Skip to content

[Optimizer] Fix reinterpretation of strings in _get_numpy_value #2514

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
justinchuby merged 3 commits into from
Aug 27, 2025
+12 −3
Merged

[Optimizer] Fix reinterpretation of strings in _get_numpy_value #2514

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7277ffb
[Optimizer] Fix reinterpretation of strings in _get_numpy_value
iksnagreb Aug 26, 2025
d739409
[Optimizer] Skip calling .view for string type in _get_numpy_value
iksnagreb Aug 27, 2025
1ed42ab
[Optimizer] Do not convert strings to object in _get_numpy_value
iksnagreb Aug 27, 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
17 changes: 14 additions & 3 deletions onnxscript/optimizer/_constant_folding.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -278,9 +278,20 @@ def _get_numpy_value(
if size_limit is not None and const_value.size > size_limit:
return None
try:
# Reinterpret the array with `.view()` because some implementations of
# ir.TensorProtocol (e.g. PyTorch<=2.7) do not use ml_dtypes for bfloat16 etc.
array = const_value.numpy().view(const_value.dtype.numpy())
# Turn the constant value into a numpy array representation with the
# specifics of this conversion handled by the tensor type (might a
# yield result which needs to be reinterpreted)
array = const_value.numpy()
# Make sure strings are converted to object type first (might be
# some fixed width string representation which .view cannot
# convert, resulting in "TypeError: Cannot change data-type for
# array of references.")
if const_value.dtype == ir.DataType.STRING:
array = array.astype(np.object_)
# Reinterpret the array with `.view()` because some implementations
# of ir.TensorProtocol (e.g. PyTorch<=2.7) do not use ml_dtypes for
# bfloat16 etc.
array = array.view(const_value.dtype.numpy())
except FileNotFoundError:
# External data is not available.
logger.warning(
Expand Down