trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9983206544030585,
  "global_step": 774,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.94,
      "learning_rate": 0.00010620155038759689,
      "loss": 1.6514,
      "step": 500
    },
    {
      "epoch": 3.0,
      "step": 774,
      "total_flos": 4.195821253201335e+17,
      "train_loss": 1.602422630448058,
      "train_runtime": 56205.4133,
      "train_samples_per_second": 110.352,
      "train_steps_per_second": 0.014
    }
  ],
  "max_steps": 774,
  "num_train_epochs": 3,
  "total_flos": 4.195821253201335e+17,
  "trial_name": null,
  "trial_params": null
}