{
  "best_metric": NaN,
  "best_model_checkpoint": "miner_id_24/checkpoint-25",
  "epoch": 1.81941309255079,
  "eval_steps": 25,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03611738148984198,
      "grad_norm": NaN,
      "learning_rate": 5e-05,
      "loss": 27108.3242,
      "step": 1
    },
    {
      "epoch": 0.03611738148984198,
      "eval_loss": NaN,
      "eval_runtime": 2.4092,
      "eval_samples_per_second": 20.753,
      "eval_steps_per_second": 5.396,
      "step": 1
    },
    {
      "epoch": 0.07223476297968397,
      "grad_norm": NaN,
      "learning_rate": 0.0001,
      "loss": 0.0,
      "step": 2
    },
    {
      "epoch": 0.10835214446952596,
      "grad_norm": NaN,
      "learning_rate": 9.990365154573717e-05,
      "loss": 0.0,
      "step": 3
    },
    {
      "epoch": 0.14446952595936793,
      "grad_norm": NaN,
      "learning_rate": 9.961501876182148e-05,
      "loss": 0.0,
      "step": 4
    },
    {
      "epoch": 0.18058690744920994,
      "grad_norm": NaN,
      "learning_rate": 9.913533761814537e-05,
      "loss": 0.0,
      "step": 5
    },
    {
      "epoch": 0.21670428893905191,
      "grad_norm": NaN,
      "learning_rate": 9.846666218300807e-05,
      "loss": 0.0,
      "step": 6
    },
    {
      "epoch": 0.2528216704288939,
      "grad_norm": NaN,
      "learning_rate": 9.761185582727977e-05,
      "loss": 0.0,
      "step": 7
    },
    {
      "epoch": 0.28893905191873587,
      "grad_norm": NaN,
      "learning_rate": 9.657457896300791e-05,
      "loss": 0.0,
      "step": 8
    },
    {
      "epoch": 0.32505643340857787,
      "grad_norm": NaN,
      "learning_rate": 9.535927336897098e-05,
      "loss": 0.0,
      "step": 9
    },
    {
      "epoch": 0.3611738148984199,
      "grad_norm": NaN,
      "learning_rate": 9.397114317029975e-05,
      "loss": 0.0,
      "step": 10
    },
    {
      "epoch": 0.3972911963882618,
      "grad_norm": NaN,
      "learning_rate": 9.241613255361455e-05,
      "loss": 0.0,
      "step": 11
    },
    {
      "epoch": 0.43340857787810383,
      "grad_norm": NaN,
      "learning_rate": 9.070090031310558e-05,
      "loss": 0.0,
      "step": 12
    },
    {
      "epoch": 0.46952595936794583,
      "grad_norm": NaN,
      "learning_rate": 8.883279133655399e-05,
      "loss": 0.0,
      "step": 13
    },
    {
      "epoch": 0.5056433408577878,
      "grad_norm": NaN,
      "learning_rate": 8.681980515339464e-05,
      "loss": 0.0,
      "step": 14
    },
    {
      "epoch": 0.5417607223476298,
      "grad_norm": NaN,
      "learning_rate": 8.467056167950311e-05,
      "loss": 0.0,
      "step": 15
    },
    {
      "epoch": 0.5778781038374717,
      "grad_norm": NaN,
      "learning_rate": 8.239426430539243e-05,
      "loss": 0.0,
      "step": 16
    },
    {
      "epoch": 0.6139954853273137,
      "grad_norm": NaN,
      "learning_rate": 8.000066048588211e-05,
      "loss": 0.0,
      "step": 17
    },
    {
      "epoch": 0.6501128668171557,
      "grad_norm": NaN,
      "learning_rate": 7.75e-05,
      "loss": 0.0,
      "step": 18
    },
    {
      "epoch": 0.6862302483069977,
      "grad_norm": NaN,
      "learning_rate": 7.490299105985507e-05,
      "loss": 0.0,
      "step": 19
    },
    {
      "epoch": 0.7223476297968398,
      "grad_norm": NaN,
      "learning_rate": 7.222075445642904e-05,
      "loss": 0.0,
      "step": 20
    },
    {
      "epoch": 0.7584650112866818,
      "grad_norm": NaN,
      "learning_rate": 6.946477593864228e-05,
      "loss": 0.0,
      "step": 21
    },
    {
      "epoch": 0.7945823927765236,
      "grad_norm": NaN,
      "learning_rate": 6.664685702961344e-05,
      "loss": 0.0,
      "step": 22
    },
    {
      "epoch": 0.8306997742663657,
      "grad_norm": NaN,
      "learning_rate": 6.377906449072578e-05,
      "loss": 0.0,
      "step": 23
    },
    {
      "epoch": 0.8668171557562077,
      "grad_norm": NaN,
      "learning_rate": 6.087367864990233e-05,
      "loss": 0.0,
      "step": 24
    },
    {
      "epoch": 0.9029345372460497,
      "grad_norm": NaN,
      "learning_rate": 5.794314081535644e-05,
      "loss": 0.0,
      "step": 25
    },
    {
      "epoch": 0.9029345372460497,
      "eval_loss": NaN,
      "eval_runtime": 1.7749,
      "eval_samples_per_second": 28.171,
      "eval_steps_per_second": 7.324,
      "step": 25
    },
    {
      "epoch": 0.9390519187358917,
      "grad_norm": NaN,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.0,
      "step": 26
    },
    {
      "epoch": 0.9751693002257337,
      "grad_norm": NaN,
      "learning_rate": 5.205685918464356e-05,
      "loss": 0.0,
      "step": 27
    },
    {
      "epoch": 1.0248306997742664,
      "grad_norm": NaN,
      "learning_rate": 4.912632135009769e-05,
      "loss": 0.0,
      "step": 28
    },
    {
      "epoch": 1.0609480812641083,
      "grad_norm": NaN,
      "learning_rate": 4.6220935509274235e-05,
      "loss": 0.0,
      "step": 29
    },
    {
      "epoch": 1.0970654627539504,
      "grad_norm": NaN,
      "learning_rate": 4.3353142970386564e-05,
      "loss": 0.0,
      "step": 30
    },
    {
      "epoch": 1.1331828442437923,
      "grad_norm": NaN,
      "learning_rate": 4.053522406135775e-05,
      "loss": 0.0,
      "step": 31
    },
    {
      "epoch": 1.1693002257336342,
      "grad_norm": NaN,
      "learning_rate": 3.777924554357096e-05,
      "loss": 0.0,
      "step": 32
    },
    {
      "epoch": 1.2054176072234764,
      "grad_norm": NaN,
      "learning_rate": 3.509700894014496e-05,
      "loss": 0.0,
      "step": 33
    },
    {
      "epoch": 1.2415349887133182,
      "grad_norm": NaN,
      "learning_rate": 3.250000000000001e-05,
      "loss": 0.0,
      "step": 34
    },
    {
      "epoch": 1.2776523702031604,
      "grad_norm": NaN,
      "learning_rate": 2.9999339514117912e-05,
      "loss": 0.0,
      "step": 35
    },
    {
      "epoch": 1.3137697516930023,
      "grad_norm": NaN,
      "learning_rate": 2.760573569460757e-05,
      "loss": 0.0,
      "step": 36
    },
    {
      "epoch": 1.3498871331828441,
      "grad_norm": NaN,
      "learning_rate": 2.53294383204969e-05,
      "loss": 0.0,
      "step": 37
    },
    {
      "epoch": 1.3860045146726863,
      "grad_norm": NaN,
      "learning_rate": 2.3180194846605367e-05,
      "loss": 0.0,
      "step": 38
    },
    {
      "epoch": 1.4221218961625282,
      "grad_norm": NaN,
      "learning_rate": 2.1167208663446025e-05,
      "loss": 0.0,
      "step": 39
    },
    {
      "epoch": 1.4582392776523703,
      "grad_norm": NaN,
      "learning_rate": 1.9299099686894423e-05,
      "loss": 0.0,
      "step": 40
    },
    {
      "epoch": 1.4943566591422122,
      "grad_norm": NaN,
      "learning_rate": 1.758386744638546e-05,
      "loss": 0.0,
      "step": 41
    },
    {
      "epoch": 1.530474040632054,
      "grad_norm": NaN,
      "learning_rate": 1.602885682970026e-05,
      "loss": 0.0,
      "step": 42
    },
    {
      "epoch": 1.5665914221218962,
      "grad_norm": NaN,
      "learning_rate": 1.464072663102903e-05,
      "loss": 0.0,
      "step": 43
    },
    {
      "epoch": 1.6027088036117383,
      "grad_norm": NaN,
      "learning_rate": 1.3425421036992098e-05,
      "loss": 0.0,
      "step": 44
    },
    {
      "epoch": 1.6388261851015802,
      "grad_norm": NaN,
      "learning_rate": 1.2388144172720251e-05,
      "loss": 0.0,
      "step": 45
    },
    {
      "epoch": 1.674943566591422,
      "grad_norm": NaN,
      "learning_rate": 1.1533337816991932e-05,
      "loss": 0.0,
      "step": 46
    },
    {
      "epoch": 1.711060948081264,
      "grad_norm": NaN,
      "learning_rate": 1.0864662381854632e-05,
      "loss": 0.0,
      "step": 47
    },
    {
      "epoch": 1.747178329571106,
      "grad_norm": NaN,
      "learning_rate": 1.0384981238178534e-05,
      "loss": 0.0,
      "step": 48
    },
    {
      "epoch": 1.7832957110609482,
      "grad_norm": NaN,
      "learning_rate": 1.0096348454262845e-05,
      "loss": 0.0,
      "step": 49
    },
    {
      "epoch": 1.81941309255079,
      "grad_norm": NaN,
      "learning_rate": 1e-05,
      "loss": 0.0,
      "step": 50
    },
    {
      "epoch": 1.81941309255079,
      "eval_loss": NaN,
      "eval_runtime": 1.7715,
      "eval_samples_per_second": 28.224,
      "eval_steps_per_second": 7.338,
      "step": 50
    }
  ],
  "logging_steps": 1,
  "max_steps": 50,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 25,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 1,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.658027780734976e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}