LOFT: Low-Rank Orthogonal Fine-Tuning via Task-Aware Support Selection

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Lanxin, Mishra, Bamdev, Jawanpuria, Pratik, Lin, Lequan, Shi, Dai, Gao, Junbin, Han, Andi
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917484599181312
author Zhao, Lanxin
Mishra, Bamdev
Jawanpuria, Pratik
Lin, Lequan
Shi, Dai
Gao, Junbin
Han, Andi
author_facet Zhao, Lanxin
Mishra, Bamdev
Jawanpuria, Pratik
Lin, Lequan
Shi, Dai
Gao, Junbin
Han, Andi
contents Orthogonal parameter-efficient fine-tuning (PEFT) adapts pretrained weights through structure-preserving multiplicative transformations, but existing methods often conflate two distinct design choices: the subspace in which adaptation occurs and the transformation applied within that subspace. This paper introduces LOFT, a low-rank orthogonal fine-tuning framework that explicitly separates these two components. By viewing orthogonal adaptation as a multiplicative subspace rotation, LOFT provides a unified formulation that recovers representative orthogonal PEFT methods, including coordinate-, butterfly-, Householder-, and principal-subspace-based variants. More importantly, this perspective exposes support selection as a central design axis rather than a byproduct of a particular parameterization. We develop a first-order analysis showing that useful adaptation supports should be informed by the downstream training signal, motivating practical task-aware support selection strategies. Across language understanding, visual transfer, mathematical reasoning, and multilingual out-of-distribution adaptation, LOFT recovers principal-subspace orthogonal adaptation while gradient-informed supports improve the efficiency-performance trade-off under matched parameter, memory, and compute budgets. These results suggest that principled support selection is an important direction for improving orthogonal PEFT.
format Preprint
id arxiv_https___arxiv_org_abs_2605_11872
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle LOFT: Low-Rank Orthogonal Fine-Tuning via Task-Aware Support Selection
Zhao, Lanxin
Mishra, Bamdev
Jawanpuria, Pratik
Lin, Lequan
Shi, Dai
Gao, Junbin
Han, Andi
Machine Learning
Orthogonal parameter-efficient fine-tuning (PEFT) adapts pretrained weights through structure-preserving multiplicative transformations, but existing methods often conflate two distinct design choices: the subspace in which adaptation occurs and the transformation applied within that subspace. This paper introduces LOFT, a low-rank orthogonal fine-tuning framework that explicitly separates these two components. By viewing orthogonal adaptation as a multiplicative subspace rotation, LOFT provides a unified formulation that recovers representative orthogonal PEFT methods, including coordinate-, butterfly-, Householder-, and principal-subspace-based variants. More importantly, this perspective exposes support selection as a central design axis rather than a byproduct of a particular parameterization. We develop a first-order analysis showing that useful adaptation supports should be informed by the downstream training signal, motivating practical task-aware support selection strategies. Across language understanding, visual transfer, mathematical reasoning, and multilingual out-of-distribution adaptation, LOFT recovers principal-subspace orthogonal adaptation while gradient-informed supports improve the efficiency-performance trade-off under matched parameter, memory, and compute budgets. These results suggest that principled support selection is an important direction for improving orthogonal PEFT.
title LOFT: Low-Rank Orthogonal Fine-Tuning via Task-Aware Support Selection
topic Machine Learning
url https://arxiv.org/abs/2605.11872