Multi-objective Large Language Model Alignment with Hierarchical Experts
By: Zhuo Li , Guodong Du , Weiyang Guo and more
Potential Business Impact:
Makes AI understand many different wishes at once.
Aligning large language models (LLMs) to simultaneously satisfy multiple objectives remains a significant challenge, especially given the diverse and often conflicting nature of human preferences. Existing alignment methods struggle to balance trade-offs effectively, often requiring costly retraining or yielding suboptimal results across the Pareto frontier of preferences. In this paper, we introduce \textit{HoE}(Hierarchical Mixture-of-Experts), a \textit{lightweight}, \textit{parameter-efficient}, and \textit{plug-and-play} approach that eliminates the need for model training, while enabling LLMs to adapt across the entire Pareto frontier and accommodate diverse user preferences. In particular, \textit{HoE} consists of three hierarchical components: LoRA Experts, Router Experts and Preference Routing, reaching optimal Pareto frontiers and achieving a trade-off between parameter size, training cost, and performance. We evaluate \textit{HoE} across various tasks on 14 objectives and 200 different preferences among 6 benchmarks, demonstrating superior performance over 15 recent baselines. Code is available in the supplementary materials.
Similar Papers
Breaking the MoE LLM Trilemma: Dynamic Expert Clustering with Structured Compression
Computation and Language
Makes AI smarter, faster, and use less memory.
OrdMoE: Preference Alignment via Hierarchical Expert Group Ranking in Multimodal Mixture-of-Experts LLMs
Machine Learning (CS)
Teaches AI to judge its own answers better.
Hierarchical LoRA MoE for Efficient CTR Model Scaling
Machine Learning (CS)
Helps websites show you ads you'll click.