trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "global_step": 780,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 12.82,
      "learning_rate": 1.794871794871795e-05,
      "loss": 1.4127,
      "step": 500
    },
    {
      "epoch": 20.0,
      "step": 780,
      "total_flos": 2017174487040000.0,
      "train_loss": 1.2078121283115484,
      "train_runtime": 699.037,
      "train_samples_per_second": 5.522,
      "train_steps_per_second": 1.116
    }
  ],
  "max_steps": 780,
  "num_train_epochs": 20,
  "total_flos": 2017174487040000.0,
  "trial_name": null,
  "trial_params": null
}