trainer_state.json

{
  "best_metric": 0.04020816832780838,
  "best_model_checkpoint": "BioM-ELECTRA-Large-SQuAD2-XEF/checkpoint-201",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 201,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_loss": 0.3198401629924774,
      "eval_runtime": 12.6078,
      "eval_samples_per_second": 42.275,
      "eval_steps_per_second": 5.314,
      "step": 67
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.07152208685874939,
      "eval_runtime": 12.6377,
      "eval_samples_per_second": 42.175,
      "eval_steps_per_second": 5.302,
      "step": 134
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.04020816832780838,
      "eval_runtime": 12.7124,
      "eval_samples_per_second": 41.928,
      "eval_steps_per_second": 5.27,
      "step": 201
    }
  ],
  "logging_steps": 500,
  "max_steps": 201,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1113751860309504.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}