Enregistré dans:
Détails bibliographiques
Auteurs principaux: Li, Shiwei, Luo, Xiandi, Wang, Haozhao, Tang, Xing, Cui, Ziqiang, Liu, Dugang, Li, Yuhua, He, Xiuqiang, Li, Ruixuan
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2508.06953
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911100485763072
author Li, Shiwei
Luo, Xiandi
Wang, Haozhao
Tang, Xing
Cui, Ziqiang
Liu, Dugang
Li, Yuhua
He, Xiuqiang
Li, Ruixuan
author_facet Li, Shiwei
Luo, Xiandi
Wang, Haozhao
Tang, Xing
Cui, Ziqiang
Liu, Dugang
Li, Yuhua
He, Xiuqiang
Li, Ruixuan
contents Low-rank adaptation (LoRA) is a parameter-efficient fine-tuning (PEFT) method widely used in large language models (LLMs). It approximates the update of a pretrained weight matrix $W\in\mathbb{R}^{m\times n}$ by the product of two low-rank matrices, $BA$, where $A \in\mathbb{R}^{r\times n}$ and $B\in\mathbb{R}^{m\times r} (r\ll\min\{m,n\})$. Increasing the dimension $r$ can raise the rank of LoRA weights (i.e., $BA$), which typically improves fine-tuning performance but also significantly increases the number of trainable parameters. In this paper, we propose Block Diversified Low-Rank Adaptation (BoRA), which improves the rank of LoRA weights with a small number of additional parameters. Specifically, BoRA treats the product $BA$ as a block matrix multiplication, where $A$ and $B$ are partitioned into $b$ blocks along the columns and rows, respectively (i.e., $A=[A_1,\dots,A_b]$ and $B=[B_1,\dots,B_b]^\top$). Consequently, the product $BA$ becomes the concatenation of the block products $B_iA_j$ for $i,j\in[b]$. To enhance the diversity of different block products, BoRA introduces a unique diagonal matrix $Σ_{i,j} \in \mathbb{R}^{r\times r}$ for each block multiplication, resulting in $B_i Σ_{i,j} A_j$. By leveraging these block-wise diagonal matrices, BoRA increases the rank of LoRA weights by a factor of $b$ while only requiring $b^2r$ additional parameters. Extensive experiments across multiple datasets and models demonstrate the superiority of BoRA, and ablation studies further validate its scalability.
format Preprint
id arxiv_https___arxiv_org_abs_2508_06953
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle BoRA: Towards More Expressive Low-Rank Adaptation with Block Diversity
Li, Shiwei
Luo, Xiandi
Wang, Haozhao
Tang, Xing
Cui, Ziqiang
Liu, Dugang
Li, Yuhua
He, Xiuqiang
Li, Ruixuan
Machine Learning
Low-rank adaptation (LoRA) is a parameter-efficient fine-tuning (PEFT) method widely used in large language models (LLMs). It approximates the update of a pretrained weight matrix $W\in\mathbb{R}^{m\times n}$ by the product of two low-rank matrices, $BA$, where $A \in\mathbb{R}^{r\times n}$ and $B\in\mathbb{R}^{m\times r} (r\ll\min\{m,n\})$. Increasing the dimension $r$ can raise the rank of LoRA weights (i.e., $BA$), which typically improves fine-tuning performance but also significantly increases the number of trainable parameters. In this paper, we propose Block Diversified Low-Rank Adaptation (BoRA), which improves the rank of LoRA weights with a small number of additional parameters. Specifically, BoRA treats the product $BA$ as a block matrix multiplication, where $A$ and $B$ are partitioned into $b$ blocks along the columns and rows, respectively (i.e., $A=[A_1,\dots,A_b]$ and $B=[B_1,\dots,B_b]^\top$). Consequently, the product $BA$ becomes the concatenation of the block products $B_iA_j$ for $i,j\in[b]$. To enhance the diversity of different block products, BoRA introduces a unique diagonal matrix $Σ_{i,j} \in \mathbb{R}^{r\times r}$ for each block multiplication, resulting in $B_i Σ_{i,j} A_j$. By leveraging these block-wise diagonal matrices, BoRA increases the rank of LoRA weights by a factor of $b$ while only requiring $b^2r$ additional parameters. Extensive experiments across multiple datasets and models demonstrate the superiority of BoRA, and ablation studies further validate its scalability.
title BoRA: Towards More Expressive Low-Rank Adaptation with Block Diversity
topic Machine Learning
url https://arxiv.org/abs/2508.06953