Skip to content

Conversation

@shewu-quic
Copy link
Collaborator

Summary:

  • Add a pass to decompose triu op during quantization and export stage so that aten.ge.Scalar does not appear after to_edge.
  • Refactor linear op build to support non-constant weight and bias
  • Fixed typo in doc
  • Fixed RMSNorm without eps

@shewu-quic shewu-quic requested a review from cccclai as a code owner December 1, 2025 09:44
@pytorch-bot
Copy link

pytorch-bot bot commented Dec 1, 2025

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16014

Note: Links to docs will display an error until the docs builds have been completed.

⏳ No Failures, 3 Pending

As of commit 52358a1 with merge base 1864fb0 (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@meta-cla meta-cla bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Dec 1, 2025
@shewu-quic
Copy link
Collaborator Author

Hi @cccclai ,
This PR addresses the op error in this issue #15734 and also resolves the QNN lowering error about RMSNorm failure reported by a Discord user.

When you have some time, could you please review this?

@shewu-quic
Copy link
Collaborator Author

@pytorchbot label "release notes: qualcomm"

@pytorch-bot pytorch-bot bot added the release notes: qualcomm Changes to the Qualcomm backend delegate label Dec 1, 2025
Copy link
Contributor

@cccclai cccclai left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks! Can you rebase so we can test CI? Our CI was disabled for a while

…nstant weights

Summary:
- Add a pass to decompose triu op during quantization and export stage so that aten.ge.Scalar does not appear after to_edge.
- Refactor linear op build to support non-constant weight and bias

- Fixed doc

- Fixed RMSNorm without eps
@shewu-quic shewu-quic force-pushed the dev1/hutton/support_triu branch from ee4023a to 16b9c46 Compare December 2, 2025 00:50
@shewu-quic
Copy link
Collaborator Author

Thanks! Can you rebase so we can test CI? Our CI was disabled for a while

I have rebased. Thanks.

@meta-codesync
Copy link

meta-codesync bot commented Dec 2, 2025

@cccclai has imported this pull request. If you are a Meta employee, you can view this in D88116321.

@cccclai cccclai merged commit a042763 into pytorch:main Dec 2, 2025
143 checks passed
AdrianLundell pushed a commit to AdrianLundell/executorch that referenced this pull request Dec 2, 2025
…nstant weights (pytorch#16014)

Summary:
- Add a pass to decompose triu op during quantization and export stage
so that aten.ge.Scalar does not appear after to_edge.
- Refactor linear op build to support non-constant weight and bias
- Fixed typo in doc
- Fixed RMSNorm without eps
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. release notes: qualcomm Changes to the Qualcomm backend delegate

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants