Standardize config.

2026-05-23 19:51:33 +00:00
parent e2e3f730a5
commit 6b5325c6f3
87 changed files with 426 additions and 531 deletions
--- a/vllm/deployment.yaml
+++ b/vllm/deployment.yaml
@@ -22,7 +22,7 @@ spec:
        nvidia.com/gpu.product: "{{ .gpuProduct }}"
      containers:
        - name: vllm
-          image: "{{ .image }}"
+          image: vllm/vllm-openai:v0.5.4
          imagePullPolicy: IfNotPresent
          securityContext:
            allowPrivilegeEscalation: false
@@ -33,11 +33,9 @@ spec:
          args:
            - --model={{ .model }}
            - --max-model-len={{ .maxModelLen }}
-            - --tensor-parallel-size={{ .tensorParallelSize }}
+            - --tensor-parallel-size=1
            - --gpu-memory-utilization={{ .gpuMemoryUtilization }}
-            {{- if .apps.vllm.enforceEager }}
            - --enforce-eager=True
-            {{- end }}
          env:
            - name: VLLM_TORCH_DTYPE
              value: "auto"