-
Notifications
You must be signed in to change notification settings - Fork 21.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[MPS] Fix batchnorm forward and backward pass #94351
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/94351
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ✅ No FailuresAs of commit 604b8ad: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
1611b36
to
4c94a24
Compare
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
The merge job was canceled. If you believe this is a mistake,then you can re trigger it through pytorch-bot. |
@pytorchbot merge -g |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
float primary = 1.0f; | ||
MPSGraphTensor *primaryTensor = [mpsGraph constantWithScalar:primary dataType:MPSDataTypeFloat32]; | ||
|
||
scaledInverseSqrtVariance = [mpsGraph divisionWithPrimaryTensor:primaryTensor | ||
secondaryTensor:sqrtVariance | ||
name:nil]; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is there a reason not to use reciprocalWithTensor:
here instead? It should be faster, shouldn't it? Or does it come with performance implications?
(PR is coming)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
With the precise mode being default, there won't be any major perf implications.
Fixes batchnorm forward/backward pass and layer_norm:
Batchnorm Forward pass:
Batchnorm Backward pass:
Layer norm: