deepspeedai / deepspeedai/DeepSpeed

Is "Hierarchical All-to-all" feat available in current version?

Open
#6,957 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
43.1k
Forks
5k
Avg merge
4d 15h
Merged PRs (30d)
112

Description

According to https://arxiv.org/pdf/2201.05596, we are very interested on "Hierarchical All-to-all" and think this is an amazing feature. But after reading source code, we could not find any part related to this. I found that team said that they did not have a ETA for this in https://github.com/microsoft/DeepSpeed/issues/2039, will this be supported in the future?

Contributor guide

Open the contributing guide

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 reviewing the current source code for distributed communication and the discussion in issue #2039, then compare it with the hierarchical all-to-all design described in the linked paper. The issue is complete only when current support or a concrete implementation path and scope are established.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
distributed-systems, 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.