deepspeedai / deepspeedai/DeepSpeed
Does ds inference support op fusion for multi-head attention?
Open
Nobody has claimed this yet.
enhancement
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
As the title says, currently I am not able to find the implementation of fused multi-head attention in ds inference.
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 surveying the DeepSpeed inference implementation for multi-head attention and checking whether an op-fusion entry point is present. The issue does not name files or tests, so first determine whether fused multi-head attention already exists and clarify whether the expected work is documentation or implementation. Done should mean the supported behavior and any required change are clearly established.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning, performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100