Meta has developed an AI tool which can translate over 200 languages, according to the company.
Meta says that its FLORES-200 database connects to its AI model of over 200 different datasets of languages to compare languages and ensure that users receive high-quality translations.
The databases have been made available to the public on Github in the spirit of letting the online community contribute, allowing the translations to improve as the database ages.
According to Meta, machine translation technology has solidified itself as a key focus of artificial research today.
This demo of NLLB shows its potential as it is able to translate texts into multiple different languages.
In Association with
While NLLB has text-to-text translation technology, common across other language tools, the technology has also enabled both speech-to-text translation and as well as speech-to-speech translation.
According to Meta, this is an essential part of this service as “text data is important but not sufficient for building translation tools for everyone’s needs”.
“Speech translation benchmark data was previously available for a handful of languages, so we created CoVoST 2, which covers 22 languages and 36 language directions with different resource conditions,” the company stated
Meta says that in developing the No Language Left Behind programme, the company firstly contextualised the need for low-resource language translation support through exploratory interviews with native speakers.
The company then created datasets and models aimed at narrowing the performance gap between low and high-resource languages.
Meta says it evaluated the performance of over 40,000 different translation directions using a human-translated benchmark, Flores-200, and combined human evaluation with a novel toxicity benchmark covering all languages in Flores-200 to assess translation safety.
“Our model achieves an improvement of 44% BLEU relative to the previous state-of-the-art, laying important groundwork towards realising a universal translation system," the company claims.
Meta is also awarding up to $200,000 of grants for impactful uses of NLLB-200 to researchers and nonprofit organisations with initiatives focused on the UN Sustainable Development Goals.
Nonprofits interested in using NLLB-200 to translate two or more African languages, as well as researchers working in linguistics, machine translation and language technology, are invited to apply.
These research advancements will support more than 25 billion translations served every day in Feed on Facebook, Instagram and Meta's other technologies, according to the company.
