deepspeedai / deepspeedai/DeepSpeed
[Questions] Why Ulysess need all2all for QKV, but RingAttention just need KV under context parallel ?
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
Hi,
I have one question, Why Ulysess need all2all for QKV, but RingAttention just need exchange KV under context parallel ?
Thanks
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
No file, test, or entry point is named. Start by locating the implementations or documentation for Ulysess, RingAttention, QKV, and KV exchange in DeepSpeed, then compare their context-parallel communication requirements; done means providing a clear, referenced explanation of why the communication patterns differ.
Written by the indexing model from the issue text.
Assessment
- Domain
- distributed-systems, machine-learning
- Issue type
- Documentation
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 15/100