Self-Alignment with Principle-Following Reward Models - View it on GitHub