deepspeedai / deepspeedai/DeepSpeed
The sequence length is not divisible by Sequence Parallel World Size
Open
Nobody has claimed this yet.
bug
training
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
Many thanks for open-sourcing the sequence parallel solution DeepSpeed Ulysses. I’d like to inquire about how to handle situations where the sequence length is not divisible by Sequence Parallel World Size.
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
The issue does not name any files, tests, or entry points. First locate the DeepSpeed Ulysses sequence-parallel implementation and determine how it currently handles sequence lengths relative to the world size. Done should include an agreed behavior for non-divisible lengths and coverage demonstrating that behavior.
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