trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.125703564727955,
  "global_step": 3000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.38,
      "learning_rate": 9.966216216216216e-08,
      "loss": 6.0802,
      "step": 1000
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.928678678678677e-08,
      "loss": 4.0971,
      "step": 2000
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.891141141141141e-08,
      "loss": 3.2402,
      "step": 3000
    }
  ],
  "max_steps": 266500,
  "num_train_epochs": 100,
  "total_flos": 5.22238550605824e+16,
  "trial_name": null,
  "trial_params": null
}