vllm 1

Dockerfile

@@ -27,4 +27,9 @@ RUN pip install vllm==0.7.2
 # breaking changes in master
 RUN pip install --upgrade git+https://github.com/huggingface/transformers.git@336dc69d63d56f232a183a3e7f52790429b871ef
 
-ENTRYPOINT ["sh", "-c", "python -m vllm.entrypoints.openai.api_server --model convergence-ai/proxy-lite --trust-remote-code --tokenizer-pool-size 10 --limit-mm-per-prompt image=2 --enable-auto-tool-choice --tool-call-parser hermes --port 7860"]
+RUN pip install ray[adag]
+RUN pip install cupy
+
+RUN export VLLM_USE_V1=1
+
+ENTRYPOINT ["sh", "-c", "python -m vllm.entrypoints.openai.api_server --model convergence-ai/proxy-lite --trust-remote-code --tokenizer-pool-size 20 --limit-mm-per-prompt image=1 --enable-auto-tool-choice --tool-call-parser hermes --port 7860"]