{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 100.0,
  "eval_steps": 500,
  "global_step": 900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.1111111111111111,
      "eval_loss": 1.3930704593658447,
      "eval_runtime": 0.7497,
      "eval_samples_per_second": 24.01,
      "eval_steps_per_second": 4.002,
      "step": 9
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.16666666666666666,
      "eval_loss": 1.376913070678711,
      "eval_runtime": 0.5755,
      "eval_samples_per_second": 31.277,
      "eval_steps_per_second": 5.213,
      "step": 18
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.5555555555555556,
      "eval_loss": 1.3498488664627075,
      "eval_runtime": 0.5792,
      "eval_samples_per_second": 31.078,
      "eval_steps_per_second": 5.18,
      "step": 27
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6666666666666666,
      "eval_loss": 1.3331588506698608,
      "eval_runtime": 0.5535,
      "eval_samples_per_second": 32.518,
      "eval_steps_per_second": 5.42,
      "step": 36
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.3144757747650146,
      "eval_runtime": 0.5818,
      "eval_samples_per_second": 30.94,
      "eval_steps_per_second": 5.157,
      "step": 45
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.300213098526001,
      "eval_runtime": 0.5539,
      "eval_samples_per_second": 32.496,
      "eval_steps_per_second": 5.416,
      "step": 54
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.2916626930236816,
      "eval_runtime": 0.5489,
      "eval_samples_per_second": 32.791,
      "eval_steps_per_second": 5.465,
      "step": 63
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.270052194595337,
      "eval_runtime": 0.5746,
      "eval_samples_per_second": 31.328,
      "eval_steps_per_second": 5.221,
      "step": 72
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.2555011510849,
      "eval_runtime": 0.5489,
      "eval_samples_per_second": 32.791,
      "eval_steps_per_second": 5.465,
      "step": 81
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.2477619647979736,
      "eval_runtime": 0.5687,
      "eval_samples_per_second": 31.651,
      "eval_steps_per_second": 5.275,
      "step": 90
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.2382668256759644,
      "eval_runtime": 0.5502,
      "eval_samples_per_second": 32.716,
      "eval_steps_per_second": 5.453,
      "step": 99
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.223071575164795,
      "eval_runtime": 0.5476,
      "eval_samples_per_second": 32.873,
      "eval_steps_per_second": 5.479,
      "step": 108
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.2033451795578003,
      "eval_runtime": 0.5538,
      "eval_samples_per_second": 32.503,
      "eval_steps_per_second": 5.417,
      "step": 117
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.186124563217163,
      "eval_runtime": 0.5442,
      "eval_samples_per_second": 33.075,
      "eval_steps_per_second": 5.512,
      "step": 126
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.182234287261963,
      "eval_runtime": 0.5735,
      "eval_samples_per_second": 31.387,
      "eval_steps_per_second": 5.231,
      "step": 135
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.1588956117630005,
      "eval_runtime": 0.5533,
      "eval_samples_per_second": 32.534,
      "eval_steps_per_second": 5.422,
      "step": 144
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.1478044986724854,
      "eval_runtime": 0.5826,
      "eval_samples_per_second": 30.897,
      "eval_steps_per_second": 5.15,
      "step": 153
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.1322474479675293,
      "eval_runtime": 0.5883,
      "eval_samples_per_second": 30.598,
      "eval_steps_per_second": 5.1,
      "step": 162
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.110813021659851,
      "eval_runtime": 0.5594,
      "eval_samples_per_second": 32.177,
      "eval_steps_per_second": 5.363,
      "step": 171
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.1011286973953247,
      "eval_runtime": 0.5793,
      "eval_samples_per_second": 31.073,
      "eval_steps_per_second": 5.179,
      "step": 180
    },
    {
      "epoch": 21.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.093163251876831,
      "eval_runtime": 0.5654,
      "eval_samples_per_second": 31.836,
      "eval_steps_per_second": 5.306,
      "step": 189
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.0637242794036865,
      "eval_runtime": 0.5679,
      "eval_samples_per_second": 31.695,
      "eval_steps_per_second": 5.283,
      "step": 198
    },
    {
      "epoch": 23.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.0390856266021729,
      "eval_runtime": 0.574,
      "eval_samples_per_second": 31.358,
      "eval_steps_per_second": 5.226,
      "step": 207
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.0522559881210327,
      "eval_runtime": 0.6203,
      "eval_samples_per_second": 29.02,
      "eval_steps_per_second": 4.837,
      "step": 216
    },
    {
      "epoch": 25.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 1.0473227500915527,
      "eval_runtime": 0.5963,
      "eval_samples_per_second": 30.186,
      "eval_steps_per_second": 5.031,
      "step": 225
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 0.9998855590820312,
      "eval_runtime": 0.6467,
      "eval_samples_per_second": 27.834,
      "eval_steps_per_second": 4.639,
      "step": 234
    },
    {
      "epoch": 27.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.0170878171920776,
      "eval_runtime": 0.5994,
      "eval_samples_per_second": 30.032,
      "eval_steps_per_second": 5.005,
      "step": 243
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.028573989868164,
      "eval_runtime": 0.6332,
      "eval_samples_per_second": 28.427,
      "eval_steps_per_second": 4.738,
      "step": 252
    },
    {
      "epoch": 29.0,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 1.0290330648422241,
      "eval_runtime": 0.5676,
      "eval_samples_per_second": 31.712,
      "eval_steps_per_second": 5.285,
      "step": 261
    },
    {
      "epoch": 30.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.9571393132209778,
      "eval_runtime": 0.5753,
      "eval_samples_per_second": 31.286,
      "eval_steps_per_second": 5.214,
      "step": 270
    },
    {
      "epoch": 31.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.9450912475585938,
      "eval_runtime": 0.5556,
      "eval_samples_per_second": 32.4,
      "eval_steps_per_second": 5.4,
      "step": 279
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.85059654712677,
      "eval_runtime": 0.5543,
      "eval_samples_per_second": 32.473,
      "eval_steps_per_second": 5.412,
      "step": 288
    },
    {
      "epoch": 33.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.8480208516120911,
      "eval_runtime": 0.5762,
      "eval_samples_per_second": 31.241,
      "eval_steps_per_second": 5.207,
      "step": 297
    },
    {
      "epoch": 34.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.8499312400817871,
      "eval_runtime": 0.555,
      "eval_samples_per_second": 32.433,
      "eval_steps_per_second": 5.405,
      "step": 306
    },
    {
      "epoch": 35.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.8403282165527344,
      "eval_runtime": 0.5709,
      "eval_samples_per_second": 31.532,
      "eval_steps_per_second": 5.255,
      "step": 315
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.7771400809288025,
      "eval_runtime": 0.5569,
      "eval_samples_per_second": 32.319,
      "eval_steps_per_second": 5.387,
      "step": 324
    },
    {
      "epoch": 37.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.7591123580932617,
      "eval_runtime": 0.6487,
      "eval_samples_per_second": 27.747,
      "eval_steps_per_second": 4.625,
      "step": 333
    },
    {
      "epoch": 38.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.7476389408111572,
      "eval_runtime": 0.5654,
      "eval_samples_per_second": 31.835,
      "eval_steps_per_second": 5.306,
      "step": 342
    },
    {
      "epoch": 39.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.7831047773361206,
      "eval_runtime": 0.5458,
      "eval_samples_per_second": 32.977,
      "eval_steps_per_second": 5.496,
      "step": 351
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.7049207091331482,
      "eval_runtime": 0.5872,
      "eval_samples_per_second": 30.653,
      "eval_steps_per_second": 5.109,
      "step": 360
    },
    {
      "epoch": 41.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.6811972856521606,
      "eval_runtime": 0.5667,
      "eval_samples_per_second": 31.762,
      "eval_steps_per_second": 5.294,
      "step": 369
    },
    {
      "epoch": 42.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.6736953258514404,
      "eval_runtime": 0.5538,
      "eval_samples_per_second": 32.502,
      "eval_steps_per_second": 5.417,
      "step": 378
    },
    {
      "epoch": 43.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.6515324711799622,
      "eval_runtime": 0.5632,
      "eval_samples_per_second": 31.958,
      "eval_steps_per_second": 5.326,
      "step": 387
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.6634184122085571,
      "eval_runtime": 1.1806,
      "eval_samples_per_second": 15.246,
      "eval_steps_per_second": 2.541,
      "step": 396
    },
    {
      "epoch": 45.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6234365105628967,
      "eval_runtime": 0.6418,
      "eval_samples_per_second": 28.044,
      "eval_steps_per_second": 4.674,
      "step": 405
    },
    {
      "epoch": 46.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.8482791185379028,
      "eval_runtime": 0.5859,
      "eval_samples_per_second": 30.723,
      "eval_steps_per_second": 5.121,
      "step": 414
    },
    {
      "epoch": 47.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.7264916896820068,
      "eval_runtime": 0.5835,
      "eval_samples_per_second": 30.847,
      "eval_steps_per_second": 5.141,
      "step": 423
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.7383356094360352,
      "eval_runtime": 0.5549,
      "eval_samples_per_second": 32.437,
      "eval_steps_per_second": 5.406,
      "step": 432
    },
    {
      "epoch": 49.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.7005217671394348,
      "eval_runtime": 0.5608,
      "eval_samples_per_second": 32.095,
      "eval_steps_per_second": 5.349,
      "step": 441
    },
    {
      "epoch": 50.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.5740242004394531,
      "eval_runtime": 0.5536,
      "eval_samples_per_second": 32.517,
      "eval_steps_per_second": 5.42,
      "step": 450
    },
    {
      "epoch": 51.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.5622536540031433,
      "eval_runtime": 0.5935,
      "eval_samples_per_second": 30.329,
      "eval_steps_per_second": 5.055,
      "step": 459
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.557184636592865,
      "eval_runtime": 0.5545,
      "eval_samples_per_second": 32.463,
      "eval_steps_per_second": 5.41,
      "step": 468
    },
    {
      "epoch": 53.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.5769361853599548,
      "eval_runtime": 0.5844,
      "eval_samples_per_second": 30.802,
      "eval_steps_per_second": 5.134,
      "step": 477
    },
    {
      "epoch": 54.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.550247311592102,
      "eval_runtime": 0.615,
      "eval_samples_per_second": 29.266,
      "eval_steps_per_second": 4.878,
      "step": 486
    },
    {
      "epoch": 55.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.6281833052635193,
      "eval_runtime": 0.5457,
      "eval_samples_per_second": 32.988,
      "eval_steps_per_second": 5.498,
      "step": 495
    },
    {
      "epoch": 55.56,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 1.1157,
      "step": 500
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.5906974077224731,
      "eval_runtime": 0.5821,
      "eval_samples_per_second": 30.92,
      "eval_steps_per_second": 5.153,
      "step": 504
    },
    {
      "epoch": 57.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.6346855163574219,
      "eval_runtime": 0.5542,
      "eval_samples_per_second": 32.481,
      "eval_steps_per_second": 5.414,
      "step": 513
    },
    {
      "epoch": 58.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.6413730382919312,
      "eval_runtime": 0.5829,
      "eval_samples_per_second": 30.882,
      "eval_steps_per_second": 5.147,
      "step": 522
    },
    {
      "epoch": 59.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.465614378452301,
      "eval_runtime": 0.5965,
      "eval_samples_per_second": 30.178,
      "eval_steps_per_second": 5.03,
      "step": 531
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.48488152027130127,
      "eval_runtime": 0.5621,
      "eval_samples_per_second": 32.02,
      "eval_steps_per_second": 5.337,
      "step": 540
    },
    {
      "epoch": 61.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.8426976799964905,
      "eval_runtime": 0.5798,
      "eval_samples_per_second": 31.045,
      "eval_steps_per_second": 5.174,
      "step": 549
    },
    {
      "epoch": 62.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.5708574652671814,
      "eval_runtime": 0.5531,
      "eval_samples_per_second": 32.544,
      "eval_steps_per_second": 5.424,
      "step": 558
    },
    {
      "epoch": 63.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.5026788115501404,
      "eval_runtime": 0.5575,
      "eval_samples_per_second": 32.289,
      "eval_steps_per_second": 5.381,
      "step": 567
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.5724208354949951,
      "eval_runtime": 0.5481,
      "eval_samples_per_second": 32.844,
      "eval_steps_per_second": 5.474,
      "step": 576
    },
    {
      "epoch": 65.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.5301716327667236,
      "eval_runtime": 0.553,
      "eval_samples_per_second": 32.551,
      "eval_steps_per_second": 5.425,
      "step": 585
    },
    {
      "epoch": 66.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.5272272825241089,
      "eval_runtime": 0.5566,
      "eval_samples_per_second": 32.34,
      "eval_steps_per_second": 5.39,
      "step": 594
    },
    {
      "epoch": 67.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.5444329380989075,
      "eval_runtime": 0.555,
      "eval_samples_per_second": 32.435,
      "eval_steps_per_second": 5.406,
      "step": 603
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.3936518132686615,
      "eval_runtime": 0.6001,
      "eval_samples_per_second": 29.994,
      "eval_steps_per_second": 4.999,
      "step": 612
    },
    {
      "epoch": 69.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.41802236437797546,
      "eval_runtime": 0.5658,
      "eval_samples_per_second": 31.816,
      "eval_steps_per_second": 5.303,
      "step": 621
    },
    {
      "epoch": 70.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.5185115337371826,
      "eval_runtime": 0.5619,
      "eval_samples_per_second": 32.036,
      "eval_steps_per_second": 5.339,
      "step": 630
    },
    {
      "epoch": 71.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.39606520533561707,
      "eval_runtime": 0.5983,
      "eval_samples_per_second": 30.085,
      "eval_steps_per_second": 5.014,
      "step": 639
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.3859682083129883,
      "eval_runtime": 0.5757,
      "eval_samples_per_second": 31.268,
      "eval_steps_per_second": 5.211,
      "step": 648
    },
    {
      "epoch": 73.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.39656686782836914,
      "eval_runtime": 0.6158,
      "eval_samples_per_second": 29.228,
      "eval_steps_per_second": 4.871,
      "step": 657
    },
    {
      "epoch": 74.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.39676183462142944,
      "eval_runtime": 0.6324,
      "eval_samples_per_second": 28.462,
      "eval_steps_per_second": 4.744,
      "step": 666
    },
    {
      "epoch": 75.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.4546321630477905,
      "eval_runtime": 0.5605,
      "eval_samples_per_second": 32.114,
      "eval_steps_per_second": 5.352,
      "step": 675
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.4021334648132324,
      "eval_runtime": 0.553,
      "eval_samples_per_second": 32.55,
      "eval_steps_per_second": 5.425,
      "step": 684
    },
    {
      "epoch": 77.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.414422869682312,
      "eval_runtime": 0.5759,
      "eval_samples_per_second": 31.256,
      "eval_steps_per_second": 5.209,
      "step": 693
    },
    {
      "epoch": 78.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.35500773787498474,
      "eval_runtime": 0.5802,
      "eval_samples_per_second": 31.024,
      "eval_steps_per_second": 5.171,
      "step": 702
    },
    {
      "epoch": 79.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.3838707506656647,
      "eval_runtime": 0.5616,
      "eval_samples_per_second": 32.052,
      "eval_steps_per_second": 5.342,
      "step": 711
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.437086820602417,
      "eval_runtime": 0.554,
      "eval_samples_per_second": 32.49,
      "eval_steps_per_second": 5.415,
      "step": 720
    },
    {
      "epoch": 81.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.36943355202674866,
      "eval_runtime": 0.5894,
      "eval_samples_per_second": 30.537,
      "eval_steps_per_second": 5.09,
      "step": 729
    },
    {
      "epoch": 82.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.36648380756378174,
      "eval_runtime": 0.5615,
      "eval_samples_per_second": 32.058,
      "eval_steps_per_second": 5.343,
      "step": 738
    },
    {
      "epoch": 83.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.4732191264629364,
      "eval_runtime": 0.5632,
      "eval_samples_per_second": 31.959,
      "eval_steps_per_second": 5.327,
      "step": 747
    },
    {
      "epoch": 84.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.4652000665664673,
      "eval_runtime": 0.592,
      "eval_samples_per_second": 30.406,
      "eval_steps_per_second": 5.068,
      "step": 756
    },
    {
      "epoch": 85.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.670432448387146,
      "eval_runtime": 0.5801,
      "eval_samples_per_second": 31.028,
      "eval_steps_per_second": 5.171,
      "step": 765
    },
    {
      "epoch": 86.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.7354382872581482,
      "eval_runtime": 0.5599,
      "eval_samples_per_second": 32.151,
      "eval_steps_per_second": 5.359,
      "step": 774
    },
    {
      "epoch": 87.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.5439589023590088,
      "eval_runtime": 0.5486,
      "eval_samples_per_second": 32.811,
      "eval_steps_per_second": 5.468,
      "step": 783
    },
    {
      "epoch": 88.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.4809061586856842,
      "eval_runtime": 0.6101,
      "eval_samples_per_second": 29.505,
      "eval_steps_per_second": 4.917,
      "step": 792
    },
    {
      "epoch": 89.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.6704312562942505,
      "eval_runtime": 0.5954,
      "eval_samples_per_second": 30.229,
      "eval_steps_per_second": 5.038,
      "step": 801
    },
    {
      "epoch": 90.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.41327810287475586,
      "eval_runtime": 0.5783,
      "eval_samples_per_second": 31.124,
      "eval_steps_per_second": 5.187,
      "step": 810
    },
    {
      "epoch": 91.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.350969523191452,
      "eval_runtime": 0.5507,
      "eval_samples_per_second": 32.686,
      "eval_steps_per_second": 5.448,
      "step": 819
    },
    {
      "epoch": 92.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.39819759130477905,
      "eval_runtime": 0.5465,
      "eval_samples_per_second": 32.939,
      "eval_steps_per_second": 5.49,
      "step": 828
    },
    {
      "epoch": 93.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.34932640194892883,
      "eval_runtime": 0.5635,
      "eval_samples_per_second": 31.941,
      "eval_steps_per_second": 5.323,
      "step": 837
    },
    {
      "epoch": 94.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.48363664746284485,
      "eval_runtime": 0.5411,
      "eval_samples_per_second": 33.263,
      "eval_steps_per_second": 5.544,
      "step": 846
    },
    {
      "epoch": 95.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.4434005320072174,
      "eval_runtime": 0.5964,
      "eval_samples_per_second": 30.181,
      "eval_steps_per_second": 5.03,
      "step": 855
    },
    {
      "epoch": 96.0,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.4290742874145508,
      "eval_runtime": 0.5584,
      "eval_samples_per_second": 32.235,
      "eval_steps_per_second": 5.373,
      "step": 864
    },
    {
      "epoch": 97.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.34131091833114624,
      "eval_runtime": 0.5715,
      "eval_samples_per_second": 31.493,
      "eval_steps_per_second": 5.249,
      "step": 873
    },
    {
      "epoch": 98.0,
      "eval_accuracy": 0.8888888888888888,
      "eval_loss": 0.3645610511302948,
      "eval_runtime": 0.5506,
      "eval_samples_per_second": 32.692,
      "eval_steps_per_second": 5.449,
      "step": 882
    },
    {
      "epoch": 99.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.5591509938240051,
      "eval_runtime": 0.5908,
      "eval_samples_per_second": 30.468,
      "eval_steps_per_second": 5.078,
      "step": 891
    },
    {
      "epoch": 100.0,
      "eval_accuracy": 0.9444444444444444,
      "eval_loss": 0.48491573333740234,
      "eval_runtime": 0.5689,
      "eval_samples_per_second": 31.643,
      "eval_steps_per_second": 5.274,
      "step": 900
    }
  ],
  "logging_steps": 500,
  "max_steps": 900,
  "num_train_epochs": 100,
  "save_steps": 500,
  "total_flos": 1.507976427331584e+17,
  "trial_name": null,
  "trial_params": null
}