Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Bi-directional Masks for Efficient N:M Sparse Training

About

We focus on addressing the dense backward propagation issue for training efficiency of N:M fine-grained sparsity that preserves at most N out of M consecutive weights and achieves practical speedups supported by the N:M sparse tensor core. Therefore, we present a novel method of Bi-directional Masks (Bi-Mask) with its two central innovations in: 1) Separate sparse masks in the two directions of forward and backward propagation to obtain training acceleration. It disentangles the forward and backward weight sparsity and overcomes the very dense gradient computation. 2) An efficient weight row permutation method to maintain performance. It picks up the permutation candidate with the most eligible N:M weight blocks in the backward to minimize the gradient gap between traditional uni-directional masks and our bi-directional masks. Compared with existing uni-directional scenario that applies a transposable mask and enables backward acceleration, our Bi-Mask is experimentally demonstrated to be more superior in performance. Also, our Bi-Mask performs on par with or even better than methods that fail to achieve backward acceleration. Project of this paper is available at \url{https://github.com/zyxxmu/Bi-Mask}.

Yuxin Zhang, Yiting Luo, Mingbao Lin, Yunshan Zhong, Jingjing Xie, Fei Chao, Rongrong Ji• 2023

Related benchmarks

TaskDatasetResultRank
Image ClassificationImageNet 1k (test)
Top-1 Accuracy77.6
798
Machine TranslationWMT En-De 2014 (test)
BLEU25.84
379
Question AnsweringSQuAD
F182.4
127
Language ModelingGPT-2 Pre-training (val)
Validation Loss2.69
21
Machine TranslationWMT En-De 14 (val)
BLEU26.08
20
Showing 5 of 5 rows

Other info

Follow for update