Elamur Research⌕
☰
home / papers / 2025 / Stop Looking for “Important Tokens” in Multimodal Language Models: Duplication Matters More

Stop Looking for “Important Tokens” in Multimodal Language Models: Duplication Matters More

Paper · 2025 · Zichen Wen, Yifeng Gao, Shaobo Wang et al.

Summary

Vision tokens make up most of the input to multimodal LLMs, which makes inference expensive. Most training-free pruning methods score each token's "importance" from attention maps and drop the low scorers. The authors argue this approach has four weaknesses. It treats scores as fixed even after other tokens are removed. It needs explicit attention maps, which rules out FlashAttention. It inherits positional bias in attention. And it often does worse than random pruning: FastV and SparseVLM lose to random selection in about two thirds of tested cases on LLaVA-1.5-7B. As an alternative, they propose DART (Duplication-Aware Reduction of Tokens). DART picks a few pivot tokens, chosen by norm or even at random, and computes cosine similarity between the pivots and the remaining tokens. It then keeps the tokens least similar to the pivots, so redundant content is discarded first. The method needs no training, is plug-and-play, and works with efficient attention kernels. Tests on four MLLMs and more than ten image and video benchmarks show that DART keeps accuracy high under heavy compression. This suggests that removing redundancy matters more than finding "important" tokens.

Key points

  • With 88.9% of vision tokens pruned on LLaVA-1.5-7B, DART retains 93.7% of the original performance. The next-best method retains 91.5%.
  • The method runs about 1.99× faster overall and 2.99× faster in the prefilling stage. Token selection takes at most roughly 0.08 s.
  • Pivots make up no more than 2% of tokens. Using too few or too many pivots (e.g., 20+) hurts results. Pivots drawn from both visual and text tokens work better than pivots from only one modality.
  • The authors evaluated DART on LLaVA-1.5-7B, Qwen2-VL-7B and MiniCPM-V2.6, comparing mainly against FastV. They also tested Video-LLaVA on TGIF, MSVD and MSRVT with 50% of tokens kept. In the video setting, DART averages 58.0 accuracy, against 57.1 for FastV and 58.2 for the unpruned model.
  • Retained token sets that overlap by less than half can perform similarly well, and pruning may reduce hallucinations. DART cannot be applied to closed black-box models such as GPT or Claude. Code: github.com/ZichenWen1/DART.

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

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksComputer Vision and Pattern Recognition · 2,869 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 3,320 worksVision & language · 657 works

Cite

Cite as Zichen Wen, Yifeng Gao, Shaobo Wang, Junyuan Zhang, Qintong Zhang, Weijia Li, Conghui He and Linfeng Zhang. 2025. Stop Looking for “Important Tokens” in Multimodal Language Models: Duplication Matters More. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.emnlp-main.505

@inproceedings{wen2025stop,
  title = {{Stop Looking for “Important Tokens” in Multimodal Language Models: Duplication Matters More}},
  author = {Zichen Wen, Yifeng Gao, Shaobo Wang, Junyuan Zhang, Qintong Zhang, Weijia Li, Conghui He and Linfeng Zhang},
  booktitle = {Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},
  year = {2025},
  doi = {10.18653/v1/2025.emnlp-main.505},
  url = {https://doi.org/10.18653/v1/2025.emnlp-main.505}
}

References

73 references, 26 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Abdelrahman Abouelenin, Atabak Ashfaq, Adam Atkinson, Hany Awadalla, Nguyen Bach, Jianmin Bao, Alon Benhaim, Martin Cai, Vishrav Chaudhary, Congcong Chen, et al. 2025. Phi-4-mini technical report: Compact yet powerful multimodal language models via mixture-of-loras. arXiv preprint arXiv:2503.01743.
  2. Ruichuan An, Sihan Yang, Ming Lu, Kai Zeng, Yulin Luo, Ying Chen, Jiajun Cao, Hao Liang, Qi She, Shanghang Zhang, et al. 2024. Mc-llava: Multi-concept personalized vision-language model. arXiv preprint arXiv:2411.11706. — [arXiv:2411.11706](https://arxiv.org/abs/2411.11706) find
  3. Kazi Hasan Ibn Arif, JinYi Yoon, Dimitrios S Nikolopoulos, Hans Vandierendonck, Deepu John, and Bo Ji. 2024. Hired: Attention-guided token dropping for efficient inference of high-resolution vision-language models in resource-constrained environments. arXiv preprint arXiv:2408.10945. find
  4. Anas Awadalla, Irena Gao, Josh Gardner, Jack Hessel, Yusuf Hanafy, Wanrong Zhu, Kalyani Marathe, Yonatan Bitton, Samir Gadre, Shiori Sagawa, Jenia Jitsev, Simon Kornblith, Pang Wei Koh, Gabriel Ilharco, Mitchell Wortsman, and Ludwig Schmidt. 2023. Openflamingo: An open-source framework for training…
  5. Jeffrey P Bigham, Chandrika Jayant, Hanjie Ji, Greg Little, Andrew Miller, Robert C Miller, Robin Miller, Aubrey Tatarowicz, Brandyn White, Samual White, et al. 2010. Vizwiz: nearly real-time answers to visual questions. In Proceedings of the 23nd annual ACM symposium on User interface software and… find
  6. Kevin Black, Noah Brown, Danny Driess, Adnan Esmail, Michael Equi, Chelsea Finn, Niccolo Fusai, Lachy Groom, Karol Hausman, Brian Ichter, et al. 2024. π 0 : A vision-language-action flow model for general robot control. arXiv preprint arXiv:2410.24164. find
  7. Daniel Bolya, Cheng-Yang Fu, Xiaoliang Dai, Peizhao Zhang, Christoph Feichtenhofer, and Judy Hoffman. 2023. Token merging: Your ViT but faster. In International Conference on Learning Representations. find
  8. Anthony Brohan, Noah Brown, Justice Carbajal, Yevgen Chebotar, Xi Chen, Krzysztof Choromanski, Tianli Ding, Danny Driess, Avinava Dubey, Chelsea Finn, Pete Florence, Chuyuan Fu, Montse Gonzalez Arenas, Keerthana Gopalakrishnan, Kehang Han, Karol Hausman, Alexander Herzog, Jasmine Hsu, Brian Ichter,… find
  9. Jianjian Cao, Peng Ye, Shengze Li, Chong Yu, Yansong Tang, Jiwen Lu, and Tao Chen. 2024. Madtp: Multimodal alignment-guided dynamic token pruning for accelerating vision-language transformer. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 15710–15719. find
  10. 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.
  11. Yunfei Chu, Jin Xu, Qian Yang, Haojie Wei, Xipin Wei, Zhifang Guo, Yichong Leng, Yuanjun Lv, Jinzheng He, Junyang Lin, Chang Zhou, and Jingren Zhou. 2024. Qwen2-audio technical report. arXiv preprint arXiv:2407.10759.
  12. Alexis Conneau, Min Ma, Simran Khanuja, Yu Zhang, Vera Axelrod, Siddharth Dalmia, Jason Riesa, Clara Rivera, and Ankur Bapna. 2023. Fleurs: Few-shot learning evaluation of universal representations of speech. In 2022 IEEE Spoken Language Technology Workshop (SLT), pages 798–805. IEEE.
  13. Tri Dao. 2024. FlashAttention-2: Faster attention with better parallelism and work partitioning. In International Conference on Learning Representations (ICLR).
  14. Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, Ted Xiao, Ashwin Balakrishna, Suraj Nair, Rafael Rafailov, Ethan Foster, Grace Lam, Pannag Sanketi, Quan Vuong, Thomas Kollar, Benjamin Burchfiel, Russ Tedrake, Dorsa Sadigh, Sergey Levine, Percy Liang, and Chelsea Finn. 2024. Openvla: An open-source… find
  15. Tri Dao, Daniel Y. Fu, Stefano Ermon, Atri Rudra, and Christopher Ré. 2022. FlashAttention: Fast and memory-efficient exact attention with IO-awareness. In Advances in Neural Information Processing Systems (NeurIPS). find
  16. Dan Kondratyuk, Lijun Yu, Xiuye Gu, José Lezama, Jonathan Huang, Rachel Hornung, Hartwig Adam, Hassan Akbari, Yair Alon, Vighnesh Birodkar, et al. 2023. Videopoet: A large language model for zero-shot video generation. arXiv:2312.14125. — [arXiv:2312.14125](https://arxiv.org/abs/2312.14125) find
  17. Mark Endo, Xiaohan Wang, and Serena Yeung-Levy. 2024. Feather the throttle: Revisiting visual token pruning for vision-language model acceleration. arXiv preprint arXiv:2412.13180. — [arXiv:2412.13180](https://arxiv.org/abs/2412.13180) find
  18. Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023a. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In International conference on machine learning, pages 19730–19742. PMLR.
  19. Hao-Shu Fang, Hongjie Fang, Zhenyu Tang, Jirong Liu, Chenxi Wang, Junbo Wang, Haoyi Zhu, and Cewu Lu. 2024. Rh20t: A comprehensive robotic dataset for learning diverse skills in one-shot. In 2024 IEEE International Conference on Robotics and Automation (ICRA), pages 653–660. IEEE. find
  20. Chaoyou Fu, Peixian Chen, Yunhang Shen, Yulei Qin, Mengdan Zhang, Xu Lin, Jinrui Yang, Xiawu Zheng, Ke Li, Xing Sun, et al. 2023. MME: A comprehensive evaluation benchmark for multimodal large language models. arXiv:2306.13394.
  21. Chengyue Gong, Dilin Wang, Meng Li, Vikas Chandra, and Qiang Liu. 2021. Vision transformers with patch diversification. arXiv preprint arXiv:2104.12753. — [arXiv:2104.12753](https://arxiv.org/abs/2104.12753) find
  22. Yash Goyal, Tejas Khot, Douglas Summers-Stay, Dhruv Batra, and Devi Parikh. 2017. Making the v in vqa matter: Elevating the role of image understanding in visual question answering. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 6904–6913.
  23. Yuhang Han, Xuyang Liu, Pengxiang Ding, Donglin Wang, Honggang Chen, Qingsen Yan, and Siteng Huang. 2024. Rethinking token reduction in mllms: Towards a unified paradigm for training-free acceleration. arXiv preprint arXiv:2411.17686. — [arXiv:2411.17686](https://arxiv.org/abs/2411.17686) find
  24. Drew A Hudson and Christopher D Manning. 2019. GQA: A new dataset for real-world visual reasoning and compositional question answering. Conference on Computer Vision and Pattern Recognition (CVPR).
  25. Yunseok Jang, Yale Song, Youngjae Yu, Youngjin Kim, and Gunhee Kim. 2017. Tgif-qa: Toward spatiotemporal reasoning in visual question answering. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2758–2766. find
  26. Hengrui Kang, Siwei Wen, Zichen Wen, Junyan Ye, Weijia Li, Peilin Feng, Baichuan Zhou, Bin Wang, Dahua Lin, Linfeng Zhang, et al. 2025. Legion: Learning to ground and explain for synthetic image detection. arXiv preprint arXiv:2503.15264. — [arXiv:2503.15264](https://arxiv.org/abs/2503.15264) find
  27. Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. 2022. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In International conference on machine learning, pages 12888–12900. PMLR.
  28. Qixiu Li, Yaobo Liang, Zeyu Wang, Lin Luo, Xi Chen, Mozheng Liao, Fangyun Wei, Yu Deng, Sicheng Xu, Yizhong Zhang, Xiaofan Wang, Bei Liu, Jianlong Fu, Jianmin Bao, Dong Chen, Yuanchun Shi, Jiaolong Yang, and Baining Guo. 2024a. Cogact: A foundational vision-language-action model for synergizing cogn… — [arXiv:2411.19650](https://arxiv.org/abs/2411.19650) find
  29. Xuanlin Li, Kyle Hsu, Jiayuan Gu, Karl Pertsch, Oier Mees, Homer Rich Walke, Chuyuan Fu, Ishikaa Lu-nawat, Isabel Sieh, Sean Kirmani, et al. 2024b. Evaluating real-world robot manipulation policies in simulation. arXiv preprint arXiv:2405.05941. — [arXiv:2405.05941](https://arxiv.org/abs/2405.05941) find
  30. Yanwei Li, Chengyao Wang, and Jiaya Jia. 2024c. LLaMA-VID: An image is worth 2 tokens in large language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. find
  31. Yanwei Li, Yuechen Zhang, Chengyao Wang, Zhisheng Zhong, Yixin Chen, Ruihang Chu, Shaoteng Liu, and Jiaya Jia. 2024d. Mini-gemini: Mining the potential of multi-modality vision language models. arXiv:2403.18814. find
  32. Yifan Li, Yifan Du, Kun Zhou, Jinpeng Wang, Wayne Xin Zhao, and Ji-Rong Wen. 2023b. Evaluating object hallucination in large vision-language models. arXiv:2305.10355.
  33. Y Liang, C Ge, Z Tong, Y Song, P Xie, et al. 2022. Not all patches are what you need: Expediting vision transformers via token reorganizations. In ICLR. find
  34. Bin Lin, Bin Zhu, Yang Ye, Munan Ning, Peng Jin, and Li Yuan. 2023. Video-llava: Learning united visual representation by alignment before projection. arXiv:2311.10122.
  35. Hao Liu, Wilson Yan, Matei Zaharia, and Pieter Abbeel. 2024a. World model on million-length video and language with ringattention. Preprint, arXiv:2402.08268. find
  36. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2023. Improved baselines with visual instruction tuning. arXiv:2310.03744.
  37. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024b. Improved baselines with visual instruction tuning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26296–26306.
  38. Haotian Liu, Chunyuan Li, Yuheng Li, Bo Li, Yuanhan Zhang, Sheng Shen, and Yong Jae Lee. 2024c. Llava-next: Improved reasoning, ocr, and world knowledge. find
  39. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2024d. Visual instruction tuning. Advances in neural information processing systems.
  40. Ting Liu, Liangtao Shi, Richang Hong, Yue Hu, Quanjun Yin, and Linfeng Zhang. 2024e. Multi-stage vision token dropping: Towards efficient multimodal large language model. arXiv preprint arXiv:2411.10803. — [arXiv:2411.10803](https://arxiv.org/abs/2411.10803) find
  41. Xuyang Liu, Ziming Wang, Yuhang Han, Yingyao Wang, Jiale Yuan, Jun Song, Bo Zheng, Linfeng Zhang, Siteng Huang, and Honggang Chen. 2025a. Compression with global guidance: Towards training-free high-resolution mllms acceleration. arXiv preprint arXiv:2501.05179. — [arXiv:2501.05179](https://arxiv.org/abs/2501.05179) find
  42. Xuyang Liu, Zichen Wen, Shaobo Wang, Junjie Chen, Zhishan Tao, Yubo Wang, Xiangqi Jin, Chang Zou, Yiyu Wang, Chenfei Liao, et al. 2025b. Shifting ai efficiency from model-centric to data-centric compression. arXiv preprint arXiv:2505.19147. — [arXiv:2505.19147](https://arxiv.org/abs/2505.19147) find
  43. Yuan Liu, Haodong Duan, Yuanhan Zhang, Bo Li, Songyang Zhang, Wangbo Zhao, Yike Yuan, Jiaqi Wang, Conghui He, Ziwei Liu, et al. 2025c. Mmbench: Is your multi-modal model an all-around player? In European Conference on Computer Vision, pages 216–233. Springer.
  44. Yuliang Liu, Zhang Li, Mingxin Huang, Biao Yang, Wenwen Yu, Chunyuan Li, Xu-Cheng Yin, Cheng-Lin Liu, Lianwen Jin, and Xiang Bai. 2024f. Ocrbench: on the hidden mystery of ocr in large multimodal models. Science China Information Sciences, 67(12):220102.
  45. Pan Lu, Swaroop Mishra, Tanglin Xia, Liang Qiu, Kai-Wei Chang, Song-Chun Zhu, Oyvind Tafjord, Peter Clark, and Ashwin Kalyan. 2022. Learn to explain: Multimodal reasoning via thought chains for science question answering. Advances in Neural Information Processing Systems, 35:2507–2521. find
  46. Yulin Luo, Ruichuan An, Bocheng Zou, Yiming Tang, Jiaming Liu, and Shanghang Zhang. 2024. Llm as dataset analyst: Subpopulation structure discovery with large language model. In European Conference on Computer Vision, pages 235–252. Springer. find
  47. David Marr. 2010. Vision: A computational investigation into the human representation and processing of visual information. MIT press. find
  48. Tam Nguyen, Tan Nguyen, and Richard Baraniuk. 2023. Mitigating over-smoothing in transformers via regularized nonlocal functionals. Advances in Neural Information Processing Systems, 36:80233–80256. find
  49. Abby O’Neill, Abdul Rehman, Abhiram Maddukuri, Abhishek Gupta, Abhishek Padalkar, Abraham Lee, Acorn Pooley, Agrim Gupta, Ajay Mandlekar, Ajinkya Jain, et al. 2024. Open x-embodiment: Robotic learning datasets and rt-x models: Open x-embodiment collaboration 0. In 2024 IEEE International Conference… find
  50. Se Jin Park, Julian Salazar, Aren Jansen, Keisuke Kinoshita, Yong Man Ro, and RJ Skerry-Ryan. 2024. Long-form speech generation with spoken language models. arXiv preprint arXiv:2412.18603. — [arXiv:2412.18603](https://arxiv.org/abs/2412.18603) find
  51. Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learning, pages 8748–8763. PM…
  52. Yuzhang Shang, Mu Cai, Bingxin Xu, Yong Jae Lee, and Yan Yan. 2024. Llava-prumerge: Adaptive token reduction for efficient large multimodal models. arXiv preprint arXiv:2403.15388. — [arXiv:2403.15388](https://arxiv.org/abs/2403.15388) find
  53. Amanpreet Singh, Vivek Natarjan, Meet Shah, Yu Jiang, Xinlei Chen, Devi Parikh, and Marcus Rohrbach. 2019. Towards VQA models that can read. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 8317–8326. find
  54. Yunlong Tang, Jing Bi, Siting Xu, Luchuan Song, Susan Liang, Teng Wang, Daoan Zhang, Jie An, Jingyang Lin, Rongyi Zhu, et al. 2023. Video understanding with large language models: A survey. arXiv preprint arXiv:2312.17432.
  55. Gemini Team, Rohan Anil, Sebastian Borgeaud, Yonghui Wu, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, et al. 2023. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805.
  56. Kimi Team. 2024. Kimi-audio technical report. Preprint, arXiv:arXiv:placeholder.
  57. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. arXiv:1706.03762.
  58. Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, et al. 2024a. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution. arXiv preprint arXiv:2409.12191.
  59. Yi Wang, Kunchang Li, Xinhao Li, Jiashuo Yu, Yinan He, Guo Chen, Baoqi Pei, Rongkun Zheng, Jilan Xu, Zun Wang, et al. 2024b. Internvideo2: Scaling video foundation models for multimodal video understanding. Arxiv e-prints, pages arXiv–2403. find
  60. Yiqi Wang, Wentao Chen, Xiaotian Han, Xudong Lin, Haiteng Zhao, Yongfei Liu, Bohan Zhai, Jianbo Yuan, Quanzeng You, and Hongxia Yang. 2024c. Exploring the reasoning abilities of multimodal large language models (mllms): A comprehensive survey on emerging trends in multimodal reasoning. arXiv preprin… — [arXiv:2401.06805](https://arxiv.org/abs/2401.06805) find
  61. Zhehao Wang, Xian Lin, Nannan Wu, Li Yu, Kwang-Ting Cheng, and Zengqiang Yan. 2024d. Dtmformer: Dynamic token merging for boosting transformer-based medical image segmentation. In Proceedings of the AAAI Conference on Artificial Intelligence, pages 5814–5822. find
  62. 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? arXiv preprint arXiv:2502.11501. find
  63. In Proceedings of the AAAI Conference on Artificial Intelligence, pages 22128–22136. find
  64. Junyuan Zhang, Qintong Zhang, Bin Wang, Linke Ouyang, Zichen Wen, Ying Li, Ka-Ho Chow, Conghui He, and Wentao Zhang. 2024a. Ocr hinders rag: Evaluating the cascading impact of ocr on retrieval-augmented generation. arXiv preprint arXiv:2412.02592. — [arXiv:2412.02592](https://arxiv.org/abs/2412.02592) find
  65. Qizhe Zhang, Aosong Cheng, Ming Lu, Zhiyong Zhuo, Minqi Wang, Jiajun Cao, Shaobo Guo, Qi She, and Shanghang Zhang. 2024b. [cls] attention is all you need for training-free visual token pruning: Make vlm inference faster. arXiv preprint arXiv:2412.01818. — [arXiv:2412.01818](https://arxiv.org/abs/2412.01818) find
  66. Yuan Zhang, Chun-Kai Fan, Junpeng Ma, Wenzhao Zheng, Tao Huang, Kuan Cheng, Denis Gudovskiy, Tomoyuki Okuno, Yohei Nakata, Kurt Keutzer, et al. 2024c. Sparsevlm: Visual token sparsification for efficient vision-language model inference. arXiv preprint arXiv:2410.04417. find
  67. Deyao Zhu, Jun Chen, Xiaoqian Shen, Xiang Li, and Mohamed Elhoseiny. 2023. Minigpt-4: Enhancing vision-language understanding with advanced large language models. arXiv preprint arXiv:2304.10592.
  68. Long Xing, Qidong Huang, Xiaoyi Dong, Jiajie Lu, Pan Zhang, Yuhang Zang, Yuhang Cao, Conghui He, Jiaqi Wang, Feng Wu, and Dahua Lin. 2024. Pyramiddrop: Accelerating your large vision-language models via pyramid visual redundancy reduction. arXiv preprint arXiv:2410.17247. — [arXiv:2410.17247](https://arxiv.org/abs/2410.17247) find
  69. Dejing Xu, Zhou Zhao, Jun Xiao, Fei Wu, Hanwang Zhang, Xiangnan He, and Yueting Zhuang. 2017. Video question answering via gradually refined attention over appearance and motion. In Proceedings of the ACM international conference on Multimedia, pages 1645–1653. find
  70. Siyu Xu, Yunke Wang, Chenghao Xia, Dihao Zhu, Tao Huang, and Chang Xu. 2025. Vla-cache: Towards efficient vision-language-action model via adaptive token caching in robotic manipulation. arXiv preprint arXiv:2502.02175. find
  71. Linli Yao, Lei Li, Shuhuai Ren, Lean Wang, Yuanxin Liu, Xu Sun, and Lu Hou. 2024a. DeCo: Decoupling token compression from semantic abstraction in multimodal large language models. arXiv:2405.20985. — [arXiv:2405.20985](https://arxiv.org/abs/2405.20985) find
  72. Yuan Yao, Tianyu Yu, Ao Zhang, Chongyi Wang, Junbo Cui, Hongji Zhu, Tianchi Cai, Haoyu Li, Weilin Zhao, Zhihui He, et al. 2024b. Minicpm-v: A gpt-4v level mllm on your phone. arXiv preprint arXiv:2408.01800.
  73. Weihao Ye, Qiong Wu, Wenhao Lin, and Yiyi Zhou. 2025. Fit and prune: Fast and training-free visual token pruning for multi-modal large language models. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wen-2025-stop-looking-important-tokens-multimodal/