gpu memory limitation on vllm container

Author: Ahmath-Gadji

docker-compose.yaml

@@ -44,6 +44,7 @@ x-vllm: &vllm_template
   command: >
     --model ${EMBEDDER_MODEL_NAME:-jinaai/jina-embeddings-v3}
     --trust-remote-code
+    --gpu_memory_utilization 0.3
   ports:
     - ${VLLM_PORT:-8000}:8000