Part of International Conference on Representation Learning 2025 (ICLR 2025) Conference
Yuheng Zhang, Dian Yu, Baolin Peng, Linfeng Song, Ye Tian, Mingyue Huo, Nan Jiang, Haitao Mi, Dong Yu
Reinforcement Learning with Human Feedback (RLHF) has achieved great successin aligning large language models (LLMs) with human preferences. PrevalentRLHF approaches are reward-based, following the Bradley-Terry (BT) model assumption, which may not fully capture the complexity of human preferences. Inthis paper, we explore RLHF under a general preference framework and approachit from a game-theoretic perspective. Specifically, we formulate the problem asa two-player game and propose a novel online algorithm, iterative Nash policyoptimization (INPO). The key idea is to let the policy play against itself via no-regret learning, thereby approximating the Nash policy. Unlike previous methods,INPO bypasses the need for estimating the expected win rate for individual responses, which typically incurs high computational or annotation costs. Instead,we introduce a new loss objective that is directly minimized over a preferencedataset. We provide theoretical analysis for our approach and demonstrate itseffectiveness through experiments on various representative benchmarks. With anLLaMA-3-8B-based SFT model, INPO achieves a 42.6% length-controlled winrate on AlpacaEval 2.0 and a 37.8% win rate on Arena-Hard, showing substantialimprovement over the state-of-the-art online RLHF algorithms.