Article originally published on:
Today, AI translation technology still has rough edges on the surface when you hit that “translate this” function on Facebook or Instagram feeds, but Meta’s unveiling of its No Language Left Behind initiative (NLLB-200) aims to bridge that gap and “give people the opportunity to access and share web content” in 202 different languages.
Everyone on the internet can now access Meta’s NLLB-200 open-source code and toolkit enabling the research community to build various applications (written in Python) and improve the quality of low-resource languages like Asturian, Luganda, Urdu, and more. Meta says this initiative “will improve the quality of translations across its technologies by an average of 44% with that number jumping to 70%” – a benchmark determined by its proprietary BLEU metrics.
Quoting from Meta’s AI video Jean Maillard said, “Language is the key to inclusion. If you don’t understand what people are saying or writing, you can be left behind.” No Language Left Behind gave birth to Meta’s improved AI model with 55 African languages compared to other translation tools supporting only less than 25. To assess the quality of the NLLB-200 model, Meta also shared their open source evaluative dataset FLORES-200 along with the training model code.
Nonprofit organizations and researchers that want to use NLLB-200 for their meaningful and high-impact projects can be eligible for Meta’s grants of up to $200,000. Projects related to “sustainability, food security, or education” are a few examples of the impactful entries that can apply for the grant. The company is specifically encouraging projects with a key focus on translating two or more African languages.
Furthermore, “Meta has begun a partnership with the Wikimedia Foundation to improve translations on Wikipedia by using NLLB-200 as its back end content translation tool,” says the article. “This is going to change the way that people live their lives … the way they do business, the way that they are educated. No language left behind really keeps that mission at the heart of what we do, as people,” said Al Youngblood, user researcher at Meta AI in a video.
To see NLLB-200’s translation capabilities through stories visit the Meta AI Demo Lab.