diff --git a/Dockerfile b/Dockerfile index 28f2e58..09ab916 100644 --- a/Dockerfile +++ b/Dockerfile @@ -64,11 +64,8 @@ RUN python -m pip install \ WORKDIR /opt ENV FLASH_ATTENTION_TRITON_AMD_ENABLE="TRUE" -RUN git clone https://github.com/ROCm/flash-attention.git &&\ - cd flash-attention &&\ - git checkout main_perf &&\ - python setup.py install && \ - cd /opt && rm -rf /opt/flash-attention +RUN pip install flash-attn --no-build-isolation \ + --index-url https://pypi.tuna.tsinghua.edu.cn/simple # 6. Clone vLLM RUN git clone https://github.com/vllm-project/vllm.git /opt/vllm