Installing CUDA on p2.xlarge (Deep Learning AMI GPU PyTorch 2.0.0 (Ubuntu 20.04))

I am trying to install CUDA to run my python script on a GPU. However, even though I am following the guidelines on the nvidia webpage, the driver is never running after installation (nvidia-smi says it "failed to communicate with the driver", even though nvcc --version shows the installed driver).

p2.xlarge has a GPU, but there no driver installed.

Is there a step by step guide?