Skip to content

Keep float64 precision of Python floats in scalar_tensor #3114

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
Muhammad Usman Mateen (usmanmateen) wants to merge 3 commits into
base: main
Choose a base branch
from
+15 −0
Open

Keep float64 precision of Python floats in scalar_tensor #3114

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
fd3f918
Add an e2e test for Python floats in float64 models
usmanmateen Oct 21, 2026
94b036b
Keep float64 precision of Python floats in scalar_tensor
usmanmateen Oct 21, 2026
aeedec3
Remove the comment in aten_scalar_tensor
usmanmateen Oct 21, 2026
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
2 changes: 2 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 @@ -9085,6 +9085,8 @@ def aten_scalar_tensor(
if dtype == -1:
dtype = FLOAT.dtype

if isinstance(s, float) and dtype in (DOUBLE.dtype, COMPLEX128.dtype):
s = op.Constant(value=ir.tensor(s, dtype=ir.DataType.DOUBLE))
return common_ops.cast_to(s, dtype=dtype)


Expand Down
13 changes: 13 additions & 0 deletions tests/function_libs/torch_lib/e2e_ops_tests.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1967,6 +1967,19 @@ def forward(self, x):
self.assertEqual(node.attributes.get_int("noop_with_empty_axes", 0), 0)
_testing.assert_onnx_program(onnx_program)

def test_python_float_keeps_float64_precision(self):
# torch.export turns the Python floats into scalar_tensor(0.1, dtype=float64).
class Model(torch.nn.Module):
def forward(self, x):
return x + 0.1, x * 0.1, x == 0.1, torch.clamp(x, max=0.1)

x = torch.tensor([1.0, 0.3, 0.1, -2.7], dtype=torch.float64)
onnx_program = torch.onnx.export(Model(), (x,), dynamo=True, verbose=False)
# assert_onnx_program allows 1e-7 for float64, more than 0.1 rounded to float32
# is off by, so compare exactly.
for actual, expected in zip(onnx_program(x), Model()(x)):
torch.testing.assert_close(actual, expected, rtol=0, atol=0)


if __name__ == "__main__":
unittest.main()