Textbook of AI

Abstract. Introduces reinforcement learning from human preferences (the technique underlying RLHF): learning a reward model from pairwise human preference judgements and then optimising a policy against it with a KL penalty toward a reference distribution.