With the emergence of reinforcement learning from human feedback, we've been applying old techniques with a new guiding function (🤫 RLHF).