1 paper · 1 filter
Zhaoqing Zhu, Chuwei Luo, Zirui Shao +4
Recent methods that integrate spatial layouts with text for document understanding in large language models (LLMs) have shown promising results. A commonly used method is to repres…