{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.870503597122302,
  "eval_steps": 500,
  "global_step": 204,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 7.142857142857143e-06,
      "loss": 3.7385,
      "step": 1
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 2.9618,
      "step": 2
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 3.7651,
      "step": 3
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.857142857142857e-05,
      "loss": 2.5572,
      "step": 4
    },
    {
      "epoch": 0.14,
      "learning_rate": 3.571428571428572e-05,
      "loss": 3.2352,
      "step": 5
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 3.7965,
      "step": 6
    },
    {
      "epoch": 0.2,
      "learning_rate": 5e-05,
      "loss": 3.4411,
      "step": 7
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.999682116415026e-05,
      "loss": 3.2539,
      "step": 8
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.998728546500082e-05,
      "loss": 3.2295,
      "step": 9
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.9971395327545466e-05,
      "loss": 3.8146,
      "step": 10
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.9949154792755286e-05,
      "loss": 2.9204,
      "step": 11
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.992056951655103e-05,
      "loss": 2.4845,
      "step": 12
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.988564676836475e-05,
      "loss": 2.3794,
      "step": 13
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.984439542929117e-05,
      "loss": 2.3144,
      "step": 14
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.979682598982912e-05,
      "loss": 2.8456,
      "step": 15
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.974295054721377e-05,
      "loss": 3.0537,
      "step": 16
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.9682782802340184e-05,
      "loss": 2.7413,
      "step": 17
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.9616338056279124e-05,
      "loss": 2.8134,
      "step": 18
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.9543633206385834e-05,
      "loss": 3.6205,
      "step": 19
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.9464686742003006e-05,
      "loss": 3.5697,
      "step": 20
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.937951873975871e-05,
      "loss": 2.8348,
      "step": 21
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.928815085846087e-05,
      "loss": 3.3747,
      "step": 22
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.9190606333589194e-05,
      "loss": 3.0038,
      "step": 23
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.9086909971386305e-05,
      "loss": 3.1921,
      "step": 24
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.8977088142549285e-05,
      "loss": 2.6579,
      "step": 25
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.886116877552347e-05,
      "loss": 3.2293,
      "step": 26
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.873918134940002e-05,
      "loss": 4.1384,
      "step": 27
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.8611156886419206e-05,
      "loss": 2.6833,
      "step": 28
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.847712794408124e-05,
      "loss": 3.4143,
      "step": 29
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.833712860686666e-05,
      "loss": 3.9069,
      "step": 30
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.8191194477568435e-05,
      "loss": 3.1464,
      "step": 31
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.803936266823792e-05,
      "loss": 3.9239,
      "step": 32
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.7881671790747e-05,
      "loss": 2.982,
      "step": 33
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.7718161946968835e-05,
      "loss": 2.9096,
      "step": 34
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.754887471857969e-05,
      "loss": 2.772,
      "step": 35
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.7373853156484406e-05,
      "loss": 2.9948,
      "step": 36
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.7193141769868265e-05,
      "loss": 2.9775,
      "step": 37
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.7006786514877997e-05,
      "loss": 2.534,
      "step": 38
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.6814834782934844e-05,
      "loss": 2.3817,
      "step": 39
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.6617335388682556e-05,
      "loss": 3.223,
      "step": 40
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.641433855757351e-05,
      "loss": 3.3482,
      "step": 41
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.620589591309603e-05,
      "loss": 2.992,
      "step": 42
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.59920604636462e-05,
      "loss": 2.4652,
      "step": 43
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.577288658904741e-05,
      "loss": 2.8348,
      "step": 44
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.554843002672129e-05,
      "loss": 4.1209,
      "step": 45
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.531874785751317e-05,
      "loss": 3.5458,
      "step": 46
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.5083898491176136e-05,
      "loss": 3.3499,
      "step": 47
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.4843941651517e-05,
      "loss": 2.5327,
      "step": 48
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.4598938361208095e-05,
      "loss": 2.9724,
      "step": 49
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.434895092626883e-05,
      "loss": 2.4149,
      "step": 50
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.409404292022081e-05,
      "loss": 2.8688,
      "step": 51
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.38342791679207e-05,
      "loss": 2.9715,
      "step": 52
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.356972572907473e-05,
      "loss": 2.6849,
      "step": 53
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.3300449881439375e-05,
      "loss": 3.185,
      "step": 54
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.302652010371205e-05,
      "loss": 3.1415,
      "step": 55
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.274800605811658e-05,
      "loss": 2.5797,
      "step": 56
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.246497857268759e-05,
      "loss": 2.2357,
      "step": 57
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.2177509623258456e-05,
      "loss": 2.8965,
      "step": 58
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.1885672315157346e-05,
      "loss": 3.1428,
      "step": 59
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.1589540864616025e-05,
      "loss": 2.3255,
      "step": 60
    },
    {
      "epoch": 1.76,
      "learning_rate": 4.128919057989622e-05,
      "loss": 3.001,
      "step": 61
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.098469784213812e-05,
      "loss": 3.0732,
      "step": 62
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.0676140085936186e-05,
      "loss": 2.7686,
      "step": 63
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.0363595779647e-05,
      "loss": 3.1581,
      "step": 64
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.0047144405434175e-05,
      "loss": 3.8875,
      "step": 65
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.972686643905558e-05,
      "loss": 3.9603,
      "step": 66
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.940284332939771e-05,
      "loss": 3.6382,
      "step": 67
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.9075157477762744e-05,
      "loss": 3.5762,
      "step": 68
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.874389221691329e-05,
      "loss": 2.7994,
      "step": 69
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.84091317898803e-05,
      "loss": 3.0237,
      "step": 70
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.8070961328539525e-05,
      "loss": 2.4911,
      "step": 71
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.772946683196179e-05,
      "loss": 3.264,
      "step": 72
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.738473514454297e-05,
      "loss": 3.1732,
      "step": 73
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.7036853933918784e-05,
      "loss": 3.5421,
      "step": 74
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.668591166867035e-05,
      "loss": 3.8679,
      "step": 75
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.633199759582596e-05,
      "loss": 3.2613,
      "step": 76
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.597520171816503e-05,
      "loss": 3.3075,
      "step": 77
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.5615614771329706e-05,
      "loss": 2.6811,
      "step": 78
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.5253328200750224e-05,
      "loss": 3.0083,
      "step": 79
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.488843413838963e-05,
      "loss": 3.3417,
      "step": 80
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.452102537931408e-05,
      "loss": 2.4357,
      "step": 81
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.4151195358094365e-05,
      "loss": 3.2885,
      "step": 82
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.377903812504487e-05,
      "loss": 3.3224,
      "step": 83
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.340464832230592e-05,
      "loss": 3.2192,
      "step": 84
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.3028121159775656e-05,
      "loss": 2.8256,
      "step": 85
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.2649552390897494e-05,
      "loss": 2.0816,
      "step": 86
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.226903828830935e-05,
      "loss": 3.1104,
      "step": 87
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.1886675619360885e-05,
      "loss": 3.0006,
      "step": 88
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.1502561621504874e-05,
      "loss": 2.4053,
      "step": 89
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.111679397756906e-05,
      "loss": 3.3762,
      "step": 90
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.072947079091472e-05,
      "loss": 2.9438,
      "step": 91
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.034069056048837e-05,
      "loss": 3.9727,
      "step": 92
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.9950552155772743e-05,
      "loss": 3.3063,
      "step": 93
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.95591547916436e-05,
      "loss": 2.9004,
      "step": 94
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.9166598003138766e-05,
      "loss": 3.149,
      "step": 95
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.8772981620145623e-05,
      "loss": 2.7673,
      "step": 96
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.8378405742013713e-05,
      "loss": 2.4541,
      "step": 97
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.7982970712098794e-05,
      "loss": 2.7791,
      "step": 98
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.7586777092244804e-05,
      "loss": 3.8837,
      "step": 99
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.7189925637210323e-05,
      "loss": 2.5179,
      "step": 100
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.6792517269045943e-05,
      "loss": 3.7649,
      "step": 101
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.6394653051429064e-05,
      "loss": 3.5913,
      "step": 102
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.5996434163962763e-05,
      "loss": 5.0055,
      "step": 103
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.5597961876445077e-05,
      "loss": 2.6826,
      "step": 104
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.5199337523115418e-05,
      "loss": 3.1971,
      "step": 105
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.480066247688459e-05,
      "loss": 3.0641,
      "step": 106
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.4402038123554933e-05,
      "loss": 2.9264,
      "step": 107
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.4003565836037246e-05,
      "loss": 2.5069,
      "step": 108
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.360534694857094e-05,
      "loss": 4.365,
      "step": 109
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.3207482730954063e-05,
      "loss": 3.4356,
      "step": 110
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.2810074362789676e-05,
      "loss": 3.2688,
      "step": 111
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.2413222907755195e-05,
      "loss": 2.7195,
      "step": 112
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.2017029287901212e-05,
      "loss": 2.6272,
      "step": 113
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.162159425798629e-05,
      "loss": 2.8288,
      "step": 114
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.1227018379854383e-05,
      "loss": 2.0868,
      "step": 115
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.083340199686124e-05,
      "loss": 3.5304,
      "step": 116
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.0440845208356402e-05,
      "loss": 3.0078,
      "step": 117
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.0049447844227266e-05,
      "loss": 3.5883,
      "step": 118
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.9659309439511628e-05,
      "loss": 2.6867,
      "step": 119
    },
    {
      "epoch": 3.45,
      "learning_rate": 1.927052920908528e-05,
      "loss": 2.3407,
      "step": 120
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.8883206022430956e-05,
      "loss": 3.6757,
      "step": 121
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.849743837849513e-05,
      "loss": 4.9815,
      "step": 122
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.8113324380639117e-05,
      "loss": 3.0576,
      "step": 123
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.7730961711690655e-05,
      "loss": 2.4852,
      "step": 124
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.735044760910251e-05,
      "loss": 2.6612,
      "step": 125
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.6971878840224346e-05,
      "loss": 2.9253,
      "step": 126
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.6595351677694083e-05,
      "loss": 3.4379,
      "step": 127
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.6220961874955134e-05,
      "loss": 2.8997,
      "step": 128
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.5848804641905634e-05,
      "loss": 3.2326,
      "step": 129
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.547897462068592e-05,
      "loss": 2.8364,
      "step": 130
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.5111565861610378e-05,
      "loss": 2.5979,
      "step": 131
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.4746671799249784e-05,
      "loss": 3.2094,
      "step": 132
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.4384385228670288e-05,
      "loss": 2.2841,
      "step": 133
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.4024798281834966e-05,
      "loss": 3.333,
      "step": 134
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.3668002404174047e-05,
      "loss": 3.1955,
      "step": 135
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.331408833132966e-05,
      "loss": 1.9727,
      "step": 136
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.2963146066081217e-05,
      "loss": 2.5208,
      "step": 137
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.2615264855457038e-05,
      "loss": 2.4915,
      "step": 138
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.2270533168038217e-05,
      "loss": 2.6322,
      "step": 139
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.1929038671460486e-05,
      "loss": 3.6504,
      "step": 140
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.1590868210119692e-05,
      "loss": 2.9824,
      "step": 141
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.125610778308672e-05,
      "loss": 2.6982,
      "step": 142
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.0924842522237267e-05,
      "loss": 4.0364,
      "step": 143
    },
    {
      "epoch": 4.14,
      "learning_rate": 1.0597156670602299e-05,
      "loss": 2.9809,
      "step": 144
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.027313356094443e-05,
      "loss": 1.7772,
      "step": 145
    },
    {
      "epoch": 4.2,
      "learning_rate": 9.95285559456583e-06,
      "loss": 1.9795,
      "step": 146
    },
    {
      "epoch": 4.23,
      "learning_rate": 9.636404220353013e-06,
      "loss": 2.0211,
      "step": 147
    },
    {
      "epoch": 4.26,
      "learning_rate": 9.323859914063814e-06,
      "loss": 2.3984,
      "step": 148
    },
    {
      "epoch": 4.29,
      "learning_rate": 9.015302157861883e-06,
      "loss": 2.1658,
      "step": 149
    },
    {
      "epoch": 4.32,
      "learning_rate": 8.710809420103789e-06,
      "loss": 2.0633,
      "step": 150
    },
    {
      "epoch": 4.35,
      "learning_rate": 8.41045913538398e-06,
      "loss": 2.5502,
      "step": 151
    },
    {
      "epoch": 4.37,
      "learning_rate": 8.11432768484267e-06,
      "loss": 2.2849,
      "step": 152
    },
    {
      "epoch": 4.4,
      "learning_rate": 7.822490376741554e-06,
      "loss": 2.1318,
      "step": 153
    },
    {
      "epoch": 4.43,
      "learning_rate": 7.535021427312417e-06,
      "loss": 3.1577,
      "step": 154
    },
    {
      "epoch": 4.46,
      "learning_rate": 7.251993941883428e-06,
      "loss": 2.4369,
      "step": 155
    },
    {
      "epoch": 4.49,
      "learning_rate": 6.9734798962879575e-06,
      "loss": 2.7381,
      "step": 156
    },
    {
      "epoch": 4.52,
      "learning_rate": 6.699550118560632e-06,
      "loss": 2.8945,
      "step": 157
    },
    {
      "epoch": 4.55,
      "learning_rate": 6.430274270925271e-06,
      "loss": 1.8788,
      "step": 158
    },
    {
      "epoch": 4.58,
      "learning_rate": 6.1657208320793054e-06,
      "loss": 3.3476,
      "step": 159
    },
    {
      "epoch": 4.6,
      "learning_rate": 5.905957079779187e-06,
      "loss": 2.3672,
      "step": 160
    },
    {
      "epoch": 4.63,
      "learning_rate": 5.6510490737311735e-06,
      "loss": 3.0826,
      "step": 161
    },
    {
      "epoch": 4.66,
      "learning_rate": 5.4010616387919095e-06,
      "loss": 2.4247,
      "step": 162
    },
    {
      "epoch": 4.69,
      "learning_rate": 5.156058348483006e-06,
      "loss": 2.9849,
      "step": 163
    },
    {
      "epoch": 4.72,
      "learning_rate": 4.916101508823873e-06,
      "loss": 3.2181,
      "step": 164
    },
    {
      "epoch": 4.75,
      "learning_rate": 4.681252142486841e-06,
      "loss": 2.756,
      "step": 165
    },
    {
      "epoch": 4.78,
      "learning_rate": 4.451569973278719e-06,
      "loss": 2.5613,
      "step": 166
    },
    {
      "epoch": 4.81,
      "learning_rate": 4.227113410952585e-06,
      "loss": 3.7162,
      "step": 167
    },
    {
      "epoch": 4.83,
      "learning_rate": 4.007939536353805e-06,
      "loss": 2.9746,
      "step": 168
    },
    {
      "epoch": 4.86,
      "learning_rate": 3.7941040869039714e-06,
      "loss": 2.4908,
      "step": 169
    },
    {
      "epoch": 4.89,
      "learning_rate": 3.585661442426494e-06,
      "loss": 2.2047,
      "step": 170
    },
    {
      "epoch": 4.92,
      "learning_rate": 3.382664611317446e-06,
      "loss": 2.5563,
      "step": 171
    },
    {
      "epoch": 4.95,
      "learning_rate": 3.1851652170651584e-06,
      "loss": 2.804,
      "step": 172
    },
    {
      "epoch": 4.98,
      "learning_rate": 2.9932134851220038e-06,
      "loss": 3.0241,
      "step": 173
    },
    {
      "epoch": 5.01,
      "learning_rate": 2.8068582301317425e-06,
      "loss": 2.2791,
      "step": 174
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.6261468435155978e-06,
      "loss": 2.997,
      "step": 175
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.4511252814203107e-06,
      "loss": 2.8508,
      "step": 176
    },
    {
      "epoch": 5.09,
      "learning_rate": 2.2818380530311655e-06,
      "loss": 2.4406,
      "step": 177
    },
    {
      "epoch": 5.12,
      "learning_rate": 2.1183282092530065e-06,
      "loss": 2.5544,
      "step": 178
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.960637331762091e-06,
      "loss": 2.9358,
      "step": 179
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.8088055224315697e-06,
      "loss": 2.2767,
      "step": 180
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.6628713931333445e-06,
      "loss": 3.5501,
      "step": 181
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.5228720559187642e-06,
      "loss": 2.3521,
      "step": 182
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.3888431135807956e-06,
      "loss": 2.6828,
      "step": 183
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.2608186505999847e-06,
      "loss": 3.3842,
      "step": 184
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.138831224476533e-06,
      "loss": 3.7512,
      "step": 185
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.0229118574507174e-06,
      "loss": 2.5144,
      "step": 186
    },
    {
      "epoch": 5.38,
      "learning_rate": 9.130900286137001e-07,
      "loss": 1.8449,
      "step": 187
    },
    {
      "epoch": 5.41,
      "learning_rate": 8.093936664108071e-07,
      "loss": 2.3796,
      "step": 188
    },
    {
      "epoch": 5.44,
      "learning_rate": 7.118491415391337e-07,
      "loss": 3.1911,
      "step": 189
    },
    {
      "epoch": 5.47,
      "learning_rate": 6.204812602412902e-07,
      "loss": 3.3982,
      "step": 190
    },
    {
      "epoch": 5.5,
      "learning_rate": 5.353132579969972e-07,
      "loss": 2.5562,
      "step": 191
    },
    {
      "epoch": 5.53,
      "learning_rate": 4.56366793614163e-07,
      "loss": 3.1778,
      "step": 192
    },
    {
      "epoch": 5.55,
      "learning_rate": 3.8366194372088384e-07,
      "loss": 2.8666,
      "step": 193
    },
    {
      "epoch": 5.58,
      "learning_rate": 3.1721719765981926e-07,
      "loss": 1.849,
      "step": 194
    },
    {
      "epoch": 5.61,
      "learning_rate": 2.5704945278623436e-07,
      "loss": 2.9817,
      "step": 195
    },
    {
      "epoch": 5.64,
      "learning_rate": 2.0317401017088122e-07,
      "loss": 2.8708,
      "step": 196
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.5560457070883105e-07,
      "loss": 1.9914,
      "step": 197
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.1435323163525025e-07,
      "loss": 2.631,
      "step": 198
    },
    {
      "epoch": 5.73,
      "learning_rate": 7.94304834489723e-08,
      "loss": 3.1964,
      "step": 199
    },
    {
      "epoch": 5.76,
      "learning_rate": 5.0845207244715196e-08,
      "loss": 3.1362,
      "step": 200
    },
    {
      "epoch": 5.78,
      "learning_rate": 2.8604672454538018e-08,
      "loss": 2.4684,
      "step": 201
    },
    {
      "epoch": 5.81,
      "learning_rate": 1.2714534999183625e-08,
      "loss": 2.9911,
      "step": 202
    },
    {
      "epoch": 5.84,
      "learning_rate": 3.1788358497431005e-09,
      "loss": 2.7763,
      "step": 203
    },
    {
      "epoch": 5.87,
      "learning_rate": 0.0,
      "loss": 3.1651,
      "step": 204
    },
    {
      "epoch": 5.87,
      "step": 204,
      "total_flos": 0.0,
      "train_loss": 2.9658100090774835,
      "train_runtime": 18093.1922,
      "train_samples_per_second": 23.54,
      "train_steps_per_second": 0.011
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 204,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 50,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}