Tailored Primitive Initialization is the Secret Key to Reinforcement LearningYihang YaoGuangtao Zenget al.2026ACL 2026Conference paper