Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization

Zhanhui Zhou, Jie Liu, Jing Shao, Xiangyu Yue, Chao Yang, Wanli Ouyang, Yu Qiao


Abstract
A single language model, even when aligned with labelers through reinforcement learning from human feedback (RLHF), may not suit all human preferences. Recent approaches therefore prefer customization, gathering multi-dimensional feedback, and creating distinct reward models for each dimension.Different language models are then optimized for various preferences using multi-objective RLHF (MORLHF) with varying reward weights.However, RL fine-tuning is unstable and resource-heavy, especially with diverse and usually conflicting objectives.In this paper, we present Multi-Objective Direct Preference Optimization (MODPO), an RL-free extension of Direct Preference Optimization (DPO) for multiple alignment objectives.Essentially, MODPO folds language modeling directly into reward modeling, training language models as implicit collective reward models that combine all objectives with specific weights. MODPO theoretically yields the same optimal solutions as MORLHF but is practically more stable and efficient.Empirical results in safety alignment and long-form question answering show that MODPO matches or outperforms existing methods, producing a Pareto front of language models catering to diverse preferences with three times less computational resources compared to MORLHF.Code is available at https://github.com/ZHZisZZ/modpo.
Anthology ID:
2024.findings-acl.630
Volume:
Findings of the Association for Computational Linguistics: ACL 2024
Month:
August
Year:
2024
Address:
Bangkok, Thailand
Editors:
Lun-Wei Ku, Andre Martins, Vivek Srikumar
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
10586–10613
Language:
URL:
https://aclanthology.org/2024.findings-acl.630
DOI:
10.18653/v1/2024.findings-acl.630
Bibkey:
Cite (ACL):
Zhanhui Zhou, Jie Liu, Jing Shao, Xiangyu Yue, Chao Yang, Wanli Ouyang, and Yu Qiao. 2024. Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization. In Findings of the Association for Computational Linguistics: ACL 2024, pages 10586–10613, Bangkok, Thailand. Association for Computational Linguistics.
Cite (Informal):
Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization (Zhou et al., Findings 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.findings-acl.630.pdf