deepspeedai / deepspeedai/DeepSpeed
[REQUEST] Why con't Pipeline Parallelism do sequence parallel?
Open
Nobody has claimed this yet.
enhancement
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
a little disappointed when I see assert tensor_model_parallel_size == 1 and pipeline_model_parallel_size == 1, \
'DeepSpeed\'s sequence parallel does not work with tensor parallel or pipeline parallel'
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 locating the assertion shown in the issue and reading the sequence-, tensor-, and pipeline-parallelism paths around it. Determine whether supporting their combination is intended; done would require a documented decision and, if supported, coverage for the combined configuration.
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
- 18/100