phase2_report.json

{
  "run_dir": "/kaggle/working/python-llm-grpo-new-techniques/training/output/phase2-codellama-7b-lora-kaggle-1h",
  "trainer_state_path": "/kaggle/working/python-llm-grpo-new-techniques/training/output/phase2-codellama-7b-lora-kaggle-1h/checkpoint-43/trainer_state.json",
  "metrics": {
    "global_step": 43,
    "best_metric": null,
    "last_train_loss": 0.7746126174926757,
    "best_eval_loss": null,
    "last_eval_loss": null,
    "last_learning_rate": 0.00015600000000000002,
    "last_step_logged": 40,
    "num_train_loss_logs": 4,
    "num_eval_loss_logs": 0
  }
}