Skip to content
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

Jump to bottom

Cannot import torch-spline-conv when installed from pip wheel in Torch 1.9.0 and CUDA 11.1 #2744

Closed
4 tasks done
JonnyTran opened this issue Jun 17, 2021 · 17 comments
Closed
4 tasks done

Cannot import torch-spline-conv when installed from pip wheel in Torch 1.9.0 and CUDA 11.1 #2744

JonnyTran opened this issue Jun 17, 2021 · 17 comments

Comments

@JonnyTran
Copy link

JonnyTran commented Jun 17, 2021 •
edited

📚 Installation

I had been able install all pytorch-geometric related packages via pip prior to the new 1.7.1 release and it all works well. However, with the new release, pip install seems to skip "building the wheel" for torch-spline-conv, but still give a successful installation. When importing the torch-spline-conv, it then gives this error:

---> 46 from torch_geometric.nn import MetaPath2Vec

/opt/anaconda3/lib/python3.7/site-packages/torch_geometric/nn/init.py in
3 from .data_parallel import DataParallel
4 from .reshape import Reshape
----> 5 from .conv import * # noqa
6 from .norm import * # noqa
7 from .glob import * # noqa

/opt/anaconda3/lib/python3.7/site-packages/torch_geometric/nn/conv/init.py in
22 from .point_conv import PointConv
23 from .gmm_conv import GMMConv
---> 24 from .spline_conv import SplineConv
25 from .nn_conv import NNConv, ECConv
26 from .cg_conv import CGConv

/opt/anaconda3/lib/python3.7/site-packages/torch_geometric/nn/conv/spline_conv.py in
12
13 try:
---> 14 from torch_spline_conv import spline_basis, spline_weighting
15 except ImportError:
16 spline_basis = None

/opt/anaconda3/lib/python3.7/site-packages/torch_spline_conv/init.py in
10 for library in ['_version', '_basis', 'weighting']:
11 torch.ops.load_library(importlib.machinery.PathFinder().find_spec(
---> 12 f'{library}{suffix}', [osp.dirname(file)]).origin)
13
14 if torch.cuda.is_available(): # pragma: no cover

/opt/anaconda3/lib/python3.7/site-packages/torch/_ops.py in load_library(self, path)
102 # static (global) initialization code in order to register custom
103 # operators with the JIT.
--> 104 ctypes.CDLL(path)
105 self.loaded_libraries.add(path)
106

/opt/anaconda3/lib/python3.7/ctypes/init.py in init(self, name, mode, handle, use_errno, use_last_error)
362
363 if handle is None:
--> 364 self._handle = _dlopen(self._name, mode)
365 else:
366 self._handle = handle

OSError: /lib64/libm.so.6: version `GLIBC_2.27' not found (required by /opt/anaconda3/lib/python3.7/site-packages/torch_spline_conv/_basis_cuda.so)

Environment

  • OS: Centos 7
  • Python version: 3.7
  • PyTorch version: 1.9
  • CUDA/cuDNN version: 11.1
  • GCC version: 6.3.0
  • How did you try to install PyTorch Geometric and its extensions (wheel, source): wheel
  • Any other relevant information:

CUDA="cu111"
pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-1.9.0+${CUDA}.html
pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-1.9.0+${CUDA}.html
pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-1.9.0+${CUDA}.html
pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-1.9.0+${CUDA}.html

Checklist

  • I followed the installation guide.
  • I cannot find my error message in the FAQ.
  • I set up CUDA correctly and can compile CUDA code via nvcc.
  • I do have multiple CUDA versions on my machine.

Additional context
@JonnyTran JonnyTran changed the title Wheels for Torch 1.9.0 and CUDA 11.1 doesn't build Wheels doesn't build properly with pip for Torch 1.9.0 and CUDA 11.1 Jun 17, 2021
@rusty1s
Copy link
Member

rusty1s commented Jun 17, 2021 •
edited

Can you show me the output of pip install --verbose --no-index torch-scatter -f https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html after uninstalling all previous torch-scatter versions?

@JonnyTran
Copy link
Author

JonnyTran commented Jun 17, 2021 •
edited

I got this

pip install --verbose --no-index torch-scatter -f https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html > pip install --verbose --no-index torch-scatter -f https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html > Using pip 21.1.2 from /opt/anaconda3/lib/python3.7/site-packages/pip (python 3.7) > Non-user install because site-packages writeable > Ignoring indexes: https://pypi.org/simple > Created temporary directory: /tmp/pip-ephem-wheel-cache-4tm8ouxt > Created temporary directory: /tmp/pip-req-tracker-kb9fo3av > Initialized build tracking at /tmp/pip-req-tracker-kb9fo3av > Created build tracker: /tmp/pip-req-tracker-kb9fo3av > Entered build tracker: /tmp/pip-req-tracker-kb9fo3av > Created temporary directory: /tmp/pip-install-mxk9uotc > Looking in links: https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html > 1 location(s) to search for versions of torch-scatter: > * https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html > Fetching project page and analyzing links: https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html > Getting page https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html > Looking up "https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html" in the cache > Request header has "max_age" as 0, cache bypassed > Starting new HTTPS connection (1): pytorch-geometric.com:443 > https://pytorch-geometric.com:443 "GET /whl/torch-1.9.0+cu111.html HTTP/1.1" 200 4237 > Updating cache with response from "https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html" > Caching due to etag > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_cluster-1.5.9-cp36-cp36m-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_cluster-1.5.9-cp36-cp36m-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_cluster-1.5.9-cp37-cp37m-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_cluster-1.5.9-cp37-cp37m-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_cluster-1.5.9-cp38-cp38-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_cluster-1.5.9-cp38-cp38-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_cluster-1.5.9-cp39-cp39-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_cluster-1.5.9-cp39-cp39-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: none of the wheel's tags (cp36-cp36m-linux_x86_64) are compatible (run pip debug --verbose to show compatible tags): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp36-cp36m-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: none of the wheel's tags (cp36-cp36m-win_amd64) are compatible (run pip debug --verbose to show compatible tags): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp36-cp36m-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Found link https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp37-cp37m-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html), version: 2.0.7 > Skipping link: none of the wheel's tags (cp37-cp37m-win_amd64) are compatible (run pip debug --verbose to show compatible tags): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp37-cp37m-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: none of the wheel's tags (cp38-cp38-linux_x86_64) are compatible (run pip debug --verbose to show compatible tags): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp38-cp38-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: none of the wheel's tags (cp38-cp38-win_amd64) are compatible (run pip debug --verbose to show compatible tags): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp38-cp38-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: none of the wheel's tags (cp39-cp39-linux_x86_64) are compatible (run pip debug --verbose to show compatible tags): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp39-cp39-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: none of the wheel's tags (cp39-cp39-win_amd64) are compatible (run pip debug --verbose to show compatible tags): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp39-cp39-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_sparse-0.6.10-cp36-cp36m-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_sparse-0.6.10-cp36-cp36m-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_sparse-0.6.10-cp37-cp37m-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_sparse-0.6.10-cp37-cp37m-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_sparse-0.6.10-cp38-cp38-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_sparse-0.6.10-cp38-cp38-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_sparse-0.6.10-cp39-cp39-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_sparse-0.6.10-cp39-cp39-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_spline_conv-1.2.1-cp36-cp36m-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_spline_conv-1.2.1-cp36-cp36m-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_spline_conv-1.2.1-cp37-cp37m-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_spline_conv-1.2.1-cp37-cp37m-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_spline_conv-1.2.1-cp38-cp38-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_spline_conv-1.2.1-cp38-cp38-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_spline_conv-1.2.1-cp39-cp39-linux_x86_64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: wrong project name (not torch-scatter): https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_spline_conv-1.2.1-cp39-cp39-win_amd64.whl (from https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html) > Skipping link: unsupported archive format: .html: https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html > Given no hashes to check 1 links for project 'torch-scatter': discarding no candidates > Collecting torch-scatter > Created temporary directory: /tmp/pip-unpack-6wha6z09 > Looking up "https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp37-cp37m-linux_x86_64.whl" in the cache > No cache entry available > https://pytorch-geometric.com:443 "GET /whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp37-cp37m-linux_x86_64.whl HTTP/1.1" 200 2855533 > Downloading https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp37-cp37m-linux_x86_64.whl (2.9 MB) > |████████████████████████████████| 2.8 MB 6.9 MB/s eta 0:00:01 Updating cache with response from "https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp37-cp37m-linux_x86_64.whl" > Caching due to etag > |████████████████████████████████| 2.9 MB 6.9 MB/s > Added torch-scatter from https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp37-cp37m-linux_x86_64.whl to build tracker '/tmp/pip-req-tracker-kb9fo3av' > Removed torch-scatter from https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu111/torch_scatter-2.0.7-cp37-cp37m-linux_x86_64.whl from build tracker '/tmp/pip-req-tracker-kb9fo3av' > Created temporary directory: /tmp/pip-unpack-lfv6oes1 > Installing collected packages: torch-scatter > > Successfully installed torch-scatter-2.0.7 > Removed build tracker: '/tmp/pip-req-tracker-kb9fo3av'

Btw, I always uninstall the old packages, run pip cache purge, before installing the new packages.

@rusty1s
Copy link
Member

rusty1s commented Jun 17, 2021

Ok, this looks good. And you get a similar error when running import torch_scatter?

@JonnyTran
Copy link
Author

JonnyTran commented Jun 17, 2021 •
edited

I can actually import torch_scatter, torch_sparse, torch_cluster, and torch_geometric without error. It's only import torch_spline_conv that gives the error:

OSError: /lib64/libm.so.6: version `GLIBC_2.27' not found (required by /opt/anaconda3/lib/python3.7/site-packages/torch_spline_conv/_basis_cuda.so)

@rusty1s
Copy link
Member

rusty1s commented Jun 17, 2021

Ah I see, thanks for the clarification. And torch-spline-conv is installed via wheel (you can also ensure that via --no-index --verbose)?

@JonnyTran
Copy link
Author

JonnyTran commented Jun 17, 2021 •
edited

Yes, I did install torch-spline-conv with --no-index

pip install --verbose --no-index torch-spline-conv -f https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html

Using pip 21.1.2 from /opt/anaconda3/lib/python3.7/site-packages/pip (python 3.7)
Non-user install because site-packages writeable
Ignoring indexes: https://pypi.org/simple
Created temporary directory: /tmp/pip-ephem-wheel-cache-lvg9r3x1
Created temporary directory: /tmp/pip-req-tracker-1hk47nsj
Initialized build tracking at /tmp/pip-req-tracker-1hk47nsj
Created build tracker: /tmp/pip-req-tracker-1hk47nsj
Entered build tracker: /tmp/pip-req-tracker-1hk47nsj
Created temporary directory: /tmp/pip-install-slhem3fc
Looking in links: https://pytorch-geometric.com/whl/torch-1.9.0+cu111.html
1 location(s) to search for versions of torch-spline-conv:

Successfully installed torch-spline-conv-1.2.1
Removed build tracker: '/tmp/pip-req-tracker-1hk47nsj'

@rusty1s
Copy link
Member

rusty1s commented Jun 17, 2021 •
edited

Ok, I'm trying to track this issue down, but it's unclear to me why imports of other packages work fine, while it fails for torch-spline-conv. Do you need SplineConv in your work? Other-wise feel free to uninstall it as it is an optional dependency by now.

Other-wise, you can try multiple options:

You should also check what version of glibc you have on your CentOS 7 machine.

@JonnyTran JonnyTran changed the title Wheels doesn't build properly with pip for Torch 1.9.0 and CUDA 11.1 Cannot import torch-spline-conv in Torch 1.9.0 and CUDA 11.1 Jun 17, 2021
@JonnyTran
Copy link
Author

JonnyTran commented Jun 17, 2021 •
edited

Yes, I can confirm all other packages work fine in my code when I uninstalled torch-spline-conv.

Actually when I build pytorch_spline_conv from source, I can import torch_spline_conv without errors.

Thanks very much for your help!

@JonnyTran JonnyTran changed the title Cannot import torch-spline-conv in Torch 1.9.0 and CUDA 11.1 Cannot import torch-spline-conv when installed from pip wheel in Torch 1.9.0 and CUDA 11.1 Jun 17, 2021
@saiden89
Copy link
Contributor

I'd like to reopen this issue since building from source doesn't fix the underlying problem of the wheels. It doesn't seem to be a CUDA version specific issue, since I've experienced the same with CUDA 10.2. Moreover, I've tested it with several python versions, from 3.7 to 3.9, but without success.

Reproducibility

Pytorch and dependencies installed with conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch

torch-spline-conv installed with:

❯ pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-1.9.0+cu102.html

Looking in links: https://pytorch-geometric.com/whl/torch-1.9.0+cu102.html
Collecting torch-spline-conv
  Downloading https://pytorch-geometric.com/whl/torch-1.9.0%2Bcu102/torch_spline_conv-1.2.1-cp39-cp39-linux_x86_64.whl (368 kB)
     |████████████████████████████████| 368 kB 29.4 MB/s 
Installing collected packages: torch-spline-conv
Successfully installed torch-spline-conv-1.2.1

Python output

❯ python
Python 3.9.5 | packaged by conda-forge | (default, Jun 19 2021, 00:32:32) 
[GCC 9.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch_spline_conv
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "envs/test39/lib/python3.9/site-packages/torch_spline_conv/__init__.py", line 11, in <module>
    torch.ops.load_library(importlib.machinery.PathFinder().find_spec(
  File "envs/test39/lib/python3.9/site-packages/torch/_ops.py", line 104, in load_library
    ctypes.CDLL(path)
  File "envs/test39/lib/python3.9/ctypes/__init__.py", line 382, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: /lib64/libm.so.6: version `GLIBC_2.27' not found (required by envs/test39/lib/python3.9/site-packages/torch_spline_conv/_basis_cuda.so)

The bug only shows up in a CUDA capable machine. Importing the module on a machine without CUDA support works fine.
I'll be happy to report additional details as well as submit this issue on the official torch-spline-conv repository.

@JonnyTran JonnyTran reopened this Jun 22, 2021
@rusty1s
Copy link
Member

rusty1s commented Jun 23, 2021

Ok, please submit this issue to torch-spline-conv, although I currently do not know how to fix it. Looks like a problem with PyTorch to me.

You can also check which CUDA file produces the output by checking the library value, see here.

@MrShouxingMa
Copy link

@JonnyTran @saiden89
Well, I also have the same problem that is " version `GLIBC_2.27' not found ".
Environment:
Python version: 3.6
PyTorch version: 1.9
CUDA/cuDNN version: 10.2
But I solve the problem by using the lower version.
Environment:
Python version: 3.6
PyTorch version: 1.8.0
CUDA/cuDNN version: 10.2
Maybe you can try it ! Best wish!

@saiden89 saiden89 mentioned this issue Jun 23, 2021
Closed
@saiden89
Copy link
Contributor

Issue submitted in the appropriate repository.

@rusty1s
Copy link
Member

rusty1s commented Jun 24, 2021

I updated the wheels for torch-spline-conv. Can you please try again and report if this fixes your issues?

@saiden89
Copy link
Contributor

Unfortunately, it doesn't... sorry. I've also created a new conda env from scratch and pip cache purge 😢

@rusty1s
Copy link
Member

rusty1s commented Jun 25, 2021

I'm closing this for now, see rusty1s/pytorch_spline_conv#22. Please feel free to re-open if you think this issue is not resolved yet.

@rusty1s rusty1s closed this as completed Jun 25, 2021
@mariuaas
Copy link

mariuaas commented Apr 5, 2022

FYI, it seems that this issue persists with torch 1.11.0, however uninstalling torch-spline-conv solves the problem.

@narges-tk
Copy link

The issue persists! "pip install torch-spline-conv -f https://data.pyg.org/whl/torch-2.0.0+${CUDA}.html" worked for me!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
6 participants
@JonnyTran @rusty1s @mariuaas @MrShouxingMa @saiden89 @narges-tk