README.md

metadata

license: cc-by-nc-4.0
task_categories:
  - image-text-to-text

AVerImaTeC-Filled

This dataset provides a filled knowledge store and precomputed embeddings for the AVerImaTeC shared task, supporting the system described in the paper VILLAIN at AVerImaTeC: Verifying Image-Text Claims via Multi-Agent Collaboration.

Code | Paper

Dataset Description

Generating embeddings for knowledge stores (both text and image) is computationally expensive and time-consuming. This repository releases precomputed assets to facilitate research and reproducibility for the AVerImaTeC task.

It includes:

• Filled text_related knowledge store: Data for the validation and test splits.
• Vector Store: For saving and loading persistent data.
• Precomputed text embeddings: For both the original and filled knowledge bases.
• Precomputed image embeddings: For both the original and filled knowledge bases.

Citation

If you find this dataset or the associated code useful, please cite:

@article{jung2026villain,
  title={VILLAIN at AVerImaTeC: Verifying Image-Text Claims via Multi-Agent Collaboration},
  author={Jung, Jaeyoon and Yoon, Yejun and Yoon, Seunghyun and Park, Kunwoo},
  journal={arXiv preprint arXiv:2602.04587},
  year={2026}
}