arxiv preprint - Self-Rewarding Language Models | AI Breakdown | Podwise