gpu offload in llama.cpp

apps/llama/configmap.yaml

@@ -10,4 +10,5 @@ data:
         cmd: |
           /app/llama-server
           -hf unsloth/DeepSeek-R1-0528-Qwen3-8B-GGUF:Q4_K_M
+          -ngl 37
           --port ${PORT}

apps/llama/deployment.yaml

@@ -18,6 +18,8 @@ spec:
         - name: llama-swap
           image: ghcr.io/mostlygeek/llama-swap:v139-vulkan-b5957
           imagePullPolicy: IfNotPresent
+          command:
+            - /app/llama-swap
           args:
             - --config=/config/config.yaml
             - --watch-config