Code and example data for the paper: Rule Based Rewards for Language Model Safety - View it on GitHub