mindspore-ai / mindspore-ai/hyper-parallel
Interleavead1F1B 支持F&B fusion阶段的dx dw分离。
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？
当前对于FB融合阶段，可以进一步将Dx Dw分离，让Dw 计算完grad_input之后立马send出去。增大掩盖空间
重现步骤
报错信息
schema_version: 1
source: gitcode
gitcode_repo: mindspore/hyper-parallel
gitcode_issue: 170
source_url: https://gitcode.com/mindspore/hyper-parallel/issues/170
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
Start by locating the Python implementation of the FB fusion stage and its Dx/Dw gradient communication path; the issue names no files or tests. Trace the current ordering and communication dependencies. Done means Dw can send immediately after computing grad_input while preserving correct fusion behavior and communication overlap.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- distributed-systems
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 38/100