MirageTVQA: Multilingual Table Visual Question Answering Benchmark

Citation

If you use MirageTVQA in your research, please cite:

@inproceedings{singh2025mirage,
  title={Lost in Translation and Noise: A Deep Dive into the Failure Modes of VLMs on Real-World Tables},
  author={Singh, Anshul and Chaudhary, Rohan and Singh, Gagneet and Kumar, Abhay},
  booktitle={AI for Tabular Data Workshop at EurIPS},
  year={2025}
}

Acknowledgments

We thank the creators of WikiSQL, FinQA, and the arXiv dataset for providing the source tables. We also acknowledge the use of Qwen, Gemini, and other open-source models in our data generation pipeline.

---

⭐ If you find MirageTVQA useful, please consider starring the repository!