Saved in:
Bibliographic Details
Main Authors: Zhang, Baoquan, Xu, Guangning, Ng, Michael. K.
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.11042
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915340519211008
author Zhang, Baoquan
Xu, Guangning
Ng, Michael. K.
author_facet Zhang, Baoquan
Xu, Guangning
Ng, Michael. K.
contents Pretrained Foundation Models (PFMs) have transformed numerous applications by enabling efficient adaptation to customized tasks. Parameter-Efficient Fine-Tuning (PEFT) has emerged as a resource-efficient alternative to full fine-tuning, especially leveraging reparameterized weights $ΔW$ to adapt models for downstream tasks. However, a critical yet underexplored question remains: can we utilize well-pretrained weights $W_0$ to guide the update of task-specific $ΔW$, avoiding inefficient training it from scratch? To end this, we propose Generative Parameter-Efficient Fine-Tuning (GenFT), a novel method that extracts structured, transferable information from $W_0$ for efficient $ΔW$ training. To extract row and column structure information, GenFT applies row and column transformations to distill essential patterns from $W_0$. A tailored policy further decomposes $ΔW$ into layer-shared and layer-specific components, balancing information reuse and individualized flexibility. GenFT is simple yet effective, achieving superior performance across CV and NLP tasks. Extensive experiments on VTAB-1K, FGVC, and GLUE benchmarks demonstrate that GenFT outperforms state-of-the-art PEFT methods, offering a new perspective for efficient model adaptation.
format Preprint
id arxiv_https___arxiv_org_abs_2506_11042
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle GenFT: A Generative Parameter-Efficient Fine-Tuning Method for Pretrained Foundation Models
Zhang, Baoquan
Xu, Guangning
Ng, Michael. K.
Machine Learning
Pretrained Foundation Models (PFMs) have transformed numerous applications by enabling efficient adaptation to customized tasks. Parameter-Efficient Fine-Tuning (PEFT) has emerged as a resource-efficient alternative to full fine-tuning, especially leveraging reparameterized weights $ΔW$ to adapt models for downstream tasks. However, a critical yet underexplored question remains: can we utilize well-pretrained weights $W_0$ to guide the update of task-specific $ΔW$, avoiding inefficient training it from scratch? To end this, we propose Generative Parameter-Efficient Fine-Tuning (GenFT), a novel method that extracts structured, transferable information from $W_0$ for efficient $ΔW$ training. To extract row and column structure information, GenFT applies row and column transformations to distill essential patterns from $W_0$. A tailored policy further decomposes $ΔW$ into layer-shared and layer-specific components, balancing information reuse and individualized flexibility. GenFT is simple yet effective, achieving superior performance across CV and NLP tasks. Extensive experiments on VTAB-1K, FGVC, and GLUE benchmarks demonstrate that GenFT outperforms state-of-the-art PEFT methods, offering a new perspective for efficient model adaptation.
title GenFT: A Generative Parameter-Efficient Fine-Tuning Method for Pretrained Foundation Models
topic Machine Learning
url https://arxiv.org/abs/2506.11042