Sparsity May Be All You Need: Sparse Random Parameter Adaptation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Rios, Jesus, Dognin, Pierre, Luss, Ronny, Ramamurthy, Karthikeyan N.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915501710508032
author Rios, Jesus
Dognin, Pierre
Luss, Ronny
Ramamurthy, Karthikeyan N.
author_facet Rios, Jesus
Dognin, Pierre
Luss, Ronny
Ramamurthy, Karthikeyan N.
contents Full fine-tuning of large language models for alignment and task adaptation has become prohibitively expensive as models have grown in size. Parameter-Efficient Fine-Tuning (PEFT) methods aim at significantly reducing the computational and memory resources needed for fine-tuning these models by only training on a small number of parameters instead of all model parameters. Currently, the most popular PEFT method is the Low-Rank Adaptation (LoRA), which freezes the parameters of the model and introduces a small set of trainable parameters in the form of low-rank matrices. We propose simply reducing the number of trainable parameters by randomly selecting a small proportion of the model parameters to train on, while fixing all other parameters, without any additional prior assumptions such as low-rank structures. In this paper, we compare the efficiency and performance of our proposed approach to other PEFT methods as well as full parameter fine-tuning. We find our method to be competitive with LoRA when using a similar number of trainable parameters. Our findings suggest that what truly matters for a PEFT technique to perform well is not necessarily the specific adapter structure, but rather the number of trainable parameters being used.
format Preprint
id arxiv_https___arxiv_org_abs_2502_15975
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Sparsity May Be All You Need: Sparse Random Parameter Adaptation
Rios, Jesus
Dognin, Pierre
Luss, Ronny
Ramamurthy, Karthikeyan N.
Computation and Language
Artificial Intelligence
Full fine-tuning of large language models for alignment and task adaptation has become prohibitively expensive as models have grown in size. Parameter-Efficient Fine-Tuning (PEFT) methods aim at significantly reducing the computational and memory resources needed for fine-tuning these models by only training on a small number of parameters instead of all model parameters. Currently, the most popular PEFT method is the Low-Rank Adaptation (LoRA), which freezes the parameters of the model and introduces a small set of trainable parameters in the form of low-rank matrices. We propose simply reducing the number of trainable parameters by randomly selecting a small proportion of the model parameters to train on, while fixing all other parameters, without any additional prior assumptions such as low-rank structures. In this paper, we compare the efficiency and performance of our proposed approach to other PEFT methods as well as full parameter fine-tuning. We find our method to be competitive with LoRA when using a similar number of trainable parameters. Our findings suggest that what truly matters for a PEFT technique to perform well is not necessarily the specific adapter structure, but rather the number of trainable parameters being used.
title Sparsity May Be All You Need: Sparse Random Parameter Adaptation
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2502.15975