- 
                Notifications
    You must be signed in to change notification settings 
- Fork 706
Arm backend: Use reshape instead of view before edge #15269
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
Conversation
The view operator can't handle non-contigious strides, such as the result of an expand. These are normalized after to_edge, but in the transform_for_annotation_pipeline we shouldn't use views for that reason. Reshape is the equivalent operator that can handle such strides. This issue was found in the roberta model. Signed-off-by: Erik Lundell <erik.lundell@arm.com> Change-Id: I042b76ecf8b99e4e65bf951c00ec53a9d0d36c80
| 🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15269
 Note: Links to docs will display an error until the docs builds have been completed. ❌ 10 New Failures, 3 Unrelated FailuresAs of commit 1354c37 with merge base b24c39a ( NEW FAILURES - The following jobs have failed:
 
 BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures 
 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.
Unrelated CI failures
The view operator can't handle non-contigious strides, such as the result of an expand. These are normalized after to_edge, but in the transform_for_annotation_pipeline we shouldn't use views for that reason.
Reshape is the equivalent operator that can handle such strides.
cc @freddan80 @per @zingo @oscarandersson8218 @digantdesai