This is the repository for the paper: Format Matters: The Robustness of Multimodal LLMs in Reviewing Evidence from Tables and Charts (AAAI 2026)
- download data.zip
- download outputs.zip
Edit base_path in Line 49 to obtain all results in Table 2.
python3 run_eval.pypython3 run_claim_table.pypython3 run_claim_img.pypython3 run_claim_combine.pypython3 run_eval.pyPlease cite our paper as follows:
@article{Ho_Wu_Kumar_Boudin_Takasu_Aizawa_2026,
title={Format Matters: The Robustness of Multimodal LLMs in Reviewing Evidence from Tables and Charts},
volume={40},
url={https://ojs.aaai.org/index.php/AAAI/article/view/40361},
DOI={10.1609/aaai.v40i37.40361},
number={37}, journal={Proceedings of the AAAI Conference on Artificial Intelligence},
author={Ho, Xanh and Wu, Yun-Ang and Kumar, Sunisth and Boudin, Florian and Takasu, Atsuhiro and Aizawa, Akiko},
year={2026},
month={Mar.},
pages={31014-31022}
}
The structure of the code in this repository is based on: https://github.com/Alab-NII/SciTabAlign