NVIDIA / NVIDIA/apex

can ASP be used to form NM sparsity other than 2:4?

Open
#1,283 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
9k
Forks
1.5k
Avg merge
2d 4h
Merged PRs (30d)
3

Description

I have read the paper "Channel permutation for NM sparsity" . I guess it is a great job.
However, I am confused whether or not the codes that are publically available can be used to prune a network into the NM sparsity other than 2:4? E.g., 1:4, or 1:8.
If the codes can be used to realize the function as I say above, either directly or only with minor modifiction, please let me know.
Thank you anyway.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by locating the public ASP implementation and its pruning entry points, then check whether the code or documentation defines supported N:M ratios beyond 2:4. The issue would be complete when support for 1:4 or 1:8 is confirmed with a reproducible example, or the required scope and limitations are documented.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.