Scholarlink

Pre-training followed by fine-tuning is widely adopted among practitioners. The performance can be improved by “model soups” [46] 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 .

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

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

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Tao Li,
  • Weisen Jiang,
  • Fanghui Liu,
  • Xiaolin Huang,
  • James T. Kwok

摘要

Pre-training followed by fine-tuning is widely adopted among practitioners. The performance can be improved by “model soups” [46] 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 .