Skip to content

[CUDA] Update Flash Attention to support head_sink for smooth softmax in GQA#25358

Closed
tianleiwu wants to merge 2 commits into
mainfrom
tlwu/gqa_head_sink_cuda
Closed

[CUDA] Update Flash Attention to support head_sink for smooth softmax in GQA#25358
tianleiwu wants to merge 2 commits into
mainfrom
tlwu/gqa_head_sink_cuda

add test

f8cd59e
Select commit
Loading
Failed to load commit list.
GitHub Advanced Security / CodeQL failed Jul 10, 2025 in 2s

1 configuration not found

Warning: Code scanning may not have found all the alerts introduced by this pull request, because 1 configuration present on refs/heads/main was not found:

API upload

  • ❓  <default>

New alerts in code changed by this pull request

  • 1 error
  • 1 note

See annotations below for details.

View all branch alerts.

Annotations

Check notice on line 194 in onnxruntime/test/python/transformers/test_gqa.py

See this annotation in the file changed.

Code scanning / CodeQL

Unused local variable Note test

Variable use_cuda_triton is not used.

Check failure on line 748 in onnxruntime/test/python/transformers/test_gqa.py

See this annotation in the file changed.

Code scanning / CodeQL

Potentially uninitialized local variable Error test

Local variable 'local_mask' may be used before it is initialized.