paper-with-me

홈 › Papers

RandLoRA: Full-rank parameter-efficient fine-tuning of large models

2025-02-03 · Paul Albert, Frederic Z. Zhang, Hemanth Saratchandran, Cristian Rodriguez-Opazo, Anton Van Den Hengel, Ehsan Abbasnejad

Low-Rank Adaptation (LoRA) and its variants have shown impressive results in reducing the number of trainable parameters and memory requirements of large transformer networks while maintaining fine-tuning performance. The low-rank nature of the weight update inherently limits the representation power of fine-tuned models, however, thus potentially compromising performance on complex tasks. This raises a critical question: when a performance gap between LoRA and standard fine-tuning is observed, is it due to the reduced number of trainable parameters or the rank deficiency? This paper aims to answer this question by introducing RandLoRA, a parameter-efficient method that performs full-rank updates using a learned linear combinations of low-rank, non-trainable random matrices. Our method limits the number of trainable parameters by restricting optimization to diagonal scaling matrices applied to the fixed random matrices. This allows us to effectively overcome the low-rank limitations while maintaining parameter and memory efficiency during training. Through extensive experimentation across vision, language, and vision-language benchmarks, we systematically evaluate the limitations of LoRA and existing random basis methods. Our findings reveal that full-rank updates are beneficial across vision and language tasks individually, and even more so for vision-language tasks, where RandLoRA significantly reduces -- and sometimes eliminates -- the performance gap between standard fine-tuning and LoRA, demonstrating its efficacy.

📄 PDF Abstract BibTeX arXiv:2502.00987

Code (0)

등록된 구현이 없습니다.

Tasks

parameter-efficient fine-tuning

Similar Papers 제목 키워드 기반

LoFT: Low-Rank Adaptation That Behaves Like Full Fine-Tuning

2025-05-27 · Nurbek Tastan, Stefanos Laskaridis, Martin Takac, Karthik Nandakumar 외

Large pre-trained models are commonly adapted to downstream tasks using parameter-efficient fine-tuning methods such as Low-Rank Adaptation (LoRA), which injects small trainable low-rank matrices instead of updating all …

parameter-efficient fine-tuning

Parameter-Efficient Fine-Tuning for Medical Text Summarization: A Comparative Study of Lora, Prompt Tuning, and Full Fine-Tuning

2026-03-23 · Ulugbek Shernazarov, Rostislav Svitsov, Bin Shi arxiv

Fine-tuning large language models for domain-specific tasks such as medical text summarization demands substantial computational resources. Parameter-efficient fine-tuning (PEFT) methods offer promising alternatives by u…

parameter-efficient fine-tuningText Summarization

FuRA: Full-Rank Parameter-Efficient Fine-Tuning with Spectral Preconditioning

2026-05-19 · Yequan Zhao, Ruijie Zhang, Liyan Tan, Niall Moran 외 arxiv

Both full fine-tuning (Full FT) and parameter-efficient fine-tuning methods such as LoRA introduce weight updates without accounting for the spectral structure established during pretraining. As a result, noisy gradients…

parameter-efficient fine-tuningReinforcement LearningMathematical Reasoning

LoRA-Pro: Are Low-Rank Adapters Properly Optimized?

2024-07-25 · Zhengbo Wang, Jian Liang, Ran He, Zilei Wang 외

Low-rank adaptation, also known as LoRA, has emerged as a prominent method for parameter-efficient fine-tuning of foundation models. Despite its computational efficiency, LoRA still yields inferior performance compared t…

Code GenerationComputational EfficiencyDialogue Generationimage-classification+4

SDS-LoRA: Overcoming Anisotropic Gradient Scaling in Low-Rank Adaptation

2026-06-15 · Junghun Oh, Sungyong Baik, Kyoung Mu Lee arxiv

Low-Rank Adaptation (LoRA) enables efficient adaptation of large pretrained models to downstream tasks by parameterizing weight updates with low-rank matrices. In this paper, we investigate the limitations of the LoRA pa…