fix: remove vllm embedding

2025-11-27 01:12:57 +01:00
parent 4d7c811a46
commit 90fa8a073c
3 changed files with 5 additions and 25 deletions
--- a/services/vllm/config_llama.yaml
+++ b/services/vllm/config_llama.yaml
@@ -2,5 +2,7 @@ model: meta-llama/Llama-3.1-8B-Instruct
 host: "0.0.0.0"
 port: 8001
 uvicorn-log-level: "info"
-gpu-memory-utilization: 0.9
-max-model-len: 20480
+gpu-memory-utilization: 0.95
+max-model-len: 32768
+dtype: auto
+enforce-eager: false