Return to Issue Details Efficient and Interpretable Multimodal Document Understanding: A Survey of Layout Transformers, Vision–Language Models, Multi-Agent Frameworks, and Conditional Computation Download Download PDF