Justin Chu c5701d0ab5 [ONNX] Create fake implementations for onnx ops; fix boolean mask in attention (#165780) ... Previously we rely on the concreate implementation to generate fake implementation. This makes the fake implementation overly complicated and breaks in some cases when there are dynamic shapes. This PR updates onnx op registration to instead take a dedicated fake implementation. **Also fixed: When boolean mask is supplied to torch sdpa, it was previously taken the negation, which is incorrect.** Fix https://github.com/pytorch/pytorch/issues/164909 Also taken changes from https://github.com/pytorch/pytorch/pull/156635 Pull Request resolved: https://github.com/pytorch/pytorch/pull/165780 Approved by: https://github.com/titaiwangms		2025-10-29 04:51:49 +00:00
..
test_ops.py	[ONNX] Create fake implementations for onnx ops; fix boolean mask in attention (#165780)	2025-10-29 04:51:49 +00:00