PapersBuilders13:25Chinese contentCLIFT:用保形自验证训练开源网页智能体并替代外部裁判开源网页智能体训练时奖励信号太稀疏?这篇 CLIFT 用保形自验证把裁判反馈变成可复用信号,测试时还能不调 API 就做轨迹筛选,WebArena Infinity 上开源里最強。#CLIFT#WebArena#agents#GPT-5.5aarXiv cs.AI@Yifan Zhang et al. (6)SourceRead laterReadWorth followingUsefulFollow CLIFT