{
  "best_metric": 10.362070083618164,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 0.1718213058419244,
  "eval_steps": 25,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003436426116838488,
      "grad_norm": 0.03398178890347481,
      "learning_rate": 2e-05,
      "loss": 10.3798,
      "step": 1
    },
    {
      "epoch": 0.003436426116838488,
      "eval_loss": 10.373637199401855,
      "eval_runtime": 0.5919,
      "eval_samples_per_second": 827.779,
      "eval_steps_per_second": 104.739,
      "step": 1
    },
    {
      "epoch": 0.006872852233676976,
      "grad_norm": 0.03234674409031868,
      "learning_rate": 4e-05,
      "loss": 10.3804,
      "step": 2
    },
    {
      "epoch": 0.010309278350515464,
      "grad_norm": 0.032289400696754456,
      "learning_rate": 6e-05,
      "loss": 10.3799,
      "step": 3
    },
    {
      "epoch": 0.013745704467353952,
      "grad_norm": 0.034552861005067825,
      "learning_rate": 8e-05,
      "loss": 10.3785,
      "step": 4
    },
    {
      "epoch": 0.01718213058419244,
      "grad_norm": 0.03398998826742172,
      "learning_rate": 0.0001,
      "loss": 10.3772,
      "step": 5
    },
    {
      "epoch": 0.020618556701030927,
      "grad_norm": 0.03427039831876755,
      "learning_rate": 9.987820251299122e-05,
      "loss": 10.3804,
      "step": 6
    },
    {
      "epoch": 0.024054982817869417,
      "grad_norm": 0.040781695395708084,
      "learning_rate": 9.951340343707852e-05,
      "loss": 10.3784,
      "step": 7
    },
    {
      "epoch": 0.027491408934707903,
      "grad_norm": 0.043919749557971954,
      "learning_rate": 9.890738003669029e-05,
      "loss": 10.3761,
      "step": 8
    },
    {
      "epoch": 0.030927835051546393,
      "grad_norm": 0.05272490903735161,
      "learning_rate": 9.806308479691595e-05,
      "loss": 10.3792,
      "step": 9
    },
    {
      "epoch": 0.03436426116838488,
      "grad_norm": 0.049420926719903946,
      "learning_rate": 9.698463103929542e-05,
      "loss": 10.375,
      "step": 10
    },
    {
      "epoch": 0.037800687285223365,
      "grad_norm": 0.05374721437692642,
      "learning_rate": 9.567727288213005e-05,
      "loss": 10.3756,
      "step": 11
    },
    {
      "epoch": 0.041237113402061855,
      "grad_norm": 0.07004938274621964,
      "learning_rate": 9.414737964294636e-05,
      "loss": 10.3723,
      "step": 12
    },
    {
      "epoch": 0.044673539518900345,
      "grad_norm": 0.06476572155952454,
      "learning_rate": 9.24024048078213e-05,
      "loss": 10.3742,
      "step": 13
    },
    {
      "epoch": 0.048109965635738834,
      "grad_norm": 0.08800965547561646,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.3756,
      "step": 14
    },
    {
      "epoch": 0.05154639175257732,
      "grad_norm": 0.08712270855903625,
      "learning_rate": 8.83022221559489e-05,
      "loss": 10.3735,
      "step": 15
    },
    {
      "epoch": 0.054982817869415807,
      "grad_norm": 0.09331571310758591,
      "learning_rate": 8.596699001693255e-05,
      "loss": 10.3761,
      "step": 16
    },
    {
      "epoch": 0.058419243986254296,
      "grad_norm": 0.0855708047747612,
      "learning_rate": 8.345653031794292e-05,
      "loss": 10.3708,
      "step": 17
    },
    {
      "epoch": 0.061855670103092786,
      "grad_norm": 0.09578525274991989,
      "learning_rate": 8.07830737662829e-05,
      "loss": 10.3726,
      "step": 18
    },
    {
      "epoch": 0.06529209621993128,
      "grad_norm": 0.10996372997760773,
      "learning_rate": 7.795964517353735e-05,
      "loss": 10.3733,
      "step": 19
    },
    {
      "epoch": 0.06872852233676977,
      "grad_norm": 0.10964931547641754,
      "learning_rate": 7.500000000000001e-05,
      "loss": 10.3683,
      "step": 20
    },
    {
      "epoch": 0.07216494845360824,
      "grad_norm": 0.11921316385269165,
      "learning_rate": 7.191855733945387e-05,
      "loss": 10.3703,
      "step": 21
    },
    {
      "epoch": 0.07560137457044673,
      "grad_norm": 0.12095928937196732,
      "learning_rate": 6.873032967079561e-05,
      "loss": 10.3716,
      "step": 22
    },
    {
      "epoch": 0.07903780068728522,
      "grad_norm": 0.16240659356117249,
      "learning_rate": 6.545084971874738e-05,
      "loss": 10.3659,
      "step": 23
    },
    {
      "epoch": 0.08247422680412371,
      "grad_norm": 0.13121552765369415,
      "learning_rate": 6.209609477998338e-05,
      "loss": 10.3651,
      "step": 24
    },
    {
      "epoch": 0.0859106529209622,
      "grad_norm": 0.13949643075466156,
      "learning_rate": 5.868240888334653e-05,
      "loss": 10.3644,
      "step": 25
    },
    {
      "epoch": 0.0859106529209622,
      "eval_loss": 10.366501808166504,
      "eval_runtime": 0.5892,
      "eval_samples_per_second": 831.698,
      "eval_steps_per_second": 105.235,
      "step": 25
    },
    {
      "epoch": 0.08934707903780069,
      "grad_norm": 0.13870833814144135,
      "learning_rate": 5.522642316338268e-05,
      "loss": 10.3673,
      "step": 26
    },
    {
      "epoch": 0.09278350515463918,
      "grad_norm": 0.1562601774930954,
      "learning_rate": 5.174497483512506e-05,
      "loss": 10.366,
      "step": 27
    },
    {
      "epoch": 0.09621993127147767,
      "grad_norm": 0.1659892499446869,
      "learning_rate": 4.825502516487497e-05,
      "loss": 10.3652,
      "step": 28
    },
    {
      "epoch": 0.09965635738831616,
      "grad_norm": 0.1718658059835434,
      "learning_rate": 4.477357683661734e-05,
      "loss": 10.3627,
      "step": 29
    },
    {
      "epoch": 0.10309278350515463,
      "grad_norm": 0.1496630758047104,
      "learning_rate": 4.131759111665349e-05,
      "loss": 10.3659,
      "step": 30
    },
    {
      "epoch": 0.10652920962199312,
      "grad_norm": 0.1727280616760254,
      "learning_rate": 3.790390522001662e-05,
      "loss": 10.3641,
      "step": 31
    },
    {
      "epoch": 0.10996563573883161,
      "grad_norm": 0.16806212067604065,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.3642,
      "step": 32
    },
    {
      "epoch": 0.1134020618556701,
      "grad_norm": 0.1888965219259262,
      "learning_rate": 3.12696703292044e-05,
      "loss": 10.3581,
      "step": 33
    },
    {
      "epoch": 0.11683848797250859,
      "grad_norm": 0.15063142776489258,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 10.3586,
      "step": 34
    },
    {
      "epoch": 0.12027491408934708,
      "grad_norm": 0.200958251953125,
      "learning_rate": 2.500000000000001e-05,
      "loss": 10.3599,
      "step": 35
    },
    {
      "epoch": 0.12371134020618557,
      "grad_norm": 0.18877319991588593,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 10.3539,
      "step": 36
    },
    {
      "epoch": 0.12714776632302405,
      "grad_norm": 0.248277947306633,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 10.3544,
      "step": 37
    },
    {
      "epoch": 0.13058419243986255,
      "grad_norm": 0.19337622821331024,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 10.3549,
      "step": 38
    },
    {
      "epoch": 0.13402061855670103,
      "grad_norm": 0.20499683916568756,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 10.353,
      "step": 39
    },
    {
      "epoch": 0.13745704467353953,
      "grad_norm": 0.22483669221401215,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 10.3539,
      "step": 40
    },
    {
      "epoch": 0.140893470790378,
      "grad_norm": 0.21703113615512848,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.3546,
      "step": 41
    },
    {
      "epoch": 0.14432989690721648,
      "grad_norm": 0.21860122680664062,
      "learning_rate": 7.597595192178702e-06,
      "loss": 10.3515,
      "step": 42
    },
    {
      "epoch": 0.14776632302405499,
      "grad_norm": 0.2903410494327545,
      "learning_rate": 5.852620357053651e-06,
      "loss": 10.354,
      "step": 43
    },
    {
      "epoch": 0.15120274914089346,
      "grad_norm": 0.3075595498085022,
      "learning_rate": 4.322727117869951e-06,
      "loss": 10.3498,
      "step": 44
    },
    {
      "epoch": 0.15463917525773196,
      "grad_norm": 0.36717870831489563,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 10.3494,
      "step": 45
    },
    {
      "epoch": 0.15807560137457044,
      "grad_norm": 0.25866419076919556,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 10.3518,
      "step": 46
    },
    {
      "epoch": 0.16151202749140894,
      "grad_norm": 0.3514833152294159,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 10.3519,
      "step": 47
    },
    {
      "epoch": 0.16494845360824742,
      "grad_norm": 0.3934487998485565,
      "learning_rate": 4.865965629214819e-07,
      "loss": 10.3589,
      "step": 48
    },
    {
      "epoch": 0.16838487972508592,
      "grad_norm": 0.35140520334243774,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 10.3506,
      "step": 49
    },
    {
      "epoch": 0.1718213058419244,
      "grad_norm": 0.5064478516578674,
      "learning_rate": 0.0,
      "loss": 10.3369,
      "step": 50
    },
    {
      "epoch": 0.1718213058419244,
      "eval_loss": 10.362070083618164,
      "eval_runtime": 0.5985,
      "eval_samples_per_second": 818.69,
      "eval_steps_per_second": 103.589,
      "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": 2886373343232.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}