Fix type annotation for torch.backends.cudnn.allow_tf32 (#72757)

Summary:
Fixes https://github.com/pytorch/pytorch/issues/72753

Pull Request resolved: https://github.com/pytorch/pytorch/pull/72757

Reviewed By: samdow

Differential Revision: D34204436

Pulled By: ngimel

fbshipit-source-id: 3528efd7bdf72c1d9338806555ecb643ab94ffeb
(cherry picked from commit 7036c2e6e6)
This commit is contained in:
lkct 2022-02-14 09:21:12 -08:00 committed by PyTorch MergeBot
parent 89c934f4b8
commit 7d542a4f2b

View File

@ -133,3 +133,4 @@ sys.modules[__name__] = CudnnModule(sys.modules[__name__], __name__)
enabled: bool
deterministic: bool
benchmark: bool
allow_tf32: bool