Saved in:
Bibliographic Details
Main Authors: Bai, Guangji, Li, Yijiang, Ling, Chen, Kim, Kibaek, Zhao, Liang
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.17946
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910679578968064
author Bai, Guangji
Li, Yijiang
Ling, Chen
Kim, Kibaek
Zhao, Liang
author_facet Bai, Guangji
Li, Yijiang
Ling, Chen
Kim, Kibaek
Zhao, Liang
contents The transformative impact of large language models (LLMs) like LLaMA and GPT on natural language processing is countered by their prohibitive computational demands. Pruning has emerged as a pivotal compression strategy, introducing sparsity to enhance both memory and computational efficiency. Yet, traditional global pruning is impractical for LLMs due to scalability issues, while local pruning, despite its efficiency, leads to suboptimal solutions. Addressing these challenges, we propose SparseLLM, a novel framework that redefines the global pruning process into manageable, coordinated subproblems, allowing for resource-efficient optimization with global optimality. SparseLLM's approach, which conceptualizes LLMs as a chain of modular functions and leverages auxiliary variables for problem decomposition, not only facilitates a pragmatic application on LLMs but also demonstrates significant performance improvements, particularly in high-sparsity regimes where it surpasses current state-of-the-art methods.
format Preprint
id arxiv_https___arxiv_org_abs_2402_17946
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SparseLLM: Towards Global Pruning for Pre-trained Language Models
Bai, Guangji
Li, Yijiang
Ling, Chen
Kim, Kibaek
Zhao, Liang
Computation and Language
The transformative impact of large language models (LLMs) like LLaMA and GPT on natural language processing is countered by their prohibitive computational demands. Pruning has emerged as a pivotal compression strategy, introducing sparsity to enhance both memory and computational efficiency. Yet, traditional global pruning is impractical for LLMs due to scalability issues, while local pruning, despite its efficiency, leads to suboptimal solutions. Addressing these challenges, we propose SparseLLM, a novel framework that redefines the global pruning process into manageable, coordinated subproblems, allowing for resource-efficient optimization with global optimality. SparseLLM's approach, which conceptualizes LLMs as a chain of modular functions and leverages auxiliary variables for problem decomposition, not only facilitates a pragmatic application on LLMs but also demonstrates significant performance improvements, particularly in high-sparsity regimes where it surpasses current state-of-the-art methods.
title SparseLLM: Towards Global Pruning for Pre-trained Language Models
topic Computation and Language
url https://arxiv.org/abs/2402.17946