Mingye Zhu


2024

pdf bib
FlipGuard: Defending Preference Alignment against Update Regression with Constrained Optimization
Mingye Zhu | Yi Liu | Quan Wang | Junbo Guo | Zhendong Mao
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing

pdf bib
LIRE: listwise reward enhancement for preference alignment
Mingye Zhu | Yi Liu | Lei Zhang | Junbo Guo | Zhendong Mao
Findings of the Association for Computational Linguistics: ACL 2024

Recently, tremendous strides have been made to align the generation of Large Language Models (LLMs) with human values to mitigate toxic or unhelpful content. Leveraging Reinforcement Learning from Human Feedback (RLHF) proves effective and is widely adopted by researchers. However, implementing RLHF is complex, and its sensitivity to hyperparameters renders achieving stable performance and scalability challenging. Furthermore, prevailing approaches to preference alignment primarily concentrate on pairwise comparisons, with limited exploration into multi-response scenarios, thereby overlooking the potential richness within the candidate pool. For the above reasons, we propose a new approach: Listwise Reward Enhancement for Preference Alignment (LIRE), a gradient-based reward optimization approach that incorporates the offline rewards of multiple responses into a streamlined listwise framework, thus eliminating the need for online sampling during training. LIRE is straightforward to implement, requiring minimal parameter tuning, and seamlessly aligns with the pairwise paradigm while naturally extending to multi-response scenarios. Moreover, we introduce a self-enhancement algorithm aimed at iteratively refining the reward during training. Our experiments demonstrate that LIRE consistently outperforms existing methods across several benchmarks on dialogue and summarization tasks, with good transferability to out-of-distribution data, assessed using proxy reward models and human annotators.