": 31549
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.043309688568115,
      "learning_rate": 4.612976392039102e-07,
      "loss": 1.3526,
      "step": 31550
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.494683027267456,
      "learning_rate": 4.6065872280612085e-07,
      "loss": 0.9487,
      "step": 31551
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6086416244506836,
      "learning_rate": 4.600198064083315e-07,
      "loss": 0.9079,
      "step": 31552
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.9251034259796143,
      "learning_rate": 4.593808900105422e-07,
      "loss": 1.0394,
      "step": 31553
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.103532314300537,
      "learning_rate": 4.5874197361275285e-07,
      "loss": 1.0475,
      "step": 31554
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0068681240081787,
      "learning_rate": 4.581030572149635e-07,
      "loss": 0.8378,
      "step": 31555
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9074621200561523,
      "learning_rate": 4.5746414081717414e-07,
      "loss": 0.9871,
      "step": 31556
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.691449522972107,
      "learning_rate": 4.568252244193848e-07,
      "loss": 0.7112,
      "step": 31557
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.4766151905059814,
      "learning_rate": 4.5618630802159543e-07,
      "loss": 1.0027,
      "step": 31558
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.4246230125427246,
      "learning_rate": 4.555473916238061e-07,
      "loss": 0.4793,
      "step": 31559
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9038876295089722,
      "learning_rate": 4.549084752260167e-07,
      "loss": 0.9213,
      "step": 31560
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.0570356845855713,
      "learning_rate": 4.542695588282273e-07,
      "loss": 0.8789,
      "step": 31561
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.890252709388733,
      "learning_rate": 4.5363064243043797e-07,
      "loss": 0.8299,
      "step": 31562
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0163486003875732,
      "learning_rate": 4.529917260326486e-07,
      "loss": 0.7948,
      "step": 31563
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8428596258163452,
      "learning_rate": 4.523528096348593e-07,
      "loss": 0.7913,
      "step": 31564
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.198523759841919,
      "learning_rate": 4.5171389323706996e-07,
      "loss": 0.5452,
      "step": 31565
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.76041579246521,
      "learning_rate": 4.510749768392806e-07,
      "loss": 0.7802,
      "step": 31566
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3662242889404297,
      "learning_rate": 4.5043606044149126e-07,
      "loss": 0.3227,
      "step": 31567
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5520938634872437,
      "learning_rate": 4.497971440437019e-07,
      "loss": 0.8114,
      "step": 31568
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8211013078689575,
      "learning_rate": 4.4915822764591255e-07,
      "loss": 0.914,
      "step": 31569
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.714084506034851,
      "learning_rate": 4.485193112481232e-07,
      "loss": 0.8371,
      "step": 31570
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.677438735961914,
      "learning_rate": 4.4788039485033385e-07,
      "loss": 0.8467,
      "step": 31571
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.516684889793396,
      "learning_rate": 4.4724147845254455e-07,
      "loss": 0.8223,
      "step": 31572
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7472468614578247,
      "learning_rate": 4.466025620547552e-07,
      "loss": 0.7648,
      "step": 31573
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3471460342407227,
      "learning_rate": 4.4596364565696584e-07,
      "loss": 0.8585,
      "step": 31574
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.064077138900757,
      "learning_rate": 4.453247292591765e-07,
      "loss": 0.8709,
      "step": 31575
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.108182907104492,
      "learning_rate": 4.4468581286138713e-07,
      "loss": 0.8081,
      "step": 31576
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.839822769165039,
      "learning_rate": 4.440468964635978e-07,
      "loss": 1.0232,
      "step": 31577
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7544149160385132,
      "learning_rate": 4.4340798006580843e-07,
      "loss": 0.8077,
      "step": 31578
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1527812480926514,
      "learning_rate": 4.427690636680191e-07,
      "loss": 0.8477,
      "step": 31579
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6839357614517212,
      "learning_rate": 4.421301472702298e-07,
      "loss": 0.8685,
      "step": 31580
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.01884388923645,
      "learning_rate": 4.414912308724404e-07,
      "loss": 0.7229,
      "step": 31581
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.878330111503601,
      "learning_rate": 4.4085231447465107e-07,
      "loss": 0.8213,
      "step": 31582
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8743679523468018,
      "learning_rate": 4.402133980768617e-07,
      "loss": 0.7999,
      "step": 31583
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4535688161849976,
      "learning_rate": 4.395744816790723e-07,
      "loss": 0.7239,
      "step": 31584
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6285234689712524,
      "learning_rate": 4.3893556528128296e-07,
      "loss": 0.8443,
      "step": 31585
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.033123016357422,
      "learning_rate": 4.382966488834936e-07,
      "loss": 0.7778,
      "step": 31586
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.8917906284332275,
      "learning_rate": 4.3765773248570425e-07,
      "loss": 0.4568,
      "step": 31587
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.912154197692871,
      "learning_rate": 4.370188160879149e-07,
      "loss": 0.9292,
      "step": 31588
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3537700176239014,
      "learning_rate": 4.3637989969012554e-07,
      "loss": 0.9164,
      "step": 31589
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.075882911682129,
      "learning_rate": 4.357409832923362e-07,
      "loss": 0.5899,
      "step": 31590
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9073965549468994,
      "learning_rate": 4.351020668945469e-07,
      "loss": 0.7685,
      "step": 31591
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.226649045944214,
      "learning_rate": 4.3446315049675754e-07,
      "loss": 0.8537,
      "step": 31592
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.1271941661834717,
      "learning_rate": 4.338242340989682e-07,
      "loss": 0.52,
      "step": 31593
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7546104192733765,
      "learning_rate": 4.3318531770117883e-07,
      "loss": 0.7846,
      "step": 31594
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.381587266921997,
      "learning_rate": 4.325464013033895e-07,
      "loss": 0.8857,
      "step": 31595
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8003263473510742,
      "learning_rate": 4.3190748490560013e-07,
      "loss": 0.8413,
      "step": 31596
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.263613224029541,
      "learning_rate": 4.312685685078108e-07,
      "loss": 0.562,
      "step": 31597
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1982007026672363,
      "learning_rate": 4.306296521100214e-07,
      "loss": 0.7412,
      "step": 31598
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5408252477645874,
      "learning_rate": 4.299907357122321e-07,
      "loss": 0.7076,
      "step": 31599
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2123584747314453,
      "learning_rate": 4.2935181931444277e-07,
      "loss": 0.8544,
      "step": 31600
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.7324538230896,
      "learning_rate": 4.287129029166534e-07,
      "loss": 1.1451,
      "step": 31601
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.0848917961120605,
      "learning_rate": 4.2807398651886406e-07,
      "loss": 1.4913,
      "step": 31602
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6335487365722656,
      "learning_rate": 4.274350701210747e-07,
      "loss": 0.8674,
      "step": 31603
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.082810640335083,
      "learning_rate": 4.2679615372328536e-07,
      "loss": 0.8898,
      "step": 31604
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7633349895477295,
      "learning_rate": 4.26157237325496e-07,
      "loss": 0.7592,
      "step": 31605
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0158560276031494,
      "learning_rate": 4.2551832092770665e-07,
      "loss": 1.1431,
      "step": 31606
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7145586013793945,
      "learning_rate": 4.2487940452991735e-07,
      "loss": 0.7993,
      "step": 31607
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.9833359718322754,
      "learning_rate": 4.242404881321279e-07,
      "loss": 0.6583,
      "step": 31608
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.1457767486572266,
      "learning_rate": 4.236015717343386e-07,
      "loss": 0.7025,
      "step": 31609
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.904664158821106,
      "learning_rate": 4.2296265533654924e-07,
      "loss": 0.3955,
      "step": 31610
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.4020702838897705,
      "learning_rate": 4.223237389387599e-07,
      "loss": 0.8704,
      "step": 31611
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3364784717559814,
      "learning_rate": 4.2168482254097053e-07,
      "loss": 0.3202,
      "step": 31612
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6769195795059204,
      "learning_rate": 4.210459061431812e-07,
      "loss": 0.7964,
      "step": 31613
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4964349269866943,
      "learning_rate": 4.204069897453918e-07,
      "loss": 0.7442,
      "step": 31614
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3323585987091064,
      "learning_rate": 4.1976807334760247e-07,
      "loss": 1.0029,
      "step": 31615
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2810897827148438,
      "learning_rate": 4.191291569498131e-07,
      "loss": 0.8893,
      "step": 31616
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7074252367019653,
      "learning_rate": 4.184902405520238e-07,
      "loss": 1.0605,
      "step": 31617
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5620980262756348,
      "learning_rate": 4.1785132415423447e-07,
      "loss": 0.7966,
      "step": 31618
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7140456438064575,
      "learning_rate": 4.172124077564451e-07,
      "loss": 0.8579,
      "step": 31619
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.969053268432617,
      "learning_rate": 4.1657349135865576e-07,
      "loss": 1.4857,
      "step": 31620
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.774808406829834,
      "learning_rate": 4.159345749608664e-07,
      "loss": 0.5636,
      "step": 31621
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.5752294063568115,
      "learning_rate": 4.1529565856307706e-07,
      "loss": 1.2151,
      "step": 31622
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9795081615447998,
      "learning_rate": 4.146567421652877e-07,
      "loss": 0.7142,
      "step": 31623
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9821720123291016,
      "learning_rate": 4.1401782576749835e-07,
      "loss": 0.8288,
      "step": 31624
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5613617897033691,
      "learning_rate": 4.13378909369709e-07,
      "loss": 0.9309,
      "step": 31625
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.40643835067749,
      "learning_rate": 4.127399929719197e-07,
      "loss": 1.0544,
      "step": 31626
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.308060884475708,
      "learning_rate": 4.1210107657413034e-07,
      "loss": 0.6714,
      "step": 31627
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6824630498886108,
      "learning_rate": 4.11462160176341e-07,
      "loss": 0.7445,
      "step": 31628
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.703984022140503,
      "learning_rate": 4.1082324377855164e-07,
      "loss": 0.7297,
      "step": 31629
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.150970458984375,
      "learning_rate": 4.101843273807623e-07,
      "loss": 0.8728,
      "step": 31630
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5918114185333252,
      "learning_rate": 4.0954541098297293e-07,
      "loss": 0.7661,
      "step": 31631
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.103266477584839,
      "learning_rate": 4.089064945851835e-07,
      "loss": 0.8356,
      "step": 31632
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9315264225006104,
      "learning_rate": 4.0826757818739417e-07,
      "loss": 0.8454,
      "step": 31633
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.121349334716797,
      "learning_rate": 4.076286617896048e-07,
      "loss": 1.1857,
      "step": 31634
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8263168334960938,
      "learning_rate": 4.0698974539181547e-07,
      "loss": 0.8836,
      "step": 31635
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.090904951095581,
      "learning_rate": 4.0635082899402617e-07,
      "loss": 0.843,
      "step": 31636
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.151210069656372,
      "learning_rate": 4.057119125962368e-07,
      "loss": 0.7938,
      "step": 31637
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.753037214279175,
      "learning_rate": 4.0507299619844746e-07,
      "loss": 0.5425,
      "step": 31638
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.1737232208251953,
      "learning_rate": 4.044340798006581e-07,
      "loss": 0.9284,
      "step": 31639
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9079972505569458,
      "learning_rate": 4.0379516340286875e-07,
      "loss": 0.7327,
      "step": 31640
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.936981439590454,
      "learning_rate": 4.031562470050794e-07,
      "loss": 1.1413,
      "step": 31641
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5532156229019165,
      "learning_rate": 4.0251733060729005e-07,
      "loss": 0.8089,
      "step": 31642
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.5221080780029297,
      "learning_rate": 4.018784142095007e-07,
      "loss": 0.7527,
      "step": 31643
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7739582061767578,
      "learning_rate": 4.012394978117114e-07,
      "loss": 0.9248,
      "step": 31644
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.89873206615448,
      "learning_rate": 4.0060058141392204e-07,
      "loss": 0.7845,
      "step": 31645
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.6995060443878174,
      "learning_rate": 3.999616650161327e-07,
      "loss": 1.0136,
      "step": 31646
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0749356746673584,
      "learning_rate": 3.9932274861834334e-07,
      "loss": 1.0632,
      "step": 31647
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8701380491256714,
      "learning_rate": 3.98683832220554e-07,
      "loss": 0.7979,
      "step": 31648
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1858768463134766,
      "learning_rate": 3.9804491582276463e-07,
      "loss": 0.8759,
      "step": 31649
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5493762493133545,
      "learning_rate": 3.974059994249753e-07,
      "loss": 0.756,
      "step": 31650
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.042851686477661,
      "learning_rate": 3.967670830271859e-07,
      "loss": 0.9383,
      "step": 31651
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.733397006988525,
      "learning_rate": 3.961281666293966e-07,
      "loss": 1.2169,
      "step": 31652
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.734715461730957,
      "learning_rate": 3.9548925023160727e-07,
      "loss": 0.7822,
      "step": 31653
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8601837158203125,
      "learning_rate": 3.948503338338179e-07,
      "loss": 1.0562,
      "step": 31654
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8878761529922485,
      "learning_rate": 3.9421141743602857e-07,
      "loss": 0.9362,
      "step": 31655
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.8562538623809814,
      "learning_rate": 3.935725010382392e-07,
      "loss": 1.0228,
      "step": 31656
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.508154273033142,
      "learning_rate": 3.929335846404498e-07,
      "loss": 0.7329,
      "step": 31657
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5484869480133057,
      "learning_rate": 3.9229466824266045e-07,
      "loss": 0.7662,
      "step": 31658
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6763384342193604,
      "learning_rate": 3.916557518448711e-07,
      "loss": 0.7947,
      "step": 31659
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3255667686462402,
      "learning_rate": 3.9101683544708175e-07,
      "loss": 1.1291,
      "step": 31660
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8309043645858765,
      "learning_rate": 3.903779190492924e-07,
      "loss": 0.8531,
      "step": 31661
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.6356046199798584,
      "learning_rate": 3.8973900265150304e-07,
      "loss": 0.7681,
      "step": 31662
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.0201854705810547,
      "learning_rate": 3.8910008625371374e-07,
      "loss": 1.1541,
      "step": 31663
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5610750913619995,
      "learning_rate": 3.884611698559244e-07,
      "loss": 0.8255,
      "step": 31664
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.150981903076172,
      "learning_rate": 3.8782225345813504e-07,
      "loss": 0.8986,
      "step": 31665
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5221306085586548,
      "learning_rate": 3.871833370603457e-07,
      "loss": 0.7285,
      "step": 31666
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.499345302581787,
      "learning_rate": 3.8654442066255633e-07,
      "loss": 0.5572,
      "step": 31667
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5592377185821533,
      "learning_rate": 3.85905504264767e-07,
      "loss": 0.7385,
      "step": 31668
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.9103338718414307,
      "learning_rate": 3.852665878669776e-07,
      "loss": 0.7764,
      "step": 31669
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8173261880874634,
      "learning_rate": 3.8462767146918827e-07,
      "loss": 0.7784,
      "step": 31670
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1081318855285645,
      "learning_rate": 3.8398875507139897e-07,
      "loss": 0.8275,
      "step": 31671
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.709669589996338,
      "learning_rate": 3.833498386736096e-07,
      "loss": 0.7843,
      "step": 31672
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9890294075012207,
      "learning_rate": 3.8271092227582026e-07,
      "loss": 0.8009,
      "step": 31673
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9891170263290405,
      "learning_rate": 3.820720058780309e-07,
      "loss": 0.9599,
      "step": 31674
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.6296608448028564,
      "learning_rate": 3.8143308948024156e-07,
      "loss": 1.0132,
      "step": 31675
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.4004509449005127,
      "learning_rate": 3.807941730824522e-07,
      "loss": 0.9338,
      "step": 31676
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.524256467819214,
      "learning_rate": 3.8015525668466285e-07,
      "loss": 0.6079,
      "step": 31677
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.020313024520874,
      "learning_rate": 3.795163402868735e-07,
      "loss": 0.7714,
      "step": 31678
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6910488605499268,
      "learning_rate": 3.788774238890842e-07,
      "loss": 0.9095,
      "step": 31679
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.506010890007019,
      "learning_rate": 3.7823850749129485e-07,
      "loss": 0.7002,
      "step": 31680
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.376991033554077,
      "learning_rate": 3.775995910935054e-07,
      "loss": 0.738,
      "step": 31681
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.6778509616851807,
      "learning_rate": 3.769606746957161e-07,
      "loss": 0.5062,
      "step": 31682
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8506184816360474,
      "learning_rate": 3.7632175829792674e-07,
      "loss": 0.7199,
      "step": 31683
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5873241424560547,
      "learning_rate": 3.756828419001374e-07,
      "loss": 0.6948,
      "step": 31684
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3708620071411133,
      "learning_rate": 3.7504392550234803e-07,
      "loss": 0.9212,
      "step": 31685
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5179482698440552,
      "learning_rate": 3.744050091045587e-07,
      "loss": 0.8053,
      "step": 31686
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.505888819694519,
      "learning_rate": 3.737660927067693e-07,
      "loss": 0.7907,
      "step": 31687
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.975074052810669,
      "learning_rate": 3.7312717630897997e-07,
      "loss": 0.7973,
      "step": 31688
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2824113368988037,
      "learning_rate": 3.724882599111906e-07,
      "loss": 0.7184,
      "step": 31689
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.832215666770935,
      "learning_rate": 3.718493435134013e-07,
      "loss": 0.7888,
      "step": 31690
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.636147141456604,
      "learning_rate": 3.7121042711561196e-07,
      "loss": 0.8525,
      "step": 31691
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5587376356124878,
      "learning_rate": 3.705715107178226e-07,
      "loss": 0.8237,
      "step": 31692
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.015347480773926,
      "learning_rate": 3.6993259432003326e-07,
      "loss": 0.9747,
      "step": 31693
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2808971405029297,
      "learning_rate": 3.692936779222439e-07,
      "loss": 0.9785,
      "step": 31694
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7422350645065308,
      "learning_rate": 3.6865476152445455e-07,
      "loss": 0.783,
      "step": 31695
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9080638885498047,
      "learning_rate": 3.680158451266652e-07,
      "loss": 0.8331,
      "step": 31696
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.361936092376709,
      "learning_rate": 3.6737692872887585e-07,
      "loss": 0.8215,
      "step": 31697
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8439418077468872,
      "learning_rate": 3.6673801233108655e-07,
      "loss": 0.9085,
      "step": 31698
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8671793937683105,
      "learning_rate": 3.660990959332972e-07,
      "loss": 0.7649,
      "step": 31699
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.702249050140381,
      "learning_rate": 3.6546017953550784e-07,
      "loss": 1.1542,
      "step": 31700
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9638655185699463,
      "learning_rate": 3.648212631377185e-07,
      "loss": 0.9392,
      "step": 31701
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8381521701812744,
      "learning_rate": 3.6418234673992913e-07,
      "loss": 0.9886,
      "step": 31702
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.045701503753662,
      "learning_rate": 3.635434303421398e-07,
      "loss": 0.8273,
      "step": 31703
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9889779090881348,
      "learning_rate": 3.6290451394435043e-07,
      "loss": 0.8724,
      "step": 31704
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.036313772201538,
      "learning_rate": 3.62265597546561e-07,
      "loss": 0.7355,
      "step": 31705
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6117286682128906,
      "learning_rate": 3.6162668114877167e-07,
      "loss": 0.7808,
      "step": 31706
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8611152172088623,
      "learning_rate": 3.609877647509823e-07,
      "loss": 0.9923,
      "step": 31707
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7472209930419922,
      "learning_rate": 3.6034884835319296e-07,
      "loss": 0.8077,
      "step": 31708
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7999979257583618,
      "learning_rate": 3.5970993195540366e-07,
      "loss": 0.8241,
      "step": 31709
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.969092607498169,
      "learning_rate": 3.590710155576143e-07,
      "loss": 0.9083,
      "step": 31710
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.400777816772461,
      "learning_rate": 3.5843209915982496e-07,
      "loss": 0.5916,
      "step": 31711
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.603048801422119,
      "learning_rate": 3.577931827620356e-07,
      "loss": 0.959,
      "step": 31712
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.8748526573181152,
      "learning_rate": 3.5715426636424625e-07,
      "loss": 0.9991,
      "step": 31713
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.629387617111206,
      "learning_rate": 3.565153499664569e-07,
      "loss": 0.7325,
      "step": 31714
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4672163724899292,
      "learning_rate": 3.5587643356866755e-07,
      "loss": 0.7594,
      "step": 31715
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7574667930603027,
      "learning_rate": 3.552375171708782e-07,
      "loss": 0.8296,
      "step": 31716
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9238070249557495,
      "learning_rate": 3.545986007730889e-07,
      "loss": 0.8635,
      "step": 31717
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4590370655059814,
      "learning_rate": 3.5395968437529954e-07,
      "loss": 0.7357,
      "step": 31718
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.727449655532837,
      "learning_rate": 3.533207679775102e-07,
      "loss": 0.8433,
      "step": 31719
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.459254503250122,
      "learning_rate": 3.5268185157972083e-07,
      "loss": 0.792,
      "step": 31720
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3232066631317139,
      "learning_rate": 3.520429351819315e-07,
      "loss": 0.3299,
      "step": 31721
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8160253763198853,
      "learning_rate": 3.5140401878414213e-07,
      "loss": 0.8428,
      "step": 31722
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9500397443771362,
      "learning_rate": 3.507651023863528e-07,
      "loss": 0.8686,
      "step": 31723
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.691004991531372,
      "learning_rate": 3.501261859885634e-07,
      "loss": 0.7658,
      "step": 31724
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.387921094894409,
      "learning_rate": 3.494872695907741e-07,
      "loss": 1.1392,
      "step": 31725
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.2860827445983887,
      "learning_rate": 3.4884835319298477e-07,
      "loss": 1.5363,
      "step": 31726
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8147505521774292,
      "learning_rate": 3.482094367951954e-07,
      "loss": 0.7443,
      "step": 31727
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.752377986907959,
      "learning_rate": 3.4757052039740606e-07,
      "loss": 0.7601,
      "step": 31728
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3723230361938477,
      "learning_rate": 3.4693160399961666e-07,
      "loss": 0.7346,
      "step": 31729
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.499159812927246,
      "learning_rate": 3.462926876018273e-07,
      "loss": 1.1085,
      "step": 31730
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.9116170406341553,
      "learning_rate": 3.4565377120403795e-07,
      "loss": 1.2532,
      "step": 31731
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5481975078582764,
      "learning_rate": 3.450148548062486e-07,
      "loss": 0.7881,
      "step": 31732
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3574867248535156,
      "learning_rate": 3.4437593840845924e-07,
      "loss": 0.9637,
      "step": 31733
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9607162475585938,
      "learning_rate": 3.437370220106699e-07,
      "loss": 1.2436,
      "step": 31734
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.930029034614563,
      "learning_rate": 3.430981056128806e-07,
      "loss": 0.8884,
      "step": 31735
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9799448251724243,
      "learning_rate": 3.4245918921509124e-07,
      "loss": 0.8618,
      "step": 31736
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7114137411117554,
      "learning_rate": 3.418202728173019e-07,
      "loss": 0.6762,
      "step": 31737
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8713676929473877,
      "learning_rate": 3.4118135641951253e-07,
      "loss": 0.7311,
      "step": 31738
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9812848567962646,
      "learning_rate": 3.405424400217232e-07,
      "loss": 0.7418,
      "step": 31739
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.950793981552124,
      "learning_rate": 3.3990352362393383e-07,
      "loss": 0.7727,
      "step": 31740
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.226360559463501,
      "learning_rate": 3.392646072261445e-07,
      "loss": 0.785,
      "step": 31741
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4559905529022217,
      "learning_rate": 3.386256908283551e-07,
      "loss": 0.4481,
      "step": 31742
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6567808389663696,
      "learning_rate": 3.379867744305658e-07,
      "loss": 0.8793,
      "step": 31743
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.586069107055664,
      "learning_rate": 3.3734785803277647e-07,
      "loss": 1.3756,
      "step": 31744
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0483057498931885,
      "learning_rate": 3.367089416349871e-07,
      "loss": 0.8999,
      "step": 31745
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0586228370666504,
      "learning_rate": 3.3607002523719776e-07,
      "loss": 1.113,
      "step": 31746
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.097879409790039,
      "learning_rate": 3.354311088394084e-07,
      "loss": 0.7204,
      "step": 31747
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.267507553100586,
      "learning_rate": 3.3479219244161906e-07,
      "loss": 0.7,
      "step": 31748
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.357574224472046,
      "learning_rate": 3.341532760438297e-07,
      "loss": 1.3308,
      "step": 31749
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3633222579956055,
      "learning_rate": 3.3351435964604035e-07,
      "loss": 0.7549,
      "step": 31750
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7674517631530762,
      "learning_rate": 3.32875443248251e-07,
      "loss": 0.791,
      "step": 31751
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.756652593612671,
      "learning_rate": 3.322365268504617e-07,
      "loss": 0.7724,
      "step": 31752
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.871971845626831,
      "learning_rate": 3.3159761045267224e-07,
      "loss": 0.9045,
      "step": 31753
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6841182708740234,
      "learning_rate": 3.3095869405488294e-07,
      "loss": 0.8386,
      "step": 31754
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6497422456741333,
      "learning_rate": 3.303197776570936e-07,
      "loss": 0.8718,
      "step": 31755
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6609865427017212,
      "learning_rate": 3.2968086125930423e-07,
      "loss": 0.3626,
      "step": 31756
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1346757411956787,
      "learning_rate": 3.290419448615149e-07,
      "loss": 0.7836,
      "step": 31757
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.15315580368042,
      "learning_rate": 3.284030284637255e-07,
      "loss": 0.5793,
      "step": 31758
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4968379735946655,
      "learning_rate": 3.2776411206593617e-07,
      "loss": 0.664,
      "step": 31759
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3095992803573608,
      "learning_rate": 3.271251956681468e-07,
      "loss": 1.0452,
      "step": 31760
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.564784049987793,
      "learning_rate": 3.2648627927035747e-07,
      "loss": 0.6141,
      "step": 31761
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.118328332901001,
      "learning_rate": 3.2584736287256817e-07,
      "loss": 0.9772,
      "step": 31762
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.5398945808410645,
      "learning_rate": 3.252084464747788e-07,
      "loss": 0.7012,
      "step": 31763
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8124029636383057,
      "learning_rate": 3.2456953007698946e-07,
      "loss": 0.8748,
      "step": 31764
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8232741355895996,
      "learning_rate": 3.239306136792001e-07,
      "loss": 0.8933,
      "step": 31765
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.299350261688232,
      "learning_rate": 3.2329169728141076e-07,
      "loss": 0.938,
      "step": 31766
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6886719465255737,
      "learning_rate": 3.226527808836214e-07,
      "loss": 0.7436,
      "step": 31767
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.009920597076416,
      "learning_rate": 3.2201386448583205e-07,
      "loss": 0.8109,
      "step": 31768
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8739851713180542,
      "learning_rate": 3.213749480880427e-07,
      "loss": 0.9037,
      "step": 31769
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8157093524932861,
      "learning_rate": 3.207360316902534e-07,
      "loss": 0.7823,
      "step": 31770
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.9240851402282715,
      "learning_rate": 3.2009711529246404e-07,
      "loss": 0.6132,
      "step": 31771
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.7191855907440186,
      "learning_rate": 3.194581988946747e-07,
      "loss": 0.3896,
      "step": 31772
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8006621599197388,
      "learning_rate": 3.1881928249688534e-07,
      "loss": 0.8301,
      "step": 31773
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9104949235916138,
      "learning_rate": 3.18180366099096e-07,
      "loss": 0.9654,
      "step": 31774
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.8012535572052,
      "learning_rate": 3.1754144970130663e-07,
      "loss": 0.6952,
      "step": 31775
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8422693014144897,
      "learning_rate": 3.169025333035173e-07,
      "loss": 0.8846,
      "step": 31776
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6324489116668701,
      "learning_rate": 3.1626361690572787e-07,
      "loss": 0.7657,
      "step": 31777
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6133445501327515,
      "learning_rate": 3.156247005079385e-07,
      "loss": 0.8553,
      "step": 31778
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.7729926109313965,
      "learning_rate": 3.1498578411014917e-07,
      "loss": 0.7624,
      "step": 31779
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7639422416687012,
      "learning_rate": 3.143468677123598e-07,
      "loss": 0.7703,
      "step": 31780
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1603689193725586,
      "learning_rate": 3.137079513145705e-07,
      "loss": 0.966,
      "step": 31781
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7202788591384888,
      "learning_rate": 3.1306903491678116e-07,
      "loss": 0.8882,
      "step": 31782
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1909091472625732,
      "learning_rate": 3.124301185189918e-07,
      "loss": 0.8441,
      "step": 31783
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.362950801849365,
      "learning_rate": 3.1179120212120245e-07,
      "loss": 1.36,
      "step": 31784
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.81601619720459,
      "learning_rate": 3.111522857234131e-07,
      "loss": 1.2543,
      "step": 31785
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9389945268630981,
      "learning_rate": 3.1051336932562375e-07,
      "loss": 0.8264,
      "step": 31786
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8833351135253906,
      "learning_rate": 3.098744529278344e-07,
      "loss": 0.8832,
      "step": 31787
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.077439546585083,
      "learning_rate": 3.0923553653004504e-07,
      "loss": 1.0997,
      "step": 31788
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.689652681350708,
      "learning_rate": 3.0859662013225574e-07,
      "loss": 0.4509,
      "step": 31789
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0244715213775635,
      "learning_rate": 3.079577037344664e-07,
      "loss": 0.9229,
      "step": 31790
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3362095355987549,
      "learning_rate": 3.0731878733667704e-07,
      "loss": 0.3312,
      "step": 31791
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.702436923980713,
      "learning_rate": 3.066798709388877e-07,
      "loss": 0.7577,
      "step": 31792
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4419715404510498,
      "learning_rate": 3.0604095454109833e-07,
      "loss": 0.3978,
      "step": 31793
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.258512496948242,
      "learning_rate": 3.05402038143309e-07,
      "loss": 0.9655,
      "step": 31794
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.09609317779541,
      "learning_rate": 3.047631217455196e-07,
      "loss": 1.2564,
      "step": 31795
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8316593170166016,
      "learning_rate": 3.0412420534773027e-07,
      "loss": 0.9078,
      "step": 31796
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6731138229370117,
      "learning_rate": 3.034852889499409e-07,
      "loss": 0.7055,
      "step": 31797
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.336623668670654,
      "learning_rate": 3.0284637255215157e-07,
      "loss": 1.1801,
      "step": 31798
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4716132879257202,
      "learning_rate": 3.022074561543622e-07,
      "loss": 0.7701,
      "step": 31799
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.401657223701477,
      "learning_rate": 3.0156853975657286e-07,
      "loss": 0.3439,
      "step": 31800
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8730965852737427,
      "learning_rate": 3.009296233587835e-07,
      "loss": 0.8217,
      "step": 31801
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8833237886428833,
      "learning_rate": 3.002907069609942e-07,
      "loss": 0.9087,
      "step": 31802
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7570984363555908,
      "learning_rate": 2.9965179056320485e-07,
      "loss": 0.8335,
      "step": 31803
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9177685976028442,
      "learning_rate": 2.990128741654155e-07,
      "loss": 0.8965,
      "step": 31804
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4630852937698364,
      "learning_rate": 2.9837395776762615e-07,
      "loss": 0.6851,
      "step": 31805
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9061410427093506,
      "learning_rate": 2.977350413698368e-07,
      "loss": 0.8585,
      "step": 31806
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.665804147720337,
      "learning_rate": 2.9709612497204744e-07,
      "loss": 0.8031,
      "step": 31807
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3250465393066406,
      "learning_rate": 2.964572085742581e-07,
      "loss": 0.9612,
      "step": 31808
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6312251091003418,
      "learning_rate": 2.9581829217646874e-07,
      "loss": 0.8286,
      "step": 31809
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.9750025272369385,
      "learning_rate": 2.951793757786794e-07,
      "loss": 0.9766,
      "step": 31810
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4145007133483887,
      "learning_rate": 2.9454045938089003e-07,
      "loss": 0.7428,
      "step": 31811
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.6775994300842285,
      "learning_rate": 2.939015429831007e-07,
      "loss": 0.7664,
      "step": 31812
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2792437076568604,
      "learning_rate": 2.932626265853113e-07,
      "loss": 0.3468,
      "step": 31813
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.125626564025879,
      "learning_rate": 2.9262371018752197e-07,
      "loss": 0.866,
      "step": 31814
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8009347915649414,
      "learning_rate": 2.919847937897326e-07,
      "loss": 0.8274,
      "step": 31815
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6489306688308716,
      "learning_rate": 2.913458773919433e-07,
      "loss": 0.8046,
      "step": 31816
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.3363234996795654,
      "learning_rate": 2.9070696099415397e-07,
      "loss": 0.9502,
      "step": 31817
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7058566808700562,
      "learning_rate": 2.900680445963646e-07,
      "loss": 0.8239,
      "step": 31818
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.831532597541809,
      "learning_rate": 2.8942912819857526e-07,
      "loss": 0.8438,
      "step": 31819
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0417802333831787,
      "learning_rate": 2.8879021180078585e-07,
      "loss": 0.8137,
      "step": 31820
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8774367570877075,
      "learning_rate": 2.8815129540299655e-07,
      "loss": 0.7632,
      "step": 31821
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.163454294204712,
      "learning_rate": 2.875123790052072e-07,
      "loss": 0.8959,
      "step": 31822
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.094167947769165,
      "learning_rate": 2.8687346260741785e-07,
      "loss": 0.6355,
      "step": 31823
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5890473127365112,
      "learning_rate": 2.862345462096285e-07,
      "loss": 0.7217,
      "step": 31824
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4644412994384766,
      "learning_rate": 2.8559562981183914e-07,
      "loss": 0.7279,
      "step": 31825
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8566699028015137,
      "learning_rate": 2.849567134140498e-07,
      "loss": 0.8186,
      "step": 31826
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1692018508911133,
      "learning_rate": 2.8431779701626044e-07,
      "loss": 0.9193,
      "step": 31827
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1507420539855957,
      "learning_rate": 2.836788806184711e-07,
      "loss": 0.8119,
      "step": 31828
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9982357025146484,
      "learning_rate": 2.830399642206818e-07,
      "loss": 0.7549,
      "step": 31829
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.450183629989624,
      "learning_rate": 2.8240104782289243e-07,
      "loss": 1.0558,
      "step": 31830
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5897228717803955,
      "learning_rate": 2.817621314251031e-07,
      "loss": 0.7942,
      "step": 31831
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5038554668426514,
      "learning_rate": 2.8112321502731367e-07,
      "loss": 0.8403,
      "step": 31832
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8369956016540527,
      "learning_rate": 2.804842986295243e-07,
      "loss": 0.7309,
      "step": 31833
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6314817667007446,
      "learning_rate": 2.79845382231735e-07,
      "loss": 0.8091,
      "step": 31834
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.1559464931488037,
      "learning_rate": 2.7920646583394566e-07,
      "loss": 0.8453,
      "step": 31835
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.500577211380005,
      "learning_rate": 2.785675494361563e-07,
      "loss": 0.9503,
      "step": 31836
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.60544753074646,
      "learning_rate": 2.7792863303836696e-07,
      "loss": 1.1141,
      "step": 31837
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.0612359046936035,
      "learning_rate": 2.772897166405776e-07,
      "loss": 0.923,
      "step": 31838
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6826921701431274,
      "learning_rate": 2.7665080024278825e-07,
      "loss": 0.8944,
      "step": 31839
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5309454202651978,
      "learning_rate": 2.760118838449989e-07,
      "loss": 0.7987,
      "step": 31840
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.440451979637146,
      "learning_rate": 2.7537296744720955e-07,
      "loss": 0.3268,
      "step": 31841
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.043213129043579,
      "learning_rate": 2.747340510494202e-07,
      "loss": 0.7516,
      "step": 31842
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8481669425964355,
      "learning_rate": 2.740951346516309e-07,
      "loss": 0.9728,
      "step": 31843
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.435387372970581,
      "learning_rate": 2.734562182538415e-07,
      "loss": 0.9792,
      "step": 31844
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3292731046676636,
      "learning_rate": 2.7281730185605213e-07,
      "loss": 0.3711,
      "step": 31845
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8817403316497803,
      "learning_rate": 2.721783854582628e-07,
      "loss": 1.0434,
      "step": 31846
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.846152901649475,
      "learning_rate": 2.7153946906047343e-07,
      "loss": 0.7607,
      "step": 31847
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7254056930541992,
      "learning_rate": 2.7090055266268413e-07,
      "loss": 0.7763,
      "step": 31848
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.171677589416504,
      "learning_rate": 2.702616362648948e-07,
      "loss": 0.887,
      "step": 31849
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.591431140899658,
      "learning_rate": 2.696227198671054e-07,
      "loss": 0.9663,
      "step": 31850
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.707663536071777,
      "learning_rate": 2.6898380346931607e-07,
      "loss": 1.1615,
      "step": 31851
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.60831618309021,
      "learning_rate": 2.683448870715267e-07,
      "loss": 0.7917,
      "step": 31852
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.776041865348816,
      "learning_rate": 2.6770597067373736e-07,
      "loss": 0.8916,
      "step": 31853
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0385494232177734,
      "learning_rate": 2.67067054275948e-07,
      "loss": 0.7406,
      "step": 31854
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5935527086257935,
      "learning_rate": 2.6642813787815866e-07,
      "loss": 0.8533,
      "step": 31855
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.761493444442749,
      "learning_rate": 2.657892214803693e-07,
      "loss": 0.8112,
      "step": 31856
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5805073976516724,
      "learning_rate": 2.6515030508257995e-07,
      "loss": 0.7771,
      "step": 31857
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.050885200500488,
      "learning_rate": 2.645113886847906e-07,
      "loss": 0.9859,
      "step": 31858
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9022334814071655,
      "learning_rate": 2.6387247228700125e-07,
      "loss": 0.8398,
      "step": 31859
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7022294998168945,
      "learning_rate": 2.632335558892119e-07,
      "loss": 0.8438,
      "step": 31860
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.413720965385437,
      "learning_rate": 2.625946394914226e-07,
      "loss": 0.4543,
      "step": 31861
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.9914376735687256,
      "learning_rate": 2.6195572309363324e-07,
      "loss": 0.9174,
      "step": 31862
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9348981380462646,
      "learning_rate": 2.613168066958439e-07,
      "loss": 0.6867,
      "step": 31863
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7656424045562744,
      "learning_rate": 2.6067789029805453e-07,
      "loss": 1.0272,
      "step": 31864
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.17294979095459,
      "learning_rate": 2.600389739002652e-07,
      "loss": 0.7099,
      "step": 31865
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.6496713161468506,
      "learning_rate": 2.5940005750247583e-07,
      "loss": 0.8154,
      "step": 31866
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5476100444793701,
      "learning_rate": 2.587611411046865e-07,
      "loss": 0.7636,
      "step": 31867
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.823180913925171,
      "learning_rate": 2.581222247068971e-07,
      "loss": 0.8816,
      "step": 31868
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8502204418182373,
      "learning_rate": 2.5748330830910777e-07,
      "loss": 0.7816,
      "step": 31869
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.626335859298706,
      "learning_rate": 2.568443919113184e-07,
      "loss": 0.8799,
      "step": 31870
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.213440418243408,
      "learning_rate": 2.5620547551352906e-07,
      "loss": 0.833,
      "step": 31871
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.043954610824585,
      "learning_rate": 2.555665591157397e-07,
      "loss": 0.8598,
      "step": 31872
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.334804058074951,
      "learning_rate": 2.5492764271795036e-07,
      "loss": 1.0156,
      "step": 31873
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2770886421203613,
      "learning_rate": 2.54288726320161e-07,
      "loss": 0.9008,
      "step": 31874
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6999647617340088,
      "learning_rate": 2.536498099223717e-07,
      "loss": 0.7803,
      "step": 31875
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8755779266357422,
      "learning_rate": 2.5301089352458235e-07,
      "loss": 0.7544,
      "step": 31876
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.245694160461426,
      "learning_rate": 2.52371977126793e-07,
      "loss": 1.2759,
      "step": 31877
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2604806423187256,
      "learning_rate": 2.5173306072900365e-07,
      "loss": 0.7546,
      "step": 31878
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9414223432540894,
      "learning_rate": 2.510941443312143e-07,
      "loss": 0.9445,
      "step": 31879
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7365119457244873,
      "learning_rate": 2.5045522793342494e-07,
      "loss": 0.9214,
      "step": 31880
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.054741144180298,
      "learning_rate": 2.498163115356356e-07,
      "loss": 0.8134,
      "step": 31881
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9018898010253906,
      "learning_rate": 2.4917739513784623e-07,
      "loss": 0.8824,
      "step": 31882
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.24800705909729,
      "learning_rate": 2.485384787400569e-07,
      "loss": 0.7986,
      "step": 31883
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.12400484085083,
      "learning_rate": 2.4789956234226753e-07,
      "loss": 1.0418,
      "step": 31884
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.919777274131775,
      "learning_rate": 2.472606459444782e-07,
      "loss": 0.8941,
      "step": 31885
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8725366592407227,
      "learning_rate": 2.466217295466888e-07,
      "loss": 0.8207,
      "step": 31886
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7080341577529907,
      "learning_rate": 2.4598281314889947e-07,
      "loss": 0.7126,
      "step": 31887
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8215906620025635,
      "learning_rate": 2.4534389675111017e-07,
      "loss": 0.8335,
      "step": 31888
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.078733205795288,
      "learning_rate": 2.447049803533208e-07,
      "loss": 0.9978,
      "step": 31889
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.689208745956421,
      "learning_rate": 2.4406606395553146e-07,
      "loss": 0.9665,
      "step": 31890
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7653541564941406,
      "learning_rate": 2.434271475577421e-07,
      "loss": 0.8761,
      "step": 31891
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0237417221069336,
      "learning_rate": 2.427882311599527e-07,
      "loss": 0.5714,
      "step": 31892
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8324203491210938,
      "learning_rate": 2.421493147621634e-07,
      "loss": 0.7988,
      "step": 31893
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6870770454406738,
      "learning_rate": 2.4151039836437405e-07,
      "loss": 0.7905,
      "step": 31894
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.361948847770691,
      "learning_rate": 2.408714819665847e-07,
      "loss": 0.3769,
      "step": 31895
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8444650173187256,
      "learning_rate": 2.4023256556879534e-07,
      "loss": 0.9566,
      "step": 31896
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3128299713134766,
      "learning_rate": 2.39593649171006e-07,
      "loss": 0.9227,
      "step": 31897
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.264408111572266,
      "learning_rate": 2.3895473277321664e-07,
      "loss": 0.8129,
      "step": 31898
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2609753608703613,
      "learning_rate": 2.383158163754273e-07,
      "loss": 0.9297,
      "step": 31899
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6614458560943604,
      "learning_rate": 2.3767689997763796e-07,
      "loss": 0.7048,
      "step": 31900
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.123837471008301,
      "learning_rate": 2.370379835798486e-07,
      "loss": 0.8218,
      "step": 31901
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6002365350723267,
      "learning_rate": 2.3639906718205925e-07,
      "loss": 0.9148,
      "step": 31902
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8356777429580688,
      "learning_rate": 2.357601507842699e-07,
      "loss": 1.1237,
      "step": 31903
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8135218620300293,
      "learning_rate": 2.3512123438648052e-07,
      "loss": 0.7914,
      "step": 31904
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1741271018981934,
      "learning_rate": 2.344823179886912e-07,
      "loss": 0.7819,
      "step": 31905
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.170283317565918,
      "learning_rate": 2.3384340159090184e-07,
      "loss": 0.8884,
      "step": 31906
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.710456371307373,
      "learning_rate": 2.332044851931125e-07,
      "loss": 0.7478,
      "step": 31907
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.943251132965088,
      "learning_rate": 2.3256556879532314e-07,
      "loss": 1.3502,
      "step": 31908
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.139700412750244,
      "learning_rate": 2.319266523975338e-07,
      "loss": 0.5748,
      "step": 31909
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.5416676998138428,
      "learning_rate": 2.3128773599974446e-07,
      "loss": 0.7102,
      "step": 31910
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.696542501449585,
      "learning_rate": 2.306488196019551e-07,
      "loss": 0.9472,
      "step": 31911
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.358067035675049,
      "learning_rate": 2.3000990320416575e-07,
      "loss": 1.2711,
      "step": 31912
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9964145421981812,
      "learning_rate": 2.2937098680637642e-07,
      "loss": 0.5415,
      "step": 31913
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0097408294677734,
      "learning_rate": 2.2873207040858707e-07,
      "loss": 0.9035,
      "step": 31914
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4533356428146362,
      "learning_rate": 2.2809315401079772e-07,
      "loss": 0.7971,
      "step": 31915
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3479933738708496,
      "learning_rate": 2.2745423761300834e-07,
      "loss": 0.8291,
      "step": 31916
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.965040922164917,
      "learning_rate": 2.2681532121521898e-07,
      "loss": 1.1754,
      "step": 31917
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.228472948074341,
      "learning_rate": 2.2617640481742966e-07,
      "loss": 0.6099,
      "step": 31918
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7391542196273804,
      "learning_rate": 2.255374884196403e-07,
      "loss": 0.7458,
      "step": 31919
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.4275529384613037,
      "learning_rate": 2.2489857202185095e-07,
      "loss": 0.9083,
      "step": 31920
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1230177879333496,
      "learning_rate": 2.242596556240616e-07,
      "loss": 0.9266,
      "step": 31921
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.882695198059082,
      "learning_rate": 2.2362073922627227e-07,
      "loss": 1.0546,
      "step": 31922
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.209890842437744,
      "learning_rate": 2.2298182282848292e-07,
      "loss": 1.3732,
      "step": 31923
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.177316188812256,
      "learning_rate": 2.2234290643069357e-07,
      "loss": 0.7735,
      "step": 31924
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8233188390731812,
      "learning_rate": 2.2170399003290421e-07,
      "loss": 0.8176,
      "step": 31925
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7032055854797363,
      "learning_rate": 2.210650736351149e-07,
      "loss": 0.7981,
      "step": 31926
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.5320982933044434,
      "learning_rate": 2.2042615723732553e-07,
      "loss": 0.8965,
      "step": 31927
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0860049724578857,
      "learning_rate": 2.1978724083953615e-07,
      "loss": 0.7834,
      "step": 31928
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1367626190185547,
      "learning_rate": 2.191483244417468e-07,
      "loss": 0.8781,
      "step": 31929
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0161075592041016,
      "learning_rate": 2.1850940804395745e-07,
      "loss": 0.8781,
      "step": 31930
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0435256958007812,
      "learning_rate": 2.178704916461681e-07,
      "loss": 0.7505,
      "step": 31931
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4934126138687134,
      "learning_rate": 2.1723157524837877e-07,
      "loss": 0.8211,
      "step": 31932
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.9033172130584717,
      "learning_rate": 2.1659265885058942e-07,
      "loss": 0.7671,
      "step": 31933
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3303608894348145,
      "learning_rate": 2.1595374245280006e-07,
      "loss": 0.8434,
      "step": 31934
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9448957443237305,
      "learning_rate": 2.153148260550107e-07,
      "loss": 0.6728,
      "step": 31935
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7959438562393188,
      "learning_rate": 2.1467590965722138e-07,
      "loss": 0.7493,
      "step": 31936
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.144993305206299,
      "learning_rate": 2.1403699325943203e-07,
      "loss": 0.8489,
      "step": 31937
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.468717098236084,
      "learning_rate": 2.1339807686164268e-07,
      "loss": 0.6427,
      "step": 31938
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.079007863998413,
      "learning_rate": 2.1275916046385332e-07,
      "loss": 0.8939,
      "step": 31939
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1528475284576416,
      "learning_rate": 2.1212024406606395e-07,
      "loss": 0.8526,
      "step": 31940
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2955665588378906,
      "learning_rate": 2.1148132766827462e-07,
      "loss": 0.9848,
      "step": 31941
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.849735140800476,
      "learning_rate": 2.1084241127048527e-07,
      "loss": 0.8357,
      "step": 31942
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8384220600128174,
      "learning_rate": 2.102034948726959e-07,
      "loss": 0.8591,
      "step": 31943
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0363118648529053,
      "learning_rate": 2.0956457847490656e-07,
      "loss": 0.8062,
      "step": 31944
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.73026967048645,
      "learning_rate": 2.0892566207711723e-07,
      "loss": 1.4021,
      "step": 31945
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.151104688644409,
      "learning_rate": 2.0828674567932788e-07,
      "loss": 0.7611,
      "step": 31946
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6854311227798462,
      "learning_rate": 2.0764782928153853e-07,
      "loss": 0.8476,
      "step": 31947
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9024999141693115,
      "learning_rate": 2.0700891288374917e-07,
      "loss": 0.9171,
      "step": 31948
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1524436473846436,
      "learning_rate": 2.0636999648595985e-07,
      "loss": 0.7169,
      "step": 31949
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7979071140289307,
      "learning_rate": 2.057310800881705e-07,
      "loss": 1.0797,
      "step": 31950
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.023491859436035,
      "learning_rate": 2.0509216369038114e-07,
      "loss": 0.852,
      "step": 31951
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.282963514328003,
      "learning_rate": 2.0445324729259176e-07,
      "loss": 0.7451,
      "step": 31952
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0877325534820557,
      "learning_rate": 2.038143308948024e-07,
      "loss": 1.0319,
      "step": 31953
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6653884649276733,
      "learning_rate": 2.0317541449701308e-07,
      "loss": 0.7361,
      "step": 31954
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4295597076416016,
      "learning_rate": 2.0253649809922373e-07,
      "loss": 0.4009,
      "step": 31955
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.6536672115325928,
      "learning_rate": 2.0189758170143438e-07,
      "loss": 0.7769,
      "step": 31956
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.710317611694336,
      "learning_rate": 2.0125866530364502e-07,
      "loss": 0.7832,
      "step": 31957
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5906972885131836,
      "learning_rate": 2.006197489058557e-07,
      "loss": 0.3908,
      "step": 31958
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0670032501220703,
      "learning_rate": 1.9998083250806634e-07,
      "loss": 0.8758,
      "step": 31959
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7027199268341064,
      "learning_rate": 1.99341916110277e-07,
      "loss": 0.9448,
      "step": 31960
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.6224656105041504,
      "learning_rate": 1.9870299971248764e-07,
      "loss": 1.0001,
      "step": 31961
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.297884464263916,
      "learning_rate": 1.980640833146983e-07,
      "loss": 1.2678,
      "step": 31962
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8510538339614868,
      "learning_rate": 1.9742516691690896e-07,
      "loss": 0.785,
      "step": 31963
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.556615948677063,
      "learning_rate": 1.967862505191196e-07,
      "loss": 0.8241,
      "step": 31964
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.122925281524658,
      "learning_rate": 1.9614733412133023e-07,
      "loss": 0.8436,
      "step": 31965
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.894460916519165,
      "learning_rate": 1.9550841772354087e-07,
      "loss": 0.8147,
      "step": 31966
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8095886707305908,
      "learning_rate": 1.9486950132575152e-07,
      "loss": 0.6381,
      "step": 31967
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.710647463798523,
      "learning_rate": 1.942305849279622e-07,
      "loss": 0.7278,
      "step": 31968
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.164778232574463,
      "learning_rate": 1.9359166853017284e-07,
      "loss": 1.0442,
      "step": 31969
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8116276264190674,
      "learning_rate": 1.929527521323835e-07,
      "loss": 0.7616,
      "step": 31970
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5673975944519043,
      "learning_rate": 1.9231383573459414e-07,
      "loss": 0.7937,
      "step": 31971
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.483591079711914,
      "learning_rate": 1.916749193368048e-07,
      "loss": 0.8787,
      "step": 31972
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9055404663085938,
      "learning_rate": 1.9103600293901546e-07,
      "loss": 1.2326,
      "step": 31973
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8950361013412476,
      "learning_rate": 1.903970865412261e-07,
      "loss": 0.901,
      "step": 31974
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6713619232177734,
      "learning_rate": 1.8975817014343675e-07,
      "loss": 0.7188,
      "step": 31975
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3040869235992432,
      "learning_rate": 1.8911925374564742e-07,
      "loss": 0.2774,
      "step": 31976
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.612383484840393,
      "learning_rate": 1.8848033734785804e-07,
      "loss": 0.8061,
      "step": 31977
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7509853839874268,
      "learning_rate": 1.878414209500687e-07,
      "loss": 0.6165,
      "step": 31978
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.805867910385132,
      "learning_rate": 1.8720250455227934e-07,
      "loss": 0.9614,
      "step": 31979
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.554103374481201,
      "learning_rate": 1.8656358815448999e-07,
      "loss": 1.0184,
      "step": 31980
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.035682201385498,
      "learning_rate": 1.8592467175670066e-07,
      "loss": 0.8627,
      "step": 31981
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2574033737182617,
      "learning_rate": 1.852857553589113e-07,
      "loss": 0.9036,
      "step": 31982
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.046718597412109,
      "learning_rate": 1.8464683896112195e-07,
      "loss": 0.8767,
      "step": 31983
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.29263973236084,
      "learning_rate": 1.840079225633326e-07,
      "loss": 1.5129,
      "step": 31984
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.01774001121521,
      "learning_rate": 1.8336900616554327e-07,
      "loss": 0.7733,
      "step": 31985
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8072956800460815,
      "learning_rate": 1.8273008976775392e-07,
      "loss": 0.8677,
      "step": 31986
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4860249757766724,
      "learning_rate": 1.8209117336996457e-07,
      "loss": 0.7451,
      "step": 31987
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9670124053955078,
      "learning_rate": 1.8145225697217521e-07,
      "loss": 0.869,
      "step": 31988
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.433691143989563,
      "learning_rate": 1.8081334057438583e-07,
      "loss": 0.4788,
      "step": 31989
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.912724256515503,
      "learning_rate": 1.8017442417659648e-07,
      "loss": 1.1398,
      "step": 31990
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1689724922180176,
      "learning_rate": 1.7953550777880716e-07,
      "loss": 1.1391,
      "step": 31991
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7381880283355713,
      "learning_rate": 1.788965913810178e-07,
      "loss": 0.7496,
      "step": 31992
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.4553136825561523,
      "learning_rate": 1.7825767498322845e-07,
      "loss": 1.1057,
      "step": 31993
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.149495601654053,
      "learning_rate": 1.776187585854391e-07,
      "loss": 1.7877,
      "step": 31994
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7107017040252686,
      "learning_rate": 1.7697984218764977e-07,
      "loss": 0.7622,
      "step": 31995
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.346980333328247,
      "learning_rate": 1.7634092578986042e-07,
      "loss": 0.8789,
      "step": 31996
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5729328393936157,
      "learning_rate": 1.7570200939207106e-07,
      "loss": 0.7836,
      "step": 31997
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.716068148612976,
      "learning_rate": 1.750630929942817e-07,
      "loss": 0.7337,
      "step": 31998
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8022969961166382,
      "learning_rate": 1.7442417659649238e-07,
      "loss": 0.8869,
      "step": 31999
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1245434284210205,
      "learning_rate": 1.7378526019870303e-07,
      "loss": 0.7957,
      "step": 32000
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7422940731048584,
      "learning_rate": 1.7314634380091365e-07,
      "loss": 0.9454,
      "step": 32001
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8819546699523926,
      "learning_rate": 1.725074274031243e-07,
      "loss": 0.8132,
      "step": 32002
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6919209957122803,
      "learning_rate": 1.7186851100533495e-07,
      "loss": 0.8541,
      "step": 32003
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0423760414123535,
      "learning_rate": 1.7122959460754562e-07,
      "loss": 0.7528,
      "step": 32004
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8970736265182495,
      "learning_rate": 1.7059067820975627e-07,
      "loss": 0.704,
      "step": 32005
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.367826461791992,
      "learning_rate": 1.6995176181196691e-07,
      "loss": 1.0093,
      "step": 32006
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6707464456558228,
      "learning_rate": 1.6931284541417756e-07,
      "loss": 1.1643,
      "step": 32007
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.840650200843811,
      "learning_rate": 1.6867392901638823e-07,
      "loss": 1.0461,
      "step": 32008
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8021024465560913,
      "learning_rate": 1.6803501261859888e-07,
      "loss": 1.1114,
      "step": 32009
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8360899686813354,
      "learning_rate": 1.6739609622080953e-07,
      "loss": 1.0895,
      "step": 32010
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.971809983253479,
      "learning_rate": 1.6675717982302018e-07,
      "loss": 0.8705,
      "step": 32011
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1073005199432373,
      "learning_rate": 1.6611826342523085e-07,
      "loss": 0.9249,
      "step": 32012
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7680449485778809,
      "learning_rate": 1.6547934702744147e-07,
      "loss": 0.8746,
      "step": 32013
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2831733226776123,
      "learning_rate": 1.6484043062965212e-07,
      "loss": 0.8651,
      "step": 32014
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1931653022766113,
      "learning_rate": 1.6420151423186276e-07,
      "loss": 0.8192,
      "step": 32015
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6659893989562988,
      "learning_rate": 1.635625978340734e-07,
      "loss": 0.8464,
      "step": 32016
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1938865184783936,
      "learning_rate": 1.6292368143628408e-07,
      "loss": 0.8909,
      "step": 32017
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.980416774749756,
      "learning_rate": 1.6228476503849473e-07,
      "loss": 1.2154,
      "step": 32018
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7333686351776123,
      "learning_rate": 1.6164584864070538e-07,
      "loss": 0.8658,
      "step": 32019
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.6328277587890625,
      "learning_rate": 1.6100693224291602e-07,
      "loss": 1.7052,
      "step": 32020
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7316628694534302,
      "learning_rate": 1.603680158451267e-07,
      "loss": 0.8098,
      "step": 32021
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2553486824035645,
      "learning_rate": 1.5972909944733735e-07,
      "loss": 0.9258,
      "step": 32022
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7843105792999268,
      "learning_rate": 1.59090183049548e-07,
      "loss": 0.7951,
      "step": 32023
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3277249336242676,
      "learning_rate": 1.5845126665175864e-07,
      "loss": 0.7502,
      "step": 32024
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6908260583877563,
      "learning_rate": 1.5781235025396926e-07,
      "loss": 0.7034,
      "step": 32025
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.3849241733551025,
      "learning_rate": 1.571734338561799e-07,
      "loss": 0.6515,
      "step": 32026
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7167235612869263,
      "learning_rate": 1.5653451745839058e-07,
      "loss": 0.7638,
      "step": 32027
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0120913982391357,
      "learning_rate": 1.5589560106060123e-07,
      "loss": 1.0823,
      "step": 32028
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.209824323654175,
      "learning_rate": 1.5525668466281187e-07,
      "loss": 0.9308,
      "step": 32029
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9675341844558716,
      "learning_rate": 1.5461776826502252e-07,
      "loss": 0.8626,
      "step": 32030
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.821488618850708,
      "learning_rate": 1.539788518672332e-07,
      "loss": 0.8166,
      "step": 32031
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.513782262802124,
      "learning_rate": 1.5333993546944384e-07,
      "loss": 0.9849,
      "step": 32032
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.5875496864318848,
      "learning_rate": 1.527010190716545e-07,
      "loss": 0.806,
      "step": 32033
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8917851448059082,
      "learning_rate": 1.5206210267386514e-07,
      "loss": 0.8046,
      "step": 32034
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.894124150276184,
      "learning_rate": 1.5142318627607578e-07,
      "loss": 0.839,
      "step": 32035
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.715140342712402,
      "learning_rate": 1.5078426987828643e-07,
      "loss": 1.4289,
      "step": 32036
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7393921613693237,
      "learning_rate": 1.501453534804971e-07,
      "loss": 0.7936,
      "step": 32037
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.661651849746704,
      "learning_rate": 1.4950643708270775e-07,
      "loss": 0.8014,
      "step": 32038
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.674389362335205,
      "learning_rate": 1.488675206849184e-07,
      "loss": 0.7485,
      "step": 32039
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5893183946609497,
      "learning_rate": 1.4822860428712904e-07,
      "loss": 0.8992,
      "step": 32040
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.976414442062378,
      "learning_rate": 1.475896878893397e-07,
      "loss": 0.6815,
      "step": 32041
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7375503778457642,
      "learning_rate": 1.4695077149155034e-07,
      "loss": 0.7683,
      "step": 32042
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7680374383926392,
      "learning_rate": 1.4631185509376099e-07,
      "loss": 0.9062,
      "step": 32043
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8255937099456787,
      "learning_rate": 1.4567293869597166e-07,
      "loss": 0.9061,
      "step": 32044
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.758385419845581,
      "learning_rate": 1.450340222981823e-07,
      "loss": 1.0371,
      "step": 32045
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7258633375167847,
      "learning_rate": 1.4439510590039293e-07,
      "loss": 0.8197,
      "step": 32046
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9329301118850708,
      "learning_rate": 1.437561895026036e-07,
      "loss": 0.7874,
      "step": 32047
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.802451252937317,
      "learning_rate": 1.4311727310481425e-07,
      "loss": 0.8473,
      "step": 32048
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7947607040405273,
      "learning_rate": 1.424783567070249e-07,
      "loss": 0.8865,
      "step": 32049
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8553075790405273,
      "learning_rate": 1.4183944030923554e-07,
      "loss": 0.7799,
      "step": 32050
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.829848051071167,
      "learning_rate": 1.4120052391144621e-07,
      "loss": 0.8111,
      "step": 32051
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0871808528900146,
      "learning_rate": 1.4056160751365684e-07,
      "loss": 0.7989,
      "step": 32052
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.421684503555298,
      "learning_rate": 1.399226911158675e-07,
      "loss": 0.982,
      "step": 32053
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.0935404300689697,
      "learning_rate": 1.3928377471807816e-07,
      "loss": 0.743,
      "step": 32054
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7505345344543457,
      "learning_rate": 1.386448583202888e-07,
      "loss": 0.9251,
      "step": 32055
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7543069124221802,
      "learning_rate": 1.3800594192249945e-07,
      "loss": 0.8171,
      "step": 32056
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.210432291030884,
      "learning_rate": 1.373670255247101e-07,
      "loss": 0.7969,
      "step": 32057
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3653370141983032,
      "learning_rate": 1.3672810912692074e-07,
      "loss": 0.3623,
      "step": 32058
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5645970106124878,
      "learning_rate": 1.360891927291314e-07,
      "loss": 0.6981,
      "step": 32059
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.546473741531372,
      "learning_rate": 1.3545027633134206e-07,
      "loss": 0.8848,
      "step": 32060
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5298515558242798,
      "learning_rate": 1.348113599335527e-07,
      "loss": 0.8288,
      "step": 32061
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6138595342636108,
      "learning_rate": 1.3417244353576336e-07,
      "loss": 0.8273,
      "step": 32062
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.302312850952148,
      "learning_rate": 1.33533527137974e-07,
      "loss": 1.0861,
      "step": 32063
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0262386798858643,
      "learning_rate": 1.3289461074018465e-07,
      "loss": 0.9405,
      "step": 32064
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8954192399978638,
      "learning_rate": 1.322556943423953e-07,
      "loss": 0.812,
      "step": 32065
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9760717153549194,
      "learning_rate": 1.3161677794460595e-07,
      "loss": 0.7098,
      "step": 32066
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0754358768463135,
      "learning_rate": 1.3097786154681662e-07,
      "loss": 0.8238,
      "step": 32067
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7937616109848022,
      "learning_rate": 1.3033894514902727e-07,
      "loss": 1.1093,
      "step": 32068
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0695080757141113,
      "learning_rate": 1.2970002875123791e-07,
      "loss": 0.8657,
      "step": 32069
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.424034833908081,
      "learning_rate": 1.2906111235344856e-07,
      "loss": 0.821,
      "step": 32070
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.353787422180176,
      "learning_rate": 1.284221959556592e-07,
      "loss": 0.9458,
      "step": 32071
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7319362163543701,
      "learning_rate": 1.2778327955786985e-07,
      "loss": 0.7568,
      "step": 32072
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9382961988449097,
      "learning_rate": 1.271443631600805e-07,
      "loss": 0.777,
      "step": 32073
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1979317665100098,
      "learning_rate": 1.2650544676229118e-07,
      "loss": 0.9762,
      "step": 32074
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.632317066192627,
      "learning_rate": 1.2586653036450182e-07,
      "loss": 0.8391,
      "step": 32075
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.438770055770874,
      "learning_rate": 1.2522761396671247e-07,
      "loss": 0.6676,
      "step": 32076
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7655103206634521,
      "learning_rate": 1.2458869756892312e-07,
      "loss": 0.7923,
      "step": 32077
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.242583751678467,
      "learning_rate": 1.2394978117113376e-07,
      "loss": 1.0436,
      "step": 32078
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8192493915557861,
      "learning_rate": 1.233108647733444e-07,
      "loss": 0.831,
      "step": 32079
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.01639723777771,
      "learning_rate": 1.2267194837555508e-07,
      "loss": 0.6776,
      "step": 32080
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.43025803565979,
      "learning_rate": 1.2203303197776573e-07,
      "loss": 0.8523,
      "step": 32081
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.918096899986267,
      "learning_rate": 1.2139411557997635e-07,
      "loss": 0.7482,
      "step": 32082
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8030859231948853,
      "learning_rate": 1.2075519918218703e-07,
      "loss": 0.847,
      "step": 32083
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0672354698181152,
      "learning_rate": 1.2011628278439767e-07,
      "loss": 0.849,
      "step": 32084
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9559060335159302,
      "learning_rate": 1.1947736638660832e-07,
      "loss": 0.8616,
      "step": 32085
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.8206374645233154,
      "learning_rate": 1.1883844998881898e-07,
      "loss": 0.7857,
      "step": 32086
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.717740774154663,
      "learning_rate": 1.1819953359102963e-07,
      "loss": 0.905,
      "step": 32087
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.704822063446045,
      "learning_rate": 1.1756061719324026e-07,
      "loss": 0.8243,
      "step": 32088
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.815448760986328,
      "learning_rate": 1.1692170079545092e-07,
      "loss": 1.1724,
      "step": 32089
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7725374698638916,
      "learning_rate": 1.1628278439766157e-07,
      "loss": 0.8182,
      "step": 32090
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8021140098571777,
      "learning_rate": 1.1564386799987223e-07,
      "loss": 0.8977,
      "step": 32091
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8339756727218628,
      "learning_rate": 1.1500495160208287e-07,
      "loss": 0.9107,
      "step": 32092
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4689902067184448,
      "learning_rate": 1.1436603520429354e-07,
      "loss": 0.7423,
      "step": 32093
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3510327339172363,
      "learning_rate": 1.1372711880650417e-07,
      "loss": 0.8812,
      "step": 32094
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.25304913520813,
      "learning_rate": 1.1308820240871483e-07,
      "loss": 0.9885,
      "step": 32095
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.197625160217285,
      "learning_rate": 1.1244928601092548e-07,
      "loss": 0.6178,
      "step": 32096
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8821660280227661,
      "learning_rate": 1.1181036961313614e-07,
      "loss": 0.7896,
      "step": 32097
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.3155570030212402,
      "learning_rate": 1.1117145321534678e-07,
      "loss": 0.9146,
      "step": 32098
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5498101711273193,
      "learning_rate": 1.1053253681755744e-07,
      "loss": 0.3166,
      "step": 32099
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8418710231781006,
      "learning_rate": 1.0989362041976808e-07,
      "loss": 1.0159,
      "step": 32100
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4917337894439697,
      "learning_rate": 1.0925470402197872e-07,
      "loss": 0.8191,
      "step": 32101
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.721696376800537,
      "learning_rate": 1.0861578762418938e-07,
      "loss": 0.7826,
      "step": 32102
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0865070819854736,
      "learning_rate": 1.0797687122640003e-07,
      "loss": 0.6472,
      "step": 32103
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6664297580718994,
      "learning_rate": 1.0733795482861069e-07,
      "loss": 0.8132,
      "step": 32104
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8853298425674438,
      "learning_rate": 1.0669903843082134e-07,
      "loss": 0.8683,
      "step": 32105
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.8441147804260254,
      "learning_rate": 1.0606012203303197e-07,
      "loss": 0.916,
      "step": 32106
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0665104389190674,
      "learning_rate": 1.0542120563524263e-07,
      "loss": 0.7837,
      "step": 32107
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.305950403213501,
      "learning_rate": 1.0478228923745328e-07,
      "loss": 0.3661,
      "step": 32108
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8045743703842163,
      "learning_rate": 1.0414337283966394e-07,
      "loss": 0.8965,
      "step": 32109
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7369495630264282,
      "learning_rate": 1.0350445644187459e-07,
      "loss": 1.02,
      "step": 32110
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.65291166305542,
      "learning_rate": 1.0286554004408525e-07,
      "loss": 0.8478,
      "step": 32111
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2396693229675293,
      "learning_rate": 1.0222662364629588e-07,
      "loss": 1.1709,
      "step": 32112
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.5988497734069824,
      "learning_rate": 1.0158770724850654e-07,
      "loss": 0.8694,
      "step": 32113
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8453599214553833,
      "learning_rate": 1.0094879085071719e-07,
      "loss": 0.7198,
      "step": 32114
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.5654544830322266,
      "learning_rate": 1.0030987445292785e-07,
      "loss": 0.972,
      "step": 32115
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1205170154571533,
      "learning_rate": 9.96709580551385e-08,
      "loss": 0.8509,
      "step": 32116
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.5119240283966064,
      "learning_rate": 9.903204165734916e-08,
      "loss": 0.7454,
      "step": 32117
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6848872900009155,
      "learning_rate": 9.83931252595598e-08,
      "loss": 1.0832,
      "step": 32118
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.234292507171631,
      "learning_rate": 9.775420886177044e-08,
      "loss": 0.9154,
      "step": 32119
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0469517707824707,
      "learning_rate": 9.71152924639811e-08,
      "loss": 0.4956,
      "step": 32120
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.683212161064148,
      "learning_rate": 9.647637606619174e-08,
      "loss": 0.8206,
      "step": 32121
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.303511142730713,
      "learning_rate": 9.58374596684024e-08,
      "loss": 0.3852,
      "step": 32122
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9942065477371216,
      "learning_rate": 9.519854327061305e-08,
      "loss": 0.747,
      "step": 32123
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.895522952079773,
      "learning_rate": 9.455962687282371e-08,
      "loss": 0.9097,
      "step": 32124
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8840100765228271,
      "learning_rate": 9.392071047503435e-08,
      "loss": 0.9137,
      "step": 32125
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8338932991027832,
      "learning_rate": 9.328179407724499e-08,
      "loss": 0.3386,
      "step": 32126
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7064374685287476,
      "learning_rate": 9.264287767945565e-08,
      "loss": 0.8173,
      "step": 32127
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.028243064880371,
      "learning_rate": 9.20039612816663e-08,
      "loss": 0.7986,
      "step": 32128
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.039088726043701,
      "learning_rate": 9.136504488387696e-08,
      "loss": 0.8053,
      "step": 32129
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.380550384521484,
      "learning_rate": 9.072612848608761e-08,
      "loss": 1.1008,
      "step": 32130
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.851477026939392,
      "learning_rate": 9.008721208829824e-08,
      "loss": 0.6289,
      "step": 32131
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6449552774429321,
      "learning_rate": 8.94482956905089e-08,
      "loss": 0.834,
      "step": 32132
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6045023202896118,
      "learning_rate": 8.880937929271955e-08,
      "loss": 0.8394,
      "step": 32133
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0151865482330322,
      "learning_rate": 8.817046289493021e-08,
      "loss": 0.8717,
      "step": 32134
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.692671298980713,
      "learning_rate": 8.753154649714086e-08,
      "loss": 0.757,
      "step": 32135
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.4310343265533447,
      "learning_rate": 8.689263009935152e-08,
      "loss": 0.8573,
      "step": 32136
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.24229097366333,
      "learning_rate": 8.625371370156215e-08,
      "loss": 1.0986,
      "step": 32137
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0386557579040527,
      "learning_rate": 8.561479730377281e-08,
      "loss": 0.7841,
      "step": 32138
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.9591031074523926,
      "learning_rate": 8.497588090598346e-08,
      "loss": 1.0672,
      "step": 32139
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7654950618743896,
      "learning_rate": 8.433696450819412e-08,
      "loss": 0.9064,
      "step": 32140
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.015246629714966,
      "learning_rate": 8.369804811040476e-08,
      "loss": 1.0748,
      "step": 32141
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7676146030426025,
      "learning_rate": 8.305913171261542e-08,
      "loss": 0.7784,
      "step": 32142
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.543114423751831,
      "learning_rate": 8.242021531482606e-08,
      "loss": 0.803,
      "step": 32143
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.934856653213501,
      "learning_rate": 8.17812989170367e-08,
      "loss": 0.8696,
      "step": 32144
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.9513909816741943,
      "learning_rate": 8.114238251924737e-08,
      "loss": 0.6688,
      "step": 32145
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.104546308517456,
      "learning_rate": 8.050346612145801e-08,
      "loss": 0.7492,
      "step": 32146
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.152611255645752,
      "learning_rate": 7.986454972366867e-08,
      "loss": 1.1815,
      "step": 32147
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.7146952152252197,
      "learning_rate": 7.922563332587932e-08,
      "loss": 0.656,
      "step": 32148
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.421525001525879,
      "learning_rate": 7.858671692808995e-08,
      "loss": 1.3556,
      "step": 32149
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.495208740234375,
      "learning_rate": 7.794780053030061e-08,
      "loss": 1.2152,
      "step": 32150
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0553808212280273,
      "learning_rate": 7.730888413251126e-08,
      "loss": 0.819,
      "step": 32151
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9342721700668335,
      "learning_rate": 7.666996773472192e-08,
      "loss": 0.8928,
      "step": 32152
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5949362516403198,
      "learning_rate": 7.603105133693257e-08,
      "loss": 0.6856,
      "step": 32153
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3237978219985962,
      "learning_rate": 7.539213493914321e-08,
      "loss": 0.7743,
      "step": 32154
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.196290969848633,
      "learning_rate": 7.475321854135388e-08,
      "loss": 0.7191,
      "step": 32155
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2649447917938232,
      "learning_rate": 7.411430214356452e-08,
      "loss": 0.7497,
      "step": 32156
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.364193916320801,
      "learning_rate": 7.347538574577517e-08,
      "loss": 0.7284,
      "step": 32157
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7077094316482544,
      "learning_rate": 7.283646934798583e-08,
      "loss": 0.7612,
      "step": 32158
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.411769866943359,
      "learning_rate": 7.219755295019646e-08,
      "loss": 1.1209,
      "step": 32159
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8545801639556885,
      "learning_rate": 7.155863655240712e-08,
      "loss": 0.8818,
      "step": 32160
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1238749027252197,
      "learning_rate": 7.091972015461777e-08,
      "loss": 0.8295,
      "step": 32161
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.865909457206726,
      "learning_rate": 7.028080375682842e-08,
      "loss": 0.8195,
      "step": 32162
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6226965188980103,
      "learning_rate": 6.964188735903908e-08,
      "loss": 0.8205,
      "step": 32163
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7453560829162598,
      "learning_rate": 6.900297096124972e-08,
      "loss": 0.7985,
      "step": 32164
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9618260860443115,
      "learning_rate": 6.836405456346037e-08,
      "loss": 0.7953,
      "step": 32165
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.799057126045227,
      "learning_rate": 6.772513816567103e-08,
      "loss": 0.813,
      "step": 32166
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2517306804656982,
      "learning_rate": 6.708622176788168e-08,
      "loss": 0.9882,
      "step": 32167
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.5154800415039062,
      "learning_rate": 6.644730537009233e-08,
      "loss": 0.7334,
      "step": 32168
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.4114363193511963,
      "learning_rate": 6.580838897230297e-08,
      "loss": 0.8191,
      "step": 32169
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0089051723480225,
      "learning_rate": 6.516947257451363e-08,
      "loss": 0.9327,
      "step": 32170
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1676430702209473,
      "learning_rate": 6.453055617672428e-08,
      "loss": 0.9863,
      "step": 32171
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1409249305725098,
      "learning_rate": 6.389163977893493e-08,
      "loss": 0.8749,
      "step": 32172
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.020026206970215,
      "learning_rate": 6.325272338114559e-08,
      "loss": 0.76,
      "step": 32173
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.446122407913208,
      "learning_rate": 6.261380698335623e-08,
      "loss": 0.6675,
      "step": 32174
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8277785778045654,
      "learning_rate": 6.197489058556688e-08,
      "loss": 0.7968,
      "step": 32175
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.497811198234558,
      "learning_rate": 6.133597418777754e-08,
      "loss": 0.7257,
      "step": 32176
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1072144508361816,
      "learning_rate": 6.069705778998818e-08,
      "loss": 1.208,
      "step": 32177
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.783119559288025,
      "learning_rate": 6.005814139219884e-08,
      "loss": 0.867,
      "step": 32178
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.0192461013793945,
      "learning_rate": 5.941922499440949e-08,
      "loss": 0.871,
      "step": 32179
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.3476181030273438,
      "learning_rate": 5.878030859662013e-08,
      "loss": 1.0316,
      "step": 32180
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8489495515823364,
      "learning_rate": 5.8141392198830784e-08,
      "loss": 0.8567,
      "step": 32181
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8041507005691528,
      "learning_rate": 5.750247580104144e-08,
      "loss": 0.8586,
      "step": 32182
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5513254404067993,
      "learning_rate": 5.6863559403252084e-08,
      "loss": 0.8446,
      "step": 32183
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0792887210845947,
      "learning_rate": 5.622464300546274e-08,
      "loss": 1.2134,
      "step": 32184
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.868899941444397,
      "learning_rate": 5.558572660767339e-08,
      "loss": 0.7929,
      "step": 32185
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.366744518280029,
      "learning_rate": 5.494681020988404e-08,
      "loss": 1.0251,
      "step": 32186
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.238683223724365,
      "learning_rate": 5.430789381209469e-08,
      "loss": 0.9211,
      "step": 32187
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2057571411132812,
      "learning_rate": 5.3668977414305346e-08,
      "loss": 0.8804,
      "step": 32188
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5282995700836182,
      "learning_rate": 5.3030061016515986e-08,
      "loss": 0.868,
      "step": 32189
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5309251546859741,
      "learning_rate": 5.239114461872664e-08,
      "loss": 0.8013,
      "step": 32190
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.68459153175354,
      "learning_rate": 5.1752228220937294e-08,
      "loss": 0.7715,
      "step": 32191
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.061387300491333,
      "learning_rate": 5.111331182314794e-08,
      "loss": 1.0701,
      "step": 32192
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.679911494255066,
      "learning_rate": 5.0474395425358594e-08,
      "loss": 0.7766,
      "step": 32193
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2608389854431152,
      "learning_rate": 4.983547902756925e-08,
      "loss": 0.6281,
      "step": 32194
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9942225217819214,
      "learning_rate": 4.91965626297799e-08,
      "loss": 0.7491,
      "step": 32195
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5783274173736572,
      "learning_rate": 4.855764623199055e-08,
      "loss": 0.8359,
      "step": 32196
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.7039787769317627,
      "learning_rate": 4.79187298342012e-08,
      "loss": 0.6568,
      "step": 32197
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.768288612365723,
      "learning_rate": 4.7279813436411856e-08,
      "loss": 1.5766,
      "step": 32198
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7760082483291626,
      "learning_rate": 4.6640897038622496e-08,
      "loss": 0.5437,
      "step": 32199
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.871614694595337,
      "learning_rate": 4.600198064083315e-08,
      "loss": 0.8816,
      "step": 32200
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8532228469848633,
      "learning_rate": 4.5363064243043804e-08,
      "loss": 0.8234,
      "step": 32201
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7641818523406982,
      "learning_rate": 4.472414784525445e-08,
      "loss": 0.9789,
      "step": 32202
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8697658777236938,
      "learning_rate": 4.4085231447465104e-08,
      "loss": 0.7399,
      "step": 32203
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.515704870223999,
      "learning_rate": 4.344631504967576e-08,
      "loss": 0.9892,
      "step": 32204
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.158782720565796,
      "learning_rate": 4.2807398651886405e-08,
      "loss": 1.0173,
      "step": 32205
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.253800630569458,
      "learning_rate": 4.216848225409706e-08,
      "loss": 0.6873,
      "step": 32206
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.248927116394043,
      "learning_rate": 4.152956585630771e-08,
      "loss": 0.8547,
      "step": 32207
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3914874792099,
      "learning_rate": 4.089064945851835e-08,
      "loss": 0.7256,
      "step": 32208
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.904792070388794,
      "learning_rate": 4.0251733060729006e-08,
      "loss": 0.9038,
      "step": 32209
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.986368179321289,
      "learning_rate": 3.961281666293966e-08,
      "loss": 1.1383,
      "step": 32210
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.333554744720459,
      "learning_rate": 3.897390026515031e-08,
      "loss": 1.0939,
      "step": 32211
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.869965672492981,
      "learning_rate": 3.833498386736096e-08,
      "loss": 1.0312,
      "step": 32212
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1137661933898926,
      "learning_rate": 3.769606746957161e-08,
      "loss": 1.0369,
      "step": 32213
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8719744682312012,
      "learning_rate": 3.705715107178226e-08,
      "loss": 0.8503,
      "step": 32214
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9057952165603638,
      "learning_rate": 3.6418234673992915e-08,
      "loss": 0.8124,
      "step": 32215
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4063515663146973,
      "learning_rate": 3.577931827620356e-08,
      "loss": 0.7385,
      "step": 32216
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.327449321746826,
      "learning_rate": 3.514040187841421e-08,
      "loss": 0.8015,
      "step": 32217
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.59484601020813,
      "learning_rate": 3.450148548062486e-08,
      "loss": 0.6634,
      "step": 32218
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5010513067245483,
      "learning_rate": 3.3862569082835516e-08,
      "loss": 0.3687,
      "step": 32219
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3279075622558594,
      "learning_rate": 3.322365268504616e-08,
      "loss": 0.6914,
      "step": 32220
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.7174713611602783,
      "learning_rate": 3.258473628725682e-08,
      "loss": 1.4549,
      "step": 32221
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.870821952819824,
      "learning_rate": 3.1945819889467464e-08,
      "loss": 0.8202,
      "step": 32222
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9909405708312988,
      "learning_rate": 3.130690349167812e-08,
      "loss": 0.8494,
      "step": 32223
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.722456455230713,
      "learning_rate": 3.066798709388877e-08,
      "loss": 0.8006,
      "step": 32224
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.089521646499634,
      "learning_rate": 3.002907069609942e-08,
      "loss": 0.8112,
      "step": 32225
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.3633360862731934,
      "learning_rate": 2.9390154298310065e-08,
      "loss": 0.8331,
      "step": 32226
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4735954999923706,
      "learning_rate": 2.875123790052072e-08,
      "loss": 0.71,
      "step": 32227
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.069180488586426,
      "learning_rate": 2.811232150273137e-08,
      "loss": 0.8147,
      "step": 32228
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4427515268325806,
      "learning_rate": 2.747340510494202e-08,
      "loss": 0.3824,
      "step": 32229
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0198874473571777,
      "learning_rate": 2.6834488707152673e-08,
      "loss": 1.0533,
      "step": 32230
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.7391111850738525,
      "learning_rate": 2.619557230936332e-08,
      "loss": 1.107,
      "step": 32231
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.524698257446289,
      "learning_rate": 2.555665591157397e-08,
      "loss": 0.7273,
      "step": 32232
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.3635764122009277,
      "learning_rate": 2.4917739513784624e-08,
      "loss": 0.8733,
      "step": 32233
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6623094081878662,
      "learning_rate": 2.4278823115995274e-08,
      "loss": 0.8514,
      "step": 32234
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.395153522491455,
      "learning_rate": 2.3639906718205928e-08,
      "loss": 0.7729,
      "step": 32235
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6457446813583374,
      "learning_rate": 2.3000990320416575e-08,
      "loss": 0.8628,
      "step": 32236
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7453616857528687,
      "learning_rate": 2.2362073922627225e-08,
      "loss": 0.783,
      "step": 32237
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7596724033355713,
      "learning_rate": 2.172315752483788e-08,
      "loss": 0.9118,
      "step": 32238
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.625481128692627,
      "learning_rate": 2.108424112704853e-08,
      "loss": 0.8607,
      "step": 32239
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.360707998275757,
      "learning_rate": 2.0445324729259176e-08,
      "loss": 0.8016,
      "step": 32240
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8126935958862305,
      "learning_rate": 1.980640833146983e-08,
      "loss": 0.8258,
      "step": 32241
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7460529804229736,
      "learning_rate": 1.916749193368048e-08,
      "loss": 0.8832,
      "step": 32242
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7886919975280762,
      "learning_rate": 1.852857553589113e-08,
      "loss": 0.8282,
      "step": 32243
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.2250266075134277,
      "learning_rate": 1.788965913810178e-08,
      "loss": 0.6588,
      "step": 32244
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0948588848114014,
      "learning_rate": 1.725074274031243e-08,
      "loss": 0.8361,
      "step": 32245
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.212707757949829,
      "learning_rate": 1.661182634252308e-08,
      "loss": 0.8875,
      "step": 32246
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5003957748413086,
      "learning_rate": 1.5972909944733732e-08,
      "loss": 0.8159,
      "step": 32247
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.3276114463806152,
      "learning_rate": 1.5333993546944386e-08,
      "loss": 0.9745,
      "step": 32248
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7004855871200562,
      "learning_rate": 1.4695077149155033e-08,
      "loss": 0.7692,
      "step": 32249
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.067110776901245,
      "learning_rate": 1.4056160751365685e-08,
      "loss": 0.7967,
      "step": 32250
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0861101150512695,
      "learning_rate": 1.3417244353576336e-08,
      "loss": 0.6911,
      "step": 32251
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.5562283992767334,
      "learning_rate": 1.2778327955786985e-08,
      "loss": 0.6974,
      "step": 32252
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6340136528015137,
      "learning_rate": 1.2139411557997637e-08,
      "loss": 0.773,
      "step": 32253
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.4775140285491943,
      "learning_rate": 1.1500495160208287e-08,
      "loss": 1.0042,
      "step": 32254
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.703300952911377,
      "learning_rate": 1.086157876241894e-08,
      "loss": 0.8557,
      "step": 32255
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.996063470840454,
      "learning_rate": 1.0222662364629588e-08,
      "loss": 0.7756,
      "step": 32256
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.823939323425293,
      "learning_rate": 9.58374596684024e-09,
      "loss": 0.7995,
      "step": 32257
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.3201746940612793,
      "learning_rate": 8.94482956905089e-09,
      "loss": 0.7093,
      "step": 32258
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.8349623680114746,
      "learning_rate": 8.30591317126154e-09,
      "loss": 0.7512,
      "step": 32259
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8553169965744019,
      "learning_rate": 7.666996773472193e-09,
      "loss": 1.151,
      "step": 32260
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.097656011581421,
      "learning_rate": 7.028080375682842e-09,
      "loss": 0.9851,
      "step": 32261
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5230087041854858,
      "learning_rate": 6.3891639778934926e-09,
      "loss": 0.8328,
      "step": 32262
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8310168981552124,
      "learning_rate": 5.750247580104144e-09,
      "loss": 0.8839,
      "step": 32263
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.590511679649353,
      "learning_rate": 5.111331182314794e-09,
      "loss": 0.9137,
      "step": 32264
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.905962944030762,
      "learning_rate": 4.472414784525445e-09,
      "loss": 0.9852,
      "step": 32265
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4838916063308716,
      "learning_rate": 3.833498386736096e-09,
      "loss": 0.8185,
      "step": 32266
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9682501554489136,
      "learning_rate": 3.1945819889467463e-09,
      "loss": 0.7927,
      "step": 32267
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9250843524932861,
      "learning_rate": 2.555665591157397e-09,
      "loss": 0.8856,
      "step": 32268
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1054482460021973,
      "learning_rate": 1.916749193368048e-09,
      "loss": 0.7997,
      "step": 32269
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.6465375423431396,
      "learning_rate": 1.2778327955786985e-09,
      "loss": 1.0208,
      "step": 32270
    },
    {
      "epoch": 1.0,
      "step": 32270,
      "total_flos": 2.5051077384383496e+19,
      "train_loss": 0.9278342838096367,
      "train_runtime": 88806.679,
      "train_samples_per_second": 11.628,
      "train_steps_per_second": 0.363
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 32272,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50000,
  "total_flos": 2.5051077384383496e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}