pytorch-instance-for-colab-backend.sh

export IMAGE_FAMILY="pytorch-latest-gpu"
export ZONE="us-west1-b"
export INSTANCE_NAME="pytorch-colab-backend"
gcloud compute instances create $INSTANCE_NAME \
        --zone=$ZONE \
        --image-family=$IMAGE_FAMILY \
        --image-project=deeplearning-platform-release \
        --maintenance-policy=TERMINATE \
        --accelerator='type=nvidia-tesla-v100,count=1' \
        --metadata='install-nvidia-driver=True' \
        --preemptible