trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 20,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.25,
      "learning_rate": 0.00011710752518939715,
      "loss": 2.3651,
      "step": 5
    },
    {
      "epoch": 0.5,
      "learning_rate": 6.86e-05,
      "loss": 2.1795,
      "step": 10
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.0092474810602843e-05,
      "loss": 1.9629,
      "step": 15
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 2.1609,
      "step": 20
    }
  ],
  "max_steps": 20,
  "num_train_epochs": 1,
  "total_flos": 20511424512000.0,
  "trial_name": null,
  "trial_params": null
}