Facebook has released The NLLB project (No language left). The main characteristic of this development is the cover of more than two hundred languages, including rare languages of African and Australian peoples. In addition, Facebook has applied a new approach to the automatic learning model, where translation is carried out directly from one language to another, without intermediate translation in English.
This project will cover all modern languages to provide a means of communication with each other to all people on earth.
The model is distributed Free (for non -commercial use). To facilitate third -party projects, the open source code has been used to test the evaluation and quality of models, as well as to train models. The final model is offered in two versions: full and reduced.
Notably, Nllb is a common model for 200 languages and does not require any additional model for each language separately. The system can automatically recognize the original language and carry out the translation in the user's language.
According to the project team, the system is almost once and a half better than the previous translation systems and better translates the rare languages of seventy percent.
You can assess the quality of translation, as well as read short stories in rare languages here https://nllb.metademolab.com/
