{
  "best_metric": 11.060905456542969,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 0.3110419906687403,
  "eval_steps": 25,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006220839813374806,
      "grad_norm": 2.205610513687134,
      "learning_rate": 5e-05,
      "loss": 177.4089,
      "step": 1
    },
    {
      "epoch": 0.006220839813374806,
      "eval_loss": 11.092248916625977,
      "eval_runtime": 0.4098,
      "eval_samples_per_second": 122.009,
      "eval_steps_per_second": 31.722,
      "step": 1
    },
    {
      "epoch": 0.012441679626749611,
      "grad_norm": 2.127887725830078,
      "learning_rate": 0.0001,
      "loss": 177.391,
      "step": 2
    },
    {
      "epoch": 0.01866251944012442,
      "grad_norm": 2.02017879486084,
      "learning_rate": 9.990365154573717e-05,
      "loss": 177.3893,
      "step": 3
    },
    {
      "epoch": 0.024883359253499222,
      "grad_norm": 2.1193184852600098,
      "learning_rate": 9.961501876182148e-05,
      "loss": 177.3868,
      "step": 4
    },
    {
      "epoch": 0.03110419906687403,
      "grad_norm": 2.1466825008392334,
      "learning_rate": 9.913533761814537e-05,
      "loss": 177.3896,
      "step": 5
    },
    {
      "epoch": 0.03732503888024884,
      "grad_norm": 2.1950700283050537,
      "learning_rate": 9.846666218300807e-05,
      "loss": 177.3752,
      "step": 6
    },
    {
      "epoch": 0.04354587869362364,
      "grad_norm": 2.3872289657592773,
      "learning_rate": 9.761185582727977e-05,
      "loss": 177.4096,
      "step": 7
    },
    {
      "epoch": 0.049766718506998445,
      "grad_norm": 2.526869773864746,
      "learning_rate": 9.657457896300791e-05,
      "loss": 177.3708,
      "step": 8
    },
    {
      "epoch": 0.05598755832037325,
      "grad_norm": 2.592602252960205,
      "learning_rate": 9.535927336897098e-05,
      "loss": 177.3552,
      "step": 9
    },
    {
      "epoch": 0.06220839813374806,
      "grad_norm": 2.8431127071380615,
      "learning_rate": 9.397114317029975e-05,
      "loss": 177.3221,
      "step": 10
    },
    {
      "epoch": 0.06842923794712286,
      "grad_norm": 2.8420827388763428,
      "learning_rate": 9.241613255361455e-05,
      "loss": 177.3431,
      "step": 11
    },
    {
      "epoch": 0.07465007776049767,
      "grad_norm": 3.1785941123962402,
      "learning_rate": 9.070090031310558e-05,
      "loss": 177.3021,
      "step": 12
    },
    {
      "epoch": 0.08087091757387248,
      "grad_norm": 2.5118324756622314,
      "learning_rate": 8.883279133655399e-05,
      "loss": 177.3058,
      "step": 13
    },
    {
      "epoch": 0.08709175738724728,
      "grad_norm": 2.194812774658203,
      "learning_rate": 8.681980515339464e-05,
      "loss": 177.2497,
      "step": 14
    },
    {
      "epoch": 0.09331259720062209,
      "grad_norm": 2.296903133392334,
      "learning_rate": 8.467056167950311e-05,
      "loss": 177.229,
      "step": 15
    },
    {
      "epoch": 0.09953343701399689,
      "grad_norm": 2.238326072692871,
      "learning_rate": 8.239426430539243e-05,
      "loss": 177.2449,
      "step": 16
    },
    {
      "epoch": 0.1057542768273717,
      "grad_norm": 2.4703667163848877,
      "learning_rate": 8.000066048588211e-05,
      "loss": 177.2187,
      "step": 17
    },
    {
      "epoch": 0.1119751166407465,
      "grad_norm": 2.3876895904541016,
      "learning_rate": 7.75e-05,
      "loss": 177.215,
      "step": 18
    },
    {
      "epoch": 0.1181959564541213,
      "grad_norm": 2.6431405544281006,
      "learning_rate": 7.490299105985507e-05,
      "loss": 177.1922,
      "step": 19
    },
    {
      "epoch": 0.12441679626749612,
      "grad_norm": 2.548267364501953,
      "learning_rate": 7.222075445642904e-05,
      "loss": 177.1956,
      "step": 20
    },
    {
      "epoch": 0.13063763608087092,
      "grad_norm": 2.8234810829162598,
      "learning_rate": 6.946477593864228e-05,
      "loss": 177.1558,
      "step": 21
    },
    {
      "epoch": 0.1368584758942457,
      "grad_norm": 2.734844446182251,
      "learning_rate": 6.664685702961344e-05,
      "loss": 177.1875,
      "step": 22
    },
    {
      "epoch": 0.14307931570762053,
      "grad_norm": 2.9448533058166504,
      "learning_rate": 6.377906449072578e-05,
      "loss": 177.2152,
      "step": 23
    },
    {
      "epoch": 0.14930015552099535,
      "grad_norm": 3.150641918182373,
      "learning_rate": 6.087367864990233e-05,
      "loss": 177.207,
      "step": 24
    },
    {
      "epoch": 0.15552099533437014,
      "grad_norm": 3.422062873840332,
      "learning_rate": 5.794314081535644e-05,
      "loss": 177.1733,
      "step": 25
    },
    {
      "epoch": 0.15552099533437014,
      "eval_loss": 11.070521354675293,
      "eval_runtime": 0.0839,
      "eval_samples_per_second": 595.63,
      "eval_steps_per_second": 154.864,
      "step": 25
    },
    {
      "epoch": 0.16174183514774496,
      "grad_norm": 2.573197364807129,
      "learning_rate": 5.500000000000001e-05,
      "loss": 177.1141,
      "step": 26
    },
    {
      "epoch": 0.16796267496111975,
      "grad_norm": 2.4286153316497803,
      "learning_rate": 5.205685918464356e-05,
      "loss": 177.0893,
      "step": 27
    },
    {
      "epoch": 0.17418351477449456,
      "grad_norm": 2.4644172191619873,
      "learning_rate": 4.912632135009769e-05,
      "loss": 177.134,
      "step": 28
    },
    {
      "epoch": 0.18040435458786935,
      "grad_norm": 2.417313575744629,
      "learning_rate": 4.6220935509274235e-05,
      "loss": 177.1395,
      "step": 29
    },
    {
      "epoch": 0.18662519440124417,
      "grad_norm": 2.416379690170288,
      "learning_rate": 4.3353142970386564e-05,
      "loss": 177.0807,
      "step": 30
    },
    {
      "epoch": 0.19284603421461896,
      "grad_norm": 2.5380895137786865,
      "learning_rate": 4.053522406135775e-05,
      "loss": 177.1028,
      "step": 31
    },
    {
      "epoch": 0.19906687402799378,
      "grad_norm": 2.898738384246826,
      "learning_rate": 3.777924554357096e-05,
      "loss": 177.0989,
      "step": 32
    },
    {
      "epoch": 0.2052877138413686,
      "grad_norm": 2.7243590354919434,
      "learning_rate": 3.509700894014496e-05,
      "loss": 177.0639,
      "step": 33
    },
    {
      "epoch": 0.2115085536547434,
      "grad_norm": 2.8405022621154785,
      "learning_rate": 3.250000000000001e-05,
      "loss": 177.0149,
      "step": 34
    },
    {
      "epoch": 0.2177293934681182,
      "grad_norm": 3.0282528400421143,
      "learning_rate": 2.9999339514117912e-05,
      "loss": 177.0964,
      "step": 35
    },
    {
      "epoch": 0.223950233281493,
      "grad_norm": 3.221609354019165,
      "learning_rate": 2.760573569460757e-05,
      "loss": 177.0643,
      "step": 36
    },
    {
      "epoch": 0.2301710730948678,
      "grad_norm": 3.1520352363586426,
      "learning_rate": 2.53294383204969e-05,
      "loss": 177.0723,
      "step": 37
    },
    {
      "epoch": 0.2363919129082426,
      "grad_norm": 2.7963945865631104,
      "learning_rate": 2.3180194846605367e-05,
      "loss": 177.02,
      "step": 38
    },
    {
      "epoch": 0.24261275272161742,
      "grad_norm": 2.6416287422180176,
      "learning_rate": 2.1167208663446025e-05,
      "loss": 177.0288,
      "step": 39
    },
    {
      "epoch": 0.24883359253499224,
      "grad_norm": 2.5692756175994873,
      "learning_rate": 1.9299099686894423e-05,
      "loss": 177.0617,
      "step": 40
    },
    {
      "epoch": 0.25505443234836706,
      "grad_norm": 2.6899027824401855,
      "learning_rate": 1.758386744638546e-05,
      "loss": 177.0123,
      "step": 41
    },
    {
      "epoch": 0.26127527216174184,
      "grad_norm": 2.611703634262085,
      "learning_rate": 1.602885682970026e-05,
      "loss": 177.0352,
      "step": 42
    },
    {
      "epoch": 0.26749611197511663,
      "grad_norm": 2.60444712638855,
      "learning_rate": 1.464072663102903e-05,
      "loss": 177.0187,
      "step": 43
    },
    {
      "epoch": 0.2737169517884914,
      "grad_norm": 2.7920703887939453,
      "learning_rate": 1.3425421036992098e-05,
      "loss": 177.0074,
      "step": 44
    },
    {
      "epoch": 0.27993779160186627,
      "grad_norm": 2.844726085662842,
      "learning_rate": 1.2388144172720251e-05,
      "loss": 177.0748,
      "step": 45
    },
    {
      "epoch": 0.28615863141524106,
      "grad_norm": 2.7402775287628174,
      "learning_rate": 1.1533337816991932e-05,
      "loss": 176.9944,
      "step": 46
    },
    {
      "epoch": 0.29237947122861585,
      "grad_norm": 2.834357976913452,
      "learning_rate": 1.0864662381854632e-05,
      "loss": 176.9829,
      "step": 47
    },
    {
      "epoch": 0.2986003110419907,
      "grad_norm": 3.010420083999634,
      "learning_rate": 1.0384981238178534e-05,
      "loss": 177.0303,
      "step": 48
    },
    {
      "epoch": 0.3048211508553655,
      "grad_norm": 2.935134172439575,
      "learning_rate": 1.0096348454262845e-05,
      "loss": 176.9928,
      "step": 49
    },
    {
      "epoch": 0.3110419906687403,
      "grad_norm": 3.591810703277588,
      "learning_rate": 1e-05,
      "loss": 176.9746,
      "step": 50
    },
    {
      "epoch": 0.3110419906687403,
      "eval_loss": 11.060905456542969,
      "eval_runtime": 0.0828,
      "eval_samples_per_second": 603.794,
      "eval_steps_per_second": 156.986,
      "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": 1,
        "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": 4202692608000.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}