Saved in:
Bibliographic Details
Main Authors: Zhong, Yibo, Zhao, Jinman, Zhou, Yao
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2407.09946
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913864612839424
author Zhong, Yibo
Zhao, Jinman
Zhou, Yao
author_facet Zhong, Yibo
Zhao, Jinman
Zhou, Yao
contents Low-rank adaptation (LoRA) is a widely used parameter-efficient fine-tuning (PEFT) method that learns weight updates $ΔW = AB$ for pretrained weights $W$ through low-rank adapters $A$ and $B$. While LoRA ensures hardware efficiency, its low-rank weight updates limit adaptation performance. In this paper, we propose low-rank interconnected adaptation across layers (Lily), a novel PEFT method that introduces an interconnected framework with locally shared $A$ and globally shared $B$ experts. This structure eliminates redundant per-layer $AB$ pairs, enabling higher-rank $ΔW$ with equal or fewer parameters. To enhance expressiveness, we use data-dependent routers to determine $A$-$B$ interconnections, preventing $B$ experts from converging to the same behavior and improving representational power across domains. Experiments across modalities, architectures, and model sizes demonstrate Lily's superior performance and efficiency. GitHub: https://github.com/yibozhong/lily
format Preprint
id arxiv_https___arxiv_org_abs_2407_09946
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Low-Rank Interconnected Adaptation across Layers
Zhong, Yibo
Zhao, Jinman
Zhou, Yao
Computer Vision and Pattern Recognition
Low-rank adaptation (LoRA) is a widely used parameter-efficient fine-tuning (PEFT) method that learns weight updates $ΔW = AB$ for pretrained weights $W$ through low-rank adapters $A$ and $B$. While LoRA ensures hardware efficiency, its low-rank weight updates limit adaptation performance. In this paper, we propose low-rank interconnected adaptation across layers (Lily), a novel PEFT method that introduces an interconnected framework with locally shared $A$ and globally shared $B$ experts. This structure eliminates redundant per-layer $AB$ pairs, enabling higher-rank $ΔW$ with equal or fewer parameters. To enhance expressiveness, we use data-dependent routers to determine $A$-$B$ interconnections, preventing $B$ experts from converging to the same behavior and improving representational power across domains. Experiments across modalities, architectures, and model sizes demonstrate Lily's superior performance and efficiency. GitHub: https://github.com/yibozhong/lily
title Low-Rank Interconnected Adaptation across Layers
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2407.09946