static-reward-modeller