一作ICLR 2026Poster· General Machine Learning / Representation Learning
RLVMR: Reinforcement Learning with Verifiable Meta-Reasoning Rewards for Robust Long-Horizon Agents
Zijing Zhang · Ziyang Chen · Mingxiao Li · Zhaopeng Tu · Xiaolong Li
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。
Zijing Zhang · Ziyang Chen · Mingxiao Li · Zhaopeng Tu · Xiaolong Li
Zijing Zhang · Xiajie Yang
Long-horizon LLM agents trained with sparse terminal rewards tend to experience slow and unstable learning, and the issue is amplified by group-normalized on-policy objectives commonly used for LLM training (e.g., GRPO). When rollout groups collapse to nearly all failures early in training, withi...