Dockerfile

FROM python:3.11-slim

WORKDIR /app

# Install system dependencies
RUN apt-get update && apt-get install -y \
    build-essential \
    git \
    && rm -rf /var/lib/apt/lists/*

# Copy and install requirements
COPY requirements.txt .
RUN pip install --no-cache-dir --upgrade pip
RUN pip install --no-cache-dir -r requirements.txt
RUN pip install --no-cache-dir torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu

# ✅ Install accelerate (fixes distilgpt2 loading with device_map)
RUN pip install --no-cache-dir accelerate

# ✅ Set environment variables to a writable path
ENV HF_HOME=/app/cache
ENV TRANSFORMERS_CACHE=/app/cache
ENV TOKENIZERS_PARALLELISM=false
ENV OMP_NUM_THREADS=1

# ✅ Create cache directory with proper permissions
RUN mkdir -p /app/cache && chmod -R 777 /app/cache

# Copy application
COPY . .

# ✅ Pre-download the model during build (optional but recommended)
RUN python -c "from transformers import AutoTokenizer, AutoModelForCausalLM; \
    AutoTokenizer.from_pretrained('distilgpt2', cache_dir='/app/cache'); \
    AutoModelForCausalLM.from_pretrained('distilgpt2', cache_dir='/app/cache')" || echo "Model download failed, will retry at runtime"

# ✅ Ensure cache directory is writable after model download
RUN chmod -R 777 /app/cache

# ✅ Expose correct port
EXPOSE 7860

# ✅ Run app
CMD ["gunicorn", "--bind", "0.0.0.0:7860", "--workers", "1", "--timeout", "120", "app:app"]