Closed this issue 5 years ago · 1 comments
Figure out why torch.cuda.is_available() returns False when running via a CUDA base container on a Kubernetes node that has a NVIDIA P100 available.
torch.cuda.is_available()
False