Pytorch and torch cpu version

i don’t have NVIDIA graphic card. Nad i want to work on PyTorch geometric library. for installation of PyTorch geometric its written that torch.cuda.is_available() should return true. But its returning false in my case.
Can i work with Cuda installation of torch and cpu version of pytorch geometric?

can’t you just install the cpu packages for pytorch geometric alongside a cpu version of PyTorch? Are you getting some error?

pip install torch-scatter==latest+cpu torch-sparse==latest+cpu
pip install torch-cluster (optional)
pip install torch-spline-conv (optional)
python setup.py install or pip install torch-geometric

I have installed this but i am getting errors by writing

these two lines
import torch
from torch_geometric.data import Data

previsously i installed pytorch using following command:
conda install -c pytorch pytorch

now when i tried the command you mentioned it gives following error

Just want to share version of pytorch installed

Try with

pip install torch-scatter==latest+cpu torch-sparse==latest+cpu -f https://pytorch-geometric.com/whl/torch-1.4.0.html
1 Like