Dockerfile

FROM python:3.10-slim

WORKDIR /app

ENV HF_HOME=/app/.cache/huggingface
ENV XDG_CACHE_HOME=/app/.cache

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

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

COPY . .

# (Optional) pre-download model to image to avoid runtime download
# RUN python -c "from transformers import AutoModelForCausalLM, AutoTokenizer; m='google/gemma-2-2b-it'; AutoTokenizer.from_pretrained(m); AutoModelForCausalLM.from_pretrained(m)"

EXPOSE 7860

CMD ["python", "main.py"]