Torch.cuda.is_available return True, but torch.randn(1).cuda() stuck

I’m new to cuda and pytorch, I ran a virtual machine via kubevirt and passed through a H100 for it. I installed nvidia-driver, anaconda and pytorch in the vm. Then I got this issue, as the screenshot. How can I fix this?

apt install nvidia-driver-550
conda install -y pytorch torchvision torchaudio pytorch-cuda=12.4 -c pytorch -c nvidia

https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/latest/gpu-operator-kubevirt.html#