Many-to-English Machine Translation Tools, Data, and Pretrained Models

TL;DR

This paper introduces new tools and a multilingual neural machine translation model capable of translating from 500 source languages to English, addressing the resource disparity among languages.

Contribution

The authors present MTData, NLCodec, and RTG tools, and develop a multilingual model for 500 languages, facilitating transfer learning to low-resource languages.

Findings

Created a multilingual translation model for 500 languages

Tools are publicly available for research and transfer learning

Model can be used as a service or for further training

Abstract

While there are more than 7000 languages in the world, most translation research efforts have targeted a few high-resource languages. Commercial translation systems support only one hundred languages or fewer, and do not make these models available for transfer to low resource languages. In this work, we present useful tools for machine translation research: MTData, NLCodec, and RTG. We demonstrate their usefulness by creating a multilingual neural machine translation model capable of translating from 500 source languages to English. We make this multilingual model readily downloadable and usable as a service, or as a parent model for transfer-learning to even lower-resource languages.