RotMoLE: Enhancing Mixture of Low-Rank Experts through Rotational Gating Mechanism

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sun, Mengyang, Dou, Maochuan, Feng, Tao, Zhang, Dan, Wang, Yihao, Liu, Junpeng, Zhu, Yifan, Tang, Jie
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917531116109824
author Sun, Mengyang
Dou, Maochuan
Feng, Tao
Zhang, Dan
Wang, Yihao
Liu, Junpeng
Zhu, Yifan
Tang, Jie
author_facet Sun, Mengyang
Dou, Maochuan
Feng, Tao
Zhang, Dan
Wang, Yihao
Liu, Junpeng
Zhu, Yifan
Tang, Jie
contents While Large Language Models (LLMs) are commonly fine-tuned to handle domain-specific tasks before being applied to vertical applications, adapting them to complex scenarios with diverse specialized knowledge remains challenging. Meanwhile, Mixture-of-Experts (MoE) architecture has risen as a crucial paradigm for training LLMs, and some recent works have also incorporated MoE into Parameter-Efficient Fine-Tuning (PEFT) to propose the Mixture of Low-rank Experts (MoE-LoRA), to enhance the power of low-rank adapters for learning complicated knowledge. However, conventional gating mechanisms in MoE typically apply only a scalar reweighing to selected experts, thereby limiting their underlying capacity of representation and generalization. Motivated and enabled by the low-rank structures in MoE-LoRA, we propose RotMoLE, a specialized MoE framework for low-rank experts featuring an additional rotation gate. Beyond simple scaling, RotMoLE implements a rotation mechanism for each selected expert, enabling superior expert exploitation and specialization for learning diverse data, especially when expert candidates are limited. Empirical results on complex multi-task and multilingual training scenarios validate our effectiveness.
format Preprint
id arxiv_https___arxiv_org_abs_2605_25565
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle RotMoLE: Enhancing Mixture of Low-Rank Experts through Rotational Gating Mechanism
Sun, Mengyang
Dou, Maochuan
Feng, Tao
Zhang, Dan
Wang, Yihao
Liu, Junpeng
Zhu, Yifan
Tang, Jie
Machine Learning
Computation and Language
While Large Language Models (LLMs) are commonly fine-tuned to handle domain-specific tasks before being applied to vertical applications, adapting them to complex scenarios with diverse specialized knowledge remains challenging. Meanwhile, Mixture-of-Experts (MoE) architecture has risen as a crucial paradigm for training LLMs, and some recent works have also incorporated MoE into Parameter-Efficient Fine-Tuning (PEFT) to propose the Mixture of Low-rank Experts (MoE-LoRA), to enhance the power of low-rank adapters for learning complicated knowledge. However, conventional gating mechanisms in MoE typically apply only a scalar reweighing to selected experts, thereby limiting their underlying capacity of representation and generalization. Motivated and enabled by the low-rank structures in MoE-LoRA, we propose RotMoLE, a specialized MoE framework for low-rank experts featuring an additional rotation gate. Beyond simple scaling, RotMoLE implements a rotation mechanism for each selected expert, enabling superior expert exploitation and specialization for learning diverse data, especially when expert candidates are limited. Empirical results on complex multi-task and multilingual training scenarios validate our effectiveness.
title RotMoLE: Enhancing Mixture of Low-Rank Experts through Rotational Gating Mechanism
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2605.25565