Learning Scalable Model Soup on a Single GPU: An Efficient Subspace Training Strategy

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Li, Tao, Jiang, Weisen, Liu, Fanghui, Huang, Xiaolin, Kwok, James T.
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909265136975872
author Li, Tao
Jiang, Weisen
Liu, Fanghui
Huang, Xiaolin
Kwok, James T.
author_facet Li, Tao
Jiang, Weisen
Liu, Fanghui
Huang, Xiaolin
Kwok, James T.
contents Pre-training followed by fine-tuning is widely adopted among practitioners. The performance can be improved by "model soups"~\cite{wortsman2022model} via exploring various hyperparameter configurations.The Learned-Soup, a variant of model soups, significantly improves the performance but suffers from substantial memory and time costs due to the requirements of (i) having to load all fine-tuned models simultaneously, and (ii) a large computational graph encompassing all fine-tuned models. In this paper, we propose Memory Efficient Hyperplane Learned Soup (MEHL-Soup) to tackle this issue by formulating the learned soup as a hyperplane optimization problem and introducing block coordinate gradient descent to learn the mixing coefficients. At each iteration, MEHL-Soup only needs to load a few fine-tuned models and build a computational graph with one combined model. We further extend MEHL-Soup to MEHL-Soup+ in a layer-wise manner. Experimental results on various ViT models and data sets show that MEHL-Soup(+) outperforms Learned-Soup(+) in terms of test accuracy, and also reduces memory usage by more than $13\times$. Moreover, MEHL-Soup(+) can be run on a single GPU and achieves $9\times$ speed up in soup construction compared with the Learned-Soup. The code is released at https://github.com/nblt/MEHL-Soup.
format Preprint
id arxiv_https___arxiv_org_abs_2407_03641
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Learning Scalable Model Soup on a Single GPU: An Efficient Subspace Training Strategy
Li, Tao
Jiang, Weisen
Liu, Fanghui
Huang, Xiaolin
Kwok, James T.
Machine Learning
Pre-training followed by fine-tuning is widely adopted among practitioners. The performance can be improved by "model soups"~\cite{wortsman2022model} via exploring various hyperparameter configurations.The Learned-Soup, a variant of model soups, significantly improves the performance but suffers from substantial memory and time costs due to the requirements of (i) having to load all fine-tuned models simultaneously, and (ii) a large computational graph encompassing all fine-tuned models. In this paper, we propose Memory Efficient Hyperplane Learned Soup (MEHL-Soup) to tackle this issue by formulating the learned soup as a hyperplane optimization problem and introducing block coordinate gradient descent to learn the mixing coefficients. At each iteration, MEHL-Soup only needs to load a few fine-tuned models and build a computational graph with one combined model. We further extend MEHL-Soup to MEHL-Soup+ in a layer-wise manner. Experimental results on various ViT models and data sets show that MEHL-Soup(+) outperforms Learned-Soup(+) in terms of test accuracy, and also reduces memory usage by more than $13\times$. Moreover, MEHL-Soup(+) can be run on a single GPU and achieves $9\times$ speed up in soup construction compared with the Learned-Soup. The code is released at https://github.com/nblt/MEHL-Soup.
title Learning Scalable Model Soup on a Single GPU: An Efficient Subspace Training Strategy
topic Machine Learning
url https://arxiv.org/abs/2407.03641