1 paper
Chengwen Liu, Xiaomin Yu, Zhuoyue Chang +15
In real-world video question answering scenarios, videos often provide only localized visual cues, while verifiable answers are distributed across the open web; models therefore ne…