trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1329,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.13,
      "learning_rate": 3.118886380737397e-05,
      "loss": 2.0465,
      "step": 500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2377727614747931e-05,
      "loss": 1.9227,
      "step": 1000
    }
  ],
  "logging_steps": 500,
  "max_steps": 1329,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1329,
  "total_flos": 2775755381538816.0,
  "trial_name": null,
  "trial_params": null
}