Skip to content

Add converter for aten::_grouped_mm.default #2805

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
Sid-V5 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add converter for aten::_grouped_mm.default #2805

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
23 changes: 23 additions & 0 deletions onnxscript/function_libs/torch_lib/ops/core.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4503,6 +4503,29 @@ def aten_grid_sampler_3d_backward(
raise NotImplementedError()


@torch_op("aten::_grouped_mm")
def aten_grouped_mm(
self: TFloat,
mat2: TFloat,
offs: Optional[TInt] = None,
bias: Optional[TFloat] = None,
out_dtype: Optional[int] = None,
) -> TFloat:
"""_grouped_mm(Tensor self, Tensor mat2, *, Tensor? offs=None, Tensor? bias=None, int? out_dtype=None) -> Tensor"""

# If offs is None, it uses the "dense" / "batch" mode where groups are implicit in the batch dimension.
# self: (G, M, K), mat2: (G, K, N) -> (G, M, N)
if offs is None:
res = op.MatMul(self, mat2)
if bias is not None:
res = op.Add(res, bias)
if out_dtype is not None:
res = op.Cast(res, to=out_dtype)
return res

raise NotImplementedError("aten::_grouped_mm with 'offs' is not supported.")


def aten_gru_cell(
input: TensorType,
hx: TensorType,
Expand Down