apiVersion: apps/v1
kind: Deployment
metadata:
  name: local-ai
  namespace: local-ai-ns
  labels:
    app: local-ai
spec:
  selector:
    matchLabels:
      app: local-ai
  replicas: 1
  template:
    metadata:
      labels:
        app: local-ai
      name: local-ai
    spec:
      containers:
        - args:
          - dolphin-2.5-mixtral-8x7b
          env:
          - name: DEBUG
            value: "true"
          name: local-ai
          image: quay.io/go-skynet/local-ai:v2.14.0-cublas-cuda12-ffmpeg
          imagePullPolicy: IfNotPresent
          resources:
            limits:
              nvidia.com/gpu: 2
          volumeMounts:
            - name: models-volume
              mountPath: /build/models
      volumes:
        - name: models-volume
          persistentVolumeClaim:
            claimName: models-pvc