mindspore-ai / mindspore-ai/hyper-parallel
[bug] mf联调发现，参数初始化时，dim_map[shard_dim] index越界
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 53
- Forks
- 63
- Avg merge
- 23h 45m
- Merged PRs (30d)
- 63
Description
问题：
DTensor 参数初始化时，发现dim_map[shard_dim] index越界。
原因：
dim_map 元祖构建的dim维度取错了，期望是Tensor shape的ndim，实际取成了devicemesh的ndim
解决：
改成shape的dim
schema_version: 1
source: gitcode
gitcode_repo: mindspore/hyper-parallel
gitcode_issue: 169
source_url: https://gitcode.com/mindspore/hyper-parallel/issues/169
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
Locate the DTensor parameter-initialization code that constructs dim_map, then trace how the tensor shape and device-mesh dimensions are obtained. Update the dimension source as described so shard_dim indexing matches the Tensor shape, and verify initialization with a sharded tensor whose shape and device-mesh dimensions differ.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- distributed-systems
- Issue type
- Bug
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 65/100