fix: delay torch.multiprocessing import to prevent semaphore leak warning #114
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
There was a problem hiding this comment.
Pull Request Overview
This PR updates dependency lock files and makes several code improvements focused on import ordering, API usage fixes, and performance optimization. The changes ensure better compatibility with PyTorch's API requirements and improve code organization following import sorting best practices.
- Updates the uv.lock file to add new dependencies for workspace examples (gitpython, pillow, protobuf, pydantic, torchvision, wandb, etc.)
- Fixes PyTorch CUDA tensor creation and RNG forking API usage
- Optimizes import loading by moving torch.multiprocessing import to where it's used
- Standardizes import order to place third-party imports before local imports
Reviewed Changes
Copilot reviewed 9 out of 10 changed files in this pull request and generated no comments.
Show a summary per file
| File | Description |
|---|---|
| uv.lock | Adds new dependencies and workspace members for quickstart and step-by-step examples |
| tests/test_reproducibility/test_generator.py | Adds device="cuda" parameter to CUDA tensor creation for proper device placement |
| src/blazefl/core/client_trainer.py | Moves torch.multiprocessing import to local scope for deferred loading |
| examples/step-by-step-dsfl/pyproject.toml | Configures ruff to treat wandb as third-party import |
| examples/step-by-step-dsfl/models/selector.py | Fixes torch.random.fork_rng() to pass empty list argument |
| examples/step-by-step-dsfl/main.py | Reorders imports to place wandb before local imports |
| examples/quickstart-fedavg/pyproject.toml | Configures ruff to treat wandb as third-party import |
| examples/quickstart-fedavg/models/selector.py | Fixes torch.random.fork_rng() to pass empty list argument |
| examples/quickstart-fedavg/main.py | Reorders imports to place wandb before local imports |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
WHAT
This PR moves the
torch.multiprocessingimport from the global scope ofclient_trainer.pyto the inside of theProcessPoolClientTrainer.local_processmethod.WHY
The global import was causing
for users of
ThreadPoolClientTrainer(which does not usemultiprocessing).This change ensures that
torch.multiprocessingand its associated resources are only initialized whenProcessPoolClientTraineris actually used, resolving the leak warning forThreadPoolClientTrainerusers.