一作ICML 2026Poster· Applications / Language, Speech and Dialog
IntentRL: Training Proactive User-intent Agents for Open-ended Deep Research via Reinforcement Learning
Haohao Luo · Zexi Li · Yuexiang Xie · Wenhao Zhang · Yaliang Li · Ying Shen
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。
Haohao Luo · Zexi Li · Yuexiang Xie · Wenhao Zhang · Yaliang Li · Ying Shen
Zihao Yi · Zhenqing Ling · Delong Zeng · Haohao Luo · Zhe Xu · WEI LIU · Jian Luan · Wanxia Cao · Ying Shen
The performance of Large Language Models (LLMs) is significantly sensitive to the contextual position of information in the input. To investigate the mechanism behind this positional bias, our extensive experiments reveal a consistent phenomenon we term the attention basin: when presented with a...