Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > cs > arXiv:2503.11164

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Computer Science > Computation and Language

arXiv:2503.11164 (cs)
[Submitted on 14 Mar 2025]

Title:Towards Extreme Pruning of LLMs with Plug-and-Play Mixed Sparsity

Authors:Chi Xu, Gefei Zhang, Yantong Zhu, Luca Benini, Guosheng Hu, Yawei Li, Zhihong Zhang
View a PDF of the paper titled Towards Extreme Pruning of LLMs with Plug-and-Play Mixed Sparsity, by Chi Xu and 6 other authors
View PDF HTML (experimental)
Abstract:N:M structured pruning is essential for large language models (LLMs) because it can remove less important network weights and reduce the memory and computation requirements. Existing pruning methods mainly focus on designing metrics to measure the importance of network components to guide pruning. Apart from the impact of these metrics, we observe that different layers have different sensitivities over the network performance. Thus, we propose an efficient method based on the trace of Fisher Information Matrix (FIM) to quantitatively measure and verify the different sensitivities across layers. Based on this, we propose Mixed Sparsity Pruning (MSP) which uses a pruning-oriented evolutionary algorithm (EA) to determine the optimal sparsity levels for different layers. To guarantee fast convergence and achieve promising performance, we utilize efficient FIM-inspired layer-wise sensitivity to initialize the population of EA. In addition, our MSP can work as a plug-and-play module, ready to be integrated into existing pruning methods. Extensive experiments on LLaMA and LLaMA-2 on language modeling and zero-shot tasks demonstrate our superior performance. In particular, in extreme pruning ratio (e.g. 75%), our method significantly outperforms existing methods in terms of perplexity (PPL) by orders of magnitude (Figure 1).
Subjects: Computation and Language (cs.CL)
Cite as: arXiv:2503.11164 [cs.CL]
  (or arXiv:2503.11164v1 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2503.11164
arXiv-issued DOI via DataCite

Submission history

From: Chi Xu [view email]
[v1] Fri, 14 Mar 2025 08:05:49 UTC (120 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Towards Extreme Pruning of LLMs with Plug-and-Play Mixed Sparsity, by Chi Xu and 6 other authors
  • View PDF
  • HTML (experimental)
  • TeX Source
  • Other Formats
view license
Current browse context:
cs.CL
< prev   |   next >
new | recent | 2025-03
Change to browse by:
cs

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar
a export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

Bookmark

BibSonomy logo Reddit logo

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
  • Author
  • Venue
  • Institution
  • Topic

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)
  • About
  • Help
  • contact arXivClick here to contact arXiv Contact
  • subscribe to arXiv mailingsClick here to subscribe Subscribe
  • Copyright
  • Privacy Policy
  • Web Accessibility Assistance
  • arXiv Operational Status
    Get status notifications via email or slack