FROM python:3.11-slim # 1. Install C++17 compiler & OpenMP runtime for AVX2/FMA inference RUN apt-get update && apt-get install -y --no-install-recommends \ build-essential \ g++ \ libgomp1 \ curl \ && rm -rf /var/lib/apt/lists/* # 2. Hugging Face Spaces runs as non-root UID 1000 RUN useradd -m -u 1000 user WORKDIR /app RUN chown -R user:user /app # 3. Environment setup for HF cache, OpenMP threads & real-time logs ENV HOME=/app \ PYTHONUNBUFFERED=1 \ HF_HOME=/app/.cache/huggingface \ HF_HUB_ENABLE_HF_TRANSFER=0 \ OMP_NUM_THREADS=2 # 4. Install Python dependencies (explicitly includes psutil) COPY --chown=user:user requirements.txt . RUN pip install --no-cache-dir --upgrade pip && \ pip install --no-cache-dir -r requirements.txt psutil==6.0.0 # 5. Copy project files COPY --chown=user:user inference.cpp app.py index.html ./ # 6. Fix only line 213 (tmp[5]+t[6] -> tmp[5]+tmp[6]) & compile C++17 AVX2 binary RUN sed -i 's/tmp\[5\]+t\[6\]/tmp[5]+tmp[6]/g' inference.cpp && \ g++ -O3 -mavx2 -mfma -fopenmp -std=c++17 inference.cpp -o inference && \ chmod +x inference && \ mkdir -p /app/.cache/huggingface && \ chown -R user:user /app # 7. Switch to HF Space user & start FastAPI on port 7860 USER user EXPOSE 7860 CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]