trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_cer": 0.5721080763006914,
      "eval_loss": 4.031211853027344,
      "eval_runtime": 36.4676,
      "eval_samples_per_second": 8.583,
      "eval_steps_per_second": 0.055,
      "eval_wer": 1.1142715805095227,
      "step": 2
    },
    {
      "epoch": 2.0,
      "eval_cer": 0.5721080763006914,
      "eval_loss": 4.031211853027344,
      "eval_runtime": 35.8856,
      "eval_samples_per_second": 8.722,
      "eval_steps_per_second": 0.056,
      "eval_wer": 1.1142715805095227,
      "step": 4
    },
    {
      "epoch": 3.0,
      "eval_cer": 0.5721080763006914,
      "eval_loss": 4.031211853027344,
      "eval_runtime": 36.6017,
      "eval_samples_per_second": 8.552,
      "eval_steps_per_second": 0.055,
      "eval_wer": 1.1142715805095227,
      "step": 6
    },
    {
      "epoch": 4.0,
      "eval_cer": 0.5898102290001471,
      "eval_loss": 3.4312429428100586,
      "eval_runtime": 36.2078,
      "eval_samples_per_second": 8.645,
      "eval_steps_per_second": 0.055,
      "eval_wer": 1.1105614642592134,
      "step": 8
    },
    {
      "epoch": 5.0,
      "grad_norm": 36.06775665283203,
      "learning_rate": 6.666666666666667e-06,
      "loss": 3.8716,
      "step": 10
    },
    {
      "epoch": 5.0,
      "eval_cer": 0.5075761290638945,
      "eval_loss": 2.859005928039551,
      "eval_runtime": 35.6436,
      "eval_samples_per_second": 8.781,
      "eval_steps_per_second": 0.056,
      "eval_wer": 1.008409596834034,
      "step": 10
    },
    {
      "epoch": 5.0,
      "step": 10,
      "total_flos": 2.4906238324113408e+17,
      "train_loss": 3.8715896606445312,
      "train_runtime": 410.1322,
      "train_samples_per_second": 9.204,
      "train_steps_per_second": 0.024
    }
  ],
  "logging_steps": 10,
  "max_steps": 10,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "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": 2.4906238324113408e+17,
  "train_batch_size": 192,
  "trial_name": null,
  "trial_params": null
}