Bi-LoRA: Efficient Sharpness-Aware Minimization for Fine-Tuning Large-Scale Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Yuhang, Li, Tao, Huang, Zhehao, Yang, Zuopeng, Huang, Xiaolin
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915943973650432
author Liu, Yuhang
Li, Tao
Huang, Zhehao
Yang, Zuopeng
Huang, Xiaolin
author_facet Liu, Yuhang
Li, Tao
Huang, Zhehao
Yang, Zuopeng
Huang, Xiaolin
contents Fine-tuning large-scale pre-trained models with limited data presents significant challenges for generalization. While Sharpness-Aware Minimization (SAM) has proven effective in improving generalization by seeking flat minima, its substantial extra memory and computation overhead make it impractical for large models. Integrating SAM with parameter-efficient fine-tuning methods like Low-Rank Adaptation (LoRA) is a promising direction. However, we find that directly applying SAM to LoRA parameters limits the sharpness optimization to a restricted subspace, hindering its effectiveness. To address this limitation, we propose Bi-directional Low-Rank Adaptation (Bi-LoRA), which introduces an auxiliary LoRA module to model SAM's adversarial weight perturbations. It decouples SAM's weight perturbations from LoRA optimization: the primary LoRA module adapts to specific tasks via standard gradient descent, while the auxiliary module captures the sharpness of the loss landscape through gradient ascent. Such dual-module design enables Bi-LoRA to capture broader sharpness for achieving flatter minima while remaining memory-efficient. Another important benefit is that the dual design allows for simultaneous optimization and perturbation, eliminating SAM's doubled training costs. Extensive experiments across diverse tasks and architectures demonstrate Bi-LoRA's efficiency and effectiveness in enhancing generalization.
format Preprint
id arxiv_https___arxiv_org_abs_2508_19564
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Bi-LoRA: Efficient Sharpness-Aware Minimization for Fine-Tuning Large-Scale Models
Liu, Yuhang
Li, Tao
Huang, Zhehao
Yang, Zuopeng
Huang, Xiaolin
Machine Learning
Artificial Intelligence
Fine-tuning large-scale pre-trained models with limited data presents significant challenges for generalization. While Sharpness-Aware Minimization (SAM) has proven effective in improving generalization by seeking flat minima, its substantial extra memory and computation overhead make it impractical for large models. Integrating SAM with parameter-efficient fine-tuning methods like Low-Rank Adaptation (LoRA) is a promising direction. However, we find that directly applying SAM to LoRA parameters limits the sharpness optimization to a restricted subspace, hindering its effectiveness. To address this limitation, we propose Bi-directional Low-Rank Adaptation (Bi-LoRA), which introduces an auxiliary LoRA module to model SAM's adversarial weight perturbations. It decouples SAM's weight perturbations from LoRA optimization: the primary LoRA module adapts to specific tasks via standard gradient descent, while the auxiliary module captures the sharpness of the loss landscape through gradient ascent. Such dual-module design enables Bi-LoRA to capture broader sharpness for achieving flatter minima while remaining memory-efficient. Another important benefit is that the dual design allows for simultaneous optimization and perturbation, eliminating SAM's doubled training costs. Extensive experiments across diverse tasks and architectures demonstrate Bi-LoRA's efficiency and effectiveness in enhancing generalization.
title Bi-LoRA: Efficient Sharpness-Aware Minimization for Fine-Tuning Large-Scale Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2508.19564