rlhf = reinforcement learning from human feedback

(had to look it up)