Research map / Large language models / Retrieval and knowledge
Are We on the Right Way to Assess Document Retrieval-Augmented Generation?
Wenxuan Shen, Mingjia Wang, Yaochen Wang, Dongping Chen, Junjie Yang, Yao Wan, Weiwei Lin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| provide ground-truth evidence pages to MLLMs (Oracle) instead of no retrieval (ablation) | Qwen2.5-VL-32B · Double-Bench (single hop) | fraction fully correct: 0.200 -> 0.874 | Improved |
| provide ground-truth evidence pages to GPT-4o (Oracle) instead of no retrieval (ablation) | GPT-4o · Double-Bench (single hop) | fraction fully correct: 0.109 -> 0.678 | Improved |
| use a document-specific visual embedding model (ColQwen2.5-3B) instead of general multimodal embeddings | document retrieval · Double-Bench | hit@5: 0.720 -> 0.795 | Improved |
| use visual embeddings instead of text embeddings | document retrieval · Double-Bench | hit@5: 0.776 -> 0.795 | Mixed |
| increase number of hops in query | embedding models · Double-Bench | hit@5: 0.895 -> 0.693 (colqwen, single vs 2-hop) | Hurt |
| use a single MLLM pass with strong retriever (Colqwen-gen) instead of multi-agent pipeline (MDocAgent) | document RAG · Double-Bench (average) | retrieval hit@5: 0.688 -> 0.795 | Improved |
| use complex multi-agent frameworks (MDocAgent, ViDoRAG) instead of simple M3DocRAG | document RAG · Double-Bench | answer correct given retrieval hit: 0.451 -> 0.645 | Mixed |
Closest papers by meaning
- VisRAG: Vision-based Retrieval-augmented Generation on Multi-modality Documents (ICLR 2025 · similarity 0.91)
- SV-RAG: LoRA-Contextualizing Adaptation of MLLMs for Long Document Understanding (ICLR 2025 · similarity 0.91)
- Benchmarking Large Language Models in Retrieval-Augmented Generation (AAAI 2024 · similarity 0.91)
- UDA: A Benchmark Suite for Retrieval Augmented Generation in Real-World Document Analysis (NeurIPS 2024 · similarity 0.90)
- Docopilot: Improving Multimodal Models for Document-Level Understanding (CVPR 2025 · similarity 0.90)
- RAGGED: Towards Informed Design of Scalable and Stable RAG Systems (ICML 2025 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
