A Survey of Reinforcement Learning from Human Feedback

arXiv:2312.14925v2 Announce Type: replace
Abstract: Reinforcement learning from human feedback (RLHF) is a variant of reinforcement learning (RL) that learns from human feedback instead of relying on an engineered reward function. Building on prior work on the related setting of preference-based reinforcement learning (PbRL), it stands at the intersection of artificial intelligence and human-computer interaction. This positioning offers a promising avenue to enhance the performance and adaptability of intelligent systems while also improving the alignment of their objectives with human …

abstract artificial artificial intelligence arxiv building cs.lg feedback function human human feedback intelligence intersection prior reinforcement reinforcement learning rlhf survey type work