Reinforcement Learning with Verifiable Rewards — training a model using reward signals that come from a rule-based check of correctness, not a learned judge.
Continue to AI University →