Localized LoRA: A Structured Low-Rank Approximation for Efficient Fine-Tuning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Barazandeh, Babak, Majumdar, Subhabrata, Rajyaguru, Om, Michailidis, George
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914053325062144
author Barazandeh, Babak
Majumdar, Subhabrata
Rajyaguru, Om
Michailidis, George
author_facet Barazandeh, Babak
Majumdar, Subhabrata
Rajyaguru, Om
Michailidis, George
contents Parameter-efficient fine-tuning (PEFT) methods, such as LoRA, offer compact and effective alternatives to full model fine-tuning by introducing low-rank updates to pre-trained weights. However, most existing approaches rely on global low rank structures, which can overlook spatial patterns spread across the parameter space. In this work, we propose Localized LoRA, a generalized framework that models weight updates as a composition of low-rank matrices applied to structured blocks of the weight matrix. This formulation enables dense, localized updates throughout the parameter space without increasing the total number of trainable parameters. We provide a formal comparison between global, diagonal-local, and fully localized low-rank approximations, and show that our method consistently achieves lower approximation error under matched parameter budgets. Experiments on both synthetic and practical settings demonstrate that Localized LoRA offers a more expressive and adaptable alternative to existing methods, enabling efficient fine-tuning with improved performance.
format Preprint
id arxiv_https___arxiv_org_abs_2506_00236
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Localized LoRA: A Structured Low-Rank Approximation for Efficient Fine-Tuning
Barazandeh, Babak
Majumdar, Subhabrata
Rajyaguru, Om
Michailidis, George
Machine Learning
Artificial Intelligence
Computation and Language
Parameter-efficient fine-tuning (PEFT) methods, such as LoRA, offer compact and effective alternatives to full model fine-tuning by introducing low-rank updates to pre-trained weights. However, most existing approaches rely on global low rank structures, which can overlook spatial patterns spread across the parameter space. In this work, we propose Localized LoRA, a generalized framework that models weight updates as a composition of low-rank matrices applied to structured blocks of the weight matrix. This formulation enables dense, localized updates throughout the parameter space without increasing the total number of trainable parameters. We provide a formal comparison between global, diagonal-local, and fully localized low-rank approximations, and show that our method consistently achieves lower approximation error under matched parameter budgets. Experiments on both synthetic and practical settings demonstrate that Localized LoRA offers a more expressive and adaptable alternative to existing methods, enabling efficient fine-tuning with improved performance.
title Localized LoRA: A Structured Low-Rank Approximation for Efficient Fine-Tuning
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2506.00236