Hacktoberfest 2026: the issues maintainers tagged for October, open and beginner-friendly. Browse Hacktoberfest issues

🐛[BUG]: ValueError: infer_schema(func) in @torch.library.custom_op due to stringified type annotations (from __future__ import annotations)

Open
  • Is something wrong?
#1,990 0 comments 0 reactions 1 assignee View on GitHub

Maintainers usually reply within 1 day

@loliverhennigh is already working on this.

Since Sep 16, 2026.

  • #1991 by @DenizCanELCI — open

Assessment

This issue has not been assessed yet.

Description

? - Needs Triage bug
Version

e416350

On which installation method(s) does this occur?

Pip

Describe the issue

When importing physicsnemo (specifically through modules relying on physicsnemo.nn.functional.derivatives.mesh_green_gauss_gradient._warp_impl), a ValueError is raised during module load by PyTorch's schema inference mechanism in @torch.library.custom_op.

PyTorch's infer_schema inspects the function signature and expects actual types (such as <class 'torch.Tensor'>), but instead receives string literals (e.g. 'torch.Tensor'). This occurs when postponed evaluation of annotations is enabled (via from future import annotations or Python 3.11+ stringified annotations) without explicitly supplying the schema argument to @torch.library.custom_op.

Minimum reproducible example
python examples/structural_mechanics/deforming_plate/preprocessor.py
Relevant log output
D:\PycharmProjects\physicsnemo\venv\Scripts\python.exe D:\PycharmProjects\physicsnemo\examples\structural_mechanics\deforming_plate\preprocessor.py 
Warp CUDA warning: Could not find or load the NVIDIA CUDA driver. GPU execution will not be available.
Traceback (most recent call last):
  File "D:\PycharmProjects\physicsnemo\examples\structural_mechanics\deforming_plate\preprocessor.py", line 26, in <module>
    from deforming_plate_dataset import DeformingPlateDataset
  File "D:\PycharmProjects\physicsnemo\examples\structural_mechanics\deforming_plate\deforming_plate_dataset.py", line 31, in <module>
    from physicsnemo.datapipes.gnn.utils import load_json, save_json
  File "D:\PycharmProjects\physicsnemo\physicsnemo\datapipes\__init__.py", line 42, in <module>
    from physicsnemo.datapipes.dataset import Dataset
  File "D:\PycharmProjects\physicsnemo\physicsnemo\datapipes\dataset.py", line 40, in <module>
    from physicsnemo.datapipes.readers.base import Reader
  File "D:\PycharmProjects\physicsnemo\physicsnemo\datapipes\readers\__init__.py", line 29, in <module>
    from physicsnemo.datapipes.readers.mesh import DomainMeshReader, MeshReader
  File "D:\PycharmProjects\physicsnemo\physicsnemo\datapipes\readers\mesh.py", line 37, in <module>
    from physicsnemo.mesh import DomainMesh, Mesh
  File "D:\PycharmProjects\physicsnemo\physicsnemo\mesh\__init__.py", line 17, in <module>
    from physicsnemo.mesh.domain_mesh import DomainMesh
  File "D:\PycharmProjects\physicsnemo\physicsnemo\mesh\domain_mesh.py", line 28, in <module>
    from physicsnemo.mesh.mesh import Mesh, _requested_float_dtype
  File "D:\PycharmProjects\physicsnemo\physicsnemo\mesh\mesh.py", line 52, in <module>
    from physicsnemo.mesh.remeshing import remesh
  File "D:\PycharmProjects\physicsnemo\physicsnemo\mesh\remeshing\__init__.py", line 45, in <module>
    from physicsnemo.mesh.remeshing._partition import CellPartition, partition_cells
  File "D:\PycharmProjects\physicsnemo\physicsnemo\mesh\remeshing\_partition.py", line 42, in <module>
    from physicsnemo.nn.functional.neighbors import knn
  File "D:\PycharmProjects\physicsnemo\physicsnemo\nn\__init__.py", line 20, in <module>
    from .module.activations import (
  File "D:\PycharmProjects\physicsnemo\physicsnemo\nn\module\__init__.py", line 28, in <module>
    from .ball_query import BQWarp
  File "D:\PycharmProjects\physicsnemo\physicsnemo\nn\module\ball_query.py", line 29, in <module>
    from physicsnemo.nn.functional import radius_search
  File "D:\PycharmProjects\physicsnemo\physicsnemo\nn\functional\__init__.py", line 17, in <module>
    from .derivatives import (
  File "D:\PycharmProjects\physicsnemo\physicsnemo\nn\functional\derivatives\__init__.py", line 17, in <module>
    from .mesh_green_gauss_gradient import MeshGreenGaussGradient, mesh_green_gauss_gradient
  File "D:\PycharmProjects\physicsnemo\physicsnemo\nn\functional\derivatives\mesh_green_gauss_gradient\__init__.py", line 17, in <module>
    from .mesh_green_gauss_gradient import MeshGreenGaussGradient, mesh_green_gauss_gradient
  File "D:\PycharmProjects\physicsnemo\physicsnemo\nn\functional\derivatives\mesh_green_gauss_gradient\mesh_green_gauss_gradient.py", line 24, in <module>
    from ._warp_impl import mesh_green_gauss_gradient_warp
  File "D:\PycharmProjects\physicsnemo\physicsnemo\nn\functional\derivatives\mesh_green_gauss_gradient\_warp_impl.py", line 607, in <module>
    @torch.library.custom_op(
     ^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\PycharmProjects\physicsnemo\venv\Lib\site-packages\torch\_library\custom_ops.py", line 119, in inner
    schema_str = torch._custom_op.impl.infer_schema(fn, mutates_args)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\PycharmProjects\physicsnemo\venv\Lib\site-packages\torch\_library\infer_schema.py", line 42, in infer_schema
    error_fn(
  File "D:\PycharmProjects\physicsnemo\venv\Lib\site-packages\torch\_library\infer_schema.py", line 21, in error_fn
    raise ValueError(
ValueError: infer_schema(func): Parameter points has unsupported type torch.Tensor. The valid types are: dict_keys([<class 'torch.Tensor'>, typing.Optional[torch.Tensor], typing.Sequence[torch.Tensor], typing.List[torch.Tensor], typing.Sequence[typing.Optional[torch.Tensor]], typing.List[typing.Optional[torch.Tensor]], <class 'int'>, typing.Optional[int], typing.Sequence[int], typing.List[int], typing.Optional[typing.Sequence[int]], typing.Optional[typing.List[int]], <class 'float'>, typing.Optional[float], typing.Sequence[float], typing.List[float], typing.Optional[typing.Sequence[float]], typing.Optional[typing.List[float]], <class 'bool'>, typing.Optional[bool], typing.Sequence[bool], typing.List[bool], typing.Optional[typing.Sequence[bool]], typing.Optional[typing.List[bool]], <class 'str'>, typing.Optional[str], typing.Union[int, float, bool], typing.Union[int, float, bool, NoneType], typing.Sequence[typing.Union[int, float, bool]], typing.List[typing.Union[int, float, bool]], <class 'torch.dtype'>, typing.Optional[torch.dtype], <class 'torch.device'>, typing.Optional[torch.device]]). Got func with signature (points: 'torch.Tensor', cells: 'torch.Tensor', neighbors: 'torch.Tensor', values: 'torch.Tensor') -> 'torch.Tensor')

Process finished with exit code 1
Environment details
Dominant language
Python
Stars
3.4k
Forks
814
Avg merge
3d 14h
Merged PRs (30d)
46

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from NVIDIA/physicsnemo

All issues in NVIDIA/physicsnemo

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.