CORE: Concept-Oriented Reinforcement for Bridging the Definition-Application Gap in Mathematical ReasoningPublished in International Conference on Learning Representations (ICLR), 2026Share on Twitter Facebook LinkedIn Previous Next