test

Signed-off-by: Ylang Tsou <ylangt@google.com>

Author: Ylang Tsou

docker/Dockerfile.pypi

@@ -19,19 +19,19 @@ RUN export TPU_INFERENCE_VERSION=$(pip index versions tpu-inference --pre 2>/dev
     echo "Installing tpu-inference version: $TPU_INFERENCE_VERSION"
 
 # Clone vLLM
-WORKDIR /workspace/vllm
+WORKDIR /workspace/vllm_build
 ARG VLLM_REPO=https://github.com/vllm-project/vllm.git
-RUN git clone $VLLM_REPO /workspace/vllm
+RUN git clone $VLLM_REPO /workspace/vllm_build
 RUN pip install build
 RUN export TPU_INFERENCE_VERSION=$(cat /tmp/tpu_inference_version) && \
-    echo "Using tpu-inference version: $TPU_INFERENCE_VERSION" && \
+    echo "Using tpu-inference version: ${TPU_INFERENCE_VERSION}" && \
     sed -i "s/^tpu-inference==.*/tpu-inference==${TPU_INFERENCE_VERSION}/" requirements/tpu.txt && \
     echo "Starting build.sh" && \
     bash tools/vllm-tpu/build.sh ${TPU_INFERENCE_VERSION}
 
 # Install vllm-tpu from wheel
 RUN echo "pip installing vllm-tpu" && \
-    pip install dist/*.whl
+    pip install --no-cache-dir dist/*.whl
 # RUN export TPU_INFERENCE_VERSION=$(cat /tmp/tpu_inference_version) && \
   #   export PIP_VERSION=$(echo "$TPU_INFERENCE_VERSION" | sed 's/^v//') && \
     # echo "Installing vllm-tpu version: $PIP_VERSION" && \
@@ -46,4 +46,8 @@ RUN python3 -m pip install --no-cache-dir \
     pytest-cov \
     tblib
 
+# Preventing shadowing
+# WORKDIR /workspace
+# RUN mv /workspace/vllm /workspace/vllm_backup
+
 CMD ["/bin/bash"]