EDoRA: Efficient Weight-Decomposed Low-Rank Adaptation via Singular Value Decomposition

TL;DR

EDoRA introduces a novel parameter-efficient fine-tuning method that decomposes pre-trained weights into components, significantly reducing trainable parameters while maintaining high performance on NLP benchmarks.

Contribution

It proposes a new weight decomposition approach using SVD, enabling efficient fine-tuning with fewer parameters than existing methods like LoRA.

Findings

Achieves up to 30x fewer trainable parameters.

Maintains or surpasses state-of-the-art performance on GLUE.

Demonstrates scalability and efficiency in memory-constrained environments.

Abstract

Parameter-efficient fine-tuning methods, such as LoRA, reduces the number of trainable parameters. However, they often suffer from scalability issues and differences between their learning pattern and full fine-tuning. To overcome these limitations, we propose Efficient Weight-Decomposed Low-Rank Adaptation (EDoRA): a novel PEFT method that decomposes pre-trained weights into magnitude and directional components. By freezing low-rank matrices, initializing them by singular value decomposition, and introducing a small trainable matrix between them, EDoRA achieves substantial reduction in trainable parameters while maintaining learning capacity. Experimental results on the GLUE benchmark demonstrate that EDoRA achieves competitive or superior performance compared to state-of-the-art methods, such as LoRA and DoRA, with up to 30x fewer trainable parameters. This makes EDoRA a highly efficient solution for adapting LLMs to diverse tasks under memory-constrained settings. Code is available at https://github.com/Hamid-Nasiri/EDoRA .