Dockerfile

# Read the doc: https://huggingface.co/docs/hub/spaces-sdks-docker
# you will also find guides on how best to write your Dockerfile

FROM python:3.12.3

RUN useradd -m -u 1000 user
USER user
ENV PATH="/home/user/.local/bin:$PATH"

# Add environment variable to force CPU usage
ENV LLAMA_CPP_FORCE_CPU=1

# Create a directory for the app
WORKDIR /app

# Download the model
RUN wget -O DeepSeek-R1-Distill-Llama-8B-Q4_K_L.gguf \
    "https://huggingface.co/bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF/resolve/main/DeepSeek-R1-Distill-Llama-8B-Q4_K_L.gguf"

RUN wget -O DeepSeek-R1-Distill-Qwen-1.5B-Q2_K.gguf \
    "https://huggingface.co/bartowski/DeepSeek-R1-Distill-Qwen-1.5B-GGUF/resolve/main/DeepSeek-R1-Distill-Qwen-1.5B-Q2_K.gguf"


# Copy the files to the container
COPY --chown=user . /app

# Install the requirements
RUN pip install --no-cache-dir --upgrade -r requirements.txt

# Expose the port
EXPOSE 8000

# Command to start the server
CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]