trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 480,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 3.197399616241455,
      "learning_rate": 4e-05,
      "loss": 0.2624,
      "step": 96
    },
    {
      "epoch": 1.0,
      "eval_LOCATION_f1": 0.8585365853658538,
      "eval_LOCATION_number": 93,
      "eval_LOCATION_precision": 0.7857142857142857,
      "eval_LOCATION_recall": 0.946236559139785,
      "eval_ORGANIZATION_f1": 0.851963746223565,
      "eval_ORGANIZATION_number": 166,
      "eval_ORGANIZATION_precision": 0.8545454545454545,
      "eval_ORGANIZATION_recall": 0.8493975903614458,
      "eval_PERSON_f1": 0.9823321554770318,
      "eval_PERSON_number": 142,
      "eval_PERSON_precision": 0.9858156028368794,
      "eval_PERSON_recall": 0.9788732394366197,
      "eval_loss": 0.06350322812795639,
      "eval_overall_accuracy": 0.980246913580247,
      "eval_overall_f1": 0.8986568986568987,
      "eval_overall_precision": 0.8803827751196173,
      "eval_overall_recall": 0.9177057356608479,
      "eval_runtime": 0.6936,
      "eval_samples_per_second": 245.099,
      "eval_steps_per_second": 4.325,
      "step": 96
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.8208107948303223,
      "learning_rate": 3e-05,
      "loss": 0.054,
      "step": 192
    },
    {
      "epoch": 2.0,
      "eval_LOCATION_f1": 0.89,
      "eval_LOCATION_number": 93,
      "eval_LOCATION_precision": 0.8317757009345794,
      "eval_LOCATION_recall": 0.956989247311828,
      "eval_ORGANIZATION_f1": 0.8830409356725145,
      "eval_ORGANIZATION_number": 166,
      "eval_ORGANIZATION_precision": 0.8579545454545454,
      "eval_ORGANIZATION_recall": 0.9096385542168675,
      "eval_PERSON_f1": 0.9752650176678446,
      "eval_PERSON_number": 142,
      "eval_PERSON_precision": 0.9787234042553191,
      "eval_PERSON_recall": 0.971830985915493,
      "eval_loss": 0.05302194878458977,
      "eval_overall_accuracy": 0.9840877914951989,
      "eval_overall_f1": 0.9163636363636364,
      "eval_overall_precision": 0.8915094339622641,
      "eval_overall_recall": 0.942643391521197,
      "eval_runtime": 0.8031,
      "eval_samples_per_second": 211.67,
      "eval_steps_per_second": 3.735,
      "step": 192
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8113964796066284,
      "learning_rate": 2e-05,
      "loss": 0.0268,
      "step": 288
    },
    {
      "epoch": 3.0,
      "eval_LOCATION_f1": 0.8910891089108912,
      "eval_LOCATION_number": 93,
      "eval_LOCATION_precision": 0.8256880733944955,
      "eval_LOCATION_recall": 0.967741935483871,
      "eval_ORGANIZATION_f1": 0.8922155688622753,
      "eval_ORGANIZATION_number": 166,
      "eval_ORGANIZATION_precision": 0.8869047619047619,
      "eval_ORGANIZATION_recall": 0.8975903614457831,
      "eval_PERSON_f1": 0.9787234042553192,
      "eval_PERSON_number": 142,
      "eval_PERSON_precision": 0.9857142857142858,
      "eval_PERSON_recall": 0.971830985915493,
      "eval_loss": 0.06729695945978165,
      "eval_overall_accuracy": 0.9832647462277092,
      "eval_overall_f1": 0.9217603911980441,
      "eval_overall_precision": 0.9040767386091128,
      "eval_overall_recall": 0.940149625935162,
      "eval_runtime": 0.7494,
      "eval_samples_per_second": 226.858,
      "eval_steps_per_second": 4.003,
      "step": 288
    },
    {
      "epoch": 4.0,
      "grad_norm": 5.316173553466797,
      "learning_rate": 1e-05,
      "loss": 0.0159,
      "step": 384
    },
    {
      "epoch": 4.0,
      "eval_LOCATION_f1": 0.9312169312169313,
      "eval_LOCATION_number": 93,
      "eval_LOCATION_precision": 0.9166666666666666,
      "eval_LOCATION_recall": 0.946236559139785,
      "eval_ORGANIZATION_f1": 0.8973607038123167,
      "eval_ORGANIZATION_number": 166,
      "eval_ORGANIZATION_precision": 0.8742857142857143,
      "eval_ORGANIZATION_recall": 0.9216867469879518,
      "eval_PERSON_f1": 0.9716312056737589,
      "eval_PERSON_number": 142,
      "eval_PERSON_precision": 0.9785714285714285,
      "eval_PERSON_recall": 0.9647887323943662,
      "eval_loss": 0.054620783776044846,
      "eval_overall_accuracy": 0.9868312757201646,
      "eval_overall_f1": 0.9310344827586207,
      "eval_overall_precision": 0.9197080291970803,
      "eval_overall_recall": 0.942643391521197,
      "eval_runtime": 0.6122,
      "eval_samples_per_second": 277.708,
      "eval_steps_per_second": 4.901,
      "step": 384
    },
    {
      "epoch": 5.0,
      "grad_norm": 6.452114105224609,
      "learning_rate": 0.0,
      "loss": 0.0108,
      "step": 480
    },
    {
      "epoch": 5.0,
      "eval_LOCATION_f1": 0.9175257731958764,
      "eval_LOCATION_number": 93,
      "eval_LOCATION_precision": 0.8811881188118812,
      "eval_LOCATION_recall": 0.956989247311828,
      "eval_ORGANIZATION_f1": 0.8990825688073394,
      "eval_ORGANIZATION_number": 166,
      "eval_ORGANIZATION_precision": 0.9130434782608695,
      "eval_ORGANIZATION_recall": 0.8855421686746988,
      "eval_PERSON_f1": 0.9716312056737589,
      "eval_PERSON_number": 142,
      "eval_PERSON_precision": 0.9785714285714285,
      "eval_PERSON_recall": 0.9647887323943662,
      "eval_loss": 0.057080645114183426,
      "eval_overall_accuracy": 0.9857338820301783,
      "eval_overall_f1": 0.9290161892901619,
      "eval_overall_precision": 0.927860696517413,
      "eval_overall_recall": 0.9301745635910225,
      "eval_runtime": 0.6378,
      "eval_samples_per_second": 266.541,
      "eval_steps_per_second": 4.704,
      "step": 480
    },
    {
      "epoch": 5.0,
      "step": 480,
      "total_flos": 189589573867848.0,
      "train_loss": 0.07397198428710301,
      "train_runtime": 129.3998,
      "train_samples_per_second": 59.158,
      "train_steps_per_second": 3.709
    }
  ],
  "logging_steps": 500,
  "max_steps": 480,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 189589573867848.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}