{
  "best_metric": 1.6805955171585083,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 0.006881838827334664,
  "eval_steps": 25,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00013763677654669328,
      "grad_norm": 1.7081938982009888,
      "learning_rate": 2e-05,
      "loss": 1.7218,
      "step": 1
    },
    {
      "epoch": 0.00013763677654669328,
      "eval_loss": 2.2520158290863037,
      "eval_runtime": 107.6199,
      "eval_samples_per_second": 113.706,
      "eval_steps_per_second": 14.217,
      "step": 1
    },
    {
      "epoch": 0.00027527355309338655,
      "grad_norm": 2.0247623920440674,
      "learning_rate": 4e-05,
      "loss": 1.878,
      "step": 2
    },
    {
      "epoch": 0.00041291032964007983,
      "grad_norm": 2.2293405532836914,
      "learning_rate": 6e-05,
      "loss": 1.9059,
      "step": 3
    },
    {
      "epoch": 0.0005505471061867731,
      "grad_norm": 2.167025089263916,
      "learning_rate": 8e-05,
      "loss": 1.7918,
      "step": 4
    },
    {
      "epoch": 0.0006881838827334664,
      "grad_norm": 1.9536525011062622,
      "learning_rate": 0.0001,
      "loss": 1.7445,
      "step": 5
    },
    {
      "epoch": 0.0008258206592801597,
      "grad_norm": 3.0797555446624756,
      "learning_rate": 9.987820251299122e-05,
      "loss": 1.7011,
      "step": 6
    },
    {
      "epoch": 0.0009634574358268529,
      "grad_norm": 2.178736686706543,
      "learning_rate": 9.951340343707852e-05,
      "loss": 1.7041,
      "step": 7
    },
    {
      "epoch": 0.0011010942123735462,
      "grad_norm": 2.4224648475646973,
      "learning_rate": 9.890738003669029e-05,
      "loss": 1.7613,
      "step": 8
    },
    {
      "epoch": 0.0012387309889202395,
      "grad_norm": 1.9820811748504639,
      "learning_rate": 9.806308479691595e-05,
      "loss": 1.6051,
      "step": 9
    },
    {
      "epoch": 0.0013763677654669328,
      "grad_norm": 1.8484727144241333,
      "learning_rate": 9.698463103929542e-05,
      "loss": 1.6307,
      "step": 10
    },
    {
      "epoch": 0.001514004542013626,
      "grad_norm": 2.1186866760253906,
      "learning_rate": 9.567727288213005e-05,
      "loss": 1.6339,
      "step": 11
    },
    {
      "epoch": 0.0016516413185603193,
      "grad_norm": 1.8595067262649536,
      "learning_rate": 9.414737964294636e-05,
      "loss": 1.657,
      "step": 12
    },
    {
      "epoch": 0.0017892780951070126,
      "grad_norm": 1.826619029045105,
      "learning_rate": 9.24024048078213e-05,
      "loss": 1.658,
      "step": 13
    },
    {
      "epoch": 0.0019269148716537059,
      "grad_norm": 1.9336961507797241,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5857,
      "step": 14
    },
    {
      "epoch": 0.002064551648200399,
      "grad_norm": 1.7065236568450928,
      "learning_rate": 8.83022221559489e-05,
      "loss": 1.4756,
      "step": 15
    },
    {
      "epoch": 0.0022021884247470924,
      "grad_norm": 1.8239481449127197,
      "learning_rate": 8.596699001693255e-05,
      "loss": 1.7235,
      "step": 16
    },
    {
      "epoch": 0.0023398252012937857,
      "grad_norm": 1.83353590965271,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.6618,
      "step": 17
    },
    {
      "epoch": 0.002477461977840479,
      "grad_norm": 1.8025970458984375,
      "learning_rate": 8.07830737662829e-05,
      "loss": 1.5712,
      "step": 18
    },
    {
      "epoch": 0.0026150987543871723,
      "grad_norm": 1.8106365203857422,
      "learning_rate": 7.795964517353735e-05,
      "loss": 1.6648,
      "step": 19
    },
    {
      "epoch": 0.0027527355309338655,
      "grad_norm": 1.7610256671905518,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.6215,
      "step": 20
    },
    {
      "epoch": 0.002890372307480559,
      "grad_norm": 1.798845648765564,
      "learning_rate": 7.191855733945387e-05,
      "loss": 1.6103,
      "step": 21
    },
    {
      "epoch": 0.003028009084027252,
      "grad_norm": 1.71370267868042,
      "learning_rate": 6.873032967079561e-05,
      "loss": 1.5062,
      "step": 22
    },
    {
      "epoch": 0.0031656458605739454,
      "grad_norm": 1.6641000509262085,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.5473,
      "step": 23
    },
    {
      "epoch": 0.0033032826371206386,
      "grad_norm": 1.7582520246505737,
      "learning_rate": 6.209609477998338e-05,
      "loss": 1.6109,
      "step": 24
    },
    {
      "epoch": 0.003440919413667332,
      "grad_norm": 1.7903997898101807,
      "learning_rate": 5.868240888334653e-05,
      "loss": 1.5827,
      "step": 25
    },
    {
      "epoch": 0.003440919413667332,
      "eval_loss": 1.6997616291046143,
      "eval_runtime": 107.8002,
      "eval_samples_per_second": 113.516,
      "eval_steps_per_second": 14.193,
      "step": 25
    },
    {
      "epoch": 0.003578556190214025,
      "grad_norm": 1.8386026620864868,
      "learning_rate": 5.522642316338268e-05,
      "loss": 1.5792,
      "step": 26
    },
    {
      "epoch": 0.0037161929667607185,
      "grad_norm": 1.7905935049057007,
      "learning_rate": 5.174497483512506e-05,
      "loss": 1.4473,
      "step": 27
    },
    {
      "epoch": 0.0038538297433074117,
      "grad_norm": 1.8156318664550781,
      "learning_rate": 4.825502516487497e-05,
      "loss": 1.5866,
      "step": 28
    },
    {
      "epoch": 0.003991466519854105,
      "grad_norm": 1.9291555881500244,
      "learning_rate": 4.477357683661734e-05,
      "loss": 1.586,
      "step": 29
    },
    {
      "epoch": 0.004129103296400798,
      "grad_norm": 2.010096549987793,
      "learning_rate": 4.131759111665349e-05,
      "loss": 1.8055,
      "step": 30
    },
    {
      "epoch": 0.0042667400729474916,
      "grad_norm": 2.0822410583496094,
      "learning_rate": 3.790390522001662e-05,
      "loss": 1.7194,
      "step": 31
    },
    {
      "epoch": 0.004404376849494185,
      "grad_norm": 1.938077449798584,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.7113,
      "step": 32
    },
    {
      "epoch": 0.004542013626040878,
      "grad_norm": 2.1584365367889404,
      "learning_rate": 3.12696703292044e-05,
      "loss": 1.7429,
      "step": 33
    },
    {
      "epoch": 0.004679650402587571,
      "grad_norm": 1.9372602701187134,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 1.6595,
      "step": 34
    },
    {
      "epoch": 0.004817287179134265,
      "grad_norm": 1.986935019493103,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.6611,
      "step": 35
    },
    {
      "epoch": 0.004954923955680958,
      "grad_norm": 2.1261727809906006,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 1.6404,
      "step": 36
    },
    {
      "epoch": 0.005092560732227651,
      "grad_norm": 2.0191519260406494,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 1.8101,
      "step": 37
    },
    {
      "epoch": 0.0052301975087743445,
      "grad_norm": 2.01820707321167,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 1.6373,
      "step": 38
    },
    {
      "epoch": 0.005367834285321038,
      "grad_norm": 2.1250789165496826,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 1.8378,
      "step": 39
    },
    {
      "epoch": 0.005505471061867731,
      "grad_norm": 2.3143749237060547,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 1.8933,
      "step": 40
    },
    {
      "epoch": 0.005643107838414424,
      "grad_norm": 2.2992570400238037,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.6684,
      "step": 41
    },
    {
      "epoch": 0.005780744614961118,
      "grad_norm": 2.2429161071777344,
      "learning_rate": 7.597595192178702e-06,
      "loss": 1.7135,
      "step": 42
    },
    {
      "epoch": 0.005918381391507811,
      "grad_norm": 2.2423243522644043,
      "learning_rate": 5.852620357053651e-06,
      "loss": 1.7418,
      "step": 43
    },
    {
      "epoch": 0.006056018168054504,
      "grad_norm": 2.3367702960968018,
      "learning_rate": 4.322727117869951e-06,
      "loss": 1.9304,
      "step": 44
    },
    {
      "epoch": 0.0061936549446011974,
      "grad_norm": 2.4302263259887695,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 1.8266,
      "step": 45
    },
    {
      "epoch": 0.006331291721147891,
      "grad_norm": 2.6707515716552734,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 2.0224,
      "step": 46
    },
    {
      "epoch": 0.006468928497694584,
      "grad_norm": 3.0416553020477295,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 2.0159,
      "step": 47
    },
    {
      "epoch": 0.006606565274241277,
      "grad_norm": 3.4270143508911133,
      "learning_rate": 4.865965629214819e-07,
      "loss": 2.2476,
      "step": 48
    },
    {
      "epoch": 0.0067442020507879705,
      "grad_norm": 3.9840445518493652,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 2.3697,
      "step": 49
    },
    {
      "epoch": 0.006881838827334664,
      "grad_norm": 7.1414594650268555,
      "learning_rate": 0.0,
      "loss": 3.1481,
      "step": 50
    },
    {
      "epoch": 0.006881838827334664,
      "eval_loss": 1.6805955171585083,
      "eval_runtime": 107.8153,
      "eval_samples_per_second": 113.5,
      "eval_steps_per_second": 14.191,
      "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": 1941781210988544.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}