deepspeedai / deepspeedai/DeepSpeed

[REQUEST] BF16 in DeepSpeed Inference

Open
#3,139 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

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

Description

It seems DeepSpeed inference does not support BF16 for now. I am wondering how hard it is to support BF16? Can I just reuse all FP16 kernels? Is there any plan on it?

Thanks!

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 names no files, tests, or entry points. Start by locating DeepSpeed Inference’s existing FP16 kernel paths and the boundary where BF16 support is missing, then determine the scope of supporting BF16. Done means BF16 inference support is implemented and verified, but the issue provides no specific acceptance test.

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.