Original have torch 1.5.0 and when install PyTorch Geometric, it automatically downgraded to torch 1.4.0

Original have torch 1.5.0 and when install PyTorch Geometric, it automatically downgraded to torch 1.4.0, and also I installed it successfully, but cannot import.

Original package version:

torch-1.5.0+cu102
torchvision-0.6.0+cu102

After installing PyTorch Geometric (pip install torch-geometric):

torch 1.4.0
torch-cluster 1.5.4
torch-scatter 2.0.4
torch-sparse 0.6.1
torch-spline-conv 1.2.0
torchvision-0.5.0

After installing PyTorch Geometric, the torch version automatically downgraded to torch 1.4.0

The import error message is as follow:

Traceback (most recent call last):
  File "/home/xdwang/pytorch/lib/python3.7/site-packages/torch_sparse/__init__.py", line 15, in <module>
    library, [osp.dirname(__file__)]).origin)
  File "/home/xdwang/pytorch/lib/python3.7/site-packages/torch/_ops.py", line 106, in load_library
    ctypes.CDLL(path)
  File "/cvmfs/soft.computecanada.ca/easybuild/software/2017/Core/python/3.7.4/lib/python3.7/ctypes/__init__.py", line 364, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: libcusparse.so.10.0: cannot open shared object file: No such file or directory

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/xdwang/pytorch/lib/python3.7/site-packages/torch_geometric/__init__.py", line 2, in <module>
    import torch_geometric.nn
  File "/home/xdwang/pytorch/lib/python3.7/site-packages/torch_geometric/nn/__init__.py", line 2, in <module>
    from .data_parallel import DataParallel
  File "/home/xdwang/pytorch/lib/python3.7/site-packages/torch_geometric/nn/data_parallel.py", line 5, in <module>
    from torch_geometric.data import Batch
  File "/home/xdwang/pytorch/lib/python3.7/site-packages/torch_geometric/data/__init__.py", line 1, in <module>
    from .data import Data
  File "/home/xdwang/pytorch/lib/python3.7/site-packages/torch_geometric/data/data.py", line 7, in <module>
    from torch_sparse import coalesce
  File "/home/xdwang/pytorch/lib/python3.7/site-packages/torch_sparse/__init__.py", line 23, in <module>
    raise OSError(e)
OSError: libcusparse.so.10.0: cannot open shared object file: No such file or directory

Environment

• linux
• Python version: 3.7
• PyTorch version: 1.5.0
• CUDA/cuDNN version: 10.2
• How did you try to install PyTorch Geometric and its extensions (pip, source): pip

Checklist

• [YES ] I followed the installation guide.
• [YES ] I cannot find my error message in the FAQ.
• I set up CUDA correctly and can compile CUDA code via nvcc. I am using a computing cluster (Compute Canada) and CUDA should be set up properly
• [NO ] I have cloned the repository and tried a manual installation from source.
• I do have multiple CUDA versions on my machine. Not sure, as I cannot access to the CUDA PATH.

[CharlesGaydon]

Adding my two cents: I had pytorch==1.8.0 installed, and running conda install pyg -c pyg -c conda-forge would downgrade it to pytorch=1.10.0. My solution was to choose a specific build from the conda list of files for pyg package, and to run something like: mamba install pyg=2.0.2=py39_torch_1.9.0_cu111 -c pyg -c conda-forge resulting in

The following packages will be UPDATED:
pytorch            pytorch::pytorch-1.8.0-py3.9_cuda11.1~ --> conda-forge::pytorch-1.9.1-cuda111py39hb4a4491_3

[yulonglin]

Ok sure, yes I parsed the incompatibility error message from installing PyG and deleted the corresponding packages and it works fine now.