Tex2Py-45K: A Parallel Corpus Dataset for Bidirectional Conversion Between LaTeX and Python
Lei Li, Manni Duan, Yongheng Wang · 2024
LaTeX and Python, integral to the scientific research workflow, each offers specialized capabilities. Bidirectional LaTeX-Python conversion models could augment the efficacy and precision of scientific research. In this paper, we release the Tex2Py-45K dataset1, containing 45,668 LaTeX-Python pairs. Based on the Tex2Py-45K dataset, we propose SMML format and Hybrid training mode, which trained effective LaTeX-Python bidirectional conversion models. At last, we analyze failure cases of the LaTeX-Python bidirectional conversion, providing direction for future research.