trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.49230769230769234,
  "eval_steps": 500,
  "global_step": 1,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.49,
      "learning_rate": 2e-05,
      "loss": 1.6579,
      "step": 1
    },
    {
      "epoch": 0.49,
      "eval_loss": 1.648226261138916,
      "eval_runtime": 1525.0437,
      "eval_samples_per_second": 15.154,
      "eval_steps_per_second": 0.948,
      "step": 1
    },
    {
      "epoch": 0.49,
      "step": 1,
      "total_flos": 909231154790400.0,
      "train_loss": 2.3372400999069214,
      "train_runtime": 2103.1993,
      "train_samples_per_second": 0.988,
      "train_steps_per_second": 0.001
    }
  ],
  "logging_steps": 5,
  "max_steps": 2,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 909231154790400.0,
  "trial_name": null,
  "trial_params": null
}