trainer_state.json

{
  "best_metric": 1.2231295108795166,
  "best_model_checkpoint": "out/checkpoint-1",
  "epoch": 0.0007748934521503294,
  "eval_steps": 1,
  "global_step": 1,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 0.8480828404426575,
      "learning_rate": 0.0002,
      "loss": 1.3581,
      "step": 1
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.2231295108795166,
      "eval_runtime": 2491.4731,
      "eval_samples_per_second": 0.912,
      "eval_steps_per_second": 0.228,
      "step": 1
    }
  ],
  "logging_steps": 1,
  "max_steps": 5,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1,
  "total_flos": 350533957189632.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}