SMOL : Professionally translated parallel data for 115 under-represented languages

Isaac Caswell c {}^{\textrm{{\char 99\relax}}} * Elizabeth Nielsen c {}^{\textrm{{\char 99\relax}}} * Jiaming Luo c {}^{\textrm{{\char 99\relax}}} Colin Cherry c {}^{\textrm{{\char 99\relax}}} · 2025

Read SMOL : Professionally translated parallel data for 115 under-represented languages by Isaac Caswell c {}^{\textrm{{\char 99\relax}}} * Elizabeth Nielsen c {}^{\textrm{{\char 99\relax}}} * Jiaming Luo c {}^{\textrm{{\char 99\relax}}} Colin Cherry c {}^{\textrm{{\char 99\relax}}} (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-12301.txt.

Description

Isaac Caswell and co-authors present a 2025 paper on a new data set meant to unlock machine translation for languages with few resources. Many languages have no public training data at all, which leaves them out of modern translation tools. Their set was professionally translated into well over a hundred such languages.

The full set holds millions of translated tokens across two parts. One part picks sentences for wide coverage of distinct words, and the other covers broad topics at document level. Together with an earlier word-level set, it spans three scales of text. Tests show the data can prompt or fine-tune models for languages long left behind.

Catalog record

CREATORIsaac Caswell c {}^{\textrm{{\char 99\relax}}} * Elizabeth Nielsen c {}^{\textrm{{\char 99\relax}}} * Jiaming Luo c {}^{\textrm{{\char 99\relax}}} Colin Cherry c {}^{\textrm{{\char 99\relax}}}
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE60 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-12301

Browse the full library · About the Tower of Babel