{
  "best_global_step": 1500,
  "best_metric": 0.4973679184913635,
  "best_model_checkpoint": "/itn_neural/for_conference/llm/exaone/sft/exaone-3.5-2.4b-itn-sft_v2/checkpoint-1500",
  "epoch": 15.0,
  "eval_steps": 500,
  "global_step": 5985,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 0.7797817493630028,
      "epoch": 1.2532915360501566,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.88700738596578e-05,
      "loss": 0.8579439697265625,
      "mean_token_accuracy": 0.8582268720599876,
      "num_tokens": 752019.0,
      "step": 500
    },
    {
      "epoch": 1.2532915360501566,
      "eval_entropy": 0.4920646266577995,
      "eval_loss": 0.499345988035202,
      "eval_mean_token_accuracy": 0.9077595798936609,
      "eval_num_tokens": 752019.0,
      "eval_runtime": 51.5478,
      "eval_samples_per_second": 5.626,
      "eval_steps_per_second": 1.416,
      "step": 500
    },
    {
      "entropy": 0.41335841036844756,
      "epoch": 2.5065830721003133,
      "grad_norm": 0.1796875,
      "learning_rate": 9.435173984122025e-05,
      "loss": 0.36718594360351564,
      "mean_token_accuracy": 0.9265078916556839,
      "num_tokens": 1503555.0,
      "step": 1000
    },
    {
      "epoch": 2.5065830721003133,
      "eval_entropy": 0.4418756867924782,
      "eval_loss": 0.5064253807067871,
      "eval_mean_token_accuracy": 0.9077344348985855,
      "eval_num_tokens": 1503555.0,
      "eval_runtime": 49.1768,
      "eval_samples_per_second": 5.897,
      "eval_steps_per_second": 1.484,
      "step": 1000
    },
    {
      "entropy": 0.3867817237086151,
      "epoch": 3.7598746081504704,
      "grad_norm": 0.1923828125,
      "learning_rate": 8.669233628725197e-05,
      "loss": 0.34272763061523437,
      "mean_token_accuracy": 0.9299258076590022,
      "num_tokens": 2255834.0,
      "step": 1500
    },
    {
      "epoch": 3.7598746081504704,
      "eval_entropy": 0.4260942515445082,
      "eval_loss": 0.4973679184913635,
      "eval_mean_token_accuracy": 0.9098140972934358,
      "eval_num_tokens": 2255834.0,
      "eval_runtime": 49.6046,
      "eval_samples_per_second": 5.846,
      "eval_steps_per_second": 1.472,
      "step": 1500
    },
    {
      "entropy": 0.36877311900392307,
      "epoch": 5.012539184952978,
      "grad_norm": 0.193359375,
      "learning_rate": 7.643431585575096e-05,
      "loss": 0.3251168518066406,
      "mean_token_accuracy": 0.9323955471809204,
      "num_tokens": 3007086.0,
      "step": 2000
    },
    {
      "epoch": 5.012539184952978,
      "eval_entropy": 0.4132943349341824,
      "eval_loss": 0.5127547979354858,
      "eval_mean_token_accuracy": 0.9077546229101208,
      "eval_num_tokens": 3007086.0,
      "eval_runtime": 49.4722,
      "eval_samples_per_second": 5.862,
      "eval_steps_per_second": 1.476,
      "step": 2000
    },
    {
      "entropy": 0.35612001454549885,
      "epoch": 6.265830721003135,
      "grad_norm": 0.20703125,
      "learning_rate": 6.430416989942607e-05,
      "loss": 0.3105451049804687,
      "mean_token_accuracy": 0.9346451128107599,
      "num_tokens": 3759202.0,
      "step": 2500
    },
    {
      "epoch": 6.265830721003135,
      "eval_entropy": 0.40589027331299976,
      "eval_loss": 0.5173615217208862,
      "eval_mean_token_accuracy": 0.907476933035132,
      "eval_num_tokens": 3759202.0,
      "eval_runtime": 49.1893,
      "eval_samples_per_second": 5.896,
      "eval_steps_per_second": 1.484,
      "step": 2500
    },
    {
      "entropy": 0.3450216325284124,
      "epoch": 7.519122257053292,
      "grad_norm": 0.234375,
      "learning_rate": 5.116097704881937e-05,
      "loss": 0.29882608032226565,
      "mean_token_accuracy": 0.9364423339756922,
      "num_tokens": 4510836.0,
      "step": 3000
    },
    {
      "epoch": 7.519122257053292,
      "eval_entropy": 0.3905191029587837,
      "eval_loss": 0.52081698179245,
      "eval_mean_token_accuracy": 0.9078735692860329,
      "eval_num_tokens": 4510836.0,
      "eval_runtime": 49.1855,
      "eval_samples_per_second": 5.896,
      "eval_steps_per_second": 1.484,
      "step": 3000
    },
    {
      "entropy": 0.34002369012368683,
      "epoch": 8.772413793103448,
      "grad_norm": 0.2333984375,
      "learning_rate": 3.7935561727289265e-05,
      "loss": 0.29143359375,
      "mean_token_accuracy": 0.9378253787144712,
      "num_tokens": 5263183.0,
      "step": 3500
    },
    {
      "epoch": 8.772413793103448,
      "eval_entropy": 0.39191936630092256,
      "eval_loss": 0.5216920375823975,
      "eval_mean_token_accuracy": 0.908023670111617,
      "eval_num_tokens": 5263183.0,
      "eval_runtime": 49.3054,
      "eval_samples_per_second": 5.882,
      "eval_steps_per_second": 1.481,
      "step": 3500
    },
    {
      "entropy": 0.33561183958201557,
      "epoch": 10.025078369905955,
      "grad_norm": 0.244140625,
      "learning_rate": 2.55645715006341e-05,
      "loss": 0.2860752258300781,
      "mean_token_accuracy": 0.9386621022427285,
      "num_tokens": 6014293.0,
      "step": 4000
    },
    {
      "epoch": 10.025078369905955,
      "eval_entropy": 0.3817006227088301,
      "eval_loss": 0.5261662602424622,
      "eval_mean_token_accuracy": 0.908064066547237,
      "eval_num_tokens": 6014293.0,
      "eval_runtime": 49.2415,
      "eval_samples_per_second": 5.889,
      "eval_steps_per_second": 1.482,
      "step": 4000
    },
    {
      "entropy": 0.3327892356332747,
      "epoch": 11.278369905956113,
      "grad_norm": 0.234375,
      "learning_rate": 1.4924142021353787e-05,
      "loss": 0.28212399291992185,
      "mean_token_accuracy": 0.9393911692963772,
      "num_tokens": 6766132.0,
      "step": 4500
    },
    {
      "epoch": 11.278369905956113,
      "eval_entropy": 0.38443714257788986,
      "eval_loss": 0.5294912457466125,
      "eval_mean_token_accuracy": 0.9076788360125398,
      "eval_num_tokens": 6766132.0,
      "eval_runtime": 48.8239,
      "eval_samples_per_second": 5.94,
      "eval_steps_per_second": 1.495,
      "step": 4500
    },
    {
      "entropy": 0.33255799134830766,
      "epoch": 12.53166144200627,
      "grad_norm": 0.25,
      "learning_rate": 6.767847534832694e-06,
      "loss": 0.2817730712890625,
      "mean_token_accuracy": 0.9395757245683503,
      "num_tokens": 7518103.0,
      "step": 5000
    },
    {
      "epoch": 12.53166144200627,
      "eval_entropy": 0.38462539200913415,
      "eval_loss": 0.5295228958129883,
      "eval_mean_token_accuracy": 0.9079492965789691,
      "eval_num_tokens": 7518103.0,
      "eval_runtime": 49.2524,
      "eval_samples_per_second": 5.888,
      "eval_steps_per_second": 1.482,
      "step": 5000
    },
    {
      "entropy": 0.3324927242622786,
      "epoch": 13.784952978056426,
      "grad_norm": 0.2490234375,
      "learning_rate": 1.6733314035667803e-06,
      "loss": 0.28033297729492185,
      "mean_token_accuracy": 0.9397041946425922,
      "num_tokens": 8270092.0,
      "step": 5500
    },
    {
      "epoch": 13.784952978056426,
      "eval_entropy": 0.38496025498599223,
      "eval_loss": 0.5297314524650574,
      "eval_mean_token_accuracy": 0.9079608639625654,
      "eval_num_tokens": 8270092.0,
      "eval_runtime": 49.3,
      "eval_samples_per_second": 5.882,
      "eval_steps_per_second": 1.481,
      "step": 5500
    },
    {
      "epoch": 15.0,
      "eval_entropy": 0.3840067480524925,
      "eval_loss": 0.5302117466926575,
      "eval_mean_token_accuracy": 0.9078679248078229,
      "eval_num_tokens": 8998860.0,
      "eval_runtime": 49.1482,
      "eval_samples_per_second": 5.901,
      "eval_steps_per_second": 1.485,
      "step": 5985
    }
  ],
  "logging_steps": 500,
  "max_steps": 5985,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.2873729639986688e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}