diff --git a/other/vllm/Docker/Dockerfile.base b/other/vllm/Docker/Dockerfile.base
index 13da536..28b35ea 100644
--- a/other/vllm/Docker/Dockerfile.base
+++ b/other/vllm/Docker/Dockerfile.base
@@ -33,7 +33,7 @@ RUN /venv/bin/pip3 install torch==2.0.1 --index-url https://download.pytorch.org
 # Don't build VLLM because we don't do that on the inference server. Just install from pip.
 # RUN /venv/bin/pip install git+https://github.com/vllm-project/vllm
 
-RUN pip install vllm
+RUN /venv/bin/pip install vllm
 
 FROM nvidia/cuda:11.8.0-base-ubuntu22.04 as base