Sequential Preference Optimization: Multi-Dimensional Preference Alignment with Implicit Reward Modeling

Human preference alignment is critical in building powerful and reliable large language models (LLMs). However, current methods either ignore the multi-dimensionality of human preferences (e.g. helpfulness and harmlessness) or struggle with the complexity of managing multiple reward models. To address these issues, we propose Sequential Preference Optimization (SPO), a method that sequentially fine-tunes LLMs to align with multiple dimensions of human preferences. SPO avoids explicit reward modeling, directly optimizing the models to align with nuanced human preferences. We theoretically derive closed-form optimal SPO policy and loss function. Gradient analysis is conducted to show how SPO manages to fine-tune the LLMs while maintaining alignment on previously optimized dimensions. Empirical results on LLMs of different size and multiple evaluation datasets demonstrate that SPO successfully aligns LLMs across multiple dimensions of human preferences and significantly outperforms the baselines.

Authors

• Xingzhou Lou Institute of Automation, Chinese academy of science School of Artificial Intelligence, University of Chinese Academy of Sciences
• Junge Zhang Institute of Automation, Chinese academy of science School of Artificial Intelligence, University of Chinese Academy of Sciences
• Jian Xie
• Lifeng Liu Beijing Baichuan Intelligent Technology Co.
• Dong Yan Beijing Baichuan Intelligent Technology Co.
• Kaiqi Huang Institute of Automation, Chinese academy of science School of Artificial Intelligence, University of Chinese Academy of Sciences

Keywords

No keywords are indexed for this paper.