Fix torchToTosa lowering for avgpool2d to handle unsupported parameters #3822
Conversation
This file contains 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
|
Hi @vivekkhandelwal1 , I do not have permission to add reviewers; since you contributed some of these changes, I am tagging you here to look into the change. Could you please take a look at this change? @sahas3, @rafaelubalmw, I am Tagging you here to notify you about the changes. |
sjarus
approved these changes
Oct 31, 2024
There was a problem hiding this comment.
Looks great! Please feel free to tag me and @justin-ngo-arm on TorchToTosa updates. You can also ping us on Discord for timely feedback.
Sounds good, Suraj. Thanks for looking into the change. |
added 2 commits
October 31, 2024 15:21
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Outdated suggestions cannot be applied.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.
The existing TorchToTosa lowering logic for
torch.aten.avg_pool2ddoesn't handle some unsupported properties well, leading to a silent wrong answer(SWA) when we go throughtorch-backend-to-tosa-backend-pipeline.For instance, with the existing TOSA avgpool2d specification, we can not representcount_include_padanddivisor_override,so during TorchToTosa lowering, we silently ignore these properties which leads to SWA in some cases—the fix captured in this change errors for unsupported scenarios.For details on
count_include_padanddivisor_override,please see the below link.https://pytorch.org/docs/stable/generated/torch.nn.AvgPool2d.html