Saved in:
Bibliographic Details
Main Authors: Park, Junyoung, Kang, Myeonggu, Han, Yunki, Kim, Yanggon, Shin, Jaekang, Kim, Lee-Sup
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2407.15131
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • The attention mechanism in text generation is memory-bounded due to its sequential characteristics. Therefore, off-chip memory accesses should be minimized for faster execution. Although previous methods addressed this by pruning unimportant tokens, they fall short in selectively removing tokens with near-zero attention probabilities in each instance. Our method estimates the probability before the softmax function, effectively removing low probability tokens and achieving an 12.1x pruning ratio without fine-tuning. Additionally, we present a hardware design supporting seamless on-demand off-chip access. Our approach shows 2.6x reduced memory accesses, leading to an average 2.3x speedup and a 2.4x energy efficiency.