H. Lu

MMLDSum-LLM: Multimodal Long-Document Summarization with Visual-Alignment and Keyword-Aware

This work introduces MMLDSum-Bench, a high-quality benchmark for multimodal long-document summarization, and proposes MMLDSum-LLM, a reproducible two-stage training framework that combines supervised fine-tuning with visual-alignment weighted loss and keyword-aware weighted loss, followed by GRPO with a multi-objective...