fix memory util

2026-05-08 09:42:50 -05:00
parent 2f922279ef
commit b70f8063a8
1 changed files with 1 additions and 1 deletions
--- a/clusters/k3s-dgx/nim-service/qwen.yaml
+++ b/clusters/k3s-dgx/nim-service/qwen.yaml
@@ -45,7 +45,7 @@ spec:
    - --served-model-name
    - Qwen/Qwen3.6-27B-FP8
    - --gpu-memory-utilization
-    - "0.90"
+    - "0.85"
    - --max-model-len
    - "256000"
    - --language-model-only