deepspeedai / deepspeedai/DeepSpeed

[REQUEST] Deepspeed Inference Supports VL (vision language) model

Open
#6,917 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

Is your feature request related to a problem? Please describe.
We have been using deepspeed.init_inference API for speeding up inference for text only models (e.g. mistral, qwen 2.5 series) with success. Was hoping we can extend support for vision language models as well, e.g. qwen 2 vl, etc, which is currently not supported.

Describe the solution you'd like

  • deepspeed.init_inference to work for vision language models (for both embedding use case as well as generation use case)
  • and also make extending with our own model's tutorial clearer/cleaner.

Describe alternatives you've considered
N/A

Additional context
N/A

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 tracing the deepspeed.init_inference API and its existing text-only model support. Compare the requirements for vision-language embedding and generation use cases, then review the tutorial guidance for extending support to custom models. Done means both requested use cases work for vision-language models and the extension tutorial is clearer.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
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.