To be clear, it isn’t the large language model that engages in reinforcement learning, it’s the person who’s applying the LLM to their task. That’s all that prompt engineering is.