Skip to content

Add 1D-empty tensor support for AtenCatOp::getCanonicalizationPatterns #4341

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

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
+14 −0
Open

Add 1D-empty tensor support for AtenCatOp::getCanonicalizationPatterns #4341

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5119232
Add 1D-empty tensor support for AtenCatOp::getCanonicalizationPatterns
sakupan102 Oct 13, 2025
276571a
Add test for aten.cat with 1-D empty tensor
sakupan102 Oct 13, 2025
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
4 changes: 4 additions & 0 deletions lib/Dialect/Torch/IR/TorchOps.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4147,6 +4147,10 @@ void AtenCatOp::getCanonicalizationPatterns(RewritePatternSet &patterns,
if (!operandTy || !operandTy.hasSizes())
return failure();
int64_t adim = dim < 0 ? dim + operandTy.getSizes().size() : dim;
bool is1DEmptyTensor =
operandTy.getSizes().size() == 1 && operandTy.getSizes()[0] == 0;
if (is1DEmptyTensor)
continue;
if (operandTy.getSizes()[adim] != 0)
filtered.push_back(operand);
}
Expand Down
10 changes: 10 additions & 0 deletions test/Dialect/Torch/canonicalize.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3242,6 +3242,16 @@ func.func @aten_cat_zero(%arg0 : !torch.vtensor<[4,5,6],f32>, %arg1 : !torch.vte

// -----

// CHECK-LABEL: @aten_cat_empty
func.func @aten_cat_empty(%arg0 : !torch.vtensor<[4,5,6],f32>, %arg1 : !torch.vtensor<[0],f32>) -> !torch.vtensor<[4,5,6],f32> {
// CHECK: return %arg0 : !torch.vtensor<[4,5,6],f32>
%list = torch.prim.ListConstruct %arg0, %arg1 : (!torch.vtensor<[4,5,6],f32>, !torch.vtensor<[0],f32>) -> !torch.list<vtensor>
%dim = torch.constant.int -2
%0 = torch.aten.cat %list, %dim : !torch.list<vtensor>, !torch.int -> !torch.vtensor<[4,5,6],f32>
return %0 : !torch.vtensor<[4,5,6],f32>
}
// -----

// CHECK-LABEL: @aten_tensor_scalar_lt
func.func @aten_tensor_scalar_lt() -> (!torch.vtensor<[4],i1>, !torch.vtensor<[4],i1>) {
// CHECK: %[[CST:.+]] = torch.vtensor.literal(dense<true> : tensor<4xi1>) : !torch.vtensor<[4],i1>
Expand Down