Dockerfile

FROM python:3.10-slim

WORKDIR /app

COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

# 👇 Create custom cache directory with proper permissions
RUN mkdir -p /app/cache

# 👇 Set TRANSFORMERS_CACHE permanently
ENV TRANSFORMERS_CACHE=/app/cache

COPY llm_service.py .

EXPOSE 7860
CMD ["uvicorn", "llm_service:app", "--host", "0.0.0.0", "--port", "7860"]