Implicit Modeling for Transferability Estimation of Vision Foundation Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zheng, Yaoyan, Wang, Huiqun, Zhou, Nan, Huang, Di
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912672239321088
author Zheng, Yaoyan
Wang, Huiqun
Zhou, Nan
Huang, Di
author_facet Zheng, Yaoyan
Wang, Huiqun
Zhou, Nan
Huang, Di
contents Transferability estimation identifies the best pre-trained models for downstream tasks without incurring the high computational cost of full fine-tuning. This capability facilitates deployment and advances the pre-training and fine-tuning paradigm. However, existing methods often struggle to accurately assess transferability for emerging pre-trained models with diverse architectures, training strategies, and task alignments. In this work, we propose Implicit Transferability Modeling (ITM), a novel framework that implicitly models each model's intrinsic transferability, coupled with a Divide-and-Conquer Variational Approximation (DVA) strategy to efficiently approximate embedding space evolution. This design enables generalization across a broader range of models and downstream tasks. Extensive experiments on a comprehensive benchmark--spanning extensive training regimes and a wider variety of model types--demonstrate that ITM consistently outperforms existing methods in terms of stability, effectiveness, and efficiency.
format Preprint
id arxiv_https___arxiv_org_abs_2510_23145
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Implicit Modeling for Transferability Estimation of Vision Foundation Models
Zheng, Yaoyan
Wang, Huiqun
Zhou, Nan
Huang, Di
Computer Vision and Pattern Recognition
Transferability estimation identifies the best pre-trained models for downstream tasks without incurring the high computational cost of full fine-tuning. This capability facilitates deployment and advances the pre-training and fine-tuning paradigm. However, existing methods often struggle to accurately assess transferability for emerging pre-trained models with diverse architectures, training strategies, and task alignments. In this work, we propose Implicit Transferability Modeling (ITM), a novel framework that implicitly models each model's intrinsic transferability, coupled with a Divide-and-Conquer Variational Approximation (DVA) strategy to efficiently approximate embedding space evolution. This design enables generalization across a broader range of models and downstream tasks. Extensive experiments on a comprehensive benchmark--spanning extensive training regimes and a wider variety of model types--demonstrate that ITM consistently outperforms existing methods in terms of stability, effectiveness, and efficiency.
title Implicit Modeling for Transferability Estimation of Vision Foundation Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.23145