Skip to content

Fix linter error - #3435

Open
fheinecke wants to merge 1 commit into
NVIDIA:mainfrom
fheinecke:fred/linter-fix-1
Open

Fix linter error#3435
fheinecke wants to merge 1 commit into
NVIDIA:mainfrom
fheinecke:fred/linter-fix-1

Conversation

@fheinecke

Copy link
Copy Markdown
Collaborator

Description

Simple style fix to appease the CI linter.

Type of change

  • Documentation change (change only to the documentation, either a fix or a new content)
  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Infra/Build change
  • Code refactoring

Changes

Please list the changes introduced in this PR:

  • Fix linting issue

Checklist:

  • I have read and followed the contributing guidelines
  • The functionality is complete
  • I have commented my code, particularly in hard-to-understand areas
  • I have made corresponding changes to the documentation
  • My changes generate no new warnings
  • I have added tests that prove my fix is effective or that my feature works
  • New and existing unit tests pass locally with my changes

Signed-off-by: Fred Heinecke <fheinecke@nvidia.com>
@fheinecke
fheinecke requested a review from cyanguwa as a code owner August 27, 2026 22:22
@greptile-apps

greptile-apps Bot commented Aug 27, 2026

Copy link
Copy Markdown
Contributor

Greptile Summary

This PR performs a behavior-preserving refactor of an SM90 fused-attention eligibility condition to satisfy linting.

  • Extracts the FP16 THD training configuration checks into is_sm90_f16_thd_training.
  • Retains the existing fused-attention and deterministic-backward size guards.

Confidence Score: 5/5

The PR appears safe to merge because the refactor preserves the existing backend-selection behavior.

The extracted predicates operate on initialized, evaluation-safe values, while the size check and fused-attention fallback remain protected by the original use_fused_attention guard.

Important Files Changed

Filename Overview
transformer_engine/pytorch/attention/dot_product_attention/utils.py Extracts part of the backend-selection condition into a local boolean without changing the guarded behavior.

Reviews (1): Last reviewed commit: "Fix linter error" | Re-trigger Greptile

@cyanguwa cyanguwa added the 2.19 label Aug 28, 2026
@ptrendx

ptrendx commented Aug 28, 2026

Copy link
Copy Markdown
Member

/te-ci pytorch

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants