I'm following the example here for setting up a GPU enabled cluster in Google Kubernetes Engine. https://engineering.bitnami.com/articles/using-gpus-with-kubernetes.html
But what does step 3 mean?
Step 3: Configure Each Kubelet To Use The NVIDIA GPU
How is that different to running the steps on the Node? How do I access the kubelet?
The kubelet is not accessible on GKE. Instead, follow this guide provided by Google on how to configure GPUs on GKE.