Add optional SyncBatchNorm support to ImageNet example#1406
Add optional SyncBatchNorm support to ImageNet example#1406snehasonkusare-tech wants to merge 1 commit intopytorch:mainfrom
Conversation
|
Thank you for your pull request and welcome to our community. Action RequiredIn order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you. ProcessIn order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA. Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks! |
✅ Deploy Preview for pytorch-examples-preview canceled.
|
|
Hi, this is my first contribution to PyTorch. Please let me know if any changes are needed. |
|
Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks! |
1 participant
This PR adds optional SyncBatchNorm support to the ImageNet training script.
What’s included:
Added new CLI argument:
--sync-bnIf both (a) distributed training is enabled and (b)
--sync-bnis passed,BatchNorm layers are converted to SyncBatchNorm using:
nn.SyncBatchNorm.convert_sync_batchnorm(model)
Conversion happens before DistributedDataParallel wrapping, as required.
Why this is useful:
SyncBatchNorm improves performance and stability in multi-GPU distributed training
and is commonly used in ImageNet training recipes. Adding this option makes the
example more complete and aligned with standard training practices.
Testing:
--sync-bnis passed.Addresses Issue #793.