Optimize Dockerfile: skip expensive llama-cpp-python compilation flags

Dockerfile

@@ -2,19 +2,24 @@ FROM python:3.11-slim
 
 WORKDIR /app
 
-# Install system dependencies for llama-cpp and image processing
+# Install system dependencies
 RUN apt-get update && apt-get install -y \
     build-essential \
+    cmake \
     libopenblas-dev \
     tesseract-ocr \
     libtesseract-dev \
+    git \
     && rm -rf /var/lib/apt/lists/*
 
-# Copy requirements first for better caching
+# Pre-build llama-cpp-python with explicit flags to speed up compilation
+RUN CMAKE_ARGS="-DLLAMA_BLAS=OFF -DLLAMA_METAL=OFF" pip install --no-cache-dir --force-reinstall -v llama-cpp-python==0.2.70
+
+# Copy requirements
 COPY requirements.txt .
 
-# Install Python dependencies
-RUN pip install --no-cache-dir -r requirements.txt
+# Install remaining dependencies (llama-cpp-python already installed)
+RUN pip install --no-cache-dir -r requirements.txt --ignore-installed llama-cpp-python
 
 # Copy application code
 COPY . .
@@ -22,8 +27,8 @@ COPY . .
 # Create models directory
 RUN mkdir -p models
 
-# Expose port 7860 (HuggingFace Spaces default)
+# Expose port 7860
 EXPOSE 7860
 
-# Run Uvicorn on port 7860
+# Run app
 CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860", "--timeout-keep-alive", "75"]