{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "eval_steps": 500,
  "global_step": 60,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 11.5566,
      "step": 1
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 11.6224,
      "step": 2
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.998533413851124e-05,
      "loss": 6.494,
      "step": 3
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.9941379571543597e-05,
      "loss": 0.8094,
      "step": 4
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.9868265225415263e-05,
      "loss": 0.8526,
      "step": 5
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.976620555710087e-05,
      "loss": 7.028,
      "step": 6
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.963549992519223e-05,
      "loss": 6.659,
      "step": 7
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.9476531711828027e-05,
      "loss": 3.1902,
      "step": 8
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.9289767198167918e-05,
      "loss": 0.4599,
      "step": 9
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.9075754196709574e-05,
      "loss": 0.9921,
      "step": 10
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.883512044446023e-05,
      "loss": 0.7209,
      "step": 11
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.8568571761675893e-05,
      "loss": 0.5635,
      "step": 12
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.827688998156891e-05,
      "loss": 0.4001,
      "step": 13
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.796093065705644e-05,
      "loss": 0.282,
      "step": 14
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.7621620551276366e-05,
      "loss": 0.3168,
      "step": 15
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.725995491923131e-05,
      "loss": 0.0848,
      "step": 16
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.6876994588534234e-05,
      "loss": 0.3817,
      "step": 17
    },
    {
      "epoch": 6.0,
      "learning_rate": 1.647386284781828e-05,
      "loss": 0.1012,
      "step": 18
    },
    {
      "epoch": 6.33,
      "learning_rate": 1.6051742151937655e-05,
      "loss": 0.2289,
      "step": 19
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.5611870653623826e-05,
      "loss": 0.2038,
      "step": 20
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.515553857177022e-05,
      "loss": 0.1647,
      "step": 21
    },
    {
      "epoch": 7.33,
      "learning_rate": 1.4684084406997903e-05,
      "loss": 0.1464,
      "step": 22
    },
    {
      "epoch": 7.67,
      "learning_rate": 1.4198891015602648e-05,
      "loss": 0.1856,
      "step": 23
    },
    {
      "epoch": 8.0,
      "learning_rate": 1.3701381553399147e-05,
      "loss": 0.1491,
      "step": 24
    },
    {
      "epoch": 8.33,
      "learning_rate": 1.31930153013598e-05,
      "loss": 0.0689,
      "step": 25
    },
    {
      "epoch": 8.67,
      "learning_rate": 1.2675283385292212e-05,
      "loss": 0.1416,
      "step": 26
    },
    {
      "epoch": 9.0,
      "learning_rate": 1.2149704402110243e-05,
      "loss": 0.1034,
      "step": 27
    },
    {
      "epoch": 9.33,
      "learning_rate": 1.161781996552765e-05,
      "loss": 0.0925,
      "step": 28
    },
    {
      "epoch": 9.67,
      "learning_rate": 1.1081190184239418e-05,
      "loss": 0.0978,
      "step": 29
    },
    {
      "epoch": 10.0,
      "learning_rate": 1.0541389085854177e-05,
      "loss": 0.1098,
      "step": 30
    },
    {
      "epoch": 10.33,
      "learning_rate": 1e-05,
      "loss": 0.0948,
      "step": 31
    },
    {
      "epoch": 10.67,
      "learning_rate": 9.458610914145826e-06,
      "loss": 0.0881,
      "step": 32
    },
    {
      "epoch": 11.0,
      "learning_rate": 8.918809815760585e-06,
      "loss": 0.1011,
      "step": 33
    },
    {
      "epoch": 11.33,
      "learning_rate": 8.382180034472353e-06,
      "loss": 0.0923,
      "step": 34
    },
    {
      "epoch": 11.67,
      "learning_rate": 7.85029559788976e-06,
      "loss": 0.0799,
      "step": 35
    },
    {
      "epoch": 12.0,
      "learning_rate": 7.324716614707794e-06,
      "loss": 0.0826,
      "step": 36
    },
    {
      "epoch": 12.33,
      "learning_rate": 6.806984698640202e-06,
      "loss": 0.0695,
      "step": 37
    },
    {
      "epoch": 12.67,
      "learning_rate": 6.298618446600856e-06,
      "loss": 0.1022,
      "step": 38
    },
    {
      "epoch": 13.0,
      "learning_rate": 5.801108984397355e-06,
      "loss": 0.0734,
      "step": 39
    },
    {
      "epoch": 13.33,
      "learning_rate": 5.3159155930021e-06,
      "loss": 0.0589,
      "step": 40
    },
    {
      "epoch": 13.67,
      "learning_rate": 4.844461428229782e-06,
      "loss": 0.0674,
      "step": 41
    },
    {
      "epoch": 14.0,
      "learning_rate": 4.388129346376177e-06,
      "loss": 0.0859,
      "step": 42
    },
    {
      "epoch": 14.33,
      "learning_rate": 3.948257848062351e-06,
      "loss": 0.0899,
      "step": 43
    },
    {
      "epoch": 14.67,
      "learning_rate": 3.5261371521817247e-06,
      "loss": 0.0818,
      "step": 44
    },
    {
      "epoch": 15.0,
      "learning_rate": 3.123005411465766e-06,
      "loss": 0.0388,
      "step": 45
    },
    {
      "epoch": 15.33,
      "learning_rate": 2.740045080768694e-06,
      "loss": 0.0519,
      "step": 46
    },
    {
      "epoch": 15.67,
      "learning_rate": 2.3783794487236367e-06,
      "loss": 0.0809,
      "step": 47
    },
    {
      "epoch": 16.0,
      "learning_rate": 2.0390693429435626e-06,
      "loss": 0.0675,
      "step": 48
    },
    {
      "epoch": 16.33,
      "learning_rate": 1.7231100184310955e-06,
      "loss": 0.0788,
      "step": 49
    },
    {
      "epoch": 16.67,
      "learning_rate": 1.4314282383241097e-06,
      "loss": 0.0432,
      "step": 50
    },
    {
      "epoch": 17.0,
      "learning_rate": 1.1648795555397719e-06,
      "loss": 0.0831,
      "step": 51
    },
    {
      "epoch": 17.33,
      "learning_rate": 9.242458032904311e-07,
      "loss": 0.0524,
      "step": 52
    },
    {
      "epoch": 17.67,
      "learning_rate": 7.102328018320859e-07,
      "loss": 0.0613,
      "step": 53
    },
    {
      "epoch": 18.0,
      "learning_rate": 5.234682881719766e-07,
      "loss": 0.0767,
      "step": 54
    },
    {
      "epoch": 18.33,
      "learning_rate": 3.645000748077709e-07,
      "loss": 0.0654,
      "step": 55
    },
    {
      "epoch": 18.67,
      "learning_rate": 2.3379444289913344e-07,
      "loss": 0.0678,
      "step": 56
    },
    {
      "epoch": 19.0,
      "learning_rate": 1.317347745847386e-07,
      "loss": 0.0505,
      "step": 57
    },
    {
      "epoch": 19.33,
      "learning_rate": 5.862042845640403e-08,
      "loss": 0.0554,
      "step": 58
    },
    {
      "epoch": 19.67,
      "learning_rate": 1.4665861488761813e-08,
      "loss": 0.0765,
      "step": 59
    },
    {
      "epoch": 20.0,
      "learning_rate": 0.0,
      "loss": 0.0564,
      "step": 60
    },
    {
      "epoch": 20.0,
      "step": 60,
      "total_flos": 2666696048640.0,
      "train_loss": 0.9396996946384509,
      "train_runtime": 586.0007,
      "train_samples_per_second": 6.143,
      "train_steps_per_second": 0.102
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 60,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 50000,
  "total_flos": 2666696048640.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}