trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 1072,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.47,
      "learning_rate": 1.6007462686567165e-05,
      "loss": 2.3884,
      "step": 500
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.0149253731343284e-06,
      "loss": 1.3984,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "step": 1072,
      "total_flos": 2520207218718720.0,
      "train_loss": 1.8482726438721615,
      "train_runtime": 561.1453,
      "train_samples_per_second": 22.917,
      "train_steps_per_second": 1.91
    }
  ],
  "max_steps": 1072,
  "num_train_epochs": 1,
  "total_flos": 2520207218718720.0,
  "trial_name": null,
  "trial_params": null
}