deepspeedai / deepspeedai/DeepSpeed
How to implement CodeGen's Tensor Parallelism Inference based on DeepSpeed?
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
I want to use DeepSpeed for TP inference, but I found that AutoTP does not support the CodeGen model, I want to know the reason, maybe I can implement a way by myself.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reading DeepSpeed's AutoTP implementation and its existing model support, then compare that with the CodeGen model integration. Determine why CodeGen is unsupported and define the required tensor-parallel inference work; done means CodeGen can run with DeepSpeed AutoTP inference.
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