Signal2026-07-22
arXiv

Copy Less, Ground More: Overcoming Repetitive Copying in Long-Context Reasoning via Evidence-Aware Reinforcement Learning

Part of

Advanced Inverse Reward And Prompt Engineering