Dockerfile

FROM nvidia/cuda:12.1.1-cudnn8-runtime-ubuntu22.04
ENV DEBIAN_FRONTEND=noninteractive \
    PYTHONUNBUFFERED=1 \
    PIP_NO_CACHE_DIR=1 \
    HF_HUB_ENABLE_HF_TRANSFER=1 \
    NUMBA_CACHE_DIR=/tmp/numba_cache \
    # optional: uncomment to fully disable FS caching
    # NUMBA_DISABLE_FILE_SYSTEM_CACHING=1
    OUTLINES_CACHE_DIR=/tmp/outlines_cache

RUN apt-get update && apt-get install -y python3 python3-pip git && rm -rf /var/lib/apt/lists/*
WORKDIR /app

COPY requirements.txt /app/
RUN python3 -m pip install --upgrade pip && pip3 install -r requirements.txt

# ensure cache dirs exist & writable
RUN mkdir -p /tmp/numba_cache /tmp/outlines_cache && chmod -R 777 /tmp/numba_cache /tmp/outlines_cache

COPY app.py /app/

ENV PORT=7860
EXPOSE 7860
CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]