$\rm SP^3$: Enhancing Structured Pruning via PCA Projection

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hu, Yuxuan, Zhang, Jing, Zhao, Zhe, Zhao, Chen, Chen, Xiaodong, Li, Cuiping, Chen, Hong
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910568403697664
author Hu, Yuxuan
Zhang, Jing
Zhao, Zhe
Zhao, Chen
Chen, Xiaodong
Li, Cuiping
Chen, Hong
author_facet Hu, Yuxuan
Zhang, Jing
Zhao, Zhe
Zhao, Chen
Chen, Xiaodong
Li, Cuiping
Chen, Hong
contents Structured pruning is a widely used technique for reducing the size of pre-trained language models (PLMs), but current methods often overlook the potential of compressing the hidden dimension (d) in PLMs, a dimension critical to model size and efficiency. This paper introduces a novel structured pruning approach, Structured Pruning with PCA Projection (SP3), targeting the effective reduction of d by projecting features into a space defined by principal components before masking. Extensive experiments on benchmarks (GLUE and SQuAD) show that SP3 can reduce d by 70%, compress 94% of the BERTbase model, maintain over 96% accuracy, and outperform other methods that compress d by 6% in accuracy at the same compression ratio. SP3 has also proven effective with other models, including OPT and Llama. Our data and code are available at an anonymous repo.
format Preprint
id arxiv_https___arxiv_org_abs_2308_16475
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle $\rm SP^3$: Enhancing Structured Pruning via PCA Projection
Hu, Yuxuan
Zhang, Jing
Zhao, Zhe
Zhao, Chen
Chen, Xiaodong
Li, Cuiping
Chen, Hong
Computation and Language
Artificial Intelligence
I.2.7
Structured pruning is a widely used technique for reducing the size of pre-trained language models (PLMs), but current methods often overlook the potential of compressing the hidden dimension (d) in PLMs, a dimension critical to model size and efficiency. This paper introduces a novel structured pruning approach, Structured Pruning with PCA Projection (SP3), targeting the effective reduction of d by projecting features into a space defined by principal components before masking. Extensive experiments on benchmarks (GLUE and SQuAD) show that SP3 can reduce d by 70%, compress 94% of the BERTbase model, maintain over 96% accuracy, and outperform other methods that compress d by 6% in accuracy at the same compression ratio. SP3 has also proven effective with other models, including OPT and Llama. Our data and code are available at an anonymous repo.
title $\rm SP^3$: Enhancing Structured Pruning via PCA Projection
topic Computation and Language
Artificial Intelligence
I.2.7
url https://arxiv.org/abs/2308.16475