一作NeurIPS 2026Poster· Reinforcement Learning / Everything Else
Stabilizing Knowledge, Promoting Reasoning: Dual-Token Constraints for RLVR
Jiakang Wang · Runze Liu · Fuzheng Zhang · Xiu Li · Guorui Zhou · Kun Gai · Ling Pan
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Jiakang Wang · Runze Liu · Fuzheng Zhang · Xiu Li · Guorui Zhou · Kun Gai · Ling Pan
Zhiwei Zhang · Fei zhao · Rui Wang · Zezhong Wang · Bin Liang · Jiakang Wang · Yao Hu · Shaosheng Cao · Kam-Fai Wong
Large language models (LLMs) can call tools effectively, yet they remain brittle in multi-turn execution: after a tool-call error, smaller models often fall into repetitive invalid re-invocations instead of interpreting the feedback and recovering. This failure mode persists because current train...