OSError: dlopen.. when import torch_scatter
mlv42 opened this issue · comments
I get the error :
OSError: dlopen(/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/torch_scatter/_version_cpu.so, 0x0006): Symbol not found: __ZN3c1017RegisterOperatorsD1Ev
Referenced from: <85F5D618-9A81-323E-B8CF-4EA588421020> /Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/torch_scatter/_version_cpu.so
Expected in: /Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/torch/lib/libtorch_cpu.dylib
when I import torch_scatter.
I have:
- python 3.9.5
- torch 2.2.1
and I have installed torch_scatter using this line of code:
pip install torch-scatter -f https://data.pyg.org/whl/torch-2.2.2+cpu.html
on a Mac.
How can I fix it? I am sorry but I am not familiar with how these kind of issues work.
How did you install PyTorch? Are you using the conda-forge package?
I used pip install torch and no, I am not using conda-forge . Should I?
No, that sounds good then. Is this ARM or Intel mac?
It's an Intel Mac
Got it. What happens if you simply install via
pip uninstall torch-scatter
pip install torch-scatter
I get the error:
note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for torch-scatter
Running setup.py clean for torch-scatter
Failed to build torch-scatter
ERROR: Could not build wheels for torch-scatter, which is required to install pyproject.toml-based projects
Note: you may need to restart the kernel to use updated packages.
when I run pip install torch-scatter after I run pip uninstall torch-scatter.
Sorry for late reply:
Can you run with
pip install --verbose torch-scatter
This should give a more descriptive error message.