Reinforcement Learning from Human Feedback

It's a method of fine-tuning and aligning LLMs which requires active human input