Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > cs > arXiv:2102.08124

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Computer Science > Artificial Intelligence

arXiv:2102.08124 (cs)
[Submitted on 16 Feb 2021 (v1), last revised 20 Oct 2021 (this version, v2)]

Title:Accelerated Sparse Neural Training: A Provable and Efficient Method to Find N:M Transposable Masks

Authors:Itay Hubara, Brian Chmiel, Moshe Island, Ron Banner, Seffi Naor, Daniel Soudry
View a PDF of the paper titled Accelerated Sparse Neural Training: A Provable and Efficient Method to Find N:M Transposable Masks, by Itay Hubara and 5 other authors
View PDF
Abstract:Unstructured pruning reduces the memory footprint in deep neural networks (DNNs). Recently, researchers proposed different types of structural pruning intending to reduce also the computation complexity. In this work, we first suggest a new measure called mask-diversity which correlates with the expected accuracy of the different types of structural pruning. We focus on the recently suggested N:M fine-grained block sparsity mask, in which for each block of M weights, we have at least N zeros. While N:M fine-grained block sparsity allows acceleration in actual modern hardware, it can be used only to accelerate the inference phase. In order to allow for similar accelerations in the training phase, we suggest a novel transposable fine-grained sparsity mask, where the same mask can be used for both forward and backward passes. Our transposable mask guarantees that both the weight matrix and its transpose follow the same sparsity pattern; thus, the matrix multiplication required for passing the error backward can also be accelerated. We formulate the problem of finding the optimal transposable-mask as a minimum-cost flow problem. Additionally, to speed up the minimum-cost flow computation, we also introduce a fast linear-time approximation that can be used when the masks dynamically change during training. Our experiments suggest a 2x speed-up in the matrix multiplications with no accuracy degradation over vision and language models. Finally, to solve the problem of switching between different structure constraints, we suggest a method to convert a pre-trained model with unstructured sparsity to an N:M fine-grained block sparsity model with little to no training. A reference implementation can be found at this https URL.
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2102.08124 [cs.AI]
  (or arXiv:2102.08124v2 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2102.08124
arXiv-issued DOI via DataCite

Submission history

From: Brian Chmiel [view email]
[v1] Tue, 16 Feb 2021 12:44:16 UTC (767 KB)
[v2] Wed, 20 Oct 2021 07:16:22 UTC (962 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Accelerated Sparse Neural Training: A Provable and Efficient Method to Find N:M Transposable Masks, by Itay Hubara and 5 other authors
  • View PDF
  • TeX Source
  • Other Formats
license icon view license
Current browse context:
cs.AI
< prev   |   next >
new | recent | 2021-02
Change to browse by:
cs

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar

DBLP - CS Bibliography

listing | bibtex
Itay Hubara
Brian Chmiel
Ron Banner
Daniel Soudry
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