README.md

---
license: cc-by-nc-4.0
---

# Aftermath of DrawEduMath

This contains `answer_mapping.json`, for recreating the results of the paper titled "The Aftermath of DrawEduMath: Vision Language Models Underperform with Struggling Students and Misdiagnose Errors".

This file contains outputs from GPT-5-mini remapping teachers' answers around whether the student image contains any errors to 'yes' or 'no'.

Please consult the datacard for [DrawEduMath](https://huggingface.co/datasets/allenai/DrawEduMath) for detailed information about data source. 

Quick links: 
- Aftermath of DrawEduMath code: https://github.com/lucy3/aftermath_drawedumath/tree/main
- Aftermath of DrawEduMath paper: https://arxiv.org/abs/2603.00925
## Example Use
```
import json
with open('answer_mapping.json', 'r') as infile:
    answer_mapping = json.load(infile)
```
## License 
This dataset is licensed under CC-BY-NC-4.0. It is intended for research and educational purposes following ASSISTments's [Responsible Use Guidelines](https://sites.google.com/view/e-trials/resources/guidelines-for-drawedumath).
## Citation

```
@misc{lucy2026aftermathdrawedumathvisionlanguage,
      title={The Aftermath of DrawEduMath: Vision Language Models Underperform with Struggling Students and Misdiagnose Errors}, 
      author={Li Lucy and Albert Zhang and Nathan Anderson and Ryan Knight and Kyle Lo},
      year={2026},
      eprint={2603.00925},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2603.00925}, 
}
```