A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 17, 05:30 UTC
Update Dockerfile to support Mixtral (#2027)
Simon Mo committed
Dec 11, 2023 at 11:59 UTC
eb17212858dce5c73b8fe3e8c548861ea18f1f43
1 file changed
+2
-2
Dockerfile
+2
-2
@@ -73,9 +73,9 @@ ENTRYPOINT ["python3", "-m", "vllm.entrypoints.api_server"]
73
74
# openai api server alternative
75
FROM vllm-base AS vllm-openai
76
-# install additional dependencies for openai api server
76
+# install additional dependencies for openai api server, and mixtral
77
RUN --mount=type=cache,target=/root/.cache/pip \
78
- pip install accelerate fschat
78
+ pip install accelerate megablocks
79
80
COPY --from=build /workspace/vllm/*.so /workspace/vllm/
81
COPY vllm vllm