lukbl/LaTeX-OCR-dataset

LaTeX-OCR Dataset Summary This dataset was created to train LaTeX-OCR, a model for recognizing LaTeX code from images of mathematical formulas. Each sample consists of a synthetically rendered formula image and its corresponding LaTeX formula. The images were generated from scratch using xelatex with multiple fonts, offering more typographic variety than many other datasets that use a single font (typically Computer Modern). Data Sources Formulas… See the full description on the dataset page: https://huggingface.co/datasets/lukbl/LaTeX-OCR-dataset.