HennyJie
commented
9 months ago Hi, This issue may be due to a version mismatch between torch and torch-scatter. To resolve it, save the following command to a script. After saving, grant it execution permissions using chmod +x and then execute the script.
#!/bin/bash
TORCH=$(python -c "import torch; print(torch.__version__)")
CUDA=$(python -c "import torch; version = torch.version.cuda.replace('.', ''); print(f'cu{version}')")
pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html --upgrade
pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html --upgrade
pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html --upgrade
pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html --upgrade
pip install torch-geometric --upgradeAnother suggestion is to install the packages listed in requirements.txt one by one. Start by ensuring that torch installs successfully. Once that's done, continue with the subsequent commands (especially 'torch-scatter' and 'torch-sparse', those are the packages that depend on torch). Thanks.
Hi,
If I try to install BrainGB in a new conda environment (conda create -n braingb) on my machine (Ubuntu 20.04.3 LTS, python 3.8.10) with a GPU, via: "pip install BrainGB", I get the following output including errors:
If I instead try to install BrainGB via: "pip install -r requirements.txt" (Ubuntu 20.04.3 LTS, python 3.8.10) I get the following output:
If I instead try to create a conda env via: "conda create -n braingb anaconda" (Ubuntu 20.04.3 LTS, Python 3.8.16) then "pip install BrainGB", I get the following output with errors:
If I instead try to create a conda env via: "conda create -n braingb anaconda" (Ubuntu 20.04.3 LTS, Python 3.8.16) then "pip install -r requirements.txt", it runs almost entirely without error, but at the end the following error is returned: