UniForm: A Reuse Attention Mechanism Optimized for Efficient Vision Transformers on Edge Devices

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Yeom, Seul-Ki, Kim, Tae-Ho
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915046894862336
author Yeom, Seul-Ki
Kim, Tae-Ho
author_facet Yeom, Seul-Ki
Kim, Tae-Ho
contents Transformer-based architectures have demonstrated remarkable success across various domains, but their deployment on edge devices remains challenging due to high memory and computational demands. In this paper, we introduce a novel Reuse Attention mechanism, tailored for efficient memory access and computational optimization, enabling seamless operation on resource-constrained platforms without compromising performance. Unlike traditional multi-head attention (MHA), which redundantly computes separate attention matrices for each head, Reuse Attention consolidates these computations into a shared attention matrix, significantly reducing memory overhead and computational complexity. Comprehensive experiments on ImageNet-1K and downstream tasks show that the proposed UniForm models leveraging Reuse Attention achieve state-of-the-art imagenet classification accuracy while outperforming existing attention mechanisms, such as Linear Attention and Flash Attention, in inference speed and memory scalability. Notably, UniForm-l achieves a 76.7% Top-1 accuracy on ImageNet-1K with 21.8ms inference time on edge devices like the Jetson AGX Orin, representing up to a 5x speedup over competing benchmark methods. These results demonstrate the versatility of Reuse Attention across high-performance GPUs and edge platforms, paving the way for broader real-time applications
format Preprint
id arxiv_https___arxiv_org_abs_2412_02344
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle UniForm: A Reuse Attention Mechanism Optimized for Efficient Vision Transformers on Edge Devices
Yeom, Seul-Ki
Kim, Tae-Ho
Computer Vision and Pattern Recognition
Transformer-based architectures have demonstrated remarkable success across various domains, but their deployment on edge devices remains challenging due to high memory and computational demands. In this paper, we introduce a novel Reuse Attention mechanism, tailored for efficient memory access and computational optimization, enabling seamless operation on resource-constrained platforms without compromising performance. Unlike traditional multi-head attention (MHA), which redundantly computes separate attention matrices for each head, Reuse Attention consolidates these computations into a shared attention matrix, significantly reducing memory overhead and computational complexity. Comprehensive experiments on ImageNet-1K and downstream tasks show that the proposed UniForm models leveraging Reuse Attention achieve state-of-the-art imagenet classification accuracy while outperforming existing attention mechanisms, such as Linear Attention and Flash Attention, in inference speed and memory scalability. Notably, UniForm-l achieves a 76.7% Top-1 accuracy on ImageNet-1K with 21.8ms inference time on edge devices like the Jetson AGX Orin, representing up to a 5x speedup over competing benchmark methods. These results demonstrate the versatility of Reuse Attention across high-performance GPUs and edge platforms, paving the way for broader real-time applications
title UniForm: A Reuse Attention Mechanism Optimized for Efficient Vision Transformers on Edge Devices
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.02344