StaR-KVQA: Structured Reasoning Traces for Implicit-Knowledge Visual Question Answering
StaR-KVQA:用于隐式知识视觉问答的结构化推理轨迹
机构 * Ant International, Ant Group(蚂蚁集团国际部,蚂蚁集团) ; School of Computer Science and Technology, University of Science and Technology of China(中国科学技术大学计算机科学与技术学院) ; School of Computing and Information Systems, Singapore Management University(新加坡管理学院计算与信息系统学院) ; Anhui Provincial Key Laboratory of High Performance Computing(安徽省高性能计算重点实验室)
专题命中 逻辑推理 :reasoning(title,abstract);chain-of-thought(abstract);分类 cs.AI
AI总结 StaR-KVQA通过引入双路径结构化推理轨迹提升隐式知识视觉问答的准确性与推理透明度,采用自蒸馏方法构建轨迹增强数据集,无需外部检索工具,在OK-VQA基准上实现11.3%的精度提升。
Comments 8+3+3 pages, code: https://github.com/jianyingzhihe/StaR-KVQA