evalbench_service/k8s/vertical-autoscale.yaml

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: evalbench-eval-server-deploy
  namespace: evalbench-namespace
spec:
  targetRef:
    kind: Deployment
    name: evalbench-eval-server-deploy
    apiVersion: apps/v1
  updatePolicy:
    updateMode: Auto
  resourcePolicy:
    containerPolicies:
      - containerName: "*"
        mode: Auto
        controlledResources:
          - cpu
          - memory
        minAllowed:
          memory: 32Gi
          cpu: "2.2"
        maxAllowed: {}