{
  "best_metric": 1.5977141857147217,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 0.008104056080068074,
  "eval_steps": 25,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016208112160136148,
      "grad_norm": 24.242855072021484,
      "learning_rate": 2e-05,
      "loss": 11.3236,
      "step": 1
    },
    {
      "epoch": 0.00016208112160136148,
      "eval_loss": 2.9382340908050537,
      "eval_runtime": 749.3473,
      "eval_samples_per_second": 13.868,
      "eval_steps_per_second": 1.734,
      "step": 1
    },
    {
      "epoch": 0.00032416224320272295,
      "grad_norm": 24.139141082763672,
      "learning_rate": 4e-05,
      "loss": 11.5612,
      "step": 2
    },
    {
      "epoch": 0.00048624336480408443,
      "grad_norm": 16.70271110534668,
      "learning_rate": 6e-05,
      "loss": 11.2569,
      "step": 3
    },
    {
      "epoch": 0.0006483244864054459,
      "grad_norm": 15.517374038696289,
      "learning_rate": 8e-05,
      "loss": 10.2208,
      "step": 4
    },
    {
      "epoch": 0.0008104056080068074,
      "grad_norm": 14.588680267333984,
      "learning_rate": 0.0001,
      "loss": 9.4875,
      "step": 5
    },
    {
      "epoch": 0.0009724867296081689,
      "grad_norm": 11.657694816589355,
      "learning_rate": 9.987820251299122e-05,
      "loss": 9.1785,
      "step": 6
    },
    {
      "epoch": 0.0011345678512095303,
      "grad_norm": 10.309767723083496,
      "learning_rate": 9.951340343707852e-05,
      "loss": 9.4207,
      "step": 7
    },
    {
      "epoch": 0.0012966489728108918,
      "grad_norm": 8.698164939880371,
      "learning_rate": 9.890738003669029e-05,
      "loss": 8.5038,
      "step": 8
    },
    {
      "epoch": 0.0014587300944122533,
      "grad_norm": 8.156781196594238,
      "learning_rate": 9.806308479691595e-05,
      "loss": 8.7649,
      "step": 9
    },
    {
      "epoch": 0.0016208112160136148,
      "grad_norm": 7.506374835968018,
      "learning_rate": 9.698463103929542e-05,
      "loss": 7.5702,
      "step": 10
    },
    {
      "epoch": 0.0017828923376149762,
      "grad_norm": 6.744434356689453,
      "learning_rate": 9.567727288213005e-05,
      "loss": 7.4313,
      "step": 11
    },
    {
      "epoch": 0.0019449734592163377,
      "grad_norm": 6.5357561111450195,
      "learning_rate": 9.414737964294636e-05,
      "loss": 7.8756,
      "step": 12
    },
    {
      "epoch": 0.002107054580817699,
      "grad_norm": 6.78232479095459,
      "learning_rate": 9.24024048078213e-05,
      "loss": 7.585,
      "step": 13
    },
    {
      "epoch": 0.0022691357024190607,
      "grad_norm": 6.280641555786133,
      "learning_rate": 9.045084971874738e-05,
      "loss": 7.486,
      "step": 14
    },
    {
      "epoch": 0.002431216824020422,
      "grad_norm": 6.4468464851379395,
      "learning_rate": 8.83022221559489e-05,
      "loss": 7.0867,
      "step": 15
    },
    {
      "epoch": 0.0025932979456217836,
      "grad_norm": 5.949858665466309,
      "learning_rate": 8.596699001693255e-05,
      "loss": 7.2818,
      "step": 16
    },
    {
      "epoch": 0.002755379067223145,
      "grad_norm": 6.213530540466309,
      "learning_rate": 8.345653031794292e-05,
      "loss": 6.9205,
      "step": 17
    },
    {
      "epoch": 0.0029174601888245066,
      "grad_norm": 5.894623279571533,
      "learning_rate": 8.07830737662829e-05,
      "loss": 6.8963,
      "step": 18
    },
    {
      "epoch": 0.003079541310425868,
      "grad_norm": 5.9999470710754395,
      "learning_rate": 7.795964517353735e-05,
      "loss": 7.1314,
      "step": 19
    },
    {
      "epoch": 0.0032416224320272295,
      "grad_norm": 5.75547981262207,
      "learning_rate": 7.500000000000001e-05,
      "loss": 6.8284,
      "step": 20
    },
    {
      "epoch": 0.003403703553628591,
      "grad_norm": 5.886826038360596,
      "learning_rate": 7.191855733945387e-05,
      "loss": 6.6926,
      "step": 21
    },
    {
      "epoch": 0.0035657846752299525,
      "grad_norm": 5.543822288513184,
      "learning_rate": 6.873032967079561e-05,
      "loss": 6.6954,
      "step": 22
    },
    {
      "epoch": 0.003727865796831314,
      "grad_norm": 5.991809844970703,
      "learning_rate": 6.545084971874738e-05,
      "loss": 6.8628,
      "step": 23
    },
    {
      "epoch": 0.0038899469184326754,
      "grad_norm": 5.590277671813965,
      "learning_rate": 6.209609477998338e-05,
      "loss": 6.6366,
      "step": 24
    },
    {
      "epoch": 0.004052028040034037,
      "grad_norm": 5.658617973327637,
      "learning_rate": 5.868240888334653e-05,
      "loss": 6.2058,
      "step": 25
    },
    {
      "epoch": 0.004052028040034037,
      "eval_loss": 1.6936973333358765,
      "eval_runtime": 755.9587,
      "eval_samples_per_second": 13.747,
      "eval_steps_per_second": 1.718,
      "step": 25
    },
    {
      "epoch": 0.004214109161635398,
      "grad_norm": 5.783700942993164,
      "learning_rate": 5.522642316338268e-05,
      "loss": 6.5527,
      "step": 26
    },
    {
      "epoch": 0.00437619028323676,
      "grad_norm": 5.581294536590576,
      "learning_rate": 5.174497483512506e-05,
      "loss": 6.161,
      "step": 27
    },
    {
      "epoch": 0.004538271404838121,
      "grad_norm": 5.728160858154297,
      "learning_rate": 4.825502516487497e-05,
      "loss": 6.3117,
      "step": 28
    },
    {
      "epoch": 0.004700352526439483,
      "grad_norm": 5.748680114746094,
      "learning_rate": 4.477357683661734e-05,
      "loss": 6.4574,
      "step": 29
    },
    {
      "epoch": 0.004862433648040844,
      "grad_norm": 5.648220539093018,
      "learning_rate": 4.131759111665349e-05,
      "loss": 6.215,
      "step": 30
    },
    {
      "epoch": 0.005024514769642206,
      "grad_norm": 5.6319379806518555,
      "learning_rate": 3.790390522001662e-05,
      "loss": 5.9622,
      "step": 31
    },
    {
      "epoch": 0.005186595891243567,
      "grad_norm": 5.8718953132629395,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 6.2086,
      "step": 32
    },
    {
      "epoch": 0.005348677012844929,
      "grad_norm": 6.098597049713135,
      "learning_rate": 3.12696703292044e-05,
      "loss": 6.38,
      "step": 33
    },
    {
      "epoch": 0.00551075813444629,
      "grad_norm": 6.240828990936279,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 6.3073,
      "step": 34
    },
    {
      "epoch": 0.005672839256047652,
      "grad_norm": 6.659487247467041,
      "learning_rate": 2.500000000000001e-05,
      "loss": 6.5308,
      "step": 35
    },
    {
      "epoch": 0.005834920377649013,
      "grad_norm": 6.27744197845459,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 6.113,
      "step": 36
    },
    {
      "epoch": 0.005997001499250375,
      "grad_norm": 6.688141345977783,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 6.5809,
      "step": 37
    },
    {
      "epoch": 0.006159082620851736,
      "grad_norm": 7.0126447677612305,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 6.6365,
      "step": 38
    },
    {
      "epoch": 0.006321163742453098,
      "grad_norm": 7.171256065368652,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 6.8393,
      "step": 39
    },
    {
      "epoch": 0.006483244864054459,
      "grad_norm": 8.205148696899414,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 7.0387,
      "step": 40
    },
    {
      "epoch": 0.0066453259856558205,
      "grad_norm": 8.000205993652344,
      "learning_rate": 9.549150281252633e-06,
      "loss": 7.1483,
      "step": 41
    },
    {
      "epoch": 0.006807407107257182,
      "grad_norm": 7.7371697425842285,
      "learning_rate": 7.597595192178702e-06,
      "loss": 6.6281,
      "step": 42
    },
    {
      "epoch": 0.0069694882288585435,
      "grad_norm": 8.671666145324707,
      "learning_rate": 5.852620357053651e-06,
      "loss": 7.1924,
      "step": 43
    },
    {
      "epoch": 0.007131569350459905,
      "grad_norm": 9.656037330627441,
      "learning_rate": 4.322727117869951e-06,
      "loss": 6.8932,
      "step": 44
    },
    {
      "epoch": 0.007293650472061266,
      "grad_norm": 9.278881072998047,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 6.8363,
      "step": 45
    },
    {
      "epoch": 0.007455731593662628,
      "grad_norm": 10.156712532043457,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 7.2396,
      "step": 46
    },
    {
      "epoch": 0.007617812715263989,
      "grad_norm": 11.086377143859863,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 7.0737,
      "step": 47
    },
    {
      "epoch": 0.007779893836865351,
      "grad_norm": 12.269014358520508,
      "learning_rate": 4.865965629214819e-07,
      "loss": 7.1526,
      "step": 48
    },
    {
      "epoch": 0.007941974958466713,
      "grad_norm": 14.667993545532227,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 7.5643,
      "step": 49
    },
    {
      "epoch": 0.008104056080068074,
      "grad_norm": 21.117507934570312,
      "learning_rate": 0.0,
      "loss": 6.5986,
      "step": 50
    },
    {
      "epoch": 0.008104056080068074,
      "eval_loss": 1.5977141857147217,
      "eval_runtime": 756.8752,
      "eval_samples_per_second": 13.73,
      "eval_steps_per_second": 1.716,
      "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": 6.857571679862784e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}