---
dataset_info:
  features:
  - name: id
    dtype: int64
  - name: text
    dtype: string
  splits:
  - name: train
    num_bytes: 225647910.0
    num_examples: 2886810
  - name: test
    num_bytes: 23848817.0
    num_examples: 311298
  download_size: 131762427
  dataset_size: 249496727.0
---
# Mathematical Formulas (MF)

Mathematical dataset containing formulas based on the [AMPS](https://drive.google.com/file/d/1hQsua3TkpEmcJD_UWQx8dmNdEZPyxw23) Khan dataset and the [ARQMath](https://drive.google.com/drive/folders/1YekTVvfmYKZ8I5uiUMbs21G2mKwF9IAm) dataset V1.3. Based on the retrieved LaTeX formulas, more equivalent versions have been generated by applying randomized LaTeX printing with this [SymPy fork](https://github.com/aieng-lab/sympy-random-LaTeX)  using [Math Mutator (MAMUT)](https://github.com/aieng-lab/math-mutator). The formulas are intended to be well applicable for MLM. For instance, a masking for a formula like `(a+b)^2 = a^2 + 2ab + b^2` makes sense (e.g., `(a+[MASK])^2 = a^2 + [MASK]ab + b[MASK]2` -> masked tokens are deducable by the context), in contrast, formulas such as `f(x) = 3x+1` are not (e.g., `[MASK](x) = 3x[MASK]1` -> [MASK] tokens are ambigious).

You can find more information in [MAMUT: A Novel Framework for Modifying Mathematical Formulas for the Generation of Specialized Datasets for Language Model Training](https://arxiv.org/abs/2502.20855).

A similar dataset consisting of mathematical texts (i.e., texts containing LaTeX formulas between an inline math environment `$...$`) is [https://huggingface.co/datasets/ddrg/math_text](https://huggingface.co/datasets/ddrg/math_text).

## Citation

```
@article{
  drechsel2025mamut,
  title={{MAMUT}: A Novel Framework for Modifying Mathematical Formulas for the Generation of Specialized Datasets for Language Model Training},
  author={Jonathan Drechsel and Anja Reusch and Steffen Herbold},
  journal={Transactions on Machine Learning Research},
  issn={2835-8856},
  year={2025},
  url={https://openreview.net/forum?id=khODmRpQEx}
}
```