1What Verifiable Rewards AreVerifiable rewards turn some kinds of judgment into checks a system can run, making AI training and evaluation more grounded.Advanced 2RL for Reasoning & CodeHow reinforcement learning teaches models to reason, write code, and check their own work by rewarding useful behavior instead of memorizing answers.Advanced 3Reward Design & GamingA reward is a shortcut for what we want; gaming happens when a system learns the shortcut instead of the intent.Advanced 4Process vs Outcome SupervisionGood supervision watches both how work is done and whether the result is acceptable; confusing the two leaves gaps.Advanced 5Evaluating Reasoning GainsA practical guide to deciding whether a newer model truly reasons better, or only looks better under easier tests.Advanced