It seems as though GPU drivers do not seem to work on the Deep Learning AMI (Ubuntu 180.04) g4dn.xlarge instances I have set up for the past few hours. nvidia-smi gives output:
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.
and torch.cuda.is_available() returns false.
Does anyone know why this is happening?