AnyPrefer: An Automatic Framework for Preference Data Synthesis

Abstract

High-quality preference data is essential for aligning foundation models with human values through preference learning. However, manual annotation of such data is often time-consuming and costly. Recent methods adopt a self-rewarding approach, where the target model generates and annotates its own preference data, but this can lead to inaccuracies due to the reward model sharing weights with the target model, amplifying inherent biases. To address these issues, we propose Anyprefer, a framework designed to synthesize high-quality preference data for the target model. Anyprefer frames the data synthesis process as a cooperative two-player Markov Game, where the target model and a judge model collaborate. Here, a series of external tools are introduced to assist the judge model in accurately rewarding the target model’s responses, mitigating biases in the process. We also introduce a feedback mechanism to optimize prompts for both models, enhancing collaboration and improving data quality. The synthesized data is compiled into a new preference dataset, Anyprefer-V1, consisting of 58K high-quality preference pairs. Extensive experiments show that Anyprefer significantly improves model alignment across four applications, covering 21 datasets, achieving average improvements of 18.55 in five natural language generation datasets, 3.66 in nine vision-language understanding datasets, 30.05 in three medical image analysis datasets, and 14.50 in four visuo-motor control tasks.