{
  "best_metric": 0.7027972027972028,
  "best_model_checkpoint": "wav2vec2-5Class-train-test-finetune/checkpoint-721",
  "epoch": 323.0769230769231,
  "eval_steps": 500,
  "global_step": 1050,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.92,
      "eval_accuracy": 0.34265734265734266,
      "eval_loss": 1.59893798828125,
      "eval_runtime": 4.2802,
      "eval_samples_per_second": 66.819,
      "eval_steps_per_second": 0.701,
      "step": 3
    },
    {
      "epoch": 1.85,
      "eval_accuracy": 0.34265734265734266,
      "eval_loss": 1.5987956523895264,
      "eval_runtime": 4.8166,
      "eval_samples_per_second": 59.378,
      "eval_steps_per_second": 0.623,
      "step": 6
    },
    {
      "epoch": 2.77,
      "eval_accuracy": 0.34265734265734266,
      "eval_loss": 1.598555326461792,
      "eval_runtime": 3.989,
      "eval_samples_per_second": 71.697,
      "eval_steps_per_second": 0.752,
      "step": 9
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.34265734265734266,
      "eval_loss": 1.598075270652771,
      "eval_runtime": 4.3871,
      "eval_samples_per_second": 65.191,
      "eval_steps_per_second": 0.684,
      "step": 13
    },
    {
      "epoch": 4.92,
      "eval_accuracy": 0.3356643356643357,
      "eval_loss": 1.5975924730300903,
      "eval_runtime": 4.7955,
      "eval_samples_per_second": 59.639,
      "eval_steps_per_second": 0.626,
      "step": 16
    },
    {
      "epoch": 5.85,
      "eval_accuracy": 0.34265734265734266,
      "eval_loss": 1.5970256328582764,
      "eval_runtime": 4.4665,
      "eval_samples_per_second": 64.032,
      "eval_steps_per_second": 0.672,
      "step": 19
    },
    {
      "epoch": 6.77,
      "eval_accuracy": 0.33916083916083917,
      "eval_loss": 1.5963499546051025,
      "eval_runtime": 4.3016,
      "eval_samples_per_second": 66.488,
      "eval_steps_per_second": 0.697,
      "step": 22
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.3356643356643357,
      "eval_loss": 1.5952636003494263,
      "eval_runtime": 3.9531,
      "eval_samples_per_second": 72.347,
      "eval_steps_per_second": 0.759,
      "step": 26
    },
    {
      "epoch": 8.92,
      "eval_accuracy": 0.32867132867132864,
      "eval_loss": 1.594333291053772,
      "eval_runtime": 5.6915,
      "eval_samples_per_second": 50.25,
      "eval_steps_per_second": 0.527,
      "step": 29
    },
    {
      "epoch": 9.85,
      "eval_accuracy": 0.32867132867132864,
      "eval_loss": 1.5933252573013306,
      "eval_runtime": 4.4236,
      "eval_samples_per_second": 64.653,
      "eval_steps_per_second": 0.678,
      "step": 32
    },
    {
      "epoch": 10.77,
      "eval_accuracy": 0.32167832167832167,
      "eval_loss": 1.592211365699768,
      "eval_runtime": 4.9541,
      "eval_samples_per_second": 57.73,
      "eval_steps_per_second": 0.606,
      "step": 35
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.3181818181818182,
      "eval_loss": 1.5905568599700928,
      "eval_runtime": 5.1955,
      "eval_samples_per_second": 55.047,
      "eval_steps_per_second": 0.577,
      "step": 39
    },
    {
      "epoch": 12.92,
      "eval_accuracy": 0.3146853146853147,
      "eval_loss": 1.58920156955719,
      "eval_runtime": 3.6236,
      "eval_samples_per_second": 78.926,
      "eval_steps_per_second": 0.828,
      "step": 42
    },
    {
      "epoch": 13.85,
      "eval_accuracy": 0.3006993006993007,
      "eval_loss": 1.5877453088760376,
      "eval_runtime": 4.348,
      "eval_samples_per_second": 65.778,
      "eval_steps_per_second": 0.69,
      "step": 45
    },
    {
      "epoch": 14.77,
      "eval_accuracy": 0.2937062937062937,
      "eval_loss": 1.5862104892730713,
      "eval_runtime": 4.6902,
      "eval_samples_per_second": 60.978,
      "eval_steps_per_second": 0.64,
      "step": 48
    },
    {
      "epoch": 15.38,
      "grad_norm": 65952.0234375,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 1.5907,
      "step": 50
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.2972027972027972,
      "eval_loss": 1.5840750932693481,
      "eval_runtime": 4.547,
      "eval_samples_per_second": 62.899,
      "eval_steps_per_second": 0.66,
      "step": 52
    },
    {
      "epoch": 16.92,
      "eval_accuracy": 0.28321678321678323,
      "eval_loss": 1.5823713541030884,
      "eval_runtime": 5.3625,
      "eval_samples_per_second": 53.334,
      "eval_steps_per_second": 0.559,
      "step": 55
    },
    {
      "epoch": 17.85,
      "eval_accuracy": 0.27972027972027974,
      "eval_loss": 1.5806101560592651,
      "eval_runtime": 4.7671,
      "eval_samples_per_second": 59.995,
      "eval_steps_per_second": 0.629,
      "step": 58
    },
    {
      "epoch": 18.77,
      "eval_accuracy": 0.2692307692307692,
      "eval_loss": 1.5787912607192993,
      "eval_runtime": 4.3086,
      "eval_samples_per_second": 66.378,
      "eval_steps_per_second": 0.696,
      "step": 61
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.2692307692307692,
      "eval_loss": 1.576175332069397,
      "eval_runtime": 5.3175,
      "eval_samples_per_second": 53.784,
      "eval_steps_per_second": 0.564,
      "step": 65
    },
    {
      "epoch": 20.92,
      "eval_accuracy": 0.26573426573426573,
      "eval_loss": 1.5740149021148682,
      "eval_runtime": 4.5172,
      "eval_samples_per_second": 63.314,
      "eval_steps_per_second": 0.664,
      "step": 68
    },
    {
      "epoch": 21.85,
      "eval_accuracy": 0.25524475524475526,
      "eval_loss": 1.5717105865478516,
      "eval_runtime": 3.9011,
      "eval_samples_per_second": 73.312,
      "eval_steps_per_second": 0.769,
      "step": 71
    },
    {
      "epoch": 22.77,
      "eval_accuracy": 0.2517482517482518,
      "eval_loss": 1.5693939924240112,
      "eval_runtime": 3.9307,
      "eval_samples_per_second": 72.76,
      "eval_steps_per_second": 0.763,
      "step": 74
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.23776223776223776,
      "eval_loss": 1.566083312034607,
      "eval_runtime": 3.7134,
      "eval_samples_per_second": 77.019,
      "eval_steps_per_second": 0.808,
      "step": 78
    },
    {
      "epoch": 24.92,
      "eval_accuracy": 0.23426573426573427,
      "eval_loss": 1.5634570121765137,
      "eval_runtime": 4.5234,
      "eval_samples_per_second": 63.226,
      "eval_steps_per_second": 0.663,
      "step": 81
    },
    {
      "epoch": 25.85,
      "eval_accuracy": 0.22377622377622378,
      "eval_loss": 1.5608404874801636,
      "eval_runtime": 4.4129,
      "eval_samples_per_second": 64.81,
      "eval_steps_per_second": 0.68,
      "step": 84
    },
    {
      "epoch": 26.77,
      "eval_accuracy": 0.22377622377622378,
      "eval_loss": 1.5581375360488892,
      "eval_runtime": 4.7168,
      "eval_samples_per_second": 60.635,
      "eval_steps_per_second": 0.636,
      "step": 87
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5542311668395996,
      "eval_runtime": 5.4736,
      "eval_samples_per_second": 52.251,
      "eval_steps_per_second": 0.548,
      "step": 91
    },
    {
      "epoch": 28.92,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5511480569839478,
      "eval_runtime": 5.6532,
      "eval_samples_per_second": 50.591,
      "eval_steps_per_second": 0.531,
      "step": 94
    },
    {
      "epoch": 29.85,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5479341745376587,
      "eval_runtime": 5.2852,
      "eval_samples_per_second": 54.113,
      "eval_steps_per_second": 0.568,
      "step": 97
    },
    {
      "epoch": 30.77,
      "grad_norm": 68930.8125,
      "learning_rate": 2.857142857142857e-05,
      "loss": 1.5431,
      "step": 100
    },
    {
      "epoch": 30.77,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5448040962219238,
      "eval_runtime": 4.6157,
      "eval_samples_per_second": 61.962,
      "eval_steps_per_second": 0.65,
      "step": 100
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5407565832138062,
      "eval_runtime": 6.2131,
      "eval_samples_per_second": 46.032,
      "eval_steps_per_second": 0.483,
      "step": 104
    },
    {
      "epoch": 32.92,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5379865169525146,
      "eval_runtime": 4.645,
      "eval_samples_per_second": 61.571,
      "eval_steps_per_second": 0.646,
      "step": 107
    },
    {
      "epoch": 33.85,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5359200239181519,
      "eval_runtime": 5.5884,
      "eval_samples_per_second": 51.178,
      "eval_steps_per_second": 0.537,
      "step": 110
    },
    {
      "epoch": 34.77,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5345218181610107,
      "eval_runtime": 4.5718,
      "eval_samples_per_second": 62.557,
      "eval_steps_per_second": 0.656,
      "step": 113
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5334985256195068,
      "eval_runtime": 5.3526,
      "eval_samples_per_second": 53.432,
      "eval_steps_per_second": 0.56,
      "step": 117
    },
    {
      "epoch": 36.92,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5340909957885742,
      "eval_runtime": 4.471,
      "eval_samples_per_second": 63.967,
      "eval_steps_per_second": 0.671,
      "step": 120
    },
    {
      "epoch": 37.85,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5361381769180298,
      "eval_runtime": 3.5623,
      "eval_samples_per_second": 80.286,
      "eval_steps_per_second": 0.842,
      "step": 123
    },
    {
      "epoch": 38.77,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5397439002990723,
      "eval_runtime": 4.9023,
      "eval_samples_per_second": 58.34,
      "eval_steps_per_second": 0.612,
      "step": 126
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5478534698486328,
      "eval_runtime": 3.7352,
      "eval_samples_per_second": 76.569,
      "eval_steps_per_second": 0.803,
      "step": 130
    },
    {
      "epoch": 40.92,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5564229488372803,
      "eval_runtime": 4.3225,
      "eval_samples_per_second": 66.166,
      "eval_steps_per_second": 0.694,
      "step": 133
    },
    {
      "epoch": 41.85,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5678777694702148,
      "eval_runtime": 4.6076,
      "eval_samples_per_second": 62.072,
      "eval_steps_per_second": 0.651,
      "step": 136
    },
    {
      "epoch": 42.77,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.5821971893310547,
      "eval_runtime": 4.2697,
      "eval_samples_per_second": 66.983,
      "eval_steps_per_second": 0.703,
      "step": 139
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.6002099514007568,
      "eval_runtime": 4.533,
      "eval_samples_per_second": 63.094,
      "eval_steps_per_second": 0.662,
      "step": 143
    },
    {
      "epoch": 44.92,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.6109449863433838,
      "eval_runtime": 3.9799,
      "eval_samples_per_second": 71.861,
      "eval_steps_per_second": 0.754,
      "step": 146
    },
    {
      "epoch": 45.85,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.6145771741867065,
      "eval_runtime": 4.3613,
      "eval_samples_per_second": 65.576,
      "eval_steps_per_second": 0.688,
      "step": 149
    },
    {
      "epoch": 46.15,
      "grad_norm": 45833.69921875,
      "learning_rate": 2.857142857142857e-05,
      "loss": 1.4033,
      "step": 150
    },
    {
      "epoch": 46.77,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.6130825281143188,
      "eval_runtime": 4.2963,
      "eval_samples_per_second": 66.568,
      "eval_steps_per_second": 0.698,
      "step": 152
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.6008453369140625,
      "eval_runtime": 4.063,
      "eval_samples_per_second": 70.391,
      "eval_steps_per_second": 0.738,
      "step": 156
    },
    {
      "epoch": 48.92,
      "eval_accuracy": 0.24125874125874125,
      "eval_loss": 1.586226224899292,
      "eval_runtime": 4.5029,
      "eval_samples_per_second": 63.515,
      "eval_steps_per_second": 0.666,
      "step": 159
    },
    {
      "epoch": 49.85,
      "eval_accuracy": 0.2692307692307692,
      "eval_loss": 1.572645902633667,
      "eval_runtime": 5.0597,
      "eval_samples_per_second": 56.525,
      "eval_steps_per_second": 0.593,
      "step": 162
    },
    {
      "epoch": 50.77,
      "eval_accuracy": 0.2692307692307692,
      "eval_loss": 1.559901237487793,
      "eval_runtime": 4.4174,
      "eval_samples_per_second": 64.744,
      "eval_steps_per_second": 0.679,
      "step": 165
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.2867132867132867,
      "eval_loss": 1.5458828210830688,
      "eval_runtime": 4.357,
      "eval_samples_per_second": 65.642,
      "eval_steps_per_second": 0.689,
      "step": 169
    },
    {
      "epoch": 52.92,
      "eval_accuracy": 0.2937062937062937,
      "eval_loss": 1.5382803678512573,
      "eval_runtime": 5.6394,
      "eval_samples_per_second": 50.714,
      "eval_steps_per_second": 0.532,
      "step": 172
    },
    {
      "epoch": 53.85,
      "eval_accuracy": 0.3146853146853147,
      "eval_loss": 1.5310516357421875,
      "eval_runtime": 4.4695,
      "eval_samples_per_second": 63.989,
      "eval_steps_per_second": 0.671,
      "step": 175
    },
    {
      "epoch": 54.77,
      "eval_accuracy": 0.32517482517482516,
      "eval_loss": 1.5242317914962769,
      "eval_runtime": 3.8554,
      "eval_samples_per_second": 74.181,
      "eval_steps_per_second": 0.778,
      "step": 178
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.3356643356643357,
      "eval_loss": 1.5169461965560913,
      "eval_runtime": 3.9817,
      "eval_samples_per_second": 71.828,
      "eval_steps_per_second": 0.753,
      "step": 182
    },
    {
      "epoch": 56.92,
      "eval_accuracy": 0.34265734265734266,
      "eval_loss": 1.5103094577789307,
      "eval_runtime": 3.9287,
      "eval_samples_per_second": 72.797,
      "eval_steps_per_second": 0.764,
      "step": 185
    },
    {
      "epoch": 57.85,
      "eval_accuracy": 0.34615384615384615,
      "eval_loss": 1.5055506229400635,
      "eval_runtime": 4.3922,
      "eval_samples_per_second": 65.115,
      "eval_steps_per_second": 0.683,
      "step": 188
    },
    {
      "epoch": 58.77,
      "eval_accuracy": 0.34615384615384615,
      "eval_loss": 1.4995349645614624,
      "eval_runtime": 4.2261,
      "eval_samples_per_second": 67.675,
      "eval_steps_per_second": 0.71,
      "step": 191
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.34965034965034963,
      "eval_loss": 1.4939184188842773,
      "eval_runtime": 3.9946,
      "eval_samples_per_second": 71.597,
      "eval_steps_per_second": 0.751,
      "step": 195
    },
    {
      "epoch": 60.92,
      "eval_accuracy": 0.36013986013986016,
      "eval_loss": 1.4870301485061646,
      "eval_runtime": 4.7123,
      "eval_samples_per_second": 60.693,
      "eval_steps_per_second": 0.637,
      "step": 198
    },
    {
      "epoch": 61.54,
      "grad_norm": 27324.4609375,
      "learning_rate": 2.6984126984126984e-05,
      "loss": 1.2485,
      "step": 200
    },
    {
      "epoch": 61.85,
      "eval_accuracy": 0.36713286713286714,
      "eval_loss": 1.4828742742538452,
      "eval_runtime": 4.8484,
      "eval_samples_per_second": 58.989,
      "eval_steps_per_second": 0.619,
      "step": 201
    },
    {
      "epoch": 62.77,
      "eval_accuracy": 0.3741258741258741,
      "eval_loss": 1.4735387563705444,
      "eval_runtime": 4.203,
      "eval_samples_per_second": 68.047,
      "eval_steps_per_second": 0.714,
      "step": 204
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.3811188811188811,
      "eval_loss": 1.4612373113632202,
      "eval_runtime": 4.6341,
      "eval_samples_per_second": 61.716,
      "eval_steps_per_second": 0.647,
      "step": 208
    },
    {
      "epoch": 64.92,
      "eval_accuracy": 0.3986013986013986,
      "eval_loss": 1.4491915702819824,
      "eval_runtime": 3.9863,
      "eval_samples_per_second": 71.745,
      "eval_steps_per_second": 0.753,
      "step": 211
    },
    {
      "epoch": 65.85,
      "eval_accuracy": 0.4125874125874126,
      "eval_loss": 1.4364999532699585,
      "eval_runtime": 4.1321,
      "eval_samples_per_second": 69.214,
      "eval_steps_per_second": 0.726,
      "step": 214
    },
    {
      "epoch": 66.77,
      "eval_accuracy": 0.4230769230769231,
      "eval_loss": 1.4226809740066528,
      "eval_runtime": 4.2397,
      "eval_samples_per_second": 67.458,
      "eval_steps_per_second": 0.708,
      "step": 217
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.43356643356643354,
      "eval_loss": 1.4095807075500488,
      "eval_runtime": 3.8645,
      "eval_samples_per_second": 74.007,
      "eval_steps_per_second": 0.776,
      "step": 221
    },
    {
      "epoch": 68.92,
      "eval_accuracy": 0.4370629370629371,
      "eval_loss": 1.4010183811187744,
      "eval_runtime": 4.5348,
      "eval_samples_per_second": 63.068,
      "eval_steps_per_second": 0.662,
      "step": 224
    },
    {
      "epoch": 69.85,
      "eval_accuracy": 0.4405594405594406,
      "eval_loss": 1.3949679136276245,
      "eval_runtime": 4.4414,
      "eval_samples_per_second": 64.394,
      "eval_steps_per_second": 0.675,
      "step": 227
    },
    {
      "epoch": 70.77,
      "eval_accuracy": 0.4370629370629371,
      "eval_loss": 1.3919552564620972,
      "eval_runtime": 4.3028,
      "eval_samples_per_second": 66.468,
      "eval_steps_per_second": 0.697,
      "step": 230
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.4405594405594406,
      "eval_loss": 1.3798925876617432,
      "eval_runtime": 3.4387,
      "eval_samples_per_second": 83.17,
      "eval_steps_per_second": 0.872,
      "step": 234
    },
    {
      "epoch": 72.92,
      "eval_accuracy": 0.44755244755244755,
      "eval_loss": 1.366864800453186,
      "eval_runtime": 4.6503,
      "eval_samples_per_second": 61.502,
      "eval_steps_per_second": 0.645,
      "step": 237
    },
    {
      "epoch": 73.85,
      "eval_accuracy": 0.45454545454545453,
      "eval_loss": 1.3514918088912964,
      "eval_runtime": 4.5609,
      "eval_samples_per_second": 62.707,
      "eval_steps_per_second": 0.658,
      "step": 240
    },
    {
      "epoch": 74.77,
      "eval_accuracy": 0.47202797202797203,
      "eval_loss": 1.3400850296020508,
      "eval_runtime": 3.8017,
      "eval_samples_per_second": 75.229,
      "eval_steps_per_second": 0.789,
      "step": 243
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.4825174825174825,
      "eval_loss": 1.3286209106445312,
      "eval_runtime": 5.7477,
      "eval_samples_per_second": 49.759,
      "eval_steps_per_second": 0.522,
      "step": 247
    },
    {
      "epoch": 76.92,
      "grad_norm": 23198.236328125,
      "learning_rate": 2.5396825396825397e-05,
      "loss": 1.1198,
      "step": 250
    },
    {
      "epoch": 76.92,
      "eval_accuracy": 0.486013986013986,
      "eval_loss": 1.317462682723999,
      "eval_runtime": 4.5266,
      "eval_samples_per_second": 63.182,
      "eval_steps_per_second": 0.663,
      "step": 250
    },
    {
      "epoch": 77.85,
      "eval_accuracy": 0.48951048951048953,
      "eval_loss": 1.3067171573638916,
      "eval_runtime": 3.882,
      "eval_samples_per_second": 73.673,
      "eval_steps_per_second": 0.773,
      "step": 253
    },
    {
      "epoch": 78.77,
      "eval_accuracy": 0.4825174825174825,
      "eval_loss": 1.3013015985488892,
      "eval_runtime": 4.0902,
      "eval_samples_per_second": 69.923,
      "eval_steps_per_second": 0.733,
      "step": 256
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 0.479020979020979,
      "eval_loss": 1.2954434156417847,
      "eval_runtime": 5.4081,
      "eval_samples_per_second": 52.884,
      "eval_steps_per_second": 0.555,
      "step": 260
    },
    {
      "epoch": 80.92,
      "eval_accuracy": 0.486013986013986,
      "eval_loss": 1.289677381515503,
      "eval_runtime": 4.384,
      "eval_samples_per_second": 65.238,
      "eval_steps_per_second": 0.684,
      "step": 263
    },
    {
      "epoch": 81.85,
      "eval_accuracy": 0.486013986013986,
      "eval_loss": 1.283199667930603,
      "eval_runtime": 4.3325,
      "eval_samples_per_second": 66.013,
      "eval_steps_per_second": 0.692,
      "step": 266
    },
    {
      "epoch": 82.77,
      "eval_accuracy": 0.4825174825174825,
      "eval_loss": 1.2712346315383911,
      "eval_runtime": 4.6039,
      "eval_samples_per_second": 62.121,
      "eval_steps_per_second": 0.652,
      "step": 269
    },
    {
      "epoch": 84.0,
      "eval_accuracy": 0.493006993006993,
      "eval_loss": 1.2584125995635986,
      "eval_runtime": 4.5791,
      "eval_samples_per_second": 62.458,
      "eval_steps_per_second": 0.655,
      "step": 273
    },
    {
      "epoch": 84.92,
      "eval_accuracy": 0.4965034965034965,
      "eval_loss": 1.2516244649887085,
      "eval_runtime": 4.8825,
      "eval_samples_per_second": 58.577,
      "eval_steps_per_second": 0.614,
      "step": 276
    },
    {
      "epoch": 85.85,
      "eval_accuracy": 0.5,
      "eval_loss": 1.2455971240997314,
      "eval_runtime": 3.9744,
      "eval_samples_per_second": 71.96,
      "eval_steps_per_second": 0.755,
      "step": 279
    },
    {
      "epoch": 86.77,
      "eval_accuracy": 0.5104895104895105,
      "eval_loss": 1.2443982362747192,
      "eval_runtime": 4.5207,
      "eval_samples_per_second": 63.265,
      "eval_steps_per_second": 0.664,
      "step": 282
    },
    {
      "epoch": 88.0,
      "eval_accuracy": 0.5104895104895105,
      "eval_loss": 1.2373132705688477,
      "eval_runtime": 5.6152,
      "eval_samples_per_second": 50.933,
      "eval_steps_per_second": 0.534,
      "step": 286
    },
    {
      "epoch": 88.92,
      "eval_accuracy": 0.513986013986014,
      "eval_loss": 1.2309471368789673,
      "eval_runtime": 4.7969,
      "eval_samples_per_second": 59.622,
      "eval_steps_per_second": 0.625,
      "step": 289
    },
    {
      "epoch": 89.85,
      "eval_accuracy": 0.5209790209790209,
      "eval_loss": 1.2219436168670654,
      "eval_runtime": 4.2518,
      "eval_samples_per_second": 67.266,
      "eval_steps_per_second": 0.706,
      "step": 292
    },
    {
      "epoch": 90.77,
      "eval_accuracy": 0.5209790209790209,
      "eval_loss": 1.2145464420318604,
      "eval_runtime": 4.6368,
      "eval_samples_per_second": 61.68,
      "eval_steps_per_second": 0.647,
      "step": 295
    },
    {
      "epoch": 92.0,
      "eval_accuracy": 0.527972027972028,
      "eval_loss": 1.2054263353347778,
      "eval_runtime": 4.2071,
      "eval_samples_per_second": 67.98,
      "eval_steps_per_second": 0.713,
      "step": 299
    },
    {
      "epoch": 92.31,
      "grad_norm": 29195.7578125,
      "learning_rate": 2.380952380952381e-05,
      "loss": 0.9915,
      "step": 300
    },
    {
      "epoch": 92.92,
      "eval_accuracy": 0.534965034965035,
      "eval_loss": 1.1981616020202637,
      "eval_runtime": 4.3609,
      "eval_samples_per_second": 65.583,
      "eval_steps_per_second": 0.688,
      "step": 302
    },
    {
      "epoch": 93.85,
      "eval_accuracy": 0.5384615384615384,
      "eval_loss": 1.1913262605667114,
      "eval_runtime": 3.9073,
      "eval_samples_per_second": 73.197,
      "eval_steps_per_second": 0.768,
      "step": 305
    },
    {
      "epoch": 94.77,
      "eval_accuracy": 0.5454545454545454,
      "eval_loss": 1.185881495475769,
      "eval_runtime": 3.928,
      "eval_samples_per_second": 72.811,
      "eval_steps_per_second": 0.764,
      "step": 308
    },
    {
      "epoch": 96.0,
      "eval_accuracy": 0.548951048951049,
      "eval_loss": 1.179394006729126,
      "eval_runtime": 4.1933,
      "eval_samples_per_second": 68.204,
      "eval_steps_per_second": 0.715,
      "step": 312
    },
    {
      "epoch": 96.92,
      "eval_accuracy": 0.5454545454545454,
      "eval_loss": 1.1733678579330444,
      "eval_runtime": 5.0205,
      "eval_samples_per_second": 56.967,
      "eval_steps_per_second": 0.598,
      "step": 315
    },
    {
      "epoch": 97.85,
      "eval_accuracy": 0.5524475524475524,
      "eval_loss": 1.1637603044509888,
      "eval_runtime": 4.8886,
      "eval_samples_per_second": 58.503,
      "eval_steps_per_second": 0.614,
      "step": 318
    },
    {
      "epoch": 98.77,
      "eval_accuracy": 0.5524475524475524,
      "eval_loss": 1.1549575328826904,
      "eval_runtime": 4.9266,
      "eval_samples_per_second": 58.052,
      "eval_steps_per_second": 0.609,
      "step": 321
    },
    {
      "epoch": 100.0,
      "eval_accuracy": 0.548951048951049,
      "eval_loss": 1.1464989185333252,
      "eval_runtime": 4.7642,
      "eval_samples_per_second": 60.032,
      "eval_steps_per_second": 0.63,
      "step": 325
    },
    {
      "epoch": 100.92,
      "eval_accuracy": 0.5594405594405595,
      "eval_loss": 1.1443748474121094,
      "eval_runtime": 4.7025,
      "eval_samples_per_second": 60.819,
      "eval_steps_per_second": 0.638,
      "step": 328
    },
    {
      "epoch": 101.85,
      "eval_accuracy": 0.5629370629370629,
      "eval_loss": 1.1359333992004395,
      "eval_runtime": 4.6342,
      "eval_samples_per_second": 61.715,
      "eval_steps_per_second": 0.647,
      "step": 331
    },
    {
      "epoch": 102.77,
      "eval_accuracy": 0.5664335664335665,
      "eval_loss": 1.1271060705184937,
      "eval_runtime": 4.4245,
      "eval_samples_per_second": 64.639,
      "eval_steps_per_second": 0.678,
      "step": 334
    },
    {
      "epoch": 104.0,
      "eval_accuracy": 0.5769230769230769,
      "eval_loss": 1.109040379524231,
      "eval_runtime": 4.9047,
      "eval_samples_per_second": 58.311,
      "eval_steps_per_second": 0.612,
      "step": 338
    },
    {
      "epoch": 104.92,
      "eval_accuracy": 0.5944055944055944,
      "eval_loss": 1.0972033739089966,
      "eval_runtime": 4.5473,
      "eval_samples_per_second": 62.895,
      "eval_steps_per_second": 0.66,
      "step": 341
    },
    {
      "epoch": 105.85,
      "eval_accuracy": 0.6013986013986014,
      "eval_loss": 1.090105414390564,
      "eval_runtime": 3.7875,
      "eval_samples_per_second": 75.511,
      "eval_steps_per_second": 0.792,
      "step": 344
    },
    {
      "epoch": 106.77,
      "eval_accuracy": 0.6083916083916084,
      "eval_loss": 1.0809463262557983,
      "eval_runtime": 4.7656,
      "eval_samples_per_second": 60.014,
      "eval_steps_per_second": 0.63,
      "step": 347
    },
    {
      "epoch": 107.69,
      "grad_norm": 32308.33984375,
      "learning_rate": 2.222222222222222e-05,
      "loss": 0.8834,
      "step": 350
    },
    {
      "epoch": 108.0,
      "eval_accuracy": 0.6118881118881119,
      "eval_loss": 1.0683268308639526,
      "eval_runtime": 4.3145,
      "eval_samples_per_second": 66.288,
      "eval_steps_per_second": 0.695,
      "step": 351
    },
    {
      "epoch": 108.92,
      "eval_accuracy": 0.6223776223776224,
      "eval_loss": 1.0605404376983643,
      "eval_runtime": 4.6097,
      "eval_samples_per_second": 62.043,
      "eval_steps_per_second": 0.651,
      "step": 354
    },
    {
      "epoch": 109.85,
      "eval_accuracy": 0.6258741258741258,
      "eval_loss": 1.0562984943389893,
      "eval_runtime": 4.859,
      "eval_samples_per_second": 58.86,
      "eval_steps_per_second": 0.617,
      "step": 357
    },
    {
      "epoch": 110.77,
      "eval_accuracy": 0.6223776223776224,
      "eval_loss": 1.0537959337234497,
      "eval_runtime": 4.948,
      "eval_samples_per_second": 57.801,
      "eval_steps_per_second": 0.606,
      "step": 360
    },
    {
      "epoch": 112.0,
      "eval_accuracy": 0.6153846153846154,
      "eval_loss": 1.0491102933883667,
      "eval_runtime": 4.1434,
      "eval_samples_per_second": 69.026,
      "eval_steps_per_second": 0.724,
      "step": 364
    },
    {
      "epoch": 112.92,
      "eval_accuracy": 0.6118881118881119,
      "eval_loss": 1.044057011604309,
      "eval_runtime": 4.3774,
      "eval_samples_per_second": 65.336,
      "eval_steps_per_second": 0.685,
      "step": 367
    },
    {
      "epoch": 113.85,
      "eval_accuracy": 0.6118881118881119,
      "eval_loss": 1.0357924699783325,
      "eval_runtime": 4.7038,
      "eval_samples_per_second": 60.801,
      "eval_steps_per_second": 0.638,
      "step": 370
    },
    {
      "epoch": 114.77,
      "eval_accuracy": 0.6223776223776224,
      "eval_loss": 1.0194157361984253,
      "eval_runtime": 5.0902,
      "eval_samples_per_second": 56.187,
      "eval_steps_per_second": 0.589,
      "step": 373
    },
    {
      "epoch": 116.0,
      "eval_accuracy": 0.6293706293706294,
      "eval_loss": 1.0034115314483643,
      "eval_runtime": 4.386,
      "eval_samples_per_second": 65.208,
      "eval_steps_per_second": 0.684,
      "step": 377
    },
    {
      "epoch": 116.92,
      "eval_accuracy": 0.6258741258741258,
      "eval_loss": 0.9991269707679749,
      "eval_runtime": 5.2708,
      "eval_samples_per_second": 54.261,
      "eval_steps_per_second": 0.569,
      "step": 380
    },
    {
      "epoch": 117.85,
      "eval_accuracy": 0.6258741258741258,
      "eval_loss": 0.9959561824798584,
      "eval_runtime": 4.7556,
      "eval_samples_per_second": 60.139,
      "eval_steps_per_second": 0.631,
      "step": 383
    },
    {
      "epoch": 118.77,
      "eval_accuracy": 0.6293706293706294,
      "eval_loss": 0.9911425113677979,
      "eval_runtime": 4.0817,
      "eval_samples_per_second": 70.068,
      "eval_steps_per_second": 0.735,
      "step": 386
    },
    {
      "epoch": 120.0,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9834115505218506,
      "eval_runtime": 4.0058,
      "eval_samples_per_second": 71.396,
      "eval_steps_per_second": 0.749,
      "step": 390
    },
    {
      "epoch": 120.92,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9775691628456116,
      "eval_runtime": 4.3856,
      "eval_samples_per_second": 65.214,
      "eval_steps_per_second": 0.684,
      "step": 393
    },
    {
      "epoch": 121.85,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9772741198539734,
      "eval_runtime": 4.6976,
      "eval_samples_per_second": 60.882,
      "eval_steps_per_second": 0.639,
      "step": 396
    },
    {
      "epoch": 122.77,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9734641909599304,
      "eval_runtime": 4.6506,
      "eval_samples_per_second": 61.498,
      "eval_steps_per_second": 0.645,
      "step": 399
    },
    {
      "epoch": 123.08,
      "grad_norm": 27630.990234375,
      "learning_rate": 2.0634920634920633e-05,
      "loss": 0.7786,
      "step": 400
    },
    {
      "epoch": 124.0,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9730696082115173,
      "eval_runtime": 3.9976,
      "eval_samples_per_second": 71.542,
      "eval_steps_per_second": 0.75,
      "step": 403
    },
    {
      "epoch": 124.92,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9727755188941956,
      "eval_runtime": 4.0553,
      "eval_samples_per_second": 70.525,
      "eval_steps_per_second": 0.74,
      "step": 406
    },
    {
      "epoch": 125.85,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9657326936721802,
      "eval_runtime": 4.4666,
      "eval_samples_per_second": 64.031,
      "eval_steps_per_second": 0.672,
      "step": 409
    },
    {
      "epoch": 126.77,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9547586441040039,
      "eval_runtime": 4.6999,
      "eval_samples_per_second": 60.852,
      "eval_steps_per_second": 0.638,
      "step": 412
    },
    {
      "epoch": 128.0,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.942358136177063,
      "eval_runtime": 4.8438,
      "eval_samples_per_second": 59.045,
      "eval_steps_per_second": 0.619,
      "step": 416
    },
    {
      "epoch": 128.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9391436576843262,
      "eval_runtime": 4.4506,
      "eval_samples_per_second": 64.261,
      "eval_steps_per_second": 0.674,
      "step": 419
    },
    {
      "epoch": 129.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9418392777442932,
      "eval_runtime": 4.2912,
      "eval_samples_per_second": 66.648,
      "eval_steps_per_second": 0.699,
      "step": 422
    },
    {
      "epoch": 130.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9476207494735718,
      "eval_runtime": 4.7281,
      "eval_samples_per_second": 60.49,
      "eval_steps_per_second": 0.635,
      "step": 425
    },
    {
      "epoch": 132.0,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9457269310951233,
      "eval_runtime": 4.314,
      "eval_samples_per_second": 66.295,
      "eval_steps_per_second": 0.695,
      "step": 429
    },
    {
      "epoch": 132.92,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.941338062286377,
      "eval_runtime": 3.916,
      "eval_samples_per_second": 73.033,
      "eval_steps_per_second": 0.766,
      "step": 432
    },
    {
      "epoch": 133.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9334166049957275,
      "eval_runtime": 4.5886,
      "eval_samples_per_second": 62.329,
      "eval_steps_per_second": 0.654,
      "step": 435
    },
    {
      "epoch": 134.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9328890442848206,
      "eval_runtime": 4.1417,
      "eval_samples_per_second": 69.054,
      "eval_steps_per_second": 0.724,
      "step": 438
    },
    {
      "epoch": 136.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9333996772766113,
      "eval_runtime": 4.538,
      "eval_samples_per_second": 63.023,
      "eval_steps_per_second": 0.661,
      "step": 442
    },
    {
      "epoch": 136.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9264596700668335,
      "eval_runtime": 4.6642,
      "eval_samples_per_second": 61.318,
      "eval_steps_per_second": 0.643,
      "step": 445
    },
    {
      "epoch": 137.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9186587929725647,
      "eval_runtime": 4.5978,
      "eval_samples_per_second": 62.204,
      "eval_steps_per_second": 0.652,
      "step": 448
    },
    {
      "epoch": 138.46,
      "grad_norm": 34684.0078125,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 0.7133,
      "step": 450
    },
    {
      "epoch": 138.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.916916012763977,
      "eval_runtime": 4.1718,
      "eval_samples_per_second": 68.556,
      "eval_steps_per_second": 0.719,
      "step": 451
    },
    {
      "epoch": 140.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9141567349433899,
      "eval_runtime": 4.8158,
      "eval_samples_per_second": 59.388,
      "eval_steps_per_second": 0.623,
      "step": 455
    },
    {
      "epoch": 140.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9131244421005249,
      "eval_runtime": 4.3984,
      "eval_samples_per_second": 65.024,
      "eval_steps_per_second": 0.682,
      "step": 458
    },
    {
      "epoch": 141.85,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.9160958528518677,
      "eval_runtime": 3.9738,
      "eval_samples_per_second": 71.971,
      "eval_steps_per_second": 0.755,
      "step": 461
    },
    {
      "epoch": 142.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9223662614822388,
      "eval_runtime": 3.7836,
      "eval_samples_per_second": 75.589,
      "eval_steps_per_second": 0.793,
      "step": 464
    },
    {
      "epoch": 144.0,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.9139449000358582,
      "eval_runtime": 4.0554,
      "eval_samples_per_second": 70.522,
      "eval_steps_per_second": 0.74,
      "step": 468
    },
    {
      "epoch": 144.92,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.9089756608009338,
      "eval_runtime": 4.4989,
      "eval_samples_per_second": 63.571,
      "eval_steps_per_second": 0.667,
      "step": 471
    },
    {
      "epoch": 145.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9072948694229126,
      "eval_runtime": 3.984,
      "eval_samples_per_second": 71.788,
      "eval_steps_per_second": 0.753,
      "step": 474
    },
    {
      "epoch": 146.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9110231995582581,
      "eval_runtime": 4.596,
      "eval_samples_per_second": 62.228,
      "eval_steps_per_second": 0.653,
      "step": 477
    },
    {
      "epoch": 148.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9167369604110718,
      "eval_runtime": 4.7051,
      "eval_samples_per_second": 60.785,
      "eval_steps_per_second": 0.638,
      "step": 481
    },
    {
      "epoch": 148.92,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9118071794509888,
      "eval_runtime": 3.9295,
      "eval_samples_per_second": 72.783,
      "eval_steps_per_second": 0.763,
      "step": 484
    },
    {
      "epoch": 149.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.8996461629867554,
      "eval_runtime": 4.5063,
      "eval_samples_per_second": 63.466,
      "eval_steps_per_second": 0.666,
      "step": 487
    },
    {
      "epoch": 150.77,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.8903929591178894,
      "eval_runtime": 4.0074,
      "eval_samples_per_second": 71.369,
      "eval_steps_per_second": 0.749,
      "step": 490
    },
    {
      "epoch": 152.0,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.8889052867889404,
      "eval_runtime": 4.2482,
      "eval_samples_per_second": 67.323,
      "eval_steps_per_second": 0.706,
      "step": 494
    },
    {
      "epoch": 152.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.889894425868988,
      "eval_runtime": 4.7658,
      "eval_samples_per_second": 60.011,
      "eval_steps_per_second": 0.629,
      "step": 497
    },
    {
      "epoch": 153.85,
      "grad_norm": 27670.865234375,
      "learning_rate": 1.746031746031746e-05,
      "loss": 0.6674,
      "step": 500
    },
    {
      "epoch": 153.85,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.887377917766571,
      "eval_runtime": 4.6951,
      "eval_samples_per_second": 60.915,
      "eval_steps_per_second": 0.639,
      "step": 500
    },
    {
      "epoch": 154.77,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.8873924016952515,
      "eval_runtime": 3.8042,
      "eval_samples_per_second": 75.181,
      "eval_steps_per_second": 0.789,
      "step": 503
    },
    {
      "epoch": 156.0,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.8905075788497925,
      "eval_runtime": 3.9282,
      "eval_samples_per_second": 72.806,
      "eval_steps_per_second": 0.764,
      "step": 507
    },
    {
      "epoch": 156.92,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8881194591522217,
      "eval_runtime": 4.2085,
      "eval_samples_per_second": 67.957,
      "eval_steps_per_second": 0.713,
      "step": 510
    },
    {
      "epoch": 157.85,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.882903516292572,
      "eval_runtime": 5.345,
      "eval_samples_per_second": 53.508,
      "eval_steps_per_second": 0.561,
      "step": 513
    },
    {
      "epoch": 158.77,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8809071183204651,
      "eval_runtime": 4.4142,
      "eval_samples_per_second": 64.791,
      "eval_steps_per_second": 0.68,
      "step": 516
    },
    {
      "epoch": 160.0,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8780828714370728,
      "eval_runtime": 3.6498,
      "eval_samples_per_second": 78.361,
      "eval_steps_per_second": 0.822,
      "step": 520
    },
    {
      "epoch": 160.92,
      "eval_accuracy": 0.6818181818181818,
      "eval_loss": 0.8776365518569946,
      "eval_runtime": 3.4668,
      "eval_samples_per_second": 82.497,
      "eval_steps_per_second": 0.865,
      "step": 523
    },
    {
      "epoch": 161.85,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8795685768127441,
      "eval_runtime": 3.8004,
      "eval_samples_per_second": 75.256,
      "eval_steps_per_second": 0.789,
      "step": 526
    },
    {
      "epoch": 162.77,
      "eval_accuracy": 0.6818181818181818,
      "eval_loss": 0.8795468807220459,
      "eval_runtime": 3.8694,
      "eval_samples_per_second": 73.913,
      "eval_steps_per_second": 0.775,
      "step": 529
    },
    {
      "epoch": 164.0,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8797011971473694,
      "eval_runtime": 4.1348,
      "eval_samples_per_second": 69.169,
      "eval_steps_per_second": 0.726,
      "step": 533
    },
    {
      "epoch": 164.92,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8706856966018677,
      "eval_runtime": 4.5762,
      "eval_samples_per_second": 62.498,
      "eval_steps_per_second": 0.656,
      "step": 536
    },
    {
      "epoch": 165.85,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8697258830070496,
      "eval_runtime": 3.5794,
      "eval_samples_per_second": 79.901,
      "eval_steps_per_second": 0.838,
      "step": 539
    },
    {
      "epoch": 166.77,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8723975419998169,
      "eval_runtime": 5.761,
      "eval_samples_per_second": 49.644,
      "eval_steps_per_second": 0.521,
      "step": 542
    },
    {
      "epoch": 168.0,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.870445966720581,
      "eval_runtime": 4.2907,
      "eval_samples_per_second": 66.656,
      "eval_steps_per_second": 0.699,
      "step": 546
    },
    {
      "epoch": 168.92,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.8693636655807495,
      "eval_runtime": 4.5637,
      "eval_samples_per_second": 62.668,
      "eval_steps_per_second": 0.657,
      "step": 549
    },
    {
      "epoch": 169.23,
      "grad_norm": 67537.203125,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 0.6305,
      "step": 550
    },
    {
      "epoch": 169.85,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.8739539980888367,
      "eval_runtime": 4.5496,
      "eval_samples_per_second": 62.862,
      "eval_steps_per_second": 0.659,
      "step": 552
    },
    {
      "epoch": 170.77,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.8713040947914124,
      "eval_runtime": 4.3907,
      "eval_samples_per_second": 65.138,
      "eval_steps_per_second": 0.683,
      "step": 555
    },
    {
      "epoch": 172.0,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8682331442832947,
      "eval_runtime": 4.1777,
      "eval_samples_per_second": 68.459,
      "eval_steps_per_second": 0.718,
      "step": 559
    },
    {
      "epoch": 172.92,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.868798553943634,
      "eval_runtime": 3.5218,
      "eval_samples_per_second": 81.207,
      "eval_steps_per_second": 0.852,
      "step": 562
    },
    {
      "epoch": 173.85,
      "eval_accuracy": 0.6818181818181818,
      "eval_loss": 0.8692768216133118,
      "eval_runtime": 5.0064,
      "eval_samples_per_second": 57.127,
      "eval_steps_per_second": 0.599,
      "step": 565
    },
    {
      "epoch": 174.77,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.874369204044342,
      "eval_runtime": 4.1257,
      "eval_samples_per_second": 69.322,
      "eval_steps_per_second": 0.727,
      "step": 568
    },
    {
      "epoch": 176.0,
      "eval_accuracy": 0.6783216783216783,
      "eval_loss": 0.8759630918502808,
      "eval_runtime": 4.4848,
      "eval_samples_per_second": 63.771,
      "eval_steps_per_second": 0.669,
      "step": 572
    },
    {
      "epoch": 176.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.8696449398994446,
      "eval_runtime": 4.1683,
      "eval_samples_per_second": 68.613,
      "eval_steps_per_second": 0.72,
      "step": 575
    },
    {
      "epoch": 177.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.8668593764305115,
      "eval_runtime": 4.3889,
      "eval_samples_per_second": 65.165,
      "eval_steps_per_second": 0.684,
      "step": 578
    },
    {
      "epoch": 178.77,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.8641146421432495,
      "eval_runtime": 4.0742,
      "eval_samples_per_second": 70.197,
      "eval_steps_per_second": 0.736,
      "step": 581
    },
    {
      "epoch": 180.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.8696537613868713,
      "eval_runtime": 4.1345,
      "eval_samples_per_second": 69.173,
      "eval_steps_per_second": 0.726,
      "step": 585
    },
    {
      "epoch": 180.92,
      "eval_accuracy": 0.6748251748251748,
      "eval_loss": 0.8678367733955383,
      "eval_runtime": 3.994,
      "eval_samples_per_second": 71.607,
      "eval_steps_per_second": 0.751,
      "step": 588
    },
    {
      "epoch": 181.85,
      "eval_accuracy": 0.6818181818181818,
      "eval_loss": 0.8620542287826538,
      "eval_runtime": 4.32,
      "eval_samples_per_second": 66.204,
      "eval_steps_per_second": 0.694,
      "step": 591
    },
    {
      "epoch": 182.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8557011485099792,
      "eval_runtime": 4.7717,
      "eval_samples_per_second": 59.937,
      "eval_steps_per_second": 0.629,
      "step": 594
    },
    {
      "epoch": 184.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.848114013671875,
      "eval_runtime": 4.0948,
      "eval_samples_per_second": 69.845,
      "eval_steps_per_second": 0.733,
      "step": 598
    },
    {
      "epoch": 184.62,
      "grad_norm": 36502.2421875,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.6095,
      "step": 600
    },
    {
      "epoch": 184.92,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8428906798362732,
      "eval_runtime": 4.6887,
      "eval_samples_per_second": 60.997,
      "eval_steps_per_second": 0.64,
      "step": 601
    },
    {
      "epoch": 185.85,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8413122892379761,
      "eval_runtime": 3.8998,
      "eval_samples_per_second": 73.337,
      "eval_steps_per_second": 0.769,
      "step": 604
    },
    {
      "epoch": 186.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8402045965194702,
      "eval_runtime": 4.1508,
      "eval_samples_per_second": 68.903,
      "eval_steps_per_second": 0.723,
      "step": 607
    },
    {
      "epoch": 188.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8415275812149048,
      "eval_runtime": 4.4966,
      "eval_samples_per_second": 63.603,
      "eval_steps_per_second": 0.667,
      "step": 611
    },
    {
      "epoch": 188.92,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8409523963928223,
      "eval_runtime": 4.0007,
      "eval_samples_per_second": 71.488,
      "eval_steps_per_second": 0.75,
      "step": 614
    },
    {
      "epoch": 189.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.8388563394546509,
      "eval_runtime": 4.5212,
      "eval_samples_per_second": 63.257,
      "eval_steps_per_second": 0.664,
      "step": 617
    },
    {
      "epoch": 190.77,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.8353860378265381,
      "eval_runtime": 4.6112,
      "eval_samples_per_second": 62.023,
      "eval_steps_per_second": 0.651,
      "step": 620
    },
    {
      "epoch": 192.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8356983661651611,
      "eval_runtime": 4.6563,
      "eval_samples_per_second": 61.422,
      "eval_steps_per_second": 0.644,
      "step": 624
    },
    {
      "epoch": 192.92,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8400572538375854,
      "eval_runtime": 5.369,
      "eval_samples_per_second": 53.269,
      "eval_steps_per_second": 0.559,
      "step": 627
    },
    {
      "epoch": 193.85,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.844892144203186,
      "eval_runtime": 4.0956,
      "eval_samples_per_second": 69.831,
      "eval_steps_per_second": 0.732,
      "step": 630
    },
    {
      "epoch": 194.77,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8478845357894897,
      "eval_runtime": 4.6385,
      "eval_samples_per_second": 61.658,
      "eval_steps_per_second": 0.647,
      "step": 633
    },
    {
      "epoch": 196.0,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8454630374908447,
      "eval_runtime": 4.4423,
      "eval_samples_per_second": 64.381,
      "eval_steps_per_second": 0.675,
      "step": 637
    },
    {
      "epoch": 196.92,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8421822190284729,
      "eval_runtime": 3.8632,
      "eval_samples_per_second": 74.032,
      "eval_steps_per_second": 0.777,
      "step": 640
    },
    {
      "epoch": 197.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8425044417381287,
      "eval_runtime": 5.1031,
      "eval_samples_per_second": 56.044,
      "eval_steps_per_second": 0.588,
      "step": 643
    },
    {
      "epoch": 198.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8436546325683594,
      "eval_runtime": 4.9685,
      "eval_samples_per_second": 57.562,
      "eval_steps_per_second": 0.604,
      "step": 646
    },
    {
      "epoch": 200.0,
      "grad_norm": 66285.84375,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 0.5908,
      "step": 650
    },
    {
      "epoch": 200.0,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8366544246673584,
      "eval_runtime": 4.3292,
      "eval_samples_per_second": 66.063,
      "eval_steps_per_second": 0.693,
      "step": 650
    },
    {
      "epoch": 200.92,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.834704577922821,
      "eval_runtime": 4.7887,
      "eval_samples_per_second": 59.724,
      "eval_steps_per_second": 0.626,
      "step": 653
    },
    {
      "epoch": 201.85,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8286824226379395,
      "eval_runtime": 4.388,
      "eval_samples_per_second": 65.178,
      "eval_steps_per_second": 0.684,
      "step": 656
    },
    {
      "epoch": 202.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8259890079498291,
      "eval_runtime": 3.7365,
      "eval_samples_per_second": 76.543,
      "eval_steps_per_second": 0.803,
      "step": 659
    },
    {
      "epoch": 204.0,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8263576626777649,
      "eval_runtime": 4.9175,
      "eval_samples_per_second": 58.159,
      "eval_steps_per_second": 0.61,
      "step": 663
    },
    {
      "epoch": 204.92,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8295235633850098,
      "eval_runtime": 4.3071,
      "eval_samples_per_second": 66.401,
      "eval_steps_per_second": 0.697,
      "step": 666
    },
    {
      "epoch": 205.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8301726579666138,
      "eval_runtime": 3.7499,
      "eval_samples_per_second": 76.268,
      "eval_steps_per_second": 0.8,
      "step": 669
    },
    {
      "epoch": 206.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.828461766242981,
      "eval_runtime": 3.8022,
      "eval_samples_per_second": 75.219,
      "eval_steps_per_second": 0.789,
      "step": 672
    },
    {
      "epoch": 208.0,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.831078052520752,
      "eval_runtime": 4.2868,
      "eval_samples_per_second": 66.716,
      "eval_steps_per_second": 0.7,
      "step": 676
    },
    {
      "epoch": 208.92,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8320910334587097,
      "eval_runtime": 4.474,
      "eval_samples_per_second": 63.925,
      "eval_steps_per_second": 0.671,
      "step": 679
    },
    {
      "epoch": 209.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8305550813674927,
      "eval_runtime": 4.1246,
      "eval_samples_per_second": 69.341,
      "eval_steps_per_second": 0.727,
      "step": 682
    },
    {
      "epoch": 210.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8302868604660034,
      "eval_runtime": 4.9131,
      "eval_samples_per_second": 58.212,
      "eval_steps_per_second": 0.611,
      "step": 685
    },
    {
      "epoch": 212.0,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.8256182670593262,
      "eval_runtime": 4.5542,
      "eval_samples_per_second": 62.8,
      "eval_steps_per_second": 0.659,
      "step": 689
    },
    {
      "epoch": 212.92,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8230299353599548,
      "eval_runtime": 4.2845,
      "eval_samples_per_second": 66.752,
      "eval_steps_per_second": 0.7,
      "step": 692
    },
    {
      "epoch": 213.85,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.819442868232727,
      "eval_runtime": 4.4153,
      "eval_samples_per_second": 64.775,
      "eval_steps_per_second": 0.679,
      "step": 695
    },
    {
      "epoch": 214.77,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8183168768882751,
      "eval_runtime": 4.9672,
      "eval_samples_per_second": 57.577,
      "eval_steps_per_second": 0.604,
      "step": 698
    },
    {
      "epoch": 215.38,
      "grad_norm": 29832.03125,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.5763,
      "step": 700
    },
    {
      "epoch": 216.0,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8231977224349976,
      "eval_runtime": 4.6354,
      "eval_samples_per_second": 61.699,
      "eval_steps_per_second": 0.647,
      "step": 702
    },
    {
      "epoch": 216.92,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8236932158470154,
      "eval_runtime": 3.7182,
      "eval_samples_per_second": 76.92,
      "eval_steps_per_second": 0.807,
      "step": 705
    },
    {
      "epoch": 217.85,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.8195610642433167,
      "eval_runtime": 3.5502,
      "eval_samples_per_second": 80.56,
      "eval_steps_per_second": 0.845,
      "step": 708
    },
    {
      "epoch": 218.77,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.8142436742782593,
      "eval_runtime": 4.9155,
      "eval_samples_per_second": 58.184,
      "eval_steps_per_second": 0.61,
      "step": 711
    },
    {
      "epoch": 220.0,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.8115321397781372,
      "eval_runtime": 4.0939,
      "eval_samples_per_second": 69.86,
      "eval_steps_per_second": 0.733,
      "step": 715
    },
    {
      "epoch": 220.92,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.8130100965499878,
      "eval_runtime": 4.2197,
      "eval_samples_per_second": 67.777,
      "eval_steps_per_second": 0.711,
      "step": 718
    },
    {
      "epoch": 221.85,
      "eval_accuracy": 0.7027972027972028,
      "eval_loss": 0.8156144022941589,
      "eval_runtime": 4.2344,
      "eval_samples_per_second": 67.542,
      "eval_steps_per_second": 0.708,
      "step": 721
    },
    {
      "epoch": 222.77,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8200713992118835,
      "eval_runtime": 4.8181,
      "eval_samples_per_second": 59.36,
      "eval_steps_per_second": 0.623,
      "step": 724
    },
    {
      "epoch": 224.0,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8227414488792419,
      "eval_runtime": 4.5671,
      "eval_samples_per_second": 62.621,
      "eval_steps_per_second": 0.657,
      "step": 728
    },
    {
      "epoch": 224.92,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8232228755950928,
      "eval_runtime": 5.221,
      "eval_samples_per_second": 54.779,
      "eval_steps_per_second": 0.575,
      "step": 731
    },
    {
      "epoch": 225.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8198325634002686,
      "eval_runtime": 4.2136,
      "eval_samples_per_second": 67.875,
      "eval_steps_per_second": 0.712,
      "step": 734
    },
    {
      "epoch": 226.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8151125311851501,
      "eval_runtime": 4.8801,
      "eval_samples_per_second": 58.606,
      "eval_steps_per_second": 0.615,
      "step": 737
    },
    {
      "epoch": 228.0,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8136410713195801,
      "eval_runtime": 5.2461,
      "eval_samples_per_second": 54.516,
      "eval_steps_per_second": 0.572,
      "step": 741
    },
    {
      "epoch": 228.92,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8134062886238098,
      "eval_runtime": 3.6429,
      "eval_samples_per_second": 78.509,
      "eval_steps_per_second": 0.824,
      "step": 744
    },
    {
      "epoch": 229.85,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8123226761817932,
      "eval_runtime": 4.8374,
      "eval_samples_per_second": 59.122,
      "eval_steps_per_second": 0.62,
      "step": 747
    },
    {
      "epoch": 230.77,
      "grad_norm": 27062.134765625,
      "learning_rate": 9.523809523809523e-06,
      "loss": 0.57,
      "step": 750
    },
    {
      "epoch": 230.77,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8095433115959167,
      "eval_runtime": 3.9409,
      "eval_samples_per_second": 72.572,
      "eval_steps_per_second": 0.761,
      "step": 750
    },
    {
      "epoch": 232.0,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8082302212715149,
      "eval_runtime": 4.0933,
      "eval_samples_per_second": 69.87,
      "eval_steps_per_second": 0.733,
      "step": 754
    },
    {
      "epoch": 232.92,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8084114193916321,
      "eval_runtime": 4.4952,
      "eval_samples_per_second": 63.624,
      "eval_steps_per_second": 0.667,
      "step": 757
    },
    {
      "epoch": 233.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8113557696342468,
      "eval_runtime": 4.6955,
      "eval_samples_per_second": 60.909,
      "eval_steps_per_second": 0.639,
      "step": 760
    },
    {
      "epoch": 234.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8130276799201965,
      "eval_runtime": 4.9303,
      "eval_samples_per_second": 58.009,
      "eval_steps_per_second": 0.608,
      "step": 763
    },
    {
      "epoch": 236.0,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8153804540634155,
      "eval_runtime": 3.6663,
      "eval_samples_per_second": 78.007,
      "eval_steps_per_second": 0.818,
      "step": 767
    },
    {
      "epoch": 236.92,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8160205483436584,
      "eval_runtime": 4.6226,
      "eval_samples_per_second": 61.87,
      "eval_steps_per_second": 0.649,
      "step": 770
    },
    {
      "epoch": 237.85,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8126419186592102,
      "eval_runtime": 4.6278,
      "eval_samples_per_second": 61.801,
      "eval_steps_per_second": 0.648,
      "step": 773
    },
    {
      "epoch": 238.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8113960027694702,
      "eval_runtime": 3.8362,
      "eval_samples_per_second": 74.552,
      "eval_steps_per_second": 0.782,
      "step": 776
    },
    {
      "epoch": 240.0,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8041169047355652,
      "eval_runtime": 5.2095,
      "eval_samples_per_second": 54.9,
      "eval_steps_per_second": 0.576,
      "step": 780
    },
    {
      "epoch": 240.92,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8005608916282654,
      "eval_runtime": 4.0128,
      "eval_samples_per_second": 71.273,
      "eval_steps_per_second": 0.748,
      "step": 783
    },
    {
      "epoch": 241.85,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.7987480163574219,
      "eval_runtime": 4.8789,
      "eval_samples_per_second": 58.619,
      "eval_steps_per_second": 0.615,
      "step": 786
    },
    {
      "epoch": 242.77,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.7977189421653748,
      "eval_runtime": 4.5854,
      "eval_samples_per_second": 62.372,
      "eval_steps_per_second": 0.654,
      "step": 789
    },
    {
      "epoch": 244.0,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.8001275658607483,
      "eval_runtime": 4.7528,
      "eval_samples_per_second": 60.175,
      "eval_steps_per_second": 0.631,
      "step": 793
    },
    {
      "epoch": 244.92,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8043994903564453,
      "eval_runtime": 4.2699,
      "eval_samples_per_second": 66.98,
      "eval_steps_per_second": 0.703,
      "step": 796
    },
    {
      "epoch": 245.85,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8082275390625,
      "eval_runtime": 4.2996,
      "eval_samples_per_second": 66.518,
      "eval_steps_per_second": 0.698,
      "step": 799
    },
    {
      "epoch": 246.15,
      "grad_norm": 99001.8359375,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.5456,
      "step": 800
    },
    {
      "epoch": 246.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8120755553245544,
      "eval_runtime": 4.5242,
      "eval_samples_per_second": 63.216,
      "eval_steps_per_second": 0.663,
      "step": 802
    },
    {
      "epoch": 248.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.8106970191001892,
      "eval_runtime": 4.4479,
      "eval_samples_per_second": 64.3,
      "eval_steps_per_second": 0.674,
      "step": 806
    },
    {
      "epoch": 248.92,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.806368887424469,
      "eval_runtime": 4.1522,
      "eval_samples_per_second": 68.88,
      "eval_steps_per_second": 0.723,
      "step": 809
    },
    {
      "epoch": 249.85,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8042352199554443,
      "eval_runtime": 4.4213,
      "eval_samples_per_second": 64.687,
      "eval_steps_per_second": 0.679,
      "step": 812
    },
    {
      "epoch": 250.77,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8005724549293518,
      "eval_runtime": 4.4134,
      "eval_samples_per_second": 64.802,
      "eval_steps_per_second": 0.68,
      "step": 815
    },
    {
      "epoch": 252.0,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.7968676090240479,
      "eval_runtime": 3.8229,
      "eval_samples_per_second": 74.812,
      "eval_steps_per_second": 0.785,
      "step": 819
    },
    {
      "epoch": 252.92,
      "eval_accuracy": 0.6993006993006993,
      "eval_loss": 0.7954707741737366,
      "eval_runtime": 4.2693,
      "eval_samples_per_second": 66.99,
      "eval_steps_per_second": 0.703,
      "step": 822
    },
    {
      "epoch": 253.85,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.7973347902297974,
      "eval_runtime": 4.1401,
      "eval_samples_per_second": 69.081,
      "eval_steps_per_second": 0.725,
      "step": 825
    },
    {
      "epoch": 254.77,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.8001494407653809,
      "eval_runtime": 4.4851,
      "eval_samples_per_second": 63.767,
      "eval_steps_per_second": 0.669,
      "step": 828
    },
    {
      "epoch": 256.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.80350661277771,
      "eval_runtime": 4.4996,
      "eval_samples_per_second": 63.562,
      "eval_steps_per_second": 0.667,
      "step": 832
    },
    {
      "epoch": 256.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.8035485148429871,
      "eval_runtime": 4.5713,
      "eval_samples_per_second": 62.564,
      "eval_steps_per_second": 0.656,
      "step": 835
    },
    {
      "epoch": 257.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8012282252311707,
      "eval_runtime": 4.0638,
      "eval_samples_per_second": 70.377,
      "eval_steps_per_second": 0.738,
      "step": 838
    },
    {
      "epoch": 258.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.8000492453575134,
      "eval_runtime": 4.443,
      "eval_samples_per_second": 64.372,
      "eval_steps_per_second": 0.675,
      "step": 841
    },
    {
      "epoch": 260.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7963055968284607,
      "eval_runtime": 5.2655,
      "eval_samples_per_second": 54.316,
      "eval_steps_per_second": 0.57,
      "step": 845
    },
    {
      "epoch": 260.92,
      "eval_accuracy": 0.6958041958041958,
      "eval_loss": 0.7927840352058411,
      "eval_runtime": 5.1407,
      "eval_samples_per_second": 55.634,
      "eval_steps_per_second": 0.584,
      "step": 848
    },
    {
      "epoch": 261.54,
      "grad_norm": 24108.591796875,
      "learning_rate": 6.349206349206349e-06,
      "loss": 0.5369,
      "step": 850
    },
    {
      "epoch": 261.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7919009327888489,
      "eval_runtime": 3.8577,
      "eval_samples_per_second": 74.138,
      "eval_steps_per_second": 0.778,
      "step": 851
    },
    {
      "epoch": 262.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.791265606880188,
      "eval_runtime": 4.1966,
      "eval_samples_per_second": 68.151,
      "eval_steps_per_second": 0.715,
      "step": 854
    },
    {
      "epoch": 264.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7929325699806213,
      "eval_runtime": 4.063,
      "eval_samples_per_second": 70.391,
      "eval_steps_per_second": 0.738,
      "step": 858
    },
    {
      "epoch": 264.92,
      "eval_accuracy": 0.6818181818181818,
      "eval_loss": 0.7954928278923035,
      "eval_runtime": 4.3933,
      "eval_samples_per_second": 65.099,
      "eval_steps_per_second": 0.683,
      "step": 861
    },
    {
      "epoch": 265.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7962778210639954,
      "eval_runtime": 4.4424,
      "eval_samples_per_second": 64.38,
      "eval_steps_per_second": 0.675,
      "step": 864
    },
    {
      "epoch": 266.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7951834201812744,
      "eval_runtime": 4.2605,
      "eval_samples_per_second": 67.128,
      "eval_steps_per_second": 0.704,
      "step": 867
    },
    {
      "epoch": 268.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7936495542526245,
      "eval_runtime": 4.9467,
      "eval_samples_per_second": 57.816,
      "eval_steps_per_second": 0.606,
      "step": 871
    },
    {
      "epoch": 268.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7928897738456726,
      "eval_runtime": 4.9925,
      "eval_samples_per_second": 57.286,
      "eval_steps_per_second": 0.601,
      "step": 874
    },
    {
      "epoch": 269.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7933365702629089,
      "eval_runtime": 4.4133,
      "eval_samples_per_second": 64.804,
      "eval_steps_per_second": 0.68,
      "step": 877
    },
    {
      "epoch": 270.77,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7940818071365356,
      "eval_runtime": 4.0519,
      "eval_samples_per_second": 70.584,
      "eval_steps_per_second": 0.74,
      "step": 880
    },
    {
      "epoch": 272.0,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7939559817314148,
      "eval_runtime": 4.2845,
      "eval_samples_per_second": 66.753,
      "eval_steps_per_second": 0.7,
      "step": 884
    },
    {
      "epoch": 272.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7929409742355347,
      "eval_runtime": 4.885,
      "eval_samples_per_second": 58.546,
      "eval_steps_per_second": 0.614,
      "step": 887
    },
    {
      "epoch": 273.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7929646968841553,
      "eval_runtime": 3.7177,
      "eval_samples_per_second": 76.929,
      "eval_steps_per_second": 0.807,
      "step": 890
    },
    {
      "epoch": 274.77,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7942932844161987,
      "eval_runtime": 4.7663,
      "eval_samples_per_second": 60.004,
      "eval_steps_per_second": 0.629,
      "step": 893
    },
    {
      "epoch": 276.0,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7943535447120667,
      "eval_runtime": 4.0017,
      "eval_samples_per_second": 71.47,
      "eval_steps_per_second": 0.75,
      "step": 897
    },
    {
      "epoch": 276.92,
      "grad_norm": 30744.533203125,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 0.5388,
      "step": 900
    },
    {
      "epoch": 276.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7933218479156494,
      "eval_runtime": 4.3013,
      "eval_samples_per_second": 66.492,
      "eval_steps_per_second": 0.697,
      "step": 900
    },
    {
      "epoch": 277.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7914408445358276,
      "eval_runtime": 4.8732,
      "eval_samples_per_second": 58.689,
      "eval_steps_per_second": 0.616,
      "step": 903
    },
    {
      "epoch": 278.77,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7903594970703125,
      "eval_runtime": 4.6519,
      "eval_samples_per_second": 61.48,
      "eval_steps_per_second": 0.645,
      "step": 906
    },
    {
      "epoch": 280.0,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7888299822807312,
      "eval_runtime": 4.5788,
      "eval_samples_per_second": 62.462,
      "eval_steps_per_second": 0.655,
      "step": 910
    },
    {
      "epoch": 280.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7900360822677612,
      "eval_runtime": 4.5971,
      "eval_samples_per_second": 62.213,
      "eval_steps_per_second": 0.653,
      "step": 913
    },
    {
      "epoch": 281.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7905992865562439,
      "eval_runtime": 4.4545,
      "eval_samples_per_second": 64.205,
      "eval_steps_per_second": 0.673,
      "step": 916
    },
    {
      "epoch": 282.77,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7911333441734314,
      "eval_runtime": 4.4274,
      "eval_samples_per_second": 64.598,
      "eval_steps_per_second": 0.678,
      "step": 919
    },
    {
      "epoch": 284.0,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7906560897827148,
      "eval_runtime": 3.9207,
      "eval_samples_per_second": 72.947,
      "eval_steps_per_second": 0.765,
      "step": 923
    },
    {
      "epoch": 284.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7906984686851501,
      "eval_runtime": 4.5603,
      "eval_samples_per_second": 62.715,
      "eval_steps_per_second": 0.658,
      "step": 926
    },
    {
      "epoch": 285.85,
      "eval_accuracy": 0.6818181818181818,
      "eval_loss": 0.7905350923538208,
      "eval_runtime": 4.8134,
      "eval_samples_per_second": 59.418,
      "eval_steps_per_second": 0.623,
      "step": 929
    },
    {
      "epoch": 286.77,
      "eval_accuracy": 0.6818181818181818,
      "eval_loss": 0.7899833917617798,
      "eval_runtime": 4.0697,
      "eval_samples_per_second": 70.275,
      "eval_steps_per_second": 0.737,
      "step": 932
    },
    {
      "epoch": 288.0,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7901102304458618,
      "eval_runtime": 4.0126,
      "eval_samples_per_second": 71.276,
      "eval_steps_per_second": 0.748,
      "step": 936
    },
    {
      "epoch": 288.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7902336120605469,
      "eval_runtime": 3.8328,
      "eval_samples_per_second": 74.619,
      "eval_steps_per_second": 0.783,
      "step": 939
    },
    {
      "epoch": 289.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7909765839576721,
      "eval_runtime": 3.9497,
      "eval_samples_per_second": 72.411,
      "eval_steps_per_second": 0.76,
      "step": 942
    },
    {
      "epoch": 290.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7913976907730103,
      "eval_runtime": 4.7881,
      "eval_samples_per_second": 59.731,
      "eval_steps_per_second": 0.627,
      "step": 945
    },
    {
      "epoch": 292.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7919970750808716,
      "eval_runtime": 4.0436,
      "eval_samples_per_second": 70.729,
      "eval_steps_per_second": 0.742,
      "step": 949
    },
    {
      "epoch": 292.31,
      "grad_norm": 41198.3515625,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 0.5261,
      "step": 950
    },
    {
      "epoch": 292.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7927921414375305,
      "eval_runtime": 3.9219,
      "eval_samples_per_second": 72.923,
      "eval_steps_per_second": 0.765,
      "step": 952
    },
    {
      "epoch": 293.85,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.793153703212738,
      "eval_runtime": 4.3649,
      "eval_samples_per_second": 65.522,
      "eval_steps_per_second": 0.687,
      "step": 955
    },
    {
      "epoch": 294.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7925400733947754,
      "eval_runtime": 4.2064,
      "eval_samples_per_second": 67.992,
      "eval_steps_per_second": 0.713,
      "step": 958
    },
    {
      "epoch": 296.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7922278046607971,
      "eval_runtime": 4.03,
      "eval_samples_per_second": 70.968,
      "eval_steps_per_second": 0.744,
      "step": 962
    },
    {
      "epoch": 296.92,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7919090986251831,
      "eval_runtime": 4.4889,
      "eval_samples_per_second": 63.713,
      "eval_steps_per_second": 0.668,
      "step": 965
    },
    {
      "epoch": 297.85,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7922202348709106,
      "eval_runtime": 4.3742,
      "eval_samples_per_second": 65.383,
      "eval_steps_per_second": 0.686,
      "step": 968
    },
    {
      "epoch": 298.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7921380400657654,
      "eval_runtime": 4.27,
      "eval_samples_per_second": 66.979,
      "eval_steps_per_second": 0.703,
      "step": 971
    },
    {
      "epoch": 300.0,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7912278175354004,
      "eval_runtime": 4.209,
      "eval_samples_per_second": 67.95,
      "eval_steps_per_second": 0.713,
      "step": 975
    },
    {
      "epoch": 300.92,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7907286882400513,
      "eval_runtime": 4.5975,
      "eval_samples_per_second": 62.208,
      "eval_steps_per_second": 0.653,
      "step": 978
    },
    {
      "epoch": 301.85,
      "eval_accuracy": 0.6853146853146853,
      "eval_loss": 0.7895866632461548,
      "eval_runtime": 4.0629,
      "eval_samples_per_second": 70.394,
      "eval_steps_per_second": 0.738,
      "step": 981
    },
    {
      "epoch": 302.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7885376811027527,
      "eval_runtime": 4.0112,
      "eval_samples_per_second": 71.301,
      "eval_steps_per_second": 0.748,
      "step": 984
    },
    {
      "epoch": 304.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7877256870269775,
      "eval_runtime": 4.4199,
      "eval_samples_per_second": 64.708,
      "eval_steps_per_second": 0.679,
      "step": 988
    },
    {
      "epoch": 304.92,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7874112725257874,
      "eval_runtime": 4.0366,
      "eval_samples_per_second": 70.852,
      "eval_steps_per_second": 0.743,
      "step": 991
    },
    {
      "epoch": 305.85,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7876228094100952,
      "eval_runtime": 4.3519,
      "eval_samples_per_second": 65.718,
      "eval_steps_per_second": 0.689,
      "step": 994
    },
    {
      "epoch": 306.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7879106402397156,
      "eval_runtime": 5.3443,
      "eval_samples_per_second": 53.515,
      "eval_steps_per_second": 0.561,
      "step": 997
    },
    {
      "epoch": 307.69,
      "grad_norm": 31167.6875,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 0.5188,
      "step": 1000
    },
    {
      "epoch": 308.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7883804440498352,
      "eval_runtime": 4.1413,
      "eval_samples_per_second": 69.06,
      "eval_steps_per_second": 0.724,
      "step": 1001
    },
    {
      "epoch": 308.92,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7886692881584167,
      "eval_runtime": 4.049,
      "eval_samples_per_second": 70.634,
      "eval_steps_per_second": 0.741,
      "step": 1004
    },
    {
      "epoch": 309.85,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7890444397926331,
      "eval_runtime": 4.612,
      "eval_samples_per_second": 62.012,
      "eval_steps_per_second": 0.65,
      "step": 1007
    },
    {
      "epoch": 310.77,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7894096970558167,
      "eval_runtime": 3.8027,
      "eval_samples_per_second": 75.209,
      "eval_steps_per_second": 0.789,
      "step": 1010
    },
    {
      "epoch": 312.0,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7899323105812073,
      "eval_runtime": 4.3345,
      "eval_samples_per_second": 65.983,
      "eval_steps_per_second": 0.692,
      "step": 1014
    },
    {
      "epoch": 312.92,
      "eval_accuracy": 0.6888111888111889,
      "eval_loss": 0.7903538346290588,
      "eval_runtime": 4.5846,
      "eval_samples_per_second": 62.383,
      "eval_steps_per_second": 0.654,
      "step": 1017
    },
    {
      "epoch": 313.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7907257080078125,
      "eval_runtime": 4.136,
      "eval_samples_per_second": 69.148,
      "eval_steps_per_second": 0.725,
      "step": 1020
    },
    {
      "epoch": 314.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.790963888168335,
      "eval_runtime": 4.2526,
      "eval_samples_per_second": 67.252,
      "eval_steps_per_second": 0.705,
      "step": 1023
    },
    {
      "epoch": 316.0,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7912085056304932,
      "eval_runtime": 4.1188,
      "eval_samples_per_second": 69.437,
      "eval_steps_per_second": 0.728,
      "step": 1027
    },
    {
      "epoch": 316.92,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7911705374717712,
      "eval_runtime": 4.1524,
      "eval_samples_per_second": 68.876,
      "eval_steps_per_second": 0.722,
      "step": 1030
    },
    {
      "epoch": 317.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7911967039108276,
      "eval_runtime": 3.9058,
      "eval_samples_per_second": 73.225,
      "eval_steps_per_second": 0.768,
      "step": 1033
    },
    {
      "epoch": 318.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7912610173225403,
      "eval_runtime": 4.6095,
      "eval_samples_per_second": 62.046,
      "eval_steps_per_second": 0.651,
      "step": 1036
    },
    {
      "epoch": 320.0,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7912730574607849,
      "eval_runtime": 5.5705,
      "eval_samples_per_second": 51.342,
      "eval_steps_per_second": 0.539,
      "step": 1040
    },
    {
      "epoch": 320.92,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7911974787712097,
      "eval_runtime": 4.9154,
      "eval_samples_per_second": 58.185,
      "eval_steps_per_second": 0.61,
      "step": 1043
    },
    {
      "epoch": 321.85,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7911575436592102,
      "eval_runtime": 4.8387,
      "eval_samples_per_second": 59.107,
      "eval_steps_per_second": 0.62,
      "step": 1046
    },
    {
      "epoch": 322.77,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7911355495452881,
      "eval_runtime": 4.1368,
      "eval_samples_per_second": 69.135,
      "eval_steps_per_second": 0.725,
      "step": 1049
    },
    {
      "epoch": 323.08,
      "grad_norm": 53824.44140625,
      "learning_rate": 0.0,
      "loss": 0.5194,
      "step": 1050
    },
    {
      "epoch": 323.08,
      "eval_accuracy": 0.6923076923076923,
      "eval_loss": 0.7911302447319031,
      "eval_runtime": 4.2304,
      "eval_samples_per_second": 67.606,
      "eval_steps_per_second": 0.709,
      "step": 1050
    },
    {
      "epoch": 323.08,
      "step": 1050,
      "total_flos": 4.380490432252032e+18,
      "train_loss": 0.8143934268043155,
      "train_runtime": 4784.9132,
      "train_samples_per_second": 113.231,
      "train_steps_per_second": 0.219
    }
  ],
  "logging_steps": 50,
  "max_steps": 1050,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 350,
  "save_steps": 500,
  "total_flos": 4.380490432252032e+18,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}