COMCAT: Towards Efficient Compression and Customization of Attention-Based Vision Models

TL;DR

This paper introduces a novel, efficient compression method for attention-based vision models like ViT, outperforming pruning techniques and enabling faster training and reduced storage for text-to-image models.

Contribution

The paper proposes a new ViT compression approach based on multi-head attention insights, improving accuracy and efficiency over existing pruning methods.

Findings

Achieves higher top-1 accuracy on ImageNet with fewer parameters.

Enables up to 2.6x faster training of diffusion models.

Reduces storage costs by up to 1927.5 times.

Abstract

Attention-based vision models, such as Vision Transformer (ViT) and its variants, have shown promising performance in various computer vision tasks. However, these emerging architectures suffer from large model sizes and high computational costs, calling for efficient model compression solutions. To date, pruning ViTs has been well studied, while other compression strategies that have been widely applied in CNN compression, e.g., model factorization, is little explored in the context of ViT compression. This paper explores an efficient method for compressing vision transformers to enrich the toolset for obtaining compact attention-based vision models. Based on the new insight on the multi-head attention layer, we develop a highly efficient ViT compression solution, which outperforms the state-of-the-art pruning methods. For compressing DeiT-small and DeiT-base models on ImageNet, our proposed approach can achieve 0.45% and 0.76% higher top-1 accuracy even with fewer parameters. Our finding can also be applied to improve the customization efficiency of text-to-image diffusion models, with much faster training (up to $2.6\times$ speedup) and lower extra storage cost (up to $1927.5\times$ reduction) than the existing works.