ChartHal

# 🌀 ChartHal

**ChatHal** is a benchmark to comprehensively evaluate hallucination of visual language models (VLMs) in chart understanding.

GitHub: https://github.com/ymcui/ChartHal

Paper: https://arxiv.org/abs/2509.17481

## Usage

Please directly clone the dataset.

```
git clone https://huggingface.co/datasets/hfl/ChartHal && cd ChartHal && unzip -q images.zip
```

To perform inference, please check: [https://github.com/ymcui/ChartHal](https://github.com/ymcui/ChartHal)

## Citation

If your find our work helpful, please consider cite our paper.

```
@article{charthal,
title={{ChartHal}: A Fine-grained Framework Evaluating Hallucination of Large Vision Language Models in Chart Understanding},
author={Wang, Xingqi and Cui, Yiming and Yao, Xin and Wang, Shijin and Hu, Guoping and Qin, Xiaoyu},
year={2025},
eprint={2509.17481},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```

# 🌀 ChartHal

**ChartHal** 是一款用于全面评估视觉语言模型（Visual Language Models, VLMs）在图表理解任务中产生幻觉现象的基准评测集。

GitHub：https://github.com/ymcui/ChartHal

论文地址：https://arxiv.org/abs/2509.17481

## 使用方法

请直接克隆该数据集。

git clone https://huggingface.co/datasets/hfl/ChartHal && cd ChartHal && unzip -q images.zip

如需执行推理，请参考：[https://github.com/ymcui/ChartHal]

## 引用说明

若本研究对您的工作有所助益，请引用我们的论文。

@article{charthal,
title={{"ChartHal"}: A Fine-grained Framework Evaluating Hallucination of Large Vision Language Models in Chart Understanding},
author={Wang, Xingqi and Cui, Yiming and Yao, Xin and Wang, Shijin and Hu, Guoping and Qin, Xiaoyu},
year={2025},
eprint={2509.17481},
archivePrefix={arXiv},
primaryClass={cs.CL}
}