Sanskrit Tibetan Parallel Dataset presented in the following paper: http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.724.pdf
The eval-folder contains the three evaluation datasets in tsv-format, roman transliteration, applied sandhi split, one sentence pair a line The par-folder contains all aligned texts in same tsv-format. The json folder contains all aligned texts in a detailed json format from which the original text without Sandhi split etc. can be reconstructed. I plan to provide such files also in a separate folder.