---
title: "Attention as Selector — paper by Zhu et al. (2026)"
description: "That design has two weaknesses: the query and the page never interact before scoring, and aligning the two encoders takes… 45 references."
authors: ["Minfeng Zhu", "Linxin Bao", "Wei Chen", "Linchao Zhu"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.1117/"
doi: "10.18653/v1/2026.acl-long.1117"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/zhu-2026-attention-as-selector-unlocking-vlm/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zhu-2026-attention-as-selector-unlocking-vlm/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/23/) / Attention as Selector: Unlocking VLM Attention for Long Document Page Retrieval

# Attention as Selector: Unlocking VLM Attention for Long Document Page Retrieval

**Authors:** [Minfeng Zhu](https://md.research.elamur.ai/free/en/authors/minfeng-zhu/), Linxin Bao, [Wei Chen](https://md.research.elamur.ai/free/en/authors/wei-chen/) and [Linchao Zhu](https://md.research.elamur.ai/free/en/authors/linchao-zhu/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.acl-long.1117/) · [PDF](https://aclanthology.org/2026.acl-long.1117.pdf) · [DOI 10.18653/v1/2026.acl-long.1117](https://doi.org/10.18653/v1/2026.acl-long.1117) · [OpenAlex](https://openalex.org/W7166879874)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Long documents rarely fit into a vision-language model's context, so DocQA pipelines usually pick a few pages first. Most current retrievers, such as ColPali-style systems, embed queries and pages separately and compare them with cosine similarity. That design has two weaknesses: the query and the page never interact before scoring, and aligning the two encoders takes large contrastive datasets. The authors argue that a capable VLM can already find relevant evidence through its cross-modal attention. They introduce CAPS (Cross-modal Attention as Page Selector), which works in three steps:

- It scores each page by how strongly the final token attends to that page's visual tokens.
- It strengthens this signal through contrastive fine-tuning on a small dataset.
- It picks the single best "expert" attention head and applies an adaptive threshold to decide how many pages to keep for each query.

The retriever is a lightweight Qwen2.5-VL-3B. Across four long-document benchmarks, CAPS beats embedding-based methods on both retrieval precision and downstream DocQA accuracy, using under 10% of their training data. This makes attention-based retrieval a data-efficient alternative.

## Key points
- **Benchmarks:** MMLongBench-Doc, LongDocURL, PaperTab and FetaTab. The baselines include ColPali, M3DocRAG, MDocAgent and MoLoRAG.
- **Effect of training:** On MMLongBench-Doc without filtering, head 10 improves from 52.67 to 73.64 Recall@3 after layer-level training. Supervising that head directly reaches a nearly identical 73.88. This shows the best head can emerge without being chosen in advance.
- **Two-stage reranking:** MoLoRAG retrieves 10 candidates and CAPS reranks them. This raises Recall@5 from 72.37 to 77.49, with latency rising from 5.56s to 6.98s.
- **Thresholding:** A threshold relative to the top page score beats a fixed absolute cutoff. It raises precision and feeds fewer pages to the QA model, at a small cost in recall.
- **Limitations:**
  - Results depend on the backbone's ability to handle high-resolution inputs.
  - Retrieval stays at the page level rather than selecting regions.
  - In multi-query use, embedding methods remain faster because they can cache page representations.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Zhejiang University](https://md.research.elamur.ai/free/en/universities/zhejiang-university/) — University, Hangzhou, China · 224 works · 13,587 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 3,486 works · [🗄️ Information Systems](https://md.research.elamur.ai/free/en/shelves/computer-science/information-systems/) · 1,730 works

🧭 Topics (1): [🔍 Information retrieval & search](https://md.research.elamur.ai/free/en/topics/information-retrieval/) · 552 works

## Cite

Cite as: Minfeng Zhu, Linxin Bao, Wei Chen and Linchao Zhu. 2026. Attention as Selector: Unlocking VLM Attention for Long Document Page Retrieval. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{zhu2026attention,
  title = {{Attention as Selector: Unlocking VLM Attention for Long Document Page Retrieval}},
  author = {Minfeng Zhu, Linxin Bao, Wei Chen and Linchao Zhu},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1117},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1117}
}
```

- [← previous ATIR: Towards Audio-Text Interleaved Contextual Retrieval](https://md.research.elamur.ai/free/en/papers/zhao-2026-atir-audio-text-interleaved-contextual/)
- [next → Attention Under Attack: Analog Noise Effects and Mechanistic Vulnerabilities in Transformer Models](https://md.research.elamur.ai/free/en/papers/rahman-2026-attention-under-attack-analog-noise/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/23/)

## References

45 references, 20 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Ingeol Baek, Hwan Chang, Sunghyun Ryu, and Hwanhee Lee. 2025. How do large vision-language models see text in image? unveiling the distinctive role of OCR heads. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 20441–20453, Suzhou, China. Association f… [find](https://scholar.google.com/scholar?q=Ingeol+Baek%2C+Hwan+Chang%2C+Sunghyun+Ryu%2C+and+Hwanhee+Lee%2E+2025%2E+How+do+large+vision%2Dlanguage+models+see+text+in+image%3F+unveiling+the+distinctive+role+of+OCR+heads%2E+In+Proceedings+of+the+2025+Conference+)
2. Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, Humen Zhong, Yuanzhi Zhu, Mingkun Yang, Zhaohai Li, Jianqiang Wan, Pengfei Wang, Wei Ding, Zheren Fu, Yiheng Xu, and 8 others. 2025. [Qwen2.5-vl technical report](https://md.research.elamur.ai/free/en/papers/bai-2025-qwen2-5-vl-technical/). Preprint, arXiv:2502.13…
3. Jian Chen, Ruiyi Zhang, Yufan Zhou, Tong Yu, Franck Dernoncourt, Jiuxiang Gu, Ryan Rossi, Changyou Chen, and Tong Sun. 2025a. Sv-rag: Lora-contextualizing adaptation of mllms for long document understanding. In International Conference on Representation Learning, volume 2025, pages 4447–4462. [find](https://scholar.google.com/scholar?q=Jian+Chen%2C+Ruiyi+Zhang%2C+Yufan+Zhou%2C+Tong+Yu%2C+Franck+Dernoncourt%2C+Jiuxiang+Gu%2C+Ryan+Rossi%2C+Changyou+Chen%2C+and+Tong+Sun%2E+2025a%2E+Sv%2Drag%3A+Lora%2Dcontextualizing+adaptation+of+mllms+for+long+document+underst)
4. Yihao Ding, Zhe Huang, Runlin Wang, YanHang Zhang, Xianru Chen, Yuzhong Ma, Hyunsuk Chung, and Soyeon Caren Han. 2022. V-doc: Visual questions answers with documents. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 21492–21498. [find](https://scholar.google.com/scholar?q=Yihao+Ding%2C+Zhe+Huang%2C+Runlin+Wang%2C+YanHang+Zhang%2C+Xianru+Chen%2C+Yuzhong+Ma%2C+Hyunsuk+Chung%2C+and+Soyeon+Caren+Han%2E+2022%2E+V%2Ddoc%3A+Visual+questions+answers+with+documents%2E+In+Proceedings+of+the+IEEE%2FCVF+co)
5. Yuchen Duan, Zhe Chen, Yusong Hu, Weiyun Wang, Shenglong Ye, Botian Shi, Lewei Lu, Qibin Hou, Tong Lu, Hongsheng Li, Jifeng Dai, and Wenhai Wang. 2025. Docopilot: Improving multimodal models for document-level understanding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Re… [find](https://scholar.google.com/scholar?q=Yuchen+Duan%2C+Zhe+Chen%2C+Yusong+Hu%2C+Weiyun+Wang%2C+Shenglong+Ye%2C+Botian+Shi%2C+Lewei+Lu%2C+Qibin+Hou%2C+Tong+Lu%2C+Hongsheng+Li%2C+Jifeng+Dai%2C+and+Wenhai+Wang%2E+2025%2E+Docopilot%3A+Improving+multimodal+models+for+docum)
6. Mark Endo, Xiaohan Wang, and Serena Yeung-Levy. 2025. Feather the throttle: Revisiting visual token pruning for vision-language model acceleration. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 22826–22835. [find](https://scholar.google.com/scholar?q=Mark+Endo%2C+Xiaohan+Wang%2C+and+Serena+Yeung%2DLevy%2E+2025%2E+Feather+the+throttle%3A+Revisiting+visual+token+pruning+for+vision%2Dlanguage+model+acceleration%2E+In+Proceedings+of+the+IEEE%2FCVF+International+Confere)
7. Yixiong Fang, Tianran Sun, Yuling Shi, and Xiaodong Gu. 2025. Attentionrag: Attention-guided context pruning in retrieval-augmented generation. Preprint, arXiv:2503.10720. — [arXiv:2503.10720](https://arxiv.org/abs/2503.10720) [find](https://scholar.google.com/scholar?q=Yixiong+Fang%2C+Tianran+Sun%2C+Yuling+Shi%2C+and+Xiaodong+Gu%2E+2025%2E+Attentionrag%3A+Attention%2Dguided+context+pruning+in+retrieval%2Daugmented+generation%2E+Preprint%2C+arXiv%3A2503%2E10720%2E+%E2%80%94+%5BarXiv%3A2503%2E10720%5D%28https%3A%2F)
8. Liang Chen, Haozhe Zhao, Tianyu Liu, Shuai Bai, Junyang Lin, Chang Zhou, and Baobao Chang. 2024. [An image is worth 1/2 tokens after layer 2: Plug-and-play inference acceleration for large vision-language models](https://md.research.elamur.ai/free/en/papers/chen-2024-image-worth-1-2-tokens/). In European Conference on Computer Vision, pages 19–35. Springer.
9. Manuel Faysse, Hugues Sibille, Tony Wu, Bilel Omrani, Gautier Viaud, CELINE HUDELOT, and Pierre Colombo. 2025. [Colpali: Efficient document retrieval with vision language models](https://md.research.elamur.ai/free/en/papers/faysse-2024-colpali-efficient-document-retrieval-vision/). In International Conference on Representation Learning, volume 2025, pages 61424–61449.
10. Zhe Chen, Weiyun Wang, Yue Cao, Yangzhou Liu, Zhangwei Gao, Erfei Cui, Jinguo Zhu, Shenglong Ye, Hao Tian, Zhaoyang Liu, Lixin Gu, Xuehui Wang, Qingyun Li, Yiming Ren, Zixuan Chen, Jiapeng Luo, Jiahao Wang, Tan Jiang, Bo Wang, and 23 others. 2025b. [Expanding performance boundaries of open-source mul…](https://md.research.elamur.ai/free/en/papers/chen-2024-expanding-performance-boundaries-open/)
11. Siwei Han, Peng Xia, Ruiyi Zhang, Tong Sun, Yun Li, Hongtu Zhu, and Huaxiu Yao. 2025. Mdocagent: A multi-modal multi-agent framework for document understanding. Preprint, arXiv:2503.13964. — [arXiv:2503.13964](https://arxiv.org/abs/2503.13964) [find](https://scholar.google.com/scholar?q=Siwei+Han%2C+Peng+Xia%2C+Ruiyi+Zhang%2C+Tong+Sun%2C+Yun+Li%2C+Hongtu+Zhu%2C+and+Huaxiu+Yao%2E+2025%2E+Mdocagent%3A+A+multi%2Dmodal+multi%2Dagent+framework+for+document+understanding%2E+Preprint%2C+arXiv%3A2503%2E13964%2E+%E2%80%94+%5BarXiv%3A25)
12. Jaemin Cho, Debanjan Mahata, Ozan Irsoy, Yujie He, and Mohit Bansal. 2024. M3docrag: Multi-modal retrieval is what you need for multi-page multi-document understanding. Preprint, arXiv:2411.04952. [find](https://scholar.google.com/scholar?q=Jaemin+Cho%2C+Debanjan+Mahata%2C+Ozan+Irsoy%2C+Yujie+He%2C+and+Mohit+Bansal%2E+2024%2E+M3docrag%3A+Multi%2Dmodal+retrieval+is+what+you+need+for+multi%2Dpage+multi%2Ddocument+understanding%2E+Preprint%2C+arXiv%3A2411%2E04952%2E)
13. DeepSeek-AI, Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, Deli Chen, Dongjie Ji, Erhang Li, Fangyun Lin, Fucong Dai, and 181 others. 2025. Deepseek-v3 technical report. Preprint, arXiv:2… [find](https://scholar.google.com/scholar?q=DeepSeek%2DAI%2C+Aixin+Liu%2C+Bei+Feng%2C+Bing+Xue%2C+Bingxuan+Wang%2C+Bochao+Wu%2C+Chengda+Lu%2C+Chenggang+Zhao%2C+Chengqi+Deng%2C+Chenyu+Zhang%2C+Chong+Ruan%2C+Damai+Dai%2C+Daya+Guo%2C+Dejian+Yang%2C+Deli+Chen%2C+Dongjie+Ji%2C+Erhan)
14. Chao Deng, Jiale Yuan, Pi Bu, Peijie Wang, Zhong-Zhi Li, Jian Xu, Xiao-Hui Li, Yuan Gao, Jun Song, Bo Zheng, and Cheng-Lin Liu. 2025. [LongDocURL: a comprehensive multimodal long document benchmark integrating understanding, reasoning, and locating](https://md.research.elamur.ai/free/en/papers/deng-2025-longdocurl-comprehensive-multimodal-long-document/). In Proceedings of the 63rd Annual Meeting of the As…
15. Anwen Hu, Haiyang Xu, Liang Zhang, Jiabo Ye, Ming Yan, Ji Zhang, Qin Jin, Fei Huang, and Jingren Zhou. 2025. [mPLUG-DocOwl2: High-resolution compressing for OCR-free multi-page document understanding](https://md.research.elamur.ai/free/en/papers/hu-2025-mplug-docowl2-high-resolution-compressing/). In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1…
16. De-An Huang, Subhashree Radhakrishnan, Zhiding Yu, and Jan Kautz. 2025. Frag: Frame selection augmented generation for long video and long document understanding. Preprint, arXiv:2504.17447. — [arXiv:2504.17447](https://arxiv.org/abs/2504.17447) [find](https://scholar.google.com/scholar?q=De%2DAn+Huang%2C+Subhashree+Radhakrishnan%2C+Zhiding+Yu%2C+and+Jan+Kautz%2E+2025%2E+Frag%3A+Frame+selection+augmented+generation+for+long+video+and+long+document+understanding%2E+Preprint%2C+arXiv%3A2504%2E17447%2E+%E2%80%94+%5BarXiv%3A)
17. Yulong Hui, Yao Lu, and Huanchen Zhang. 2024. Uda: a benchmark suite for retrieval augmented generation in real-world document analysis. In Proceedings of the 38th International Conference on Neural Information Processing Systems, NIPS ’24, Red Hook, NY, USA. Curran Associates Inc. [find](https://scholar.google.com/scholar?q=Yulong+Hui%2C+Yao+Lu%2C+and+Huanchen+Zhang%2E+2024%2E+Uda%3A+a+benchmark+suite+for+retrieval+augmented+generation+in+real%2Dworld+document+analysis%2E+In+Proceedings+of+the+38th+International+Conference+on+Neural+I)
18. Zhangqi Jiang, Junkai Chen, Beier Zhu, Tingjin Luo, Yankun Shen, and Xu Yang. 2025. Devils in middle layers of large vision-language models: Interpreting, detecting and mitigating object hallucinations via attention lens. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog… [find](https://scholar.google.com/scholar?q=Zhangqi+Jiang%2C+Junkai+Chen%2C+Beier+Zhu%2C+Tingjin+Luo%2C+Yankun+Shen%2C+and+Xu+Yang%2E+2025%2E+Devils+in+middle+layers+of+large+vision%2Dlanguage+models%3A+Interpreting%2C+detecting+and+mitigating+object+hallucination)
19. Omri Kaduri, Shai Bagon, and Tali Dekel. 2025. What’s in the image? a deep-dive into the vision of vision language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 14549–14558. [find](https://scholar.google.com/scholar?q=Omri+Kaduri%2C+Shai+Bagon%2C+and+Tali+Dekel%2E+2025%2E+What%E2%80%99s+in+the+image%3F+a+deep%2Ddive+into+the+vision+of+vision+language+models%2E+In+Proceedings+of+the+IEEE%2FCVF+Conference+on+Computer+Vision+and+Pattern+Reco)
20. Minesh Mathew, Viraj Bagal, Rubèn Tito, Dimosthenis Karatzas, Ernest Valveny, and C.V. Jawahar. 2022. Infographicvqa. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pages 1697–1706. [find](https://scholar.google.com/scholar?q=Minesh+Mathew%2C+Viraj+Bagal%2C+Rub%C3%A8n+Tito%2C+Dimosthenis+Karatzas%2C+Ernest+Valveny%2C+and+C%2EV%2E+Jawahar%2E+2022%2E+Infographicvqa%2E+In+Proceedings+of+the+IEEE%2FCVF+Winter+Conference+on+Applications+of+Computer+Visio)
21. Seil Kang, Jinyeong Kim, Junhyeok Kim, and Seong Jae Hwang. 2025. See what you are told: Visual attention sink in large multimodal models. In The Thirteenth International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Seil+Kang%2C+Jinyeong+Kim%2C+Junhyeok+Kim%2C+and+Seong+Jae+Hwang%2E+2025%2E+See+what+you+are+told%3A+Visual+attention+sink+in+large+multimodal+models%2E+In+The+Thirteenth+International+Conference+on+Learning+Repres)
22. Minesh Mathew, Dimosthenis Karatzas, and C.V. Jawahar. 2021. [Docvqa: A dataset for vqa on document images](https://md.research.elamur.ai/free/en/papers/mathew-2020-docvqa-dataset-vqa-document-images/). In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pages 2200–2209.
23. Geewook Kim, Teakgyu Hong, Moonbin Yim, JeongYeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, and Seunghyun Park. 2022. [Ocr-free document understanding transformer](https://md.research.elamur.ai/free/en/papers/kim-2021-ocr-free-document-understanding-transformer/). In European Conference on Computer Vision, pages 498–517. Springer.
24. Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. [Learning transferable visual models from natural language supervision](https://md.research.elamur.ai/free/en/papers/radford-2021-learning-transferable-visual-models-natural/). In Proceedings of the 38th Internatio…
25. Kenton Lee, Mandar Joshi, Iulia Raluca Turc, Hexiang Hu, Fangyu Liu, Julian Martin Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, and Kristina Toutanova. 2023. [Pix2Struct: Screenshot parsing as pretraining for visual language understanding](https://md.research.elamur.ai/free/en/papers/lee-2022-pix2struct-screenshot-parsing-pretraining-visual/). In Proceedings of the 40th International Conf…
26. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. [Retrieval-augmented generation for knowledge-intensive nlp tasks](https://md.research.elamur.ai/free/en/papers/lewis-2020-retrieval-augmented-generation-knowledge-intensive/). In Advances in Neural Information Proc…
27. Bo Li, Kaichen Zhang, Hao Zhang, Dong Guo, Renrui Zhang, Feng Li, Yuanhan Zhang, Ziwei Liu, and Chunyuan Li. 2024. Llava-next: Stronger llms supercharge multimodal capabilities in the wild. [find](https://scholar.google.com/scholar?q=Bo+Li%2C+Kaichen+Zhang%2C+Hao+Zhang%2C+Dong+Guo%2C+Renrui+Zhang%2C+Feng+Li%2C+Yuanhan+Zhang%2C+Ziwei+Liu%2C+and+Chunyuan+Li%2E+2024%2E+Llava%2Dnext%3A+Stronger+llms+supercharge+multimodal+capabilities+in+the+wild%2E)
28. Haoyu Lu, Wen Liu, Bo Zhang, Bingxuan Wang, Kai Dong, Bo Liu, Jingxiang Sun, Tongzheng Ren, Zhuoshu Li, Hao Yang, Yaofeng Sun, Chengqi Deng, Hanwei Xu, Zhenda Xie, and Chong Ruan. 2024. [Deepseek-vl: Towards real-world vision-language understanding](https://md.research.elamur.ai/free/en/papers/lu-2024-deepseek-vl-real-world/). Preprint, arXiv:2403.05525.
29. Xueguang Ma, Shengyao Zhuang, Bevan Koopman, Guido Zuccon, Wenhu Chen, and Jimmy Lin. 2024a. [Visa: Retrieval augmented generation with visual source attribution](https://md.research.elamur.ai/free/en/papers/ma-2025-visa-retrieval-augmented-generation-visual/). Preprint, arXiv:2412.14457.
30. Yubo Ma, Yuhang Zang, Liangyu Chen, Meiqi Chen, Yizhu Jiao, Xinze Li, Xinyuan Lu, Ziyu Liu, Yan Ma, Xiaoyi Dong, Pan Zhang, Liangming Pan, Yu-Gang Jiang, Jiaqi Wang, Yixin Cao, and Aixin Sun. 2024b. [Mmlongbench-doc: Benchmarking long-context document understanding with visualizations](https://md.research.elamur.ai/free/en/papers/ma-2024-mmlongbench-doc-benchmarking-long-context/). In Advances in…
31. Manan Suri, Puneet Mathur, Franck Dernoncourt, Kanika Goswami, Ryan A. Rossi, and Dinesh Manocha. 2025. VisDoM: Multi-document QA with visually rich elements using multimodal retrieval-augmented generation. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Associati… [find](https://scholar.google.com/scholar?q=Manan+Suri%2C+Puneet+Mathur%2C+Franck+Dernoncourt%2C+Kanika+Goswami%2C+Ryan+A%2E+Rossi%2C+and+Dinesh+Manocha%2E+2025%2E+VisDoM%3A+Multi%2Ddocument+QA+with+visually+rich+elements+using+multimodal+retrieval%2Daugmented+gener)
32. Rubèn Tito, Dimosthenis Karatzas, and Ernest Valveny. 2023. Hierarchical multimodal transformers for mul-tipage docvqa. Pattern Recognition, 144:109834. [find](https://scholar.google.com/scholar?q=Rub%C3%A8n+Tito%2C+Dimosthenis+Karatzas%2C+and+Ernest+Valveny%2E+2023%2E+Hierarchical+multimodal+transformers+for+mul%2Dtipage+docvqa%2E+Pattern+Recognition%2C+144%3A109834%2E)
33. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. 2017. [Attention is all you need](https://md.research.elamur.ai/free/en/papers/vaswani-2017-attention-all-need/). In Advances in Neural Information Processing Systems, volume 30. Curran Associates, Inc.
34. Bin Wang, Chao Xu, Xiaomeng Zhao, Linke Ouyang, Fan Wu, Zhiyuan Zhao, Rui Xu, Kaiwen Liu, Yuan Qu, Fukai Shang, Bo Zhang, Liqun Wei, Zhihao Sui, Wei Li, Botian Shi, Yu Qiao, Dahua Lin, and Conghui He. 2024. [Mineru: An open-source solution for precise document content extraction](https://md.research.elamur.ai/free/en/papers/wang-2024-mineru-open-source-solution-precise/). Preprint, arXiv:2409…
35. Haoran Wei, Chenglong Liu, Jinyue Chen, Jia Wang, Lingyu Kong, Yanming Xu, Zheng Ge, Liang Zhao, Jianjian Sun, Yuang Peng, Chunrui Han, and Xiangyu Zhang. 2024. General ocr theory: Towards ocr-2.0 via a unified end-to-end model. Preprint, arXiv:2409.01704. — [arXiv:2409.01704](https://arxiv.org/abs/2409.01704) [find](https://scholar.google.com/scholar?q=Haoran+Wei%2C+Chenglong+Liu%2C+Jinyue+Chen%2C+Jia+Wang%2C+Lingyu+Kong%2C+Yanming+Xu%2C+Zheng+Ge%2C+Liang+Zhao%2C+Jianjian+Sun%2C+Yuang+Peng%2C+Chunrui+Han%2C+and+Xiangyu+Zhang%2E+2024%2E+General+ocr+theory%3A+Towards+ocr%2D2%2E0+via)
36. Haoran Wei, Yaofeng Sun, and Yukun Li. 2025. [Deepseek-ocr: Contexts optical compression](https://md.research.elamur.ai/free/en/papers/wei-2025-deepseek-ocr-contexts-optical/). Preprint, arXiv:2510.18234.
37. Zichen Wen, Yifeng Gao, Weijia Li, Conghui He, and Linfeng Zhang. 2025. Token pruning in multimodal large language models: Are we solving the right problem? In Findings of the Association for Computational Linguistics: ACL 2025, pages 15537–15549, Vienna, Austria. Association for Computational Lingu… [find](https://scholar.google.com/scholar?q=Zichen+Wen%2C+Yifeng+Gao%2C+Weijia+Li%2C+Conghui+He%2C+and+Linfeng+Zhang%2E+2025%2E+Token+pruning+in+multimodal+large+language+models%3A+Are+we+solving+the+right+problem%3F+In+Findings+of+the+Association+for+Computat)
38. Wenhao Wu, Yizhong Wang, Guangxuan Xiao, Hao Peng, and Yao Fu. 2024. [Retrieval head mechanistically explains long-context factuality](https://md.research.elamur.ai/free/en/papers/wu-2024-retrieval-head-mechanistically-explains-long/). Preprint, arXiv:2404.15574.
39. Xixi Wu, Yanchao Tan, Nan Hou, Ruiyang Zhang, and Hong Cheng. 2025. [MoLoRAG: Bootstrapping document understanding via multi-modal logic-aware retrieval](https://md.research.elamur.ai/free/en/papers/wu-2025-molorag-bootstrapping-document-understanding-multi/). In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 14035–14056, Suzhou, China. Association for Comput…
40. Long Xing, Qidong Huang, Xiaoyi Dong, Jiajie Lu, Pan Zhang, Yuhang Zang, Yuhang Cao, Conghui He, Jiaqi Wang, Feng Wu, and Dahua Lin. 2025. Conical visual concentration for efficient large vision-language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CV… [find](https://scholar.google.com/scholar?q=Long+Xing%2C+Qidong+Huang%2C+Xiaoyi+Dong%2C+Jiajie+Lu%2C+Pan+Zhang%2C+Yuhang+Zang%2C+Yuhang+Cao%2C+Conghui+He%2C+Jiaqi+Wang%2C+Feng+Wu%2C+and+Dahua+Lin%2E+2025%2E+Conical+visual+concentration+for+efficient+large+vision%2Dlangu)
41. Jiabo Ye, Anwen Hu, Haiyang Xu, Qinghao Ye, Ming Yan, Guohai Xu, Chenliang Li, Junfeng Tian, Qi Qian, Ji Zhang, Qin Jin, Liang He, Xin Lin, and Fei Huang. 2023. [UReader: Universal OCR-free visually-situated language understanding with multimodal large language model](https://md.research.elamur.ai/free/en/papers/ye-2023-ureader-universal-ocr-free-visually/). In Findings of the Association f…
42. Xiaoju Ye, Zhichun Wang, and Jingyuan Wang. 2025. Infinite retrieval: Attention enhanced llms in long-context processing. Preprint, arXiv:2502.12962. — [arXiv:2502.12962](https://arxiv.org/abs/2502.12962) [find](https://scholar.google.com/scholar?q=Xiaoju+Ye%2C+Zhichun+Wang%2C+and+Jingyuan+Wang%2E+2025%2E+Infinite+retrieval%3A+Attention+enhanced+llms+in+long%2Dcontext+processing%2E+Preprint%2C+arXiv%3A2502%2E12962%2E+%E2%80%94+%5BarXiv%3A2502%2E12962%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2502%2E12)
43. Qintong Zhang, Bin Wang, Victor Shea-Jay Huang, Junyuan Zhang, Zhengren Wang, Hao Liang, Conghui He, and Wentao Zhang. 2025a. Document parsing unveiled: Techniques, challenges, and prospects for structured information extraction. Preprint, arXiv:2410.21169. — [arXiv:2410.21169](https://arxiv.org/abs/2410.21169) [find](https://scholar.google.com/scholar?q=Qintong+Zhang%2C+Bin+Wang%2C+Victor+Shea%2DJay+Huang%2C+Junyuan+Zhang%2C+Zhengren+Wang%2C+Hao+Liang%2C+Conghui+He%2C+and+Wentao+Zhang%2E+2025a%2E+Document+parsing+unveiled%3A+Techniques%2C+challenges%2C+and+prospects+for+struc)
44. Yuan Zhang, Chun-Kai Fan, Junpeng Ma, Wenzhao Zheng, Tao Huang, Kuan Cheng, Denis Gudovskiy, Tomoyuki Okuno, Yohei Nakata, Kurt Keutzer, and 1 others. 2025b. Sparsevlm: Visual token sparsification for efficient vision-language model inference. In International Conference on Machine Learning. [find](https://scholar.google.com/scholar?q=Yuan+Zhang%2C+Chun%2DKai+Fan%2C+Junpeng+Ma%2C+Wenzhao+Zheng%2C+Tao+Huang%2C+Kuan+Cheng%2C+Denis+Gudovskiy%2C+Tomoyuki+Okuno%2C+Yohei+Nakata%2C+Kurt+Keutzer%2C+and+1+others%2E+2025b%2E+Sparsevlm%3A+Visual+token+sparsification+for)
45. Yunzhu Zhang, Yu Lu, Tianyi Wang, Fengyun Rao, Yi Yang, and Linchao Zhu. 2025c. Flexselect: Flexible token selection for efficient long video understanding. Preprint, arXiv:2506.00993. — [arXiv:2506.00993](https://arxiv.org/abs/2506.00993) [find](https://scholar.google.com/scholar?q=Yunzhu+Zhang%2C+Yu+Lu%2C+Tianyi+Wang%2C+Fengyun+Rao%2C+Yi+Yang%2C+and+Linchao+Zhu%2E+2025c%2E+Flexselect%3A+Flexible+token+selection+for+efficient+long+video+understanding%2E+Preprint%2C+arXiv%3A2506%2E00993%2E+%E2%80%94+%5BarXiv%3A2506%2E0)

*Updated 2 November 2026*