{
  "best_metric": 0.16060182452201843,
  "best_model_checkpoint": "AlexWang99/byt5_add/checkpoint-650",
  "epoch": 50.0,
  "eval_steps": 500,
  "global_step": 650,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_loss": 4.225934982299805,
      "eval_runtime": 10.7417,
      "eval_samples_per_second": 930.95,
      "eval_steps_per_second": 1.21,
      "step": 13
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.4177823066711426,
      "eval_runtime": 10.8153,
      "eval_samples_per_second": 924.617,
      "eval_steps_per_second": 1.202,
      "step": 26
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.9256454706192017,
      "eval_runtime": 10.7408,
      "eval_samples_per_second": 931.032,
      "eval_steps_per_second": 1.21,
      "step": 39
    },
    {
      "epoch": 4.0,
      "eval_loss": 1.7310363054275513,
      "eval_runtime": 11.0217,
      "eval_samples_per_second": 907.305,
      "eval_steps_per_second": 1.179,
      "step": 52
    },
    {
      "epoch": 5.0,
      "eval_loss": 1.6576673984527588,
      "eval_runtime": 10.8631,
      "eval_samples_per_second": 920.548,
      "eval_steps_per_second": 1.197,
      "step": 65
    },
    {
      "epoch": 6.0,
      "eval_loss": 1.6385396718978882,
      "eval_runtime": 10.9023,
      "eval_samples_per_second": 917.236,
      "eval_steps_per_second": 1.192,
      "step": 78
    },
    {
      "epoch": 7.0,
      "eval_loss": 1.6109799146652222,
      "eval_runtime": 11.0728,
      "eval_samples_per_second": 903.112,
      "eval_steps_per_second": 1.174,
      "step": 91
    },
    {
      "epoch": 8.0,
      "eval_loss": 1.5810920000076294,
      "eval_runtime": 10.807,
      "eval_samples_per_second": 925.329,
      "eval_steps_per_second": 1.203,
      "step": 104
    },
    {
      "epoch": 9.0,
      "eval_loss": 1.523699402809143,
      "eval_runtime": 10.8206,
      "eval_samples_per_second": 924.164,
      "eval_steps_per_second": 1.201,
      "step": 117
    },
    {
      "epoch": 10.0,
      "eval_loss": 1.480878472328186,
      "eval_runtime": 10.8089,
      "eval_samples_per_second": 925.161,
      "eval_steps_per_second": 1.203,
      "step": 130
    },
    {
      "epoch": 11.0,
      "eval_loss": 1.437759280204773,
      "eval_runtime": 10.9724,
      "eval_samples_per_second": 911.378,
      "eval_steps_per_second": 1.185,
      "step": 143
    },
    {
      "epoch": 12.0,
      "eval_loss": 1.3976322412490845,
      "eval_runtime": 10.8773,
      "eval_samples_per_second": 919.349,
      "eval_steps_per_second": 1.195,
      "step": 156
    },
    {
      "epoch": 13.0,
      "eval_loss": 1.346189260482788,
      "eval_runtime": 10.8878,
      "eval_samples_per_second": 918.455,
      "eval_steps_per_second": 1.194,
      "step": 169
    },
    {
      "epoch": 14.0,
      "eval_loss": 1.2587027549743652,
      "eval_runtime": 10.7933,
      "eval_samples_per_second": 926.498,
      "eval_steps_per_second": 1.204,
      "step": 182
    },
    {
      "epoch": 15.0,
      "eval_loss": 1.2259936332702637,
      "eval_runtime": 10.7904,
      "eval_samples_per_second": 926.753,
      "eval_steps_per_second": 1.205,
      "step": 195
    },
    {
      "epoch": 16.0,
      "eval_loss": 1.1018248796463013,
      "eval_runtime": 11.0439,
      "eval_samples_per_second": 905.473,
      "eval_steps_per_second": 1.177,
      "step": 208
    },
    {
      "epoch": 17.0,
      "eval_loss": 1.0272711515426636,
      "eval_runtime": 10.8037,
      "eval_samples_per_second": 925.605,
      "eval_steps_per_second": 1.203,
      "step": 221
    },
    {
      "epoch": 18.0,
      "eval_loss": 0.9436440467834473,
      "eval_runtime": 10.8135,
      "eval_samples_per_second": 924.773,
      "eval_steps_per_second": 1.202,
      "step": 234
    },
    {
      "epoch": 19.0,
      "eval_loss": 0.8007376194000244,
      "eval_runtime": 10.8029,
      "eval_samples_per_second": 925.679,
      "eval_steps_per_second": 1.203,
      "step": 247
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.6918877959251404,
      "eval_runtime": 10.8958,
      "eval_samples_per_second": 917.781,
      "eval_steps_per_second": 1.193,
      "step": 260
    },
    {
      "epoch": 21.0,
      "eval_loss": 0.6201093792915344,
      "eval_runtime": 10.7944,
      "eval_samples_per_second": 926.41,
      "eval_steps_per_second": 1.204,
      "step": 273
    },
    {
      "epoch": 22.0,
      "eval_loss": 0.5485938191413879,
      "eval_runtime": 10.7982,
      "eval_samples_per_second": 926.077,
      "eval_steps_per_second": 1.204,
      "step": 286
    },
    {
      "epoch": 23.0,
      "eval_loss": 0.480381041765213,
      "eval_runtime": 10.7925,
      "eval_samples_per_second": 926.572,
      "eval_steps_per_second": 1.205,
      "step": 299
    },
    {
      "epoch": 24.0,
      "eval_loss": 0.40801915526390076,
      "eval_runtime": 10.7939,
      "eval_samples_per_second": 926.447,
      "eval_steps_per_second": 1.204,
      "step": 312
    },
    {
      "epoch": 25.0,
      "eval_loss": 0.386056512594223,
      "eval_runtime": 11.0386,
      "eval_samples_per_second": 905.916,
      "eval_steps_per_second": 1.178,
      "step": 325
    },
    {
      "epoch": 26.0,
      "eval_loss": 0.34766075015068054,
      "eval_runtime": 10.8947,
      "eval_samples_per_second": 917.876,
      "eval_steps_per_second": 1.193,
      "step": 338
    },
    {
      "epoch": 27.0,
      "eval_loss": 0.3181471526622772,
      "eval_runtime": 10.7999,
      "eval_samples_per_second": 925.935,
      "eval_steps_per_second": 1.204,
      "step": 351
    },
    {
      "epoch": 28.0,
      "eval_loss": 0.2921164631843567,
      "eval_runtime": 10.8802,
      "eval_samples_per_second": 919.099,
      "eval_steps_per_second": 1.195,
      "step": 364
    },
    {
      "epoch": 29.0,
      "eval_loss": 0.2831590473651886,
      "eval_runtime": 10.9022,
      "eval_samples_per_second": 917.242,
      "eval_steps_per_second": 1.192,
      "step": 377
    },
    {
      "epoch": 30.0,
      "eval_loss": 0.2693236470222473,
      "eval_runtime": 10.8973,
      "eval_samples_per_second": 917.655,
      "eval_steps_per_second": 1.193,
      "step": 390
    },
    {
      "epoch": 31.0,
      "eval_loss": 0.24693337082862854,
      "eval_runtime": 10.8754,
      "eval_samples_per_second": 919.507,
      "eval_steps_per_second": 1.195,
      "step": 403
    },
    {
      "epoch": 32.0,
      "eval_loss": 0.2452966272830963,
      "eval_runtime": 10.8896,
      "eval_samples_per_second": 918.306,
      "eval_steps_per_second": 1.194,
      "step": 416
    },
    {
      "epoch": 33.0,
      "eval_loss": 0.2312968224287033,
      "eval_runtime": 10.8133,
      "eval_samples_per_second": 924.783,
      "eval_steps_per_second": 1.202,
      "step": 429
    },
    {
      "epoch": 34.0,
      "eval_loss": 0.213392972946167,
      "eval_runtime": 10.7935,
      "eval_samples_per_second": 926.482,
      "eval_steps_per_second": 1.204,
      "step": 442
    },
    {
      "epoch": 35.0,
      "eval_loss": 0.21386706829071045,
      "eval_runtime": 10.809,
      "eval_samples_per_second": 925.156,
      "eval_steps_per_second": 1.203,
      "step": 455
    },
    {
      "epoch": 36.0,
      "eval_loss": 0.20882292091846466,
      "eval_runtime": 10.7998,
      "eval_samples_per_second": 925.947,
      "eval_steps_per_second": 1.204,
      "step": 468
    },
    {
      "epoch": 37.0,
      "eval_loss": 0.2006647139787674,
      "eval_runtime": 10.7969,
      "eval_samples_per_second": 926.189,
      "eval_steps_per_second": 1.204,
      "step": 481
    },
    {
      "epoch": 38.0,
      "eval_loss": 0.19596967101097107,
      "eval_runtime": 10.8081,
      "eval_samples_per_second": 925.232,
      "eval_steps_per_second": 1.203,
      "step": 494
    },
    {
      "epoch": 38.46,
      "learning_rate": 1.153846153846154e-05,
      "loss": 1.3,
      "step": 500
    },
    {
      "epoch": 39.0,
      "eval_loss": 0.18297980725765228,
      "eval_runtime": 10.8038,
      "eval_samples_per_second": 925.597,
      "eval_steps_per_second": 1.203,
      "step": 507
    },
    {
      "epoch": 40.0,
      "eval_loss": 0.17824265360832214,
      "eval_runtime": 11.0511,
      "eval_samples_per_second": 904.887,
      "eval_steps_per_second": 1.176,
      "step": 520
    },
    {
      "epoch": 41.0,
      "eval_loss": 0.17460107803344727,
      "eval_runtime": 10.8972,
      "eval_samples_per_second": 917.67,
      "eval_steps_per_second": 1.193,
      "step": 533
    },
    {
      "epoch": 42.0,
      "eval_loss": 0.17409959435462952,
      "eval_runtime": 10.7994,
      "eval_samples_per_second": 925.978,
      "eval_steps_per_second": 1.204,
      "step": 546
    },
    {
      "epoch": 43.0,
      "eval_loss": 0.17077761888504028,
      "eval_runtime": 10.8856,
      "eval_samples_per_second": 918.641,
      "eval_steps_per_second": 1.194,
      "step": 559
    },
    {
      "epoch": 44.0,
      "eval_loss": 0.16676855087280273,
      "eval_runtime": 10.8901,
      "eval_samples_per_second": 918.263,
      "eval_steps_per_second": 1.194,
      "step": 572
    },
    {
      "epoch": 45.0,
      "eval_loss": 0.16503272950649261,
      "eval_runtime": 10.8935,
      "eval_samples_per_second": 917.976,
      "eval_steps_per_second": 1.193,
      "step": 585
    },
    {
      "epoch": 46.0,
      "eval_loss": 0.1650797426700592,
      "eval_runtime": 10.8935,
      "eval_samples_per_second": 917.978,
      "eval_steps_per_second": 1.193,
      "step": 598
    },
    {
      "epoch": 47.0,
      "eval_loss": 0.1628640592098236,
      "eval_runtime": 10.8953,
      "eval_samples_per_second": 917.828,
      "eval_steps_per_second": 1.193,
      "step": 611
    },
    {
      "epoch": 48.0,
      "eval_loss": 0.1627463847398758,
      "eval_runtime": 10.8324,
      "eval_samples_per_second": 923.155,
      "eval_steps_per_second": 1.2,
      "step": 624
    },
    {
      "epoch": 49.0,
      "eval_loss": 0.16095298528671265,
      "eval_runtime": 10.8256,
      "eval_samples_per_second": 923.738,
      "eval_steps_per_second": 1.201,
      "step": 637
    },
    {
      "epoch": 50.0,
      "eval_loss": 0.16060182452201843,
      "eval_runtime": 10.7991,
      "eval_samples_per_second": 926.006,
      "eval_steps_per_second": 1.204,
      "step": 650
    }
  ],
  "logging_steps": 500,
  "max_steps": 650,
  "num_train_epochs": 50,
  "save_steps": 500,
  "total_flos": 1.4355462144e+16,
  "trial_name": null,
  "trial_params": null
}