trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 540,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.26,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 1.8348,
      "step": 500
    },
    {
      "epoch": 10.0,
      "step": 540,
      "total_flos": 1.49327778152448e+17,
      "train_loss": 1.8000366281580042,
      "train_runtime": 259.9745,
      "train_samples_per_second": 32.696,
      "train_steps_per_second": 2.077
    }
  ],
  "max_steps": 540,
  "num_train_epochs": 10,
  "total_flos": 1.49327778152448e+17,
  "trial_name": null,
  "trial_params": null
}