@inproceedings{tao-etal-2023-structured,
title = "Structured Pruning for Efficient Generative Pre-trained Language Models",
author = "Tao, Chaofan and
Hou, Lu and
Bai, Haoli and
Wei, Jiansheng and
Jiang, Xin and
Liu, Qun and
Luo, Ping and
Wong, Ngai",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2023",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-acl.692",
doi = "10.18653/v1/2023.findings-acl.692",
pages = "10880--10895",
abstract = "The increasing sizes of large generative Pre-trained Language Models (PLMs) hinder their deploymentin real-world applications. To obtain efficient PLMs, previous studies mostly focus on pruning the attention heads and feed-forward networks (FFNs) of the Transformer. Nevertheless, we find that in generative PLMs, the hidden dimension shared by many other modules (e.g., embedding layer and layer normalization) contains persistent outliers regardless of the network input. This study comprehensively investigates the structured pruning of generative PLMs with all the above compressible components. To identify redundant network structures, we assign learnable masks over compressible components followed by sparse training. Various sizes of PLMs can be flexibly extracted via different thresholds, and are then task-specifically fine-tuned for further improvement. Extensive experiments on language modeling, summarization and machine translation validate the effectiveness of the proposed method. For example, the pruned BART brings 1.51x/6.96x inference speedup on GPU/CPU with 67{\%} size reduction, and can be further combined with quantization for more than 25$\times$ compression.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tao-etal-2023-structured">
<titleInfo>
<title>Structured Pruning for Efficient Generative Pre-trained Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chaofan</namePart>
<namePart type="family">Tao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Hou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haoli</namePart>
<namePart type="family">Bai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiansheng</namePart>
<namePart type="family">Wei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qun</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ping</namePart>
<namePart type="family">Luo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ngai</namePart>
<namePart type="family">Wong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2023</title>
</titleInfo>
<name type="personal">
<namePart type="given">Anna</namePart>
<namePart type="family">Rogers</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jordan</namePart>
<namePart type="family">Boyd-Graber</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Naoaki</namePart>
<namePart type="family">Okazaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Toronto, Canada</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>The increasing sizes of large generative Pre-trained Language Models (PLMs) hinder their deploymentin real-world applications. To obtain efficient PLMs, previous studies mostly focus on pruning the attention heads and feed-forward networks (FFNs) of the Transformer. Nevertheless, we find that in generative PLMs, the hidden dimension shared by many other modules (e.g., embedding layer and layer normalization) contains persistent outliers regardless of the network input. This study comprehensively investigates the structured pruning of generative PLMs with all the above compressible components. To identify redundant network structures, we assign learnable masks over compressible components followed by sparse training. Various sizes of PLMs can be flexibly extracted via different thresholds, and are then task-specifically fine-tuned for further improvement. Extensive experiments on language modeling, summarization and machine translation validate the effectiveness of the proposed method. For example, the pruned BART brings 1.51x/6.96x inference speedup on GPU/CPU with 67% size reduction, and can be further combined with quantization for more than 25\times compression.</abstract>
<identifier type="citekey">tao-etal-2023-structured</identifier>
<identifier type="doi">10.18653/v1/2023.findings-acl.692</identifier>
<location>
<url>https://aclanthology.org/2023.findings-acl.692</url>
</location>
<part>
<date>2023-07</date>
<extent unit="page">
<start>10880</start>
<end>10895</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Structured Pruning for Efficient Generative Pre-trained Language Models
%A Tao, Chaofan
%A Hou, Lu
%A Bai, Haoli
%A Wei, Jiansheng
%A Jiang, Xin
%A Liu, Qun
%A Luo, Ping
%A Wong, Ngai
%Y Rogers, Anna
%Y Boyd-Graber, Jordan
%Y Okazaki, Naoaki
%S Findings of the Association for Computational Linguistics: ACL 2023
%D 2023
%8 July
%I Association for Computational Linguistics
%C Toronto, Canada
%F tao-etal-2023-structured
%X The increasing sizes of large generative Pre-trained Language Models (PLMs) hinder their deploymentin real-world applications. To obtain efficient PLMs, previous studies mostly focus on pruning the attention heads and feed-forward networks (FFNs) of the Transformer. Nevertheless, we find that in generative PLMs, the hidden dimension shared by many other modules (e.g., embedding layer and layer normalization) contains persistent outliers regardless of the network input. This study comprehensively investigates the structured pruning of generative PLMs with all the above compressible components. To identify redundant network structures, we assign learnable masks over compressible components followed by sparse training. Various sizes of PLMs can be flexibly extracted via different thresholds, and are then task-specifically fine-tuned for further improvement. Extensive experiments on language modeling, summarization and machine translation validate the effectiveness of the proposed method. For example, the pruned BART brings 1.51x/6.96x inference speedup on GPU/CPU with 67% size reduction, and can be further combined with quantization for more than 25\times compression.
%R 10.18653/v1/2023.findings-acl.692
%U https://aclanthology.org/2023.findings-acl.692
%U https://doi.org/10.18653/v1/2023.findings-acl.692
%P 10880-10895
Markdown (Informal)
[Structured Pruning for Efficient Generative Pre-trained Language Models](https://aclanthology.org/2023.findings-acl.692) (Tao et al., Findings 2023)
ACL
- Chaofan Tao, Lu Hou, Haoli Bai, Jiansheng Wei, Xin Jiang, Qun Liu, Ping Luo, and Ngai Wong. 2023. Structured Pruning for Efficient Generative Pre-trained Language Models. In Findings of the Association for Computational Linguistics: ACL 2023, pages 10880–10895, Toronto, Canada. Association for Computational Linguistics.