FROM nvidia/cuda:12.8.1-runtime-ubuntu24.04

ENV DEBIAN_FRONTEND=noninteractive \
    PYTHONDONTWRITEBYTECODE=1 \
    PYTHONUNBUFFERED=1 \
    PORT=7860

RUN apt-get update && apt-get install -y --no-install-recommends \
    ca-certificates \
    curl \
    python3 \
    python3-pip \
    libgomp1 \
    && rm -rf /var/lib/apt/lists/*

WORKDIR /app
COPY requirements.txt .
RUN python3 -m pip install --no-cache-dir --break-system-packages -r requirements.txt
COPY . .

# Keep model acquisition in the image build, so starting a container never
# downloads multi-gigabyte weights.  curl fails the build on an incomplete or
# unavailable download and retries transient network failures.
RUN mkdir -p /app/model_weights \
    && curl --fail --location --retry 5 --retry-all-errors \
        --output /app/model_weights/rwkv7-g1i-1.5b-20260805-ctx16384.pth \
        https://huggingface.co/BlinkDL/rwkv7-g1/resolve/main/rwkv7-g1i-1.5b-20260805-ctx16384.pth \
    && curl --fail --location --retry 5 --retry-all-errors \
        --output /app/model_weights/rwkv7-g1i-2.9b-20260805-ctx16384.pth \
        https://huggingface.co/BlinkDL/rwkv7-g1/resolve/main/rwkv7-g1i-2.9b-20260805-ctx16384.pth \
    && curl --fail --location --retry 5 --retry-all-errors \
        --output /app/model_weights/rwkv7-g1i-7.2b-20260805-ctx16384.pth \
        https://huggingface.co/BlinkDL/rwkv7-g1/resolve/main/rwkv7-g1i-7.2b-20260805-ctx16384.pth

EXPOSE 7860
CMD ["python3", "app.py"]