Update Dockerfile

Dockerfile

@@ -1,33 +1,33 @@
-# Use a slim Python base
+# Use a Python image that has many tools pre-installed
 FROM python:3.11-slim
 
-# Install system dependencies (build tools, cmake for llama.cpp, nodejs for promptfoo)
+# Install Node.js (needed for promptfoo) and system tools
 RUN apt-get update && apt-get install -y \
-    build-essential \
-    cmake \
-    git \
     curl \
+    git \
     nodejs \
     npm \
     && rm -rf /var/lib/apt/lists/*
 
-# Install promptfoo globally (our security testing tool)
+# Install promptfoo globally
 RUN npm install -g promptfoo
 
 # Set working directory
 WORKDIR /app
 
-# Copy Python requirements first (better caching)
+# Copy requirements
 COPY requirements.txt .
 
 # Install Python dependencies
+# We use --extra-index-url to get the pre-built 'wheel' to avoid compilation timeout
+RUN pip install --no-cache-dir --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu llama-cpp-python
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Copy the rest of the application
+# Copy application code
 COPY . .
 
-# Expose the port Hugging Face uses
+# Expose port
 EXPOSE 7860
 
-# Run the server
+# Run server
 CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860"]