run-2/checkpoint-1006/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 1006,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.99,
      "learning_rate": 1.113638924561454e-06,
      "loss": 0.9618,
      "step": 500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.3205204639068626e-08,
      "loss": 0.8244,
      "step": 1000
    }
  ],
  "logging_steps": 500,
  "max_steps": 1006,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 1059559383550464.0,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 2.2140726444838396e-06,
    "per_device_train_batch_size": 32
  }
}