Error when importing edsnlp due to issue with numpy (possible binary incompatibility)

[paul-bssr]

Error when importing edsnlp due to issue with numpy (possible binary incompatibility)

Description

When importing edsnlp (v0.13.0), I get the following error message :

ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

The error seems to be linked to numpy version, I found a work around by updating numpy to version 1.21.6

Entire error message

ValueError                                Traceback (most recent call last)
/tmp/ipykernel_4077/213387074.py in <module>
----> 1 import edsnlp

~/.user_conda/miniconda/envs/cirrhospit_paul/lib/python3.7/site-packages/edsnlp/__init__.py in <module>
      7 import importlib.util
      8 from pathlib import Path
----> 9 from spacy import pipeline as _spacy_pipeline  # noqa: F401
     10 from . import extensions
     11 from . import patch_spacy

~/.user_conda/miniconda/envs/cirrhospit_paul/lib/python3.7/site-packages/spacy/__init__.py in <module>
      4
      5 # set library-specific custom warning handling before doing anything else
----> 6 from .errors import setup_default_warnings
      7
      8 setup_default_warnings()  # noqa: E402

~/.user_conda/miniconda/envs/cirrhospit_paul/lib/python3.7/site-packages/spacy/errors.py in <module>
      1 import warnings
      2
----> 3 from .compat import Literal
      4
      5

~/.user_conda/miniconda/envs/cirrhospit_paul/lib/python3.7/site-packages/spacy/compat.py in <module>
     37     from catalogue import _importlib_metadata as importlib_metadata  # type: ignore[no-redef]    # noqa: F401
     38
---> 39 from thinc.api import Optimizer  # noqa: F401
     40
     41 pickle = pickle

~/.user_conda/miniconda/envs/cirrhospit_paul/lib/python3.7/site-packages/thinc/api.py in <module>
----> 1 from .backends import (
      2     CupyOps,
      3     MPSOps,
      4     NumpyOps,
      5     Ops,

~/.user_conda/miniconda/envs/cirrhospit_paul/lib/python3.7/site-packages/thinc/backends/__init__.py in <module>
     15 from ._cupy_allocators import cupy_pytorch_allocator, cupy_tensorflow_allocator
     16 from ._param_server import ParamServer
---> 17 from .cupy_ops import CupyOps
     18 from .mps_ops import MPSOps
     19 from .numpy_ops import NumpyOps

~/.user_conda/miniconda/envs/cirrhospit_paul/lib/python3.7/site-packages/thinc/backends/cupy_ops.py in <module>
     14 )
     15 from . import _custom_kernels
---> 16 from .numpy_ops import NumpyOps
     17 from .ops import Ops
     18

~/.user_conda/miniconda/envs/cirrhospit_paul/lib/python3.7/site-packages/thinc/backends/numpy_ops.pyx in init thinc.backends.numpy_ops()

ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

How to reproduce the bug

import edsnlp

Your Environment

• Python Version Used: 3.7.7
• spaCy Version Used: 3.7.6
• EDS-NLP Version Used:0.13.0
• numpy: 1.19.5
• thinc: 8.2.5