README.md

metadata

license: apache-2.0
task_categories:
  - visual-question-answering
language:
  - en
size_categories:
  - 1K<n<10K

VLM-Normal-Requests

We filter the VisionArena-Chat dataset to retain only English prompts. This dataset is used for VLM fingerprint request stealthiness analysis.

Citation

@article{zhao2026sif,
  title={SIF: Semantically In-Distribution Fingerprints for Large Vision-Language Models},
  author={Zhao, Yifei and Lou, Qian and Zheng, Mengxin},
  journal={arXiv preprint arXiv:2604.17041},
  year={2026}
}

@inproceedings{chou2025visionarena,
  title={Visionarena: 230k real world user-vlm conversations with preference labels},
  author={Chou, Christopher and Dunlap, Lisa and Mashita, Koki and Mandal, Krishna and Darrell, Trevor and Stoica, Ion and Gonzalez, Joseph E and Chiang, Wei-Lin},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={3877--3887},
  year={2025}
}