I installed Pytorch
and cudatoolkit
in an Anaconda environment via the instructions provided in the official documentation here. However the output to torch.cuda.is_available()
is still False. However, Pytorch seems to be detecting CUDA correctly as it outputs
the correct version (11.8) when I prompt python -m torch.cuda.version
.
The Python version installed in the environment is 3.10.11, which I suspect to be the cause of this issue.