deepspeedai / deepspeedai/DeepSpeed

How to run DeepSpeed inside mpirun-ed singularity containers on multiple nodes

Open
#1,257 3 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

I'm trying to use DeepSpeed on our clusters, where singularity containers are run on each machine, with 8 cards installed. No Kubernetes are available. All containers are launched using mpirun. What's the proper way/cmd to use in order to run such a usecase?

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

The issue mentions DeepSpeed running under mpirun in Singularity containers across multiple nodes with eight cards per machine, and names no files or tests. Start by locating the project's documented distributed-launch entry points and determine whether they cover this setup; done means a reproducible documented command or a clear statement that it is unsupported.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
devops, distributed-systems, machine-learning
Issue type
Documentation
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.