set env

clusters/k3s-dgx/nim-service/qwen.yaml

@@ -26,12 +26,25 @@ metadata:
   name: qwen3-32b-instruct
   namespace: nim-service
 spec:
-  # args:
-  #   - --enable-auto-tool-choice
-  #   - --gpu-memory-utilization
-  #   - "0.90"
-  #   - --tool-call-parser
-  #   - hermes
+  env:
+    - name: GPU_MEMORY_UTILIZATION
+      value: "0.90"
+    - name: NIM_MAX_NUM_SEQS
+      value: "32"
+    - name: NIM_MAX_NUM_BATCHED_TOKENS
+      value: "16384"
+    - name: ENABLE_AUTO_TOOL_CHOICE
+      value: "true"
+    - name: TOOL_CALL_PARSER
+      value: "hermes"
+    - name: ENABLE_PREFIX_CACHING
+      value: "true"
+    - name: TRUNCATION_SIDE
+      value: "left"
+    - name: VLLM_LOGGING_LEVEL
+      value: "INFO"
+    - name: VLLM_ALLOW_LONG_MAX_MODEL_LEN
+      value: "true"
   image:
     repository: nvcr.io/nim/qwen/qwen3-32b-dgx-spark
     tag: "1.1.0-variant"