This repository has been archived by the owner on Nov 17, 2023. It is now read-only.
Reorder module import orders for dist-kvstore #13742
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
For distributed training, a new process is launched when importing the
kvstore_server
module. This should be imported at last so that other MXNet modules are initialized correctly. Otherwise this may result in error when unpickling custom LR scheduler/optimizers. For example, the LRScheduler in gluoncv /~https://github.com/dmlc/gluon-cv/blob/master/gluoncv/utils/lr_scheduler.py#L8 depends on a specific version of MXNet, and checks the__version__
attr of MXNet, which is not set on kvstore server due to the fact that kvstore-server module is imported before the__version__
attr is set.@hetong007
Checklist
Essentials
Please feel free to remove inapplicable items for your PR.
Changes
End to end tested with a gluon-cv example.