refactor: upgrade vLLM brain to uncensored Qwen 2.5 32B AWQ

apps/base/customer1/openclaw/vllm-gemma.yaml

@@ -23,16 +23,18 @@ spec:
           value: "present"
           effect: "NO_SCHEDULE"
       containers:
-        - name: vllm-gemma
+        - name: vllm-brain
           image: vllm/vllm-openai:latest
           command: ["python3", "-m", "vllm.entrypoints.openai.api_server"]
           args:
             - "--model"
-            - "cognitivecomputations/dolphin-2.8-gemma-2b-v0" # Uncensored/Distilled Dolphin-Gemma model
+            - "cognitivecomputations/dolphin-2.9.4-qwen2.5-32b-AWQ" # Uncensored 32B Reasoning Heavyweight (Quantized)
+            - "--quantization"
+            - "awq"
             - "--dtype"
-            - "bfloat16"
+            - "half"
             - "--max-model-len"
-            - "8192"
+            - "16384"
           ports:
             - containerPort: 8000
           resources: