Skip to content
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

[MLIR][TORCH] Add OnnxToTorch support for BlackmanWindow function #3181

Merged
merged 7 commits into from
Apr 30, 2024
Merged

[MLIR][TORCH] Add OnnxToTorch support for BlackmanWindow function #3181

Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
580887a
[MLIR][TORCH] Add OnnxToTorch support for BlackmanWindow function
vinayakdsci Apr 17, 2024
7b22a0c
Address Comments
vinayakdsci Apr 18, 2024
83e335f
Rewrite op to follow onnx implementation
vinayakdsci Apr 22, 2024
8ab21e3
Move getItemOp() to Utils.h
vinayakdsci Apr 22, 2024
2d0fcbc
Resolve conflicts
vinayakdsci Apr 22, 2024
ebeff28
Fix the resultType for variable periodicSizeFloat
vinayakdsci Apr 24, 2024
531a2ee
Add comment to explain deviation from onnx impl
vinayakdsci Apr 26, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Add comment to explain deviation from onnx impl
  • Loading branch information
@vinayakdsci
vinayakdsci committed Apr 26, 2024
commit 531a2ee727f231cd5c4a563f289c756ff9c0850b
5 changes: 5 additions & 0 deletions lib/Conversion/TorchOnnxToTorch/DefaultDomainAtoF.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2283,6 +2283,11 @@ void mlir::torch::onnx_c::populateDefaultDomainAtoF(
Value a2Component = rewriter.create<Torch::AtenMulScalarOp>(
binder.getLoc(), resultType, cosTwoRangeAngular, a2);

// AtenSubScalarOp actually requires a tensor operand as the LHS, that
// is, operand #1. Therefore, to avoid errors, the onnx implementation
// has been modified. a1 has been changed to negative half, and the
// AtenSubScalarOp has been replaced with AtenAddScalarOp, as the add
// operation is commutative.
Value subA1Component = rewriter.create<Torch::AtenAddScalarOp>(
binder.getLoc(), resultType, a1Component, a0, one);
Value result = rewriter.create<Torch::AtenAddTensorOp>(
Expand Down
Loading