Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Adaptive Preference Optimization with Uncertainty-aware Utility Anchor

About

Offline preference optimization methods are efficient for large language models (LLMs) alignment. Direct Preference optimization (DPO)-like learning, one of the most popular approaches, stands out for its efficiency in reward modeling. However, these methods typically follow the convention to use Bradley-Terry (BT) reward modeling that faces several critical assumptions, including the requirement for pairwise training data, model distribution shifting, human rationality assumption, etc. To address these limitations, we propose a general framework for offline preference optimization methods, Adaptive Preference Optimization with Utility Anchor (UAPO), which introduces an anchoring function to estimate the uncertainties brought from preference data annotation. Our method enables training even in scenarios where the data is unpaired, significantly enhancing data utilization efficiency. Moreover, the anchor design makes UAPO more robust in the training process. Experimental results demonstrate that UAPO achieves competitive outcomes without the strict dependency on data pairing, paving the way for more flexible and effective preference optimization methods.

Xiaobo Wang, Zixia Jia, Jiaqi Li, Qi Liu, Zilong Zheng• 2025

Related benchmarks

TaskDatasetResultRank
Instruction FollowingArena Hard
Win Rate59.4
263
Reward ModelingRewardBench
Chat Score94.4
216
Instruction FollowingAlpacaEval 2
LC (%)73.5
137
Reward ModelingRewardBench 2
Precise IF Score33
41
Multi-turn dialogueMT-Bench
GPT-4 Score8.9
34
General Language Understanding and ReasoningHuggingFace Open LLM Leaderboard
HellaSwag Accuracy84.22
30
Showing 6 of 6 rows

Other info

Follow for update