ThinkTwice: Jointly Optimizing Large Language Models for Reasoning and Self-Refinement Paper • 2604.01591 • Published 6 days ago • 3
