Multi-objective Large Language Model Alignment with Hierarchical Experts

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Zhuo, Du, Guodong, Guo, Weiyang, Zhou, Yigeng, Li, Xiucheng, Wang, Wenya, Liu, Fangming, Wang, Yequan, Ye, Deheng, Zhang, Min, Li, Jing
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908381324771328
author Li, Zhuo
Du, Guodong
Guo, Weiyang
Zhou, Yigeng
Li, Xiucheng
Wang, Wenya
Liu, Fangming
Wang, Yequan
Ye, Deheng
Zhang, Min
Li, Jing
author_facet Li, Zhuo
Du, Guodong
Guo, Weiyang
Zhou, Yigeng
Li, Xiucheng
Wang, Wenya
Liu, Fangming
Wang, Yequan
Ye, Deheng
Zhang, Min
Li, Jing
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2505_20925
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Multi-objective Large Language Model Alignment with Hierarchical Experts
Li, Zhuo
Du, Guodong
Guo, Weiyang
Zhou, Yigeng
Li, Xiucheng
Wang, Wenya
Liu, Fangming
Wang, Yequan
Ye, Deheng
Zhang, Min
Li, Jing
Computation and Language
Artificial Intelligence
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.
title Multi-objective Large Language Model Alignment with Hierarchical Experts
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2505.20925