Roll back vllm to 0.3.0

Author: hippothewild

vllm-online-serving/Dockerfile

@@ -1,4 +1,4 @@
-FROM quay.io/vessl-ai/python:3.10-r2
+FROM quay.io/vessl-ai/torch:2.2.0-cuda12.3-r3
 
 ENV PROMETHEUS_VERSION=2.49.1
 
@@ -10,12 +10,12 @@ RUN tar -xvf prometheus-$PROMETHEUS_VERSION.linux-amd64.tar.gz
 RUN mv prometheus-$PROMETHEUS_VERSION.linux-amd64 ./prometheus
 RUN rm prometheus-$PROMETHEUS_VERSION.linux-amd64.tar.gz
 
-# Copy prometheus.yml and grafana.json
+# Copy prometheus.yml
 COPY monitoring/prometheus.yml /app/prometheus/prometheus.yml
 
-# # Install dependencies
-# COPY requirements.txt /app/requirements.txt
-# RUN pip install -r /app/requirements.txt
+# Install dependencies
+COPY requirements.txt /app/requirements.txt
+RUN pip install -r /app/requirements.txt
 
-# # Entrypoint
-# ENTRYPOINT ["python", "-m", "api.py"]
+# Entrypoint
+ENTRYPOINT ["python", "-m", "api.py"]

vllm-online-serving/requirements.txt

@@ -1,3 +1,3 @@
 aioprometheus==23.12.0
 fastapi==0.109.2
-vllm @ git+https://github.com/vllm-project/vllm@c9b45adeeb0e5b2f597d1687e0b8f24167602395
+vllm==0.3.0