compose-vLLM.yaml aktualisiert

gpu-memory-utilization
2026-07-10 17:59:23 +00:00 · 2026-07-10 17:59:23 +00:00 · 4edd0f7a2e
parent 9254499bc6
commit 4edd0f7a2e
1 changed files with 1 additions and 1 deletions
--- a/compose-vLLM.yaml
+++ b/compose-vLLM.yaml
@ -28,7 +28,7 @@ services:
      --model Qwen/Qwen2.5-14B-Instruct-AWQ
      --host 0.0.0.0
      --port 8000
-      --gpu-memory-utilization 0.90
+      --gpu-memory-utilization 0.65
      --max-model-len 16384
      --enable-auto-tool-choice
      --tool-call-parser hermes