Skip to content

[Relax][PyTorch] Support MatrixMultiply op for ExportedProgram importer #18343

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

Merged
MasterJH5574 merged 1 commit into from
Sep 25, 2025
Merged

[Relax][PyTorch] Support MatrixMultiply op for ExportedProgram importer #18343

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
3 changes: 3 additions & 0 deletions python/tvm/relax/frontend/torch/exported_program_translator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -434,6 +434,9 @@ def create_convert_map(
"matmul.default": self._binary_op(
partial(relax.op.linear_algebra.matmul, out_dtype="float32"), operator.matmul
),
"mm.default": self._binary_op(
partial(relax.op.linear_algebra.matmul, out_dtype="float32"), operator.matmul
),
Comment on lines +437 to +439
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Sep 25, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The implementation for mm.default is identical to matmul.default (lines 434-436) and bmm.default (lines 477-479). To improve maintainability and reduce code duplication, consider defining a common handler for these matrix multiplication operations and reusing it.

For example:

    matmul_handler = self._binary_op(
        partial(relax.op.linear_algebra.matmul, out_dtype="float32"), operator.matmul
    )
    # ...
    return {
        # ...
        "matmul.default": matmul_handler,
        "mm.default": matmul_handler,
        # ...
        "bmm.default": matmul_handler,
        # ...
    }

"max.other": self._binary_op(relax.op.maximum, max),
"min.other": self._binary_op(relax.op.minimum, min),
"max.default": self._unary_op(relax.op.max),
Expand Down
26 changes: 26 additions & 0 deletions tests/python/relax/test_frontend_from_exported_program.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5914,6 +5914,32 @@ def main(
verify_model(Model(), example_args, {}, Expected)


def test_mm():
class MatrixMultiply(Module):
def forward(self, a, b):
return torch.mm(a, b)

example_args = (
torch.randn(2, 3, dtype=torch.float32),
torch.randn(3, 4, dtype=torch.float32),
)

@tvm.script.ir_module
class Expected:
@R.function
def main(
a: R.Tensor((2, 3), dtype="float32"),
b: R.Tensor((3, 4), dtype="float32"),
) -> R.Tuple(R.Tensor((2, 4), dtype="float32")):
with R.dataflow():
lv: R.Tensor((2, 4), dtype="float32") = R.matmul(a, b, out_dtype="float32")
gv: R.Tuple(R.Tensor((2, 4), dtype="float32")) = (lv,)
R.output(gv)
return gv

verify_model(MatrixMultiply(), example_args, {}, Expected)


if __name__ == "__main__":
tvm.testing.main()
1
Loading