trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 58.8235294117647,
  "global_step": 3000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.8,
      "learning_rate": 9.25925925925926e-06,
      "loss": 0.5753,
      "step": 500
    },
    {
      "epoch": 19.61,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.2734,
      "step": 1000
    },
    {
      "epoch": 29.41,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.1562,
      "step": 1500
    },
    {
      "epoch": 39.22,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.1016,
      "step": 2000
    },
    {
      "epoch": 49.02,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.071,
      "step": 2500
    },
    {
      "epoch": 58.82,
      "learning_rate": 0.0,
      "loss": 0.0559,
      "step": 3000
    }
  ],
  "max_steps": 3000,
  "num_train_epochs": 59,
  "total_flos": 3267032656318464.0,
  "trial_name": null,
  "trial_params": null
}