---
title: "Cultivating Forensic Reasoning for Generalizable Multimodal"
description: "They propose REFORM, a compact dual-decoder model that produces a forensic rationale alongside its classification and… 45 references."
authors: ["Yuchen Zhang", "Yaxiong Wang", "Kecheng Han", "Yujiao Wu", "Lianwei Wu", "Li Zhu", "Zhedong Zheng"]
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.1316/"
doi: "10.18653/v1/2026.acl-long.1316"
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/zhang-2026-cultivating-forensic-reasoning-generalizable-multimodal/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zhang-2026-cultivating-forensic-reasoning-generalizable-multimodal/)

[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/28/) / Cultivating Forensic Reasoning for Generalizable Multimodal Manipulation Detection

# Cultivating Forensic Reasoning for Generalizable Multimodal Manipulation Detection

**Authors:** [Yuchen Zhang](https://md.research.elamur.ai/free/en/authors/yuchen-zhang/), [Yaxiong Wang](https://md.research.elamur.ai/free/en/authors/yaxiong-wang/), Kecheng Han, Yujiao Wu, [Lianwei Wu](https://md.research.elamur.ai/free/en/authors/lianwei-wu/), Li Zhu and Zhedong Zheng · **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.1316/) · [PDF](https://aclanthology.org/2026.acl-long.1316.pdf) · [DOI 10.18653/v1/2026.acl-long.1316](https://doi.org/10.18653/v1/2026.acl-long.1316) · [OpenAlex](https://openalex.org/W7166809720)

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

## Summary
Generative AI makes it easy to fake image–text news, and current detectors for detecting and grounding multimodal manipulation (DGM4) generalize poorly. The authors argue this is because these detectors are trained only to predict manipulation labels, so they learn surface artifacts instead of the evidence behind a verdict. They propose REFORM, a compact dual-decoder model that produces a forensic rationale alongside its classification and grounding output. Training follows a three-stage curriculum:
- a warm-up that distills rationales from a teacher VLM,
- joint fine-tuning that ties the final judgment to the reasoning,
- reinforcement learning with GRPO that rewards reasoning consistent with a correct verdict.

To support this approach, they build ROM, a large benchmark with reasoning annotations. It extends beyond face edits to scene-level synthesis. REFORM sets new state-of-the-art results on ROM, DGM4 and MMFakeBench, with strong cross-domain and zero-shot performance. The work matters because it shows that explicit, process-oriented supervision can make misinformation detectors both more robust and more interpretable.

## Key points
- **ROM dataset:** over 704k samples with rationale annotations, covering 9 manipulation categories plus real news. The categories include FaceSwap, FaceAttribute, TextFabrication and scene-level edits. Access is controlled and research-only, under a Data Usage Agreement.
- **Main results:** ACC reaches 81.52% on ROM and 76.65% on DGM4, while F1 on MMFakeBench is 74.9.
- **Efficiency:** REFORM has only 376M parameters, compared with FKA-Owl (6.7B) and MMD-Agent (34B). A Fast Mode skips rationale generation, running at 13.17 pairs/s with no accuracy loss; the Explainable Mode runs at 1.03 pairs/s.
- **Rationale quality:** a blinded human audit of 350 samples found that the rationales recover 83.7% of ground-truth visual evidence and 82.2% of textual evidence.
- **Teacher robustness:** swapping the InternVL3.5-30B teacher for Qwen2.5-VL-3B costs only 0.84 ACC, 1.46 mAP and 0.33 mIoU.

*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

- [Xi'an Jiaotong University](https://md.research.elamur.ai/free/en/universities/xi-an-jiaotong-university/) — University, Xi'an, China · 42 works · 36,431 citations
- [University of Macau](https://md.research.elamur.ai/free/en/universities/university-of-macau/) — University, Macao, China · 43 works · 2,368 citations
- [Commonwealth Scientific and Industrial Research Organisation](https://md.research.elamur.ai/free/en/government/commonwealth-scientific-and-industrial-research-organisation/) — Government body, Canberra, Australia · 17 works · 2,064 citations
- [Hefei University of Technology](https://md.research.elamur.ai/free/en/universities/hefei-university-of-technology/) — University, Hefei, China · 17 works · 1,702 citations
- [Northwestern Polytechnical University](https://md.research.elamur.ai/free/en/universities/northwestern-polytechnical-university/) — University, Xi'an, China · 15 works · 67 citations

📚 Shelves (7): [🖥️ 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 · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,954 works · [🤝 Sociology and Political Science](https://md.research.elamur.ai/free/en/shelves/social-sciences/sociology-and-political-science/) · 626 works

🧭 Topics (2): [🧮 Reasoning](https://md.research.elamur.ai/free/en/topics/reasoning/) · 1,685 works · [🖼️ Vision & language](https://md.research.elamur.ai/free/en/topics/multimodal/) · 811 works

## Cite

Cite as: Yuchen Zhang, Yaxiong Wang, Kecheng Han, Yujiao Wu, Lianwei Wu, Li Zhu and Zhedong Zheng. 2026. Cultivating Forensic Reasoning for Generalizable Multimodal Manipulation Detection. 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{zhang2026cultivating,
  title = {{Cultivating Forensic Reasoning for Generalizable Multimodal Manipulation Detection}},
  author = {Yuchen Zhang, Yaxiong Wang, Kecheng Han, Yujiao Wu, Lianwei Wu, Li Zhu and Zhedong Zheng},
  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.1316},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1316}
}
```

- [← previous CUET_SYNTHETICA@EEUCA 2026: Gated Cross-Modal Attention with Domain-Adapted Text Encoding for Vaccine-Critical Meme Detection](https://md.research.elamur.ai/free/en/papers/zaman-2026-cuet-synthetica-eeuca-2026-gated/)
- [next → Cultural Adaptation in Large Language Models for Political Discourse](https://md.research.elamur.ai/free/en/papers/zaghouani-2026-cultural-adaptation-large-language-models/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/28/)

## References

45 references, 21 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. 01.AI. 2024. [Yi: Open foundation models by 01.ai](https://md.research.elamur.ai/free/en/papers/ai-2024-yi-open-foundation-models-01/). arXiv preprint arXiv:2403.04652v3.
2. Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, and 1 others. 2023. [Qwen technical report](https://md.research.elamur.ai/free/en/papers/bai-2023-qwen-technical-report/). arXiv preprint arXiv:2309.16609.
3. Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, and 1 others. 2025. Qwen2. 5-vl technical report. arXiv preprint arXiv:2502.13923. — [Qwen2.5-VL Technical Report](https://md.research.elamur.ai/free/en/papers/bai-2025-qwen2-5-vl-technical/)
4. Xiao Bin, Wu Haiping, Xu Weijian, Dai Xiyang, Hu Houdong, Lu Yumao, Zeng Michael, Liu Ce, and Yuan Lu. 2023. Florence-2: Advancing a unified representation for a variety of vision tasks. In CVPR. [find](https://scholar.google.com/scholar?q=Xiao+Bin%2C+Wu+Haiping%2C+Xu+Weijian%2C+Dai+Xiyang%2C+Hu+Houdong%2C+Lu+Yumao%2C+Zeng+Michael%2C+Liu+Ce%2C+and+Yuan+Lu%2E+2023%2E+Florence%2D2%3A+Advancing+a+unified+representation+for+a+variety+of+vision+tasks%2E+In+CVPR%2E)
5. Tianzhe Chu, Yuexiang Zhai, Jihan Yang, Shengbang Tong, Saining Xie, Dale Schuurmans, Quoc V. Le, Sergey Levine, and Yi Ma. 2025. [Sft memorizes, rl generalizes: A comparative study of foundation model post-training](https://md.research.elamur.ai/free/en/papers/chu-2025-sft-memorizes-rl-generalizes-comparative/). In ICML.
6. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, and 1 others. 2025. [Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities](https://md.research.elamur.ai/free/en/papers/comanici-2025-gemini-2-5-pushing-frontier/). arXiv Preprint arXiv:2507.06261.
7. Wenliang Dai, Junnan Li, DONGXU LI, Anthony Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale N Fung, and Steven Hoi. 2023a. [Instructblip: Towards general-purpose vision-language models with instruction tuning](https://md.research.elamur.ai/free/en/papers/dai-2023-instructblip-towards-general-purpose-vision/). In NeurIPS.
8. Wenliang Dai, Junnan Li, DONGXU LI, Anthony Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale N Fung, and Steven Hoi. 2023b. [Instructblip: Towards general-purpose vision-language models with instruction tuning](https://md.research.elamur.ai/free/en/papers/dai-2023-instructblip-towards-general-purpose-vision/). In NeurIPS.
9. Tri Dao. 2024. [Flashattention-2: Faster attention with better parallelism and work partitioning](https://md.research.elamur.ai/free/en/papers/dao-2023-flashattention-2-faster-attention-better/). In ICLR.
10. Mingyu Ding, Bin Xiao, Noel Codella, Ping Luo, Jingdong Wang, and Lu Yuan. 2022. Davit: Dual attention vision transformers. In ECCV. [find](https://scholar.google.com/scholar?q=Mingyu+Ding%2C+Bin+Xiao%2C+Noel+Codella%2C+Ping+Luo%2C+Jingdong+Wang%2C+and+Lu+Yuan%2E+2022%2E+Davit%3A+Dual+attention+vision+transformers%2E+In+ECCV%2E)
11. Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek, and Robin Rombach. 2024. Scaling rectified flow transformers for high-r… [find](https://scholar.google.com/scholar?q=Patrick+Esser%2C+Sumith+Kulal%2C+Andreas+Blattmann%2C+Rahim+Entezari%2C+Jonas+M%C3%BCller%2C+Harry+Saini%2C+Yam+Levi%2C+Dominik+Lorenz%2C+Axel+Sauer%2C+Frederic+Boesel%2C+Dustin+Podell%2C+Tim+Dockhorn%2C+Zion+English%2C+Kyle+Lacey%2C)
12. Jing Huang, Zhiya Tan, Shutao Gong, Fanwei Zeng, Joey Tianyi Zhou, Changtao Miao, Huazhe Tan, Weibin Yao, and Jianshu Li. 2025a. Lav-cot: Language-aware visual cot with multi-aspect reward optimization for real-world multilingual vqa. arXiv preprint arXiv:2509.10026. — [arXiv:2509.10026](https://arxiv.org/abs/2509.10026) [find](https://scholar.google.com/scholar?q=Jing+Huang%2C+Zhiya+Tan%2C+Shutao+Gong%2C+Fanwei+Zeng%2C+Joey+Tianyi+Zhou%2C+Changtao+Miao%2C+Huazhe+Tan%2C+Weibin+Yao%2C+and+Jianshu+Li%2E+2025a%2E+Lav%2Dcot%3A+Language%2Daware+visual+cot+with+multi%2Daspect+reward+optimizatio)
13. Zhenglin Huang, Tianxiao Li, Xiangtai Li, Haiquan Wen, Yiwei He, Jiangning Zhang, Hao Fei, Xi Yang, Xiaowei Huang, Bei Peng, and Guangliang Cheng. 2025b. So-fake: Benchmarking and explaining social media image forgery detection. arXiv preprint arXiv:2505.13379. — [arXiv:2505.13379](https://arxiv.org/abs/2505.13379) [find](https://scholar.google.com/scholar?q=Zhenglin+Huang%2C+Tianxiao+Li%2C+Xiangtai+Li%2C+Haiquan+Wen%2C+Yiwei+He%2C+Jiangning+Zhang%2C+Hao+Fei%2C+Xi+Yang%2C+Xiaowei+Huang%2C+Bei+Peng%2C+and+Guangliang+Cheng%2E+2025b%2E+So%2Dfake%3A+Benchmarking+and+explaining+social+me)
14. Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, and 1 others. 2024. [Gpt-4o system card](https://md.research.elamur.ai/free/en/papers/openai-2024-gpt-4o-system-card/). arXiv preprint arXiv:2410.21276.
15. Xiaoqi Jiao, Yichun Yin, Lifeng Shang, Xin Jiang, Xiao Chen, Linlin Li, Fang Wang, and Qun Liu. 2020. [TinyBERT: Distilling BERT for natural language understanding](https://md.research.elamur.ai/free/en/papers/jiao-2020-tinybert-distilling-bert-natural-language/). In EMNLP Findings. Association for Computational Linguistics.
16. Wonjae Kim, Bokyung Son, and Ildoo Kim. 2021. [Vilt: Vision-and-language transformer without convolution or region supervision](https://md.research.elamur.ai/free/en/papers/kim-2021-vilt-vision-language-transformer-convolution/). In ICML.
17. Black Forest Labs, Stephen Batifol, Andreas Blattmann, Frederic Boesel, Saksham Consul, Cyril Diagne, Tim Dockhorn, Jack English, Zion English, Patrick Esser, Sumith Kulal, Kyle Lacey, Yam Levi, Cheng Li, Dominik Lorenz, Jonas Müller, Dustin Podell, Robin Rombach, Harry Saini, and 2 others. 2025. Fl… — [arXiv:2506.15742](https://arxiv.org/abs/2506.15742) [find](https://scholar.google.com/scholar?q=Black+Forest+Labs%2C+Stephen+Batifol%2C+Andreas+Blattmann%2C+Frederic+Boesel%2C+Saksham+Consul%2C+Cyril+Diagne%2C+Tim+Dockhorn%2C+Jack+English%2C+Zion+English%2C+Patrick+Esser%2C+Sumith+Kulal%2C+Kyle+Lacey%2C+Yam+Levi%2C+Cheng)
18. Zongxia Li, Wenhao Yu, Chengsong Huang, Rui Liu, Zhenwen Liang, Fuxiao Liu, Jingxi Che, Dian Yu, Jordan Boyd-Graber, Haitao Mi, and Dong Yu. 2025. Self-rewarding vision-language model via reasoning decomposition. arXiv preprint arXiv:2508.19652. [find](https://scholar.google.com/scholar?q=Zongxia+Li%2C+Wenhao+Yu%2C+Chengsong+Huang%2C+Rui+Liu%2C+Zhenwen+Liang%2C+Fuxiao+Liu%2C+Jingxi+Che%2C+Dian+Yu%2C+Jordan+Boyd%2DGraber%2C+Haitao+Mi%2C+and+Dong+Yu%2E+2025%2E+Self%2Drewarding+vision%2Dlanguage+model+via+reasoning+de)
19. Jingchun Lian, Lingyu Liu, Yaxiong Wang, Yujiao Wu, Lianwei Wu, Li Zhu, and Zhedong Zheng. 2024. Generating attribution reports for manipulated facial images: A dataset and baseline. arXiv preprint arXiv:2412.19685. [find](https://scholar.google.com/scholar?q=Jingchun+Lian%2C+Lingyu+Liu%2C+Yaxiong+Wang%2C+Yujiao+Wu%2C+Lianwei+Wu%2C+Li+Zhu%2C+and+Zhedong+Zheng%2E+2024%2E+Generating+attribution+reports+for+manipulated+facial+images%3A+A+dataset+and+baseline%2E+arXiv+preprint+ar)
20. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023. [Visual instruction tuning](https://md.research.elamur.ai/free/en/papers/liu-2023-visual-instruction-tuning/). In NeurIPS.
21. Xuannan Liu, Peipei Li, Huaibo Huang, Zekun Li, Xing Cui, Jiahao Liang, Lixiong Qin, Weihong Deng, and Zhaofeng He. 2024. Fka-owl: Advancing multimodal fake news detection through knowledge-augmented lvlms. In ACM MM. [find](https://scholar.google.com/scholar?q=Xuannan+Liu%2C+Peipei+Li%2C+Huaibo+Huang%2C+Zekun+Li%2C+Xing+Cui%2C+Jiahao+Liang%2C+Lixiong+Qin%2C+Weihong+Deng%2C+and+Zhaofeng+He%2E+2024%2E+Fka%2Dowl%3A+Advancing+multimodal+fake+news+detection+through+knowledge%2Daugmented+)
22. Xuannan Liu, Zekun Li, Peipei Li, Huaibo Huang, Shuhan Xia, Xing Cui, Linzhi Huang, Weihong Deng, and Zhaofeng He. 2025. Mmfakebench: A mixed-source multimodal misinformation detection benchmark for lvlms. In ICLR. [find](https://scholar.google.com/scholar?q=Xuannan+Liu%2C+Zekun+Li%2C+Peipei+Li%2C+Huaibo+Huang%2C+Shuhan+Xia%2C+Xing+Cui%2C+Linzhi+Huang%2C+Weihong+Deng%2C+and+Zhaofeng+He%2E+2025%2E+Mmfakebench%3A+A+mixed%2Dsource+multimodal+misinformation+detection+benchmark+for+l)
23. Ilya Loshchilov and Frank Hutter. 2017. Decoupled weight decay regularization. In ICLR. [find](https://scholar.google.com/scholar?q=Ilya+Loshchilov+and+Frank+Hutter%2E+2017%2E+Decoupled+weight+decay+regularization%2E+In+ICLR%2E)
24. Ruotian Ma, Peisong Wang, Cheng Liu, Xingyan Liu, Jiaqi Chen, Bang Zhang, Xin Zhou, Nan Du, and Jia Li. 2025. [S 2 R: Teaching LLMs to self-verify and self-correct via reinforcement learning](https://md.research.elamur.ai/free/en/papers/ma-2025-s-2-r-teaching-llms/). In ACL. Association for Computational Linguistics.
25. Yikang Pan, Liangming Pan, Wenhu Chen, Preslav Nakov, Min-Yen Kan, and William Wang. 2023. [On the risk of misinformation pollution with large language models](https://md.research.elamur.ai/free/en/papers/pan-2023-risk-misinformation-pollution-large-language/). In EMNLP Findings. Association for Computational Linguistics.
26. Chan Young Park, Julia Mendelsohn, Anjalie Field, and Yulia Tsvetkov. 2022. Challenges and opportunities in information manipulation detection: An examination of wartime Russian media. In EMNLP Findings. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Chan+Young+Park%2C+Julia+Mendelsohn%2C+Anjalie+Field%2C+and+Yulia+Tsvetkov%2E+2022%2E+Challenges+and+opportunities+in+information+manipulation+detection%3A+An+examination+of+wartime+Russian+media%2E+In+EMNLP+Findin)
27. Rui Shao, Tianxing Wu, and Ziwei Liu. 2023. Detecting and grounding multi-modal media manipulation. In CVPR. [find](https://scholar.google.com/scholar?q=Rui+Shao%2C+Tianxing+Wu%2C+and+Ziwei+Liu%2E+2023%2E+Detecting+and+grounding+multi%2Dmodal+media+manipulation%2E+In+CVPR%2E)
28. Rui Shao, Tianxing Wu, Jianlong Wu, Liqiang Nie, and Ziwei Liu. 2024a. Detecting and grounding multimodal media manipulation and beyond. TPAMI, 46(8):5556–5574. [find](https://scholar.google.com/scholar?q=Rui+Shao%2C+Tianxing+Wu%2C+Jianlong+Wu%2C+Liqiang+Nie%2C+and+Ziwei+Liu%2E+2024a%2E+Detecting+and+grounding+multimodal+media+manipulation+and+beyond%2E+TPAMI%2C+46%288%29%3A5556%E2%80%935574%2E)
29. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, Y. K. Li, Y. Wu, and Daya Guo. 2024b. [Deepseekmath: Pushing the limits of mathematical reasoning in open language models](https://md.research.elamur.ai/free/en/papers/shao-2024-deepseekmath-pushing-limits-mathematical-reasoning/). arXiv preprint arXiv:2402.03300.
30. Jinjie Shen, Yaxiong Wang, Lechao Cheng, Nan Pu, and Zhun Zhong. 2025. Beyond artificial misalignment: Detecting and grounding semantic-coordinated multimodal manipulations. In ACM MM. [find](https://scholar.google.com/scholar?q=Jinjie+Shen%2C+Yaxiong+Wang%2C+Lechao+Cheng%2C+Nan+Pu%2C+and+Zhun+Zhong%2E+2025%2E+Beyond+artificial+misalignment%3A+Detecting+and+grounding+semantic%2Dcoordinated+multimodal+manipulations%2E+In+ACM+MM%2E)
31. StabilityAI. 2023. Introducing stable diffusion 3.5. Accessed: 2025-08-24. [find](https://scholar.google.com/scholar?q=StabilityAI%2E+2023%2E+Introducing+stable+diffusion+3%2E5%2E+Accessed%3A+2025%2D08%2D24%2E)
32. Yixuan Su, Tian Lan, Huayang Li, Jialu Xu, Yan Wang, and Deng Cai. 2023. [PandaGPT: One model to instruction-follow them all](https://md.research.elamur.ai/free/en/papers/su-2023-pandagpt-one-model-instruction-follow/). In Proceedings of the 1st Workshop on Taming Large Language Models. Association for Computational Linguistics.
33. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, and 49 others… — [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://md.research.elamur.ai/free/en/papers/touvron-2023-llama-2-open-foundation-fine/)
34. Qingyan Wang, Lianwei Wu, Yuanxia Zeng, Linyong Wang, Kang Wang, Yaxiong Wang, and Chao Gao. 2025c. Cross-modal consistency reasoning with large language models for short video-based fake news detection. In Proceedings of the 2nd International Workshop on Diffusion of Harmful Content on Online Web,… [find](https://scholar.google.com/scholar?q=Qingyan+Wang%2C+Lianwei+Wu%2C+Yuanxia+Zeng%2C+Linyong+Wang%2C+Kang+Wang%2C+Yaxiong+Wang%2C+and+Chao+Gao%2E+2025c%2E+Cross%2Dmodal+consistency+reasoning+with+large+language+models+for+short+video%2Dbased+fake+news+detecti)
35. Weiyun Wang, Zhangwei Gao, Lixin Gu, Hengjun Pu, Long Cui, Xingguang Wei, Zhaoyang Liu, Linglin Jing, Shenglong Ye, Jie Shao, and 1 others. 2025d. [Internvl3](https://md.research.elamur.ai/free/en/papers/wang-2025-internvl3-5-advancing-open/). 5: Advancing open-source multimodal models in versatility, reasoning, and efficiency. arXiv preprint arXiv:2508.18265.
36. Zhiyu Wu, Xiaokang Chen, Zizheng Pan, Xingchao Liu, Wen Liu, Damai Dai, Huazuo Gao, Yiyang Ma, Chengyue Wu, Bingxuan Wang, and 1 others. 2024. [Deepseek-vl2: Mixture-of-experts vision-language models for advanced multimodal understanding](https://md.research.elamur.ai/free/en/papers/wu-2024-deepseek-vl2-mixture-experts/). arXiv preprint arXiv:2412.10302.
37. Zehong Yan, Peng Qi, Wynne Hsu, and Mong-Li Lee. 2025. TRUST-VL: An explainable news assistant for general multimodal misinformation detection. In EMNLP. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Zehong+Yan%2C+Peng+Qi%2C+Wynne+Hsu%2C+and+Mong%2DLi+Lee%2E+2025%2E+TRUST%2DVL%3A+An+explainable+news+assistant+for+general+multimodal+misinformation+detection%2E+In+EMNLP%2E+Association+for+Computational+Linguistics%2E)
38. Jiabo Ye, Haiyang Xu, Haowei Liu, Anwen Hu, Ming Yan, Qi Qian, Ji Zhang, Fei Huang, and Jingren Zhou. 2025. [mplug-owl3: Towards long image-sequence understanding in multi-modal large language models](https://md.research.elamur.ai/free/en/papers/ye-2024-mplug-owl3-towards-long-image/). In ICLR.
39. Ruohong Zhang, Bowen Zhang, Yanghao Li, Haotian Zhang, Zhiqing Sun, Zhe Gan, Yinfei Yang, Ruoming Pang, and Yiming Yang. 2025a. [Improve vision language model chain-of-thought reasoning](https://md.research.elamur.ai/free/en/papers/zhang-2025-improve-vision-language-model-chain/). In ACL.
40. Yuchen Zhang, Yaxiong Wang, Yujiao Wu, Lianwei Wu, and Li Zhu. 2025b. The coherence trap: When mllm-crafted narratives exploit manipulated visual contexts. arXiv preprint arXiv:2505.17476. — [arXiv:2505.17476](https://arxiv.org/abs/2505.17476) [find](https://scholar.google.com/scholar?q=Yuchen+Zhang%2C+Yaxiong+Wang%2C+Yujiao+Wu%2C+Lianwei+Wu%2C+and+Li+Zhu%2E+2025b%2E+The+coherence+trap%3A+When+mllm%2Dcrafted+narratives+exploit+manipulated+visual+contexts%2E+arXiv+preprint+arXiv%3A2505%2E17476%2E+%E2%80%94+%5BarXiv%3A25)
41. Zhenxing Zhang, Yaxiong Wang, Lechao Cheng, Zhun Zhong, Dan Guo, and Meng Wang. 2025c. Asap: Advancing semantic alignment promotes multi-modal manipulation detecting and grounding. In CVPR. [find](https://scholar.google.com/scholar?q=Zhenxing+Zhang%2C+Yaxiong+Wang%2C+Lechao+Cheng%2C+Zhun+Zhong%2C+Dan+Guo%2C+and+Meng+Wang%2E+2025c%2E+Asap%3A+Advancing+semantic+alignment+promotes+multi%2Dmodal+manipulation+detecting+and+grounding%2E+In+CVPR%2E)
42. Tong Zheng, Hongming Zhang, Wenhao Yu, Xiaoyang Wang, Runpeng Dai, Rui Liu, Huiwen Bao, Chengsong Huang, Heng Huang, and Dong Yu. 2025. Parallel-r1: Towards parallel thinking via reinforcement learning. arXiv preprint arXiv:2509.07980. — [arXiv:2509.07980](https://arxiv.org/abs/2509.07980) [find](https://scholar.google.com/scholar?q=Tong+Zheng%2C+Hongming+Zhang%2C+Wenhao+Yu%2C+Xiaoyang+Wang%2C+Runpeng+Dai%2C+Rui+Liu%2C+Huiwen+Bao%2C+Chengsong+Huang%2C+Heng+Huang%2C+and+Dong+Yu%2E+2025%2E+Parallel%2Dr1%3A+Towards+parallel+thinking+via+reinforcement+learnin)
43. Junxi Wang, Jize Liu, Na Zhang, and Yaxiong Wang. 2025a. Consistency-aware fake videos detection on short video platforms. In International Conference on Intelligent Computing, pages 200–210. Springer. [find](https://scholar.google.com/scholar?q=Junxi+Wang%2C+Jize+Liu%2C+Na+Zhang%2C+and+Yaxiong+Wang%2E+2025a%2E+Consistency%2Daware+fake+videos+detection+on+short+video+platforms%2E+In+International+Conference+on+Intelligent+Computing%2C+pages+200%E2%80%93210%2E+Springer)
44. JunXi Wang, Yaxiong Wang, Lechao Cheng, and Zhun Zhong. 2025b. FakeSV-VLM: Taming VLM for detecting fake short-video news via progressive mixture-of-experts adapter. In EMNLP Findings. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=JunXi+Wang%2C+Yaxiong+Wang%2C+Lechao+Cheng%2C+and+Zhun+Zhong%2E+2025b%2E+FakeSV%2DVLM%3A+Taming+VLM+for+detecting+fake+short%2Dvideo+news+via+progressive+mixture%2Dof%2Dexperts+adapter%2E+In+EMNLP+Findings%2E+Association+for)
45. Cultivating Forensic Reasoning for Generalizable Multimodal Manipulation Detection [find](https://scholar.google.com/scholar?q=Cultivating+Forensic+Reasoning+for+Generalizable+Multimodal+Manipulation+Detection)

*Updated 2 November 2026*