trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 39,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0784313725490196,
      "grad_norm": 1.8153979778289795,
      "learning_rate": 0.00043326296795745655,
      "loss": 2.3772,
      "step": 14
    },
    {
      "epoch": 2.156862745098039,
      "grad_norm": 1.4803255796432495,
      "learning_rate": 0.00019436976651092142,
      "loss": 1.3656,
      "step": 28
    }
  ],
  "logging_steps": 14,
  "max_steps": 45,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 7.344735109099315e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}