trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 200,
  "global_step": 363,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.65,
      "eval_accuracy": 0.7214611872146118,
      "eval_loss": 0.8989303708076477,
      "eval_macro_f1": 0.5005038448277949,
      "eval_micro_f1": 0.7214611872146118,
      "eval_runtime": 1.6443,
      "eval_samples_per_second": 266.38,
      "eval_steps_per_second": 2.433,
      "step": 200
    },
    {
      "epoch": 3.0,
      "step": 363,
      "total_flos": 2222916882315228.0,
      "train_loss": 0.9607851643207644,
      "train_runtime": 121.7053,
      "train_samples_per_second": 95.172,
      "train_steps_per_second": 2.983
    }
  ],
  "logging_steps": 500,
  "max_steps": 363,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2222916882315228.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}