deepspeedai / deepspeedai/DeepSpeedExamples

Why not just use zero3 inference to generate sequence in DeepSpeed Chat stage3 training?

Open
#760 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
6.8k
Forks
1.1k
Avg merge
2d 16h
Merged PRs (30d)
1

Description

DeepSpeed Chat use tensor parallelism via hybrid engine to generate sequence in stage3 training.
I wonder if just use zero3 inference for generation is ok? So that we don't need to transform model params between train and eval mode.
Any explanations about the design of stage3 training would be appreciated. Thanks.

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 reading the DeepSpeed Chat stage3 training design and the hybrid engine's tensor-parallel generation path. Compare that flow with the proposed ZeRO-3 inference approach and document whether it is supported, including the implications for parameter transformation between training and evaluation. Done means a maintainer-approved explanation or design decision; no files or tests are identified in the issue.

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.