{
  "best_metric": 0.47584664821624756,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 0.049887752556747315,
  "eval_steps": 25,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0009977550511349464,
      "grad_norm": 14.578402519226074,
      "learning_rate": 0.0001,
      "loss": 3.1386,
      "step": 1
    },
    {
      "epoch": 0.0009977550511349464,
      "eval_loss": 3.9661507606506348,
      "eval_runtime": 30.5336,
      "eval_samples_per_second": 13.821,
      "eval_steps_per_second": 6.91,
      "step": 1
    },
    {
      "epoch": 0.001995510102269893,
      "grad_norm": 13.562771797180176,
      "learning_rate": 0.0002,
      "loss": 3.2937,
      "step": 2
    },
    {
      "epoch": 0.002993265153404839,
      "grad_norm": 10.024742126464844,
      "learning_rate": 0.00019978589232386035,
      "loss": 2.7662,
      "step": 3
    },
    {
      "epoch": 0.003991020204539786,
      "grad_norm": 6.179915428161621,
      "learning_rate": 0.00019914448613738106,
      "loss": 2.1326,
      "step": 4
    },
    {
      "epoch": 0.004988775255674732,
      "grad_norm": 5.744348049163818,
      "learning_rate": 0.00019807852804032305,
      "loss": 1.4636,
      "step": 5
    },
    {
      "epoch": 0.005986530306809678,
      "grad_norm": 8.211645126342773,
      "learning_rate": 0.00019659258262890683,
      "loss": 1.204,
      "step": 6
    },
    {
      "epoch": 0.006984285357944624,
      "grad_norm": 5.339918613433838,
      "learning_rate": 0.0001946930129495106,
      "loss": 0.8177,
      "step": 7
    },
    {
      "epoch": 0.007982040409079572,
      "grad_norm": 3.9543955326080322,
      "learning_rate": 0.0001923879532511287,
      "loss": 0.5827,
      "step": 8
    },
    {
      "epoch": 0.008979795460214518,
      "grad_norm": 3.891875743865967,
      "learning_rate": 0.00018968727415326884,
      "loss": 0.5406,
      "step": 9
    },
    {
      "epoch": 0.009977550511349464,
      "grad_norm": 2.5603177547454834,
      "learning_rate": 0.00018660254037844388,
      "loss": 0.5057,
      "step": 10
    },
    {
      "epoch": 0.010975305562484411,
      "grad_norm": 2.7920143604278564,
      "learning_rate": 0.00018314696123025454,
      "loss": 0.6804,
      "step": 11
    },
    {
      "epoch": 0.011973060613619356,
      "grad_norm": 5.024642467498779,
      "learning_rate": 0.00017933533402912354,
      "loss": 0.7344,
      "step": 12
    },
    {
      "epoch": 0.012970815664754302,
      "grad_norm": 4.325658321380615,
      "learning_rate": 0.00017518398074789775,
      "loss": 0.5581,
      "step": 13
    },
    {
      "epoch": 0.013968570715889249,
      "grad_norm": 3.3327889442443848,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.567,
      "step": 14
    },
    {
      "epoch": 0.014966325767024195,
      "grad_norm": 12.798796653747559,
      "learning_rate": 0.00016593458151000688,
      "loss": 1.4479,
      "step": 15
    },
    {
      "epoch": 0.015964080818159143,
      "grad_norm": 7.979032039642334,
      "learning_rate": 0.00016087614290087208,
      "loss": 0.9935,
      "step": 16
    },
    {
      "epoch": 0.01696183586929409,
      "grad_norm": 6.214400768280029,
      "learning_rate": 0.00015555702330196023,
      "loss": 0.5051,
      "step": 17
    },
    {
      "epoch": 0.017959590920429036,
      "grad_norm": 3.6293129920959473,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.5177,
      "step": 18
    },
    {
      "epoch": 0.018957345971563982,
      "grad_norm": 9.864937782287598,
      "learning_rate": 0.00014422886902190014,
      "loss": 0.581,
      "step": 19
    },
    {
      "epoch": 0.01995510102269893,
      "grad_norm": 4.624186038970947,
      "learning_rate": 0.000138268343236509,
      "loss": 0.6062,
      "step": 20
    },
    {
      "epoch": 0.020952856073833875,
      "grad_norm": 5.2243547439575195,
      "learning_rate": 0.00013214394653031616,
      "loss": 0.7115,
      "step": 21
    },
    {
      "epoch": 0.021950611124968822,
      "grad_norm": 6.007659435272217,
      "learning_rate": 0.00012588190451025207,
      "loss": 0.7287,
      "step": 22
    },
    {
      "epoch": 0.022948366176103765,
      "grad_norm": 2.7791993618011475,
      "learning_rate": 0.00011950903220161285,
      "loss": 0.5273,
      "step": 23
    },
    {
      "epoch": 0.02394612122723871,
      "grad_norm": 2.6890182495117188,
      "learning_rate": 0.00011305261922200519,
      "loss": 0.5876,
      "step": 24
    },
    {
      "epoch": 0.024943876278373658,
      "grad_norm": 3.2656936645507812,
      "learning_rate": 0.00010654031292301432,
      "loss": 0.49,
      "step": 25
    },
    {
      "epoch": 0.024943876278373658,
      "eval_loss": 0.5218021869659424,
      "eval_runtime": 30.5895,
      "eval_samples_per_second": 13.796,
      "eval_steps_per_second": 6.898,
      "step": 25
    },
    {
      "epoch": 0.025941631329508604,
      "grad_norm": 2.6642682552337646,
      "learning_rate": 0.0001,
      "loss": 0.3961,
      "step": 26
    },
    {
      "epoch": 0.02693938638064355,
      "grad_norm": 3.069103479385376,
      "learning_rate": 9.345968707698569e-05,
      "loss": 0.5917,
      "step": 27
    },
    {
      "epoch": 0.027937141431778497,
      "grad_norm": 2.6941702365875244,
      "learning_rate": 8.694738077799488e-05,
      "loss": 0.5521,
      "step": 28
    },
    {
      "epoch": 0.028934896482913443,
      "grad_norm": 2.3197388648986816,
      "learning_rate": 8.049096779838719e-05,
      "loss": 0.4573,
      "step": 29
    },
    {
      "epoch": 0.02993265153404839,
      "grad_norm": 1.78239107131958,
      "learning_rate": 7.411809548974792e-05,
      "loss": 0.3808,
      "step": 30
    },
    {
      "epoch": 0.030930406585183336,
      "grad_norm": 2.3100101947784424,
      "learning_rate": 6.785605346968386e-05,
      "loss": 0.4714,
      "step": 31
    },
    {
      "epoch": 0.031928161636318286,
      "grad_norm": 2.6778717041015625,
      "learning_rate": 6.173165676349103e-05,
      "loss": 0.4901,
      "step": 32
    },
    {
      "epoch": 0.03292591668745323,
      "grad_norm": 3.6796157360076904,
      "learning_rate": 5.577113097809989e-05,
      "loss": 0.4943,
      "step": 33
    },
    {
      "epoch": 0.03392367173858818,
      "grad_norm": 2.953310966491699,
      "learning_rate": 5.000000000000002e-05,
      "loss": 0.4483,
      "step": 34
    },
    {
      "epoch": 0.034921426789723126,
      "grad_norm": 2.2979049682617188,
      "learning_rate": 4.444297669803981e-05,
      "loss": 0.4089,
      "step": 35
    },
    {
      "epoch": 0.03591918184085807,
      "grad_norm": 3.4585156440734863,
      "learning_rate": 3.9123857099127936e-05,
      "loss": 0.5307,
      "step": 36
    },
    {
      "epoch": 0.03691693689199302,
      "grad_norm": 1.8488389253616333,
      "learning_rate": 3.406541848999312e-05,
      "loss": 0.4315,
      "step": 37
    },
    {
      "epoch": 0.037914691943127965,
      "grad_norm": 3.7314658164978027,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 0.4729,
      "step": 38
    },
    {
      "epoch": 0.03891244699426291,
      "grad_norm": 2.832077980041504,
      "learning_rate": 2.4816019252102273e-05,
      "loss": 0.5137,
      "step": 39
    },
    {
      "epoch": 0.03991020204539786,
      "grad_norm": 2.1612372398376465,
      "learning_rate": 2.0664665970876496e-05,
      "loss": 0.4439,
      "step": 40
    },
    {
      "epoch": 0.040907957096532804,
      "grad_norm": 3.2860889434814453,
      "learning_rate": 1.6853038769745467e-05,
      "loss": 0.5148,
      "step": 41
    },
    {
      "epoch": 0.04190571214766775,
      "grad_norm": 2.663923978805542,
      "learning_rate": 1.339745962155613e-05,
      "loss": 0.4098,
      "step": 42
    },
    {
      "epoch": 0.0429034671988027,
      "grad_norm": 1.9550365209579468,
      "learning_rate": 1.0312725846731175e-05,
      "loss": 0.3167,
      "step": 43
    },
    {
      "epoch": 0.043901222249937644,
      "grad_norm": 2.115276336669922,
      "learning_rate": 7.612046748871327e-06,
      "loss": 0.4412,
      "step": 44
    },
    {
      "epoch": 0.04489897730107259,
      "grad_norm": 2.8650126457214355,
      "learning_rate": 5.306987050489442e-06,
      "loss": 0.5174,
      "step": 45
    },
    {
      "epoch": 0.04589673235220753,
      "grad_norm": 4.486678600311279,
      "learning_rate": 3.40741737109318e-06,
      "loss": 0.8079,
      "step": 46
    },
    {
      "epoch": 0.046894487403342476,
      "grad_norm": 2.707068681716919,
      "learning_rate": 1.921471959676957e-06,
      "loss": 0.5731,
      "step": 47
    },
    {
      "epoch": 0.04789224245447742,
      "grad_norm": 3.052170753479004,
      "learning_rate": 8.555138626189618e-07,
      "loss": 0.5322,
      "step": 48
    },
    {
      "epoch": 0.04888999750561237,
      "grad_norm": 3.089545249938965,
      "learning_rate": 2.141076761396521e-07,
      "loss": 0.6806,
      "step": 49
    },
    {
      "epoch": 0.049887752556747315,
      "grad_norm": 3.840698719024658,
      "learning_rate": 0.0,
      "loss": 0.6611,
      "step": 50
    },
    {
      "epoch": 0.049887752556747315,
      "eval_loss": 0.47584664821624756,
      "eval_runtime": 30.5825,
      "eval_samples_per_second": 13.799,
      "eval_steps_per_second": 6.899,
      "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": 3604771504128000.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}