torch.cuda.is_available() is False.

[RIAHI-MED]

I'm trying to run inference using Mask3D on GPU, but I get the following error:

raise RuntimeError('Attempting to deserialize object on a CUDA '
RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

However, I am using a GPU. Any idea how to fix this?

[Vinit2244]

Make sure you are using PyTorch with GPU support (check using below given command):

python -c "import torch; print(torch.__version__, torch.version.cuda, torch.cuda.is_available())"
# should show something like: 1.12.1+cu113 11.3 True

Note that the version number for cuda and pytorch in the output of the above command might be different based on what you are using but it should not show None and False

Before downloading PyTorch with GPU support make sure your environment variables are pointing to correct cuda (system's cuda or environment cuda as per your usage)

# Point the env vars to environment cuda
export CUDA_HOME=$CONDA_PREFIX
export PATH=$CUDA_HOME/bin:$PATH
export LD_LIBRARY_PATH=$CUDA_HOME/lib:$CUDA_HOME/lib64:$LD_LIBRARY_PATH

# Point the env vars to system's cuda
export CUDA_HOME=/usr/local/cuda
export PATH=$CUDA_HOME/bin:$PATH
export LD_LIBRARY_PATH=$CUDA_HOME/lib64:$CUDA_HOME/lib:$LD_LIBRARY_PATH

Download PyTorch with GPU support using:

pip3 install torch==1.12.1+cu113 torchvision==0.13.1+cu113 --extra-index-url https://download.pytorch.org/whl/cu113