๐ฎ [2025-08] We have updated to a larger dataset, which contains nearly 80 million samples, with significant improvements in both data quality and diversity. [link]
๐ฎ [2024-02] We trained a formula recognition model, ๐๐๐ฑ๐๐๐ฅ๐ฅ๐๐ซ, using the latex-formulas dataset. It can convert LaTeX formulas into images and boasts high accuracy and strong generalization capabilities, covering most formula recognition scenarios.
For more details, please refer to theโฆ See the full description on the dataset page:
https://huggingface.co/datasets/Evanstarcraft2/latex-formulas.
