{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 1410,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5.000000000000001e-07,
      "loss": 2.9541,
      "step": 1
    },
    {
      "epoch": 0.02,
      "learning_rate": 5e-06,
      "loss": 2.9089,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 2.9232,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.5e-05,
      "loss": 2.914,
      "step": 30
    },
    {
      "epoch": 0.09,
      "learning_rate": 2e-05,
      "loss": 2.9596,
      "step": 40
    },
    {
      "epoch": 0.11,
      "learning_rate": 2.5e-05,
      "loss": 2.9106,
      "step": 50
    },
    {
      "epoch": 0.13,
      "learning_rate": 3e-05,
      "loss": 2.8611,
      "step": 60
    },
    {
      "epoch": 0.15,
      "learning_rate": 3.5e-05,
      "loss": 2.926,
      "step": 70
    },
    {
      "epoch": 0.17,
      "learning_rate": 4e-05,
      "loss": 2.8827,
      "step": 80
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.5e-05,
      "loss": 2.8972,
      "step": 90
    },
    {
      "epoch": 0.21,
      "learning_rate": 5e-05,
      "loss": 2.9258,
      "step": 100
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9618320610687025e-05,
      "loss": 2.9555,
      "step": 110
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.923664122137405e-05,
      "loss": 2.93,
      "step": 120
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.885496183206107e-05,
      "loss": 2.936,
      "step": 130
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.847328244274809e-05,
      "loss": 2.8927,
      "step": 140
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.809160305343512e-05,
      "loss": 2.8796,
      "step": 150
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.7709923664122144e-05,
      "loss": 2.9172,
      "step": 160
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.7328244274809166e-05,
      "loss": 2.8782,
      "step": 170
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.694656488549619e-05,
      "loss": 2.8833,
      "step": 180
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.656488549618321e-05,
      "loss": 2.8822,
      "step": 190
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.618320610687023e-05,
      "loss": 2.8877,
      "step": 200
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.5801526717557256e-05,
      "loss": 2.8287,
      "step": 210
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.541984732824428e-05,
      "loss": 2.8852,
      "step": 220
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.5038167938931294e-05,
      "loss": 2.8694,
      "step": 230
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.465648854961832e-05,
      "loss": 2.8847,
      "step": 240
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.4274809160305345e-05,
      "loss": 2.889,
      "step": 250
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.389312977099237e-05,
      "loss": 2.8866,
      "step": 260
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.351145038167939e-05,
      "loss": 2.8523,
      "step": 270
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.312977099236641e-05,
      "loss": 2.8664,
      "step": 280
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.2748091603053435e-05,
      "loss": 2.8683,
      "step": 290
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.236641221374046e-05,
      "loss": 2.8312,
      "step": 300
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.198473282442748e-05,
      "loss": 2.8776,
      "step": 310
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.160305343511451e-05,
      "loss": 2.8447,
      "step": 320
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.122137404580153e-05,
      "loss": 2.8987,
      "step": 330
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.0839694656488554e-05,
      "loss": 2.8599,
      "step": 340
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.0458015267175576e-05,
      "loss": 2.8863,
      "step": 350
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.00763358778626e-05,
      "loss": 2.8449,
      "step": 360
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.969465648854962e-05,
      "loss": 2.8262,
      "step": 370
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.9312977099236644e-05,
      "loss": 2.9043,
      "step": 380
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.8931297709923666e-05,
      "loss": 2.8667,
      "step": 390
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.854961832061069e-05,
      "loss": 2.8355,
      "step": 400
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.816793893129771e-05,
      "loss": 2.8561,
      "step": 410
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.778625954198473e-05,
      "loss": 2.8035,
      "step": 420
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.7404580152671756e-05,
      "loss": 2.8799,
      "step": 430
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.702290076335878e-05,
      "loss": 2.8353,
      "step": 440
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.66412213740458e-05,
      "loss": 2.8205,
      "step": 450
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.625954198473282e-05,
      "loss": 2.8467,
      "step": 460
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.5877862595419845e-05,
      "loss": 2.8587,
      "step": 470
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.549618320610687e-05,
      "loss": 2.7819,
      "step": 480
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.511450381679389e-05,
      "loss": 2.7764,
      "step": 490
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.473282442748092e-05,
      "loss": 2.7744,
      "step": 500
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.435114503816794e-05,
      "loss": 2.7788,
      "step": 510
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.3969465648854964e-05,
      "loss": 2.7927,
      "step": 520
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.358778625954199e-05,
      "loss": 2.7939,
      "step": 530
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.320610687022901e-05,
      "loss": 2.7556,
      "step": 540
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.282442748091603e-05,
      "loss": 2.7721,
      "step": 550
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.2442748091603054e-05,
      "loss": 2.766,
      "step": 560
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.2061068702290076e-05,
      "loss": 2.8024,
      "step": 570
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.16793893129771e-05,
      "loss": 2.8031,
      "step": 580
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.129770992366413e-05,
      "loss": 2.7328,
      "step": 590
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.091603053435115e-05,
      "loss": 2.7363,
      "step": 600
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.053435114503817e-05,
      "loss": 2.7882,
      "step": 610
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.0152671755725192e-05,
      "loss": 2.8019,
      "step": 620
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.9770992366412214e-05,
      "loss": 2.79,
      "step": 630
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.9389312977099237e-05,
      "loss": 2.7596,
      "step": 640
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.900763358778626e-05,
      "loss": 2.775,
      "step": 650
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.862595419847328e-05,
      "loss": 2.7885,
      "step": 660
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.824427480916031e-05,
      "loss": 2.776,
      "step": 670
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.7862595419847333e-05,
      "loss": 2.7456,
      "step": 680
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.7480916030534355e-05,
      "loss": 2.7306,
      "step": 690
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.7099236641221375e-05,
      "loss": 2.7702,
      "step": 700
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.6717557251908397e-05,
      "loss": 2.7534,
      "step": 710
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.633587786259542e-05,
      "loss": 2.7513,
      "step": 720
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.5954198473282442e-05,
      "loss": 2.8025,
      "step": 730
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.5572519083969464e-05,
      "loss": 2.7708,
      "step": 740
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.5190839694656487e-05,
      "loss": 2.7335,
      "step": 750
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.4809160305343512e-05,
      "loss": 2.7486,
      "step": 760
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.4427480916030535e-05,
      "loss": 2.7564,
      "step": 770
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.404580152671756e-05,
      "loss": 2.7758,
      "step": 780
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.3664122137404583e-05,
      "loss": 2.7942,
      "step": 790
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.3282442748091605e-05,
      "loss": 2.7317,
      "step": 800
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.2900763358778628e-05,
      "loss": 2.7783,
      "step": 810
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.2519083969465647e-05,
      "loss": 2.7836,
      "step": 820
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.2137404580152673e-05,
      "loss": 2.7881,
      "step": 830
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.1755725190839695e-05,
      "loss": 2.7586,
      "step": 840
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.1374045801526718e-05,
      "loss": 2.7601,
      "step": 850
    },
    {
      "epoch": 1.83,
      "learning_rate": 2.099236641221374e-05,
      "loss": 2.7746,
      "step": 860
    },
    {
      "epoch": 1.85,
      "learning_rate": 2.0610687022900766e-05,
      "loss": 2.7675,
      "step": 870
    },
    {
      "epoch": 1.87,
      "learning_rate": 2.0229007633587788e-05,
      "loss": 2.7414,
      "step": 880
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.984732824427481e-05,
      "loss": 2.7352,
      "step": 890
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.9465648854961833e-05,
      "loss": 2.7762,
      "step": 900
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.9083969465648855e-05,
      "loss": 2.7331,
      "step": 910
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.8702290076335878e-05,
      "loss": 2.7561,
      "step": 920
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.83206106870229e-05,
      "loss": 2.7428,
      "step": 930
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.7938931297709923e-05,
      "loss": 2.7049,
      "step": 940
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.7557251908396945e-05,
      "loss": 2.7188,
      "step": 950
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.717557251908397e-05,
      "loss": 2.699,
      "step": 960
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.6793893129770993e-05,
      "loss": 2.6526,
      "step": 970
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.6412213740458016e-05,
      "loss": 2.6924,
      "step": 980
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.6030534351145038e-05,
      "loss": 2.6693,
      "step": 990
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.5648854961832064e-05,
      "loss": 2.696,
      "step": 1000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.5267175572519086e-05,
      "loss": 2.687,
      "step": 1010
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.4885496183206107e-05,
      "loss": 2.7227,
      "step": 1020
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.450381679389313e-05,
      "loss": 2.6702,
      "step": 1030
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.4122137404580155e-05,
      "loss": 2.6786,
      "step": 1040
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.3740458015267178e-05,
      "loss": 2.6866,
      "step": 1050
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.3358778625954198e-05,
      "loss": 2.7269,
      "step": 1060
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.2977099236641221e-05,
      "loss": 2.6913,
      "step": 1070
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2595419847328243e-05,
      "loss": 2.7301,
      "step": 1080
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2213740458015267e-05,
      "loss": 2.732,
      "step": 1090
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1832061068702292e-05,
      "loss": 2.6862,
      "step": 1100
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.1450381679389314e-05,
      "loss": 2.7358,
      "step": 1110
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.1068702290076336e-05,
      "loss": 2.6862,
      "step": 1120
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0687022900763359e-05,
      "loss": 2.7138,
      "step": 1130
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.0305343511450383e-05,
      "loss": 2.7312,
      "step": 1140
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.923664122137405e-06,
      "loss": 2.7159,
      "step": 1150
    },
    {
      "epoch": 2.47,
      "learning_rate": 9.541984732824428e-06,
      "loss": 2.6882,
      "step": 1160
    },
    {
      "epoch": 2.49,
      "learning_rate": 9.16030534351145e-06,
      "loss": 2.7109,
      "step": 1170
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.778625954198473e-06,
      "loss": 2.7363,
      "step": 1180
    },
    {
      "epoch": 2.53,
      "learning_rate": 8.396946564885497e-06,
      "loss": 2.7332,
      "step": 1190
    },
    {
      "epoch": 2.55,
      "learning_rate": 8.015267175572519e-06,
      "loss": 2.6836,
      "step": 1200
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.633587786259543e-06,
      "loss": 2.7008,
      "step": 1210
    },
    {
      "epoch": 2.6,
      "learning_rate": 7.251908396946565e-06,
      "loss": 2.7129,
      "step": 1220
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.870229007633589e-06,
      "loss": 2.7233,
      "step": 1230
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.4885496183206104e-06,
      "loss": 2.6856,
      "step": 1240
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.106870229007634e-06,
      "loss": 2.7376,
      "step": 1250
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.725190839694657e-06,
      "loss": 2.7237,
      "step": 1260
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.343511450381679e-06,
      "loss": 2.698,
      "step": 1270
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.961832061068703e-06,
      "loss": 2.7422,
      "step": 1280
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.580152671755725e-06,
      "loss": 2.6834,
      "step": 1290
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.198473282442748e-06,
      "loss": 2.6675,
      "step": 1300
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.816793893129772e-06,
      "loss": 2.6683,
      "step": 1310
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.4351145038167944e-06,
      "loss": 2.7138,
      "step": 1320
    },
    {
      "epoch": 2.83,
      "learning_rate": 3.053435114503817e-06,
      "loss": 2.7272,
      "step": 1330
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.6717557251908397e-06,
      "loss": 2.7183,
      "step": 1340
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.2900763358778625e-06,
      "loss": 2.7036,
      "step": 1350
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.908396946564886e-06,
      "loss": 2.6808,
      "step": 1360
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.5267175572519084e-06,
      "loss": 2.6361,
      "step": 1370
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.1450381679389313e-06,
      "loss": 2.7164,
      "step": 1380
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.633587786259542e-07,
      "loss": 2.705,
      "step": 1390
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.816793893129771e-07,
      "loss": 2.677,
      "step": 1400
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 2.6911,
      "step": 1410
    },
    {
      "epoch": 3.0,
      "step": 1410,
      "total_flos": 6.285446300565504e+16,
      "train_loss": 0.690881545154761,
      "train_runtime": 7737.0233,
      "train_samples_per_second": 2.187,
      "train_steps_per_second": 0.182
    }
  ],
  "max_steps": 1410,
  "num_train_epochs": 3,
  "total_flos": 6.285446300565504e+16,
  "trial_name": null,
  "trial_params": null
}