SALMON: Self-Alignment with Principle-Following Reward Models

Open in new window