Models and Dataset for CVPR 2025 paper: Adaptive Markup Language Generation for Contextually-Grounded Visual Document Understanding