Guardado en:
Detalles Bibliográficos
Autores principales: Xu, Boxun, Hwang, Junyoung, Vanna-iampikul, Pruek, Yin, Yuxuan, Lim, Sung Kyu, Li, Peng
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2412.05540
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910732997623808
author Xu, Boxun
Hwang, Junyoung
Vanna-iampikul, Pruek
Yin, Yuxuan
Lim, Sung Kyu
Li, Peng
author_facet Xu, Boxun
Hwang, Junyoung
Vanna-iampikul, Pruek
Yin, Yuxuan
Lim, Sung Kyu
Li, Peng
contents Spiking Neural Networks(SNNs) provide a brain-inspired and event-driven mechanism that is believed to be critical to unlock energy-efficient deep learning. The mixture-of-experts approach mirrors the parallel distributed processing of nervous systems, introducing conditional computation policies and expanding model capacity without scaling up the number of computational operations. Additionally, spiking mixture-of-experts self-attention mechanisms enhance representation capacity, effectively capturing diverse patterns of entities and dependencies between visual or linguistic tokens. However, there is currently a lack of hardware support for highly parallel distributed processing needed by spiking transformers, which embody a brain-inspired computation. This paper introduces the first 3D hardware architecture and design methodology for Mixture-of-Experts and Multi-Head Attention spiking transformers. By leveraging 3D integration with memory-on-logic and logic-on-logic stacking, we explore such brain-inspired accelerators with spatially stackable circuitry, demonstrating significant optimization of energy efficiency and latency compared to conventional 2D CMOS integration.
format Preprint
id arxiv_https___arxiv_org_abs_2412_05540
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Towards 3D Acceleration for low-power Mixture-of-Experts and Multi-Head Attention Spiking Transformers
Xu, Boxun
Hwang, Junyoung
Vanna-iampikul, Pruek
Yin, Yuxuan
Lim, Sung Kyu
Li, Peng
Neural and Evolutionary Computing
Artificial Intelligence
Hardware Architecture
Spiking Neural Networks(SNNs) provide a brain-inspired and event-driven mechanism that is believed to be critical to unlock energy-efficient deep learning. The mixture-of-experts approach mirrors the parallel distributed processing of nervous systems, introducing conditional computation policies and expanding model capacity without scaling up the number of computational operations. Additionally, spiking mixture-of-experts self-attention mechanisms enhance representation capacity, effectively capturing diverse patterns of entities and dependencies between visual or linguistic tokens. However, there is currently a lack of hardware support for highly parallel distributed processing needed by spiking transformers, which embody a brain-inspired computation. This paper introduces the first 3D hardware architecture and design methodology for Mixture-of-Experts and Multi-Head Attention spiking transformers. By leveraging 3D integration with memory-on-logic and logic-on-logic stacking, we explore such brain-inspired accelerators with spatially stackable circuitry, demonstrating significant optimization of energy efficiency and latency compared to conventional 2D CMOS integration.
title Towards 3D Acceleration for low-power Mixture-of-Experts and Multi-Head Attention Spiking Transformers
topic Neural and Evolutionary Computing
Artificial Intelligence
Hardware Architecture
url https://arxiv.org/abs/2412.05540