Towards Understanding the Influence of Reward Margin on Preference Model Performance

arXiv:2404.04932v1 Announce Type: new
Abstract: Reinforcement Learning from Human Feedback (RLHF) is a widely used framework for the training of language models. However, the process of using RLHF to develop a language model that is well-aligned presents challenges, especially when it comes to optimizing the reward model. Our research has found that existing reward models, when trained using the traditional ranking objective based on human preference data, often struggle to effectively distinguish between responses that are more or less favorable …

abstract arxiv challenges cs.ai cs.cl feedback framework however human human feedback influence language language model language models performance process reinforcement reinforcement learning research reward model rlhf training type understanding