一作ICLR 2026Poster· Reinforcement Learning / Batch/Offline
From Curiosity to Caution: Mitigating Reward Hacking for Best-of-$N$ with Pessimism
Zhuohao Yu · Steven Wu · Adam Block
强化学习与后训练OpenReview
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。
Zhuohao Yu · Steven Wu · Adam Block
Yidong Wang · Yunze Song · Tingyuan Zhu · Xuanwang Zhang · Zhuohao Yu · Hao Chen · Chiyu Song · Qiufeng Wang · Zhen Wu · Xin-Yu Dai · Yue Zhang · Cunxiang Wang · Wei Ye · Shikun Zhang
Zhuohao Yu · Jiali Zeng · Weizheng Gu · Mengyuan Sun · Yidong Wang · Fandong Meng · Jie Zhou · Shikun Zhang · Wei Ye
Traditional reinforcement learning from human feedback (RLHF) optimizes policies on fixed training inputs, limiting the diversity of learning signals. We propose JODP (Joint Optimization of Data and Policy), a framework where the evolving policy model generates improved variants of training probl...