Robust Preference Optimization with Provable Noise Tolerance for LLMs

arXiv:2404.04102v1 Announce Type: new
Abstract: The preference alignment aims to enable large language models (LLMs) to generate responses that conform to human values, which is essential for developing general AI systems. Ranking-based methods -- a promising class of alignment approaches -- learn human preferences from datasets containing response pairs by optimizing the log-likelihood margins between preferred and dis-preferred responses. However, due to the inherent differences in annotators' preferences, ranking labels of comparisons for response pairs are unavoidably noisy. This seriously …

abstract ai systems alignment arxiv class cs.ai cs.cl cs.lg datasets general general ai generate human language language models large language large language models learn llms noise optimization ranking responses robust systems type values