assert torch!=1.12.0 for DDP training #8621
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Outdated suggestions cannot be applied.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.
π οΈ PR Summary
Made with β€οΈ by Ultralytics Actions
π Summary
Updated dependencies and streamlined Distributed Data Parallel (DDP) model instantiation.
π Key Changes
torch
andtorchvision
inrequirements.txt
, previously excluding problematic versions.train.py
with a newsmart_DDP
function.train.py
, resulting in cleaner code.smart_DDP
function inutils/torch_utils.py
that includes version checks and conditions for DDP instantiation, ensuring compatibility and preventing usage with unsupported versions.π― Purpose & Impact
torch
andtorchvision
dependencies may allow for wider compatibility with different versions of these libraries, potentially easing installation and upgrade processes.smart_DDP
function centralizes DDP logic, making it easier to manage compatibility issues with PyTorch versions and reducing potential errors during multi-GPU training setups.train.py
andutils/torch_utils.py
, the codebase becomes more maintainable and understandable, which benefits contributors and users alike.