cpu version

Author: wsmlby

server/Dockerfile.cpu

@@ -73,13 +73,13 @@ FROM base AS vllm-build
 
 ARG GIT_REPO_CHECK=0
 # Support for building with non-AVX512 vLLM: docker build --build-arg VLLM_CPU_DISABLE_AVX512="true" ...
-ARG VLLM_CPU_DISABLE_AVX512=0
+ARG VLLM_CPU_DISABLE_AVX512=true
 ENV VLLM_CPU_DISABLE_AVX512=${VLLM_CPU_DISABLE_AVX512}
 # Support for building with AVX512BF16 ISA: docker build --build-arg VLLM_CPU_AVX512BF16="true" ...
-ARG VLLM_CPU_AVX512BF16=0
+ARG VLLM_CPU_AVX512BF16=false
 ENV VLLM_CPU_AVX512BF16=${VLLM_CPU_AVX512BF16}
 # Support for building with AVX512VNNI ISA: docker build --build-arg VLLM_CPU_AVX512VNNI="true" ...
-ARG VLLM_CPU_AVX512VNNI=0
+ARG VLLM_CPU_AVX512VNNI=false
 ENV VLLM_CPU_AVX512VNNI=${VLLM_CPU_AVX512VNNI}
 
 WORKDIR /workspace/vllm

server/Dockerfile.cpu.app

@@ -23,4 +23,4 @@ ARG HOML_SERVER_VERSION=dev
 ENV HOML_SERVER_VERSION=$HOML_SERVER_VERSION
 
 # Start the server directly from main.py
-CMD ["python3", "-u", "main.py"]
+ENTRYPOINT ["python3", "-u", "main.py"]