DoubleTake: Focused Verification of Visual Evidence for Long-Document Understandingwith Kejun Xue, Ang Li, Jinghua Tang, Yongbin Gao +21 versionW_a3r7d4dkv1 · current