Models That Prove Their Own Correctness | Dark Hacker News