trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 1625,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.54,
      "learning_rate": 3.461538461538462e-05,
      "loss": 0.8601,
      "step": 500
    },
    {
      "epoch": 3.08,
      "learning_rate": 1.923076923076923e-05,
      "loss": 0.0454,
      "step": 1000
    },
    {
      "epoch": 4.62,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.0187,
      "step": 1500
    },
    {
      "epoch": 5.0,
      "step": 1625,
      "total_flos": 28064391364608.0,
      "train_loss": 0.28553352231245777,
      "train_runtime": 111.0889,
      "train_samples_per_second": 58.512,
      "train_steps_per_second": 14.628
    }
  ],
  "max_steps": 1625,
  "num_train_epochs": 5,
  "total_flos": 28064391364608.0,
  "trial_name": null,
  "trial_params": null
}