mindspore-ai / mindspore-ai/hyper-parallel
[feature][pipeline_parallel] PP + FSDP
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 53
- Forks
- 63
- Avg merge
- 23h 45m
- Merged PRs (30d)
- 63
Description
该问题是怎么引起的？
PP调度中支持FSDPModule，当PP+FSDP场景时， FSDPModule进行zero1路径，避免多个micro_batch引发频发的FSDP 通信
重现步骤
报错信息
schema_version: 1
source: gitcode
gitcode_repo: mindspore/hyper-parallel
gitcode_issue: 148
source_url: https://gitcode.com/mindspore/hyper-parallel/issues/148
Contributor guide
No contributing guide indexed for this repository
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 names pipeline-parallel scheduling and FSDPModule, but provides no file or test path. Start by locating the PP scheduler and FSDPModule integration, then verify that PP+FSDP uses the zero1 path without repeated FSDP communication across micro-batches; add or run focused coverage for that behavior.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- distributed-systems, machine-learning
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100