HealthcareGraphRAG / Dockerfile
minhthien's picture
fix: pre-download GGUF and embedding models at build time to eliminate cold-start delay
24d38de
Raw
History Blame Contribute Delete
1.08 kB
FROM python:3.11-slim
RUN apt-get update && apt-get install -y \
git git-lfs cmake build-essential \
libsm6 libxext6 libgl1 \
&& rm -rf /var/lib/apt/lists/* \
&& git lfs install
WORKDIR /app
ENV CMAKE_ARGS="-DGGML_BLAS=OFF -DGGML_METAL=OFF -DGGML_OPENMP=OFF -DGGML_NATIVE=OFF -DLLAMA_BUILD_TESTS=OFF -DLLAMA_BUILD_EXAMPLES=OFF -DLLAMA_BUILD_SERVER=OFF"
ENV CMAKE_BUILD_PARALLEL_LEVEL=4
ENV HF_HUB_CACHE=/app/model
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . .
# Pre-download GGUF model (~2.5 GB) at build time
RUN python -c "\
from huggingface_hub import hf_hub_download; \
hf_hub_download(repo_id='Jackrong/Qwen3.5-4B-Neo-GGUF', filename='Qwen3.5-4B.Q4_K_S.gguf', cache_dir='/app/model')"
# Pre-download embedding model at build time
RUN python -c "\
from huggingface_hub import snapshot_download; \
snapshot_download('BAAI/bge-small-en-v1.5')"
RUN useradd -m -u 1000 user \
&& chown -R 1000:1000 /app/model
USER user
ENV HOME=/home/user PATH=/home/user/.local/bin:$PATH
EXPOSE 7860
CMD ["python", "app.py"]