Skip to content

fix: aten::where with differing-shape inputs bugfix #1533

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.

Sign up for GitHub

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

Jump to bottom
Merged
merged 2 commits into from
Dec 12, 2022
Merged

fix: aten::where with differing-shape inputs bugfix #1533

merged 2 commits into from
Dec 12, 2022

Conversation

gs-olive
Copy link
Contributor

@gs-olive gs-olive commented Dec 7, 2022

Description

  • Behavior of Torch-TRT differs from that of Torch in the case where the input tensors to aten::where have different rank
  • Torch automatically broadcasts tensors to the highest-rank variant whereas the TRT Select layer requires tensors of the same rank and throws an error
  • Add dimension checking and unsqueeze operator to ensure broadcasting is enabled
  • Add test case to catch error

Resolves Part 1 of solution to Issue #1455

Type of change

  • Bug fix (non-breaking change which fixes an issue)

Checklist:

  • [ x ] My code follows the style guidelines of this project (You can use the linters)
  • [ x ] I have performed a self-review of my own code
  • [ x ] I have commented my code, particularly in hard-to-understand areas and hacks
  • [ x ] I have made corresponding changes to the documentation
  • [ x ] I have added tests to verify my fix or my feature
  • [ x ] New and existing unit tests pass locally with my changes
  • [ x ] I have added the relevant labels to my PR in so that relevant reviewers are notified

@github-actions github-actions bot added component: conversion Issues re: Conversion stage component: converters Issues re: Specific op converters component: core Issues re: The core compiler component: tests Issues re: Tests labels Dec 7, 2022
gs-olive
gs-olive commented Dec 7, 2022
View reviewed changes
@gs-olive gs-olive self-assigned this Dec 7, 2022
@gs-olive
fix: Bug in aten::where with differing-shape inputs
de9cbaf 
- Behavior of Torch-TRT differing from that of Torch in the case where
the input tensors to `aten::where` have different rank
- Torch automatically broadcasts tensors to the highest-rank variant
whereas the TRT Select layer requires tensors of the same rank and
throws an error
- Add dimension checking and unsqueeze operator to ensure broadcasting
is enabled
- Add test case to catch error
narendasan
narendasan approved these changes Dec 12, 2022
View reviewed changes
Copy link
Collaborator

@narendasan narendasan left a comment

Choose a reason for hiding this comment

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

LGTM

@narendasan narendasan merged commit d923805 into pytorch:master Dec 12, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@narendasan narendasan narendasan approved these changes

@bowang007 bowang007 Awaiting requested review from bowang007

@peri044 peri044 Awaiting requested review from peri044

Assignees

@gs-olive gs-olive

Labels

cla signed component: conversion Issues re: Conversion stage component: converters Issues re: Specific op converters component: core Issues re: The core compiler component: tests Issues re: Tests

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@gs-olive @narendasan @facebook-github-bot