{
  "best_metric": 1.1821939945220947,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 0.03170577045022194,
  "eval_steps": 25,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006341154090044388,
      "grad_norm": 0.30783170461654663,
      "learning_rate": 2e-05,
      "loss": 1.8121,
      "step": 1
    },
    {
      "epoch": 0.0006341154090044388,
      "eval_loss": 1.7660390138626099,
      "eval_runtime": 70.6773,
      "eval_samples_per_second": 37.579,
      "eval_steps_per_second": 4.697,
      "step": 1
    },
    {
      "epoch": 0.0012682308180088776,
      "grad_norm": 0.3460358679294586,
      "learning_rate": 4e-05,
      "loss": 1.7863,
      "step": 2
    },
    {
      "epoch": 0.0019023462270133164,
      "grad_norm": 0.32900822162628174,
      "learning_rate": 6e-05,
      "loss": 1.7817,
      "step": 3
    },
    {
      "epoch": 0.0025364616360177552,
      "grad_norm": 0.3095568120479584,
      "learning_rate": 8e-05,
      "loss": 1.7955,
      "step": 4
    },
    {
      "epoch": 0.0031705770450221942,
      "grad_norm": 0.27294236421585083,
      "learning_rate": 0.0001,
      "loss": 1.7693,
      "step": 5
    },
    {
      "epoch": 0.003804692454026633,
      "grad_norm": 0.21330417692661285,
      "learning_rate": 9.987820251299122e-05,
      "loss": 1.7208,
      "step": 6
    },
    {
      "epoch": 0.004438807863031071,
      "grad_norm": 0.21737153828144073,
      "learning_rate": 9.951340343707852e-05,
      "loss": 1.7017,
      "step": 7
    },
    {
      "epoch": 0.0050729232720355105,
      "grad_norm": 0.20402231812477112,
      "learning_rate": 9.890738003669029e-05,
      "loss": 1.6078,
      "step": 8
    },
    {
      "epoch": 0.0057070386810399495,
      "grad_norm": 0.20581334829330444,
      "learning_rate": 9.806308479691595e-05,
      "loss": 1.5612,
      "step": 9
    },
    {
      "epoch": 0.0063411540900443885,
      "grad_norm": 0.21240539848804474,
      "learning_rate": 9.698463103929542e-05,
      "loss": 1.5863,
      "step": 10
    },
    {
      "epoch": 0.006975269499048827,
      "grad_norm": 0.21864530444145203,
      "learning_rate": 9.567727288213005e-05,
      "loss": 1.5565,
      "step": 11
    },
    {
      "epoch": 0.007609384908053266,
      "grad_norm": 0.21938222646713257,
      "learning_rate": 9.414737964294636e-05,
      "loss": 1.5197,
      "step": 12
    },
    {
      "epoch": 0.008243500317057704,
      "grad_norm": 0.21815477311611176,
      "learning_rate": 9.24024048078213e-05,
      "loss": 1.4577,
      "step": 13
    },
    {
      "epoch": 0.008877615726062143,
      "grad_norm": 0.20750688016414642,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.4351,
      "step": 14
    },
    {
      "epoch": 0.009511731135066582,
      "grad_norm": 0.19422230124473572,
      "learning_rate": 8.83022221559489e-05,
      "loss": 1.42,
      "step": 15
    },
    {
      "epoch": 0.010145846544071021,
      "grad_norm": 0.1986970752477646,
      "learning_rate": 8.596699001693255e-05,
      "loss": 1.3745,
      "step": 16
    },
    {
      "epoch": 0.01077996195307546,
      "grad_norm": 0.20829378068447113,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.3615,
      "step": 17
    },
    {
      "epoch": 0.011414077362079899,
      "grad_norm": 0.20694291591644287,
      "learning_rate": 8.07830737662829e-05,
      "loss": 1.3407,
      "step": 18
    },
    {
      "epoch": 0.012048192771084338,
      "grad_norm": 0.1956637054681778,
      "learning_rate": 7.795964517353735e-05,
      "loss": 1.3417,
      "step": 19
    },
    {
      "epoch": 0.012682308180088777,
      "grad_norm": 0.1798098236322403,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.3138,
      "step": 20
    },
    {
      "epoch": 0.013316423589093214,
      "grad_norm": 0.18528112769126892,
      "learning_rate": 7.191855733945387e-05,
      "loss": 1.3492,
      "step": 21
    },
    {
      "epoch": 0.013950538998097653,
      "grad_norm": 0.17369288206100464,
      "learning_rate": 6.873032967079561e-05,
      "loss": 1.2976,
      "step": 22
    },
    {
      "epoch": 0.014584654407102092,
      "grad_norm": 0.17437110841274261,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.2588,
      "step": 23
    },
    {
      "epoch": 0.015218769816106531,
      "grad_norm": 0.17446409165859222,
      "learning_rate": 6.209609477998338e-05,
      "loss": 1.2802,
      "step": 24
    },
    {
      "epoch": 0.01585288522511097,
      "grad_norm": 0.16432027518749237,
      "learning_rate": 5.868240888334653e-05,
      "loss": 1.2186,
      "step": 25
    },
    {
      "epoch": 0.01585288522511097,
      "eval_loss": 1.246727466583252,
      "eval_runtime": 71.8404,
      "eval_samples_per_second": 36.971,
      "eval_steps_per_second": 4.621,
      "step": 25
    },
    {
      "epoch": 0.016487000634115408,
      "grad_norm": 0.1652495563030243,
      "learning_rate": 5.522642316338268e-05,
      "loss": 1.2518,
      "step": 26
    },
    {
      "epoch": 0.017121116043119847,
      "grad_norm": 0.14872929453849792,
      "learning_rate": 5.174497483512506e-05,
      "loss": 1.2467,
      "step": 27
    },
    {
      "epoch": 0.017755231452124286,
      "grad_norm": 0.15020141005516052,
      "learning_rate": 4.825502516487497e-05,
      "loss": 1.1887,
      "step": 28
    },
    {
      "epoch": 0.018389346861128725,
      "grad_norm": 0.1462440937757492,
      "learning_rate": 4.477357683661734e-05,
      "loss": 1.256,
      "step": 29
    },
    {
      "epoch": 0.019023462270133164,
      "grad_norm": 0.14527584612369537,
      "learning_rate": 4.131759111665349e-05,
      "loss": 1.2337,
      "step": 30
    },
    {
      "epoch": 0.019657577679137603,
      "grad_norm": 0.1435730755329132,
      "learning_rate": 3.790390522001662e-05,
      "loss": 1.2035,
      "step": 31
    },
    {
      "epoch": 0.020291693088142042,
      "grad_norm": 0.13261650502681732,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.2267,
      "step": 32
    },
    {
      "epoch": 0.02092580849714648,
      "grad_norm": 0.13555127382278442,
      "learning_rate": 3.12696703292044e-05,
      "loss": 1.1804,
      "step": 33
    },
    {
      "epoch": 0.02155992390615092,
      "grad_norm": 0.13484609127044678,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 1.223,
      "step": 34
    },
    {
      "epoch": 0.02219403931515536,
      "grad_norm": 0.13975246250629425,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.214,
      "step": 35
    },
    {
      "epoch": 0.022828154724159798,
      "grad_norm": 0.1460408866405487,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 1.2044,
      "step": 36
    },
    {
      "epoch": 0.023462270133164237,
      "grad_norm": 0.12898333370685577,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 1.2115,
      "step": 37
    },
    {
      "epoch": 0.024096385542168676,
      "grad_norm": 0.1304863840341568,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 1.1808,
      "step": 38
    },
    {
      "epoch": 0.024730500951173115,
      "grad_norm": 0.1284722238779068,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 1.1606,
      "step": 39
    },
    {
      "epoch": 0.025364616360177554,
      "grad_norm": 0.14207273721694946,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 1.1678,
      "step": 40
    },
    {
      "epoch": 0.02599873176918199,
      "grad_norm": 0.1767612099647522,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.2159,
      "step": 41
    },
    {
      "epoch": 0.02663284717818643,
      "grad_norm": 0.28571370244026184,
      "learning_rate": 7.597595192178702e-06,
      "loss": 1.1637,
      "step": 42
    },
    {
      "epoch": 0.027266962587190868,
      "grad_norm": 0.29650741815567017,
      "learning_rate": 5.852620357053651e-06,
      "loss": 1.1653,
      "step": 43
    },
    {
      "epoch": 0.027901077996195307,
      "grad_norm": 0.3082343637943268,
      "learning_rate": 4.322727117869951e-06,
      "loss": 1.167,
      "step": 44
    },
    {
      "epoch": 0.028535193405199746,
      "grad_norm": 0.3386911153793335,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 1.1402,
      "step": 45
    },
    {
      "epoch": 0.029169308814204185,
      "grad_norm": 0.33259084820747375,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 1.1347,
      "step": 46
    },
    {
      "epoch": 0.029803424223208624,
      "grad_norm": 0.3457428216934204,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 1.1041,
      "step": 47
    },
    {
      "epoch": 0.030437539632213063,
      "grad_norm": 0.3795575201511383,
      "learning_rate": 4.865965629214819e-07,
      "loss": 1.0849,
      "step": 48
    },
    {
      "epoch": 0.031071655041217502,
      "grad_norm": 0.3850218653678894,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 1.1325,
      "step": 49
    },
    {
      "epoch": 0.03170577045022194,
      "grad_norm": 0.41590115427970886,
      "learning_rate": 0.0,
      "loss": 1.1493,
      "step": 50
    },
    {
      "epoch": 0.03170577045022194,
      "eval_loss": 1.1821939945220947,
      "eval_runtime": 71.9855,
      "eval_samples_per_second": 36.896,
      "eval_steps_per_second": 4.612,
      "step": 50
    }
  ],
  "logging_steps": 1,
  "max_steps": 50,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5138858479910912e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}