From 4d7c811a46ec9a0ee1274a6fba835907e21e30ec Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sebastian=20Kr=C3=BCger?= Date: Thu, 27 Nov 2025 00:57:14 +0100 Subject: [PATCH] fix: vllm gpu utilization 2 --- services/vllm/config_bge.yaml | 2 +- services/vllm/config_llama.yaml | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/services/vllm/config_bge.yaml b/services/vllm/config_bge.yaml index 3b45c3d..5d61763 100644 --- a/services/vllm/config_bge.yaml +++ b/services/vllm/config_bge.yaml @@ -2,4 +2,4 @@ model: BAAI/bge-large-en-v1.5 host: "0.0.0.0" port: 8002 uvicorn-log-level: "info" -gpu-memory-utilization: 0.25 +gpu-memory-utilization: 0.1 diff --git a/services/vllm/config_llama.yaml b/services/vllm/config_llama.yaml index 7c824ef..c4f1221 100644 --- a/services/vllm/config_llama.yaml +++ b/services/vllm/config_llama.yaml @@ -2,5 +2,5 @@ model: meta-llama/Llama-3.1-8B-Instruct host: "0.0.0.0" port: 8001 uvicorn-log-level: "info" -gpu-memory-utilization: 0.75 +gpu-memory-utilization: 0.9 max-model-len: 20480