Dockerfile

FROM python:3.10-slim

# Set up non-root user
RUN useradd -m -u 1000 user
USER user
ENV HOME=/home/user \
    PATH=/home/user/.local/bin:$PATH

WORKDIR /app

# Copy and install requirements
COPY --chown=user:user requirements.txt .
RUN pip install --no-cache-dir --user -r requirements.txt

# Pre-cache the model with CPU-optimized 4-bit quantization
RUN python -c "import torch; from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig; \
    quantization_config = BitsAndBytesConfig(load_in_4bit=True, bnb_4bit_compute_dtype=torch.float16, bnb_4bit_use_double_quant=False, bnb_4bit_quant_type='nf4'); \
    AutoTokenizer.from_pretrained('Qwen/Qwen2-1.5B-Instruct'); \
    AutoModelForCausalLM.from_pretrained('Qwen/Qwen2-1.5B-Instruct', quantization_config=quantization_config, device_map='cpu')"

# Copy app
COPY --chown=user:user app.py .

# Set cache directory permissions
RUN mkdir -p $HOME/.cache && chmod -R 777 $HOME/.cache

EXPOSE 8000

CMD ["gunicorn", "--bind", "0.0.0.0:8000", "--timeout", "300", "app:app"]