trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 16,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.62,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.846,
      "step": 10
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9155555555555556,
      "eval_loss": 0.26624444127082825,
      "eval_runtime": 2.4885,
      "eval_samples_per_second": 90.415,
      "eval_steps_per_second": 3.215,
      "step": 16
    },
    {
      "epoch": 1.0,
      "step": 16,
      "total_flos": 2.04437073802752e+16,
      "train_loss": 0.6555470079183578,
      "train_runtime": 34.238,
      "train_samples_per_second": 59.145,
      "train_steps_per_second": 0.467
    }
  ],
  "max_steps": 16,
  "num_train_epochs": 1,
  "total_flos": 2.04437073802752e+16,
  "trial_name": null,
  "trial_params": null
}