Saved in:
Bibliographic Details
Main Authors: Hu, Yuezhou, Zhu, Jun, Chen, Jianfei
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.09099
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913626080673792
author Hu, Yuezhou
Zhu, Jun
Chen, Jianfei
author_facet Hu, Yuezhou
Zhu, Jun
Chen, Jianfei
contents Training deep neural networks (DNNs) is costly. Fortunately, Nvidia Ampere and Hopper GPUs can accelerate matrix multiplications twice as fast as a dense equivalent by implementing 2:4 sparsity. However, previous STE-based 2:4 pre-training methods (e.g. STE with hard-thresholding, SR-STE) suffer from optimization difficulties because of discontinuous pruning function. In this study, we comprehensively analyse the bottleneck of traditional N:M sparse training and recognize three drawbacks with discontinuity: incorrect descending direction, inability to predict the amount of descent and sparse mask oscillation. In light of this, we propose S-STE, a simple yet powerful 2:4 training method that contains two parts: to continuously project weights to be 2:4 sparse, and to rescale sparse weights with a per-tensor fixed scaling factor. Besides, we adopt minimum-variance unbiased estimation for activation gradient and FP8 quantization for whole process. Results show that our method surpasses previous 2:4 pre-training recipes and is comparable even with full parameter models. Our toolkit is available at https://github.com/huyz2023/2by4-pretrain.
format Preprint
id arxiv_https___arxiv_org_abs_2409_09099
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle S-STE: Continuous Pruning Function for Efficient 2:4 Sparse Pre-training
Hu, Yuezhou
Zhu, Jun
Chen, Jianfei
Machine Learning
Training deep neural networks (DNNs) is costly. Fortunately, Nvidia Ampere and Hopper GPUs can accelerate matrix multiplications twice as fast as a dense equivalent by implementing 2:4 sparsity. However, previous STE-based 2:4 pre-training methods (e.g. STE with hard-thresholding, SR-STE) suffer from optimization difficulties because of discontinuous pruning function. In this study, we comprehensively analyse the bottleneck of traditional N:M sparse training and recognize three drawbacks with discontinuity: incorrect descending direction, inability to predict the amount of descent and sparse mask oscillation. In light of this, we propose S-STE, a simple yet powerful 2:4 training method that contains two parts: to continuously project weights to be 2:4 sparse, and to rescale sparse weights with a per-tensor fixed scaling factor. Besides, we adopt minimum-variance unbiased estimation for activation gradient and FP8 quantization for whole process. Results show that our method surpasses previous 2:4 pre-training recipes and is comparable even with full parameter models. Our toolkit is available at https://github.com/huyz2023/2by4-pretrain.
title S-STE: Continuous Pruning Function for Efficient 2:4 Sparse Pre-training
topic Machine Learning
url https://arxiv.org/abs/2409.09099