Skip to content

[Relax][Frontend][Onnx] Cast Op special handling for ShapeExpr input #17061

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
merged 1 commit into from
Jun 4, 2024
Merged

[Relax][Frontend][Onnx] Cast Op special handling for ShapeExpr input #17061

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
6 changes: 6 additions & 0 deletions python/tvm/relax/frontend/onnx/onnx_frontend.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -442,6 +442,11 @@ class Cast(OnnxOpConverter):
@classmethod
def _impl_v13(cls, bb, inputs, attr, params):
to_type = get_type(attr["to"])
if isinstance(inputs[0], relax.ShapeExpr):
shape = inputs[0]
if all([isinstance(x, tir.IntImm) for x in shape]):
shape = [int(x) for x in shape]
return relax.const(shape, to_type)
if isinstance(inputs[0], relax.Constant):
output = inputs[0].data.numpy().astype(to_type)
return relax.const(output, to_type)
Expand Down Expand Up @@ -2210,6 +2215,7 @@ def _construct_nodes(self, graph: onnx.onnx_ml_pb2.GraphProto):
"Concat",
"Equal",
"Where",
"Cast",
]
for i, inp in enumerate(inputs):
if (
Expand Down