alibaba / alibaba/graphlearn-for-pytorch
Error handling in distributed training
Open
- Dominant language
- Python
- Stars
- 146
- Forks
- 48
- PR merge metrics
- No merged PRs in 30d
Description
### 🚀 The feature, motivation and pitch
Properly handle the errors occurred during the training:
- Stop issuing new RPC request when an previous one failed
- Terminate training once an error happened
- Properly clean up the shared graph data
Consider both the mp mode and the collocated mode.
### Alternatives
_No response_
### Additional context
_No response_
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.