Large Language Models (LLMs) integrated with Visual-rich Document Understanding (VrDU) models improve document analysis tasks.