IDCA NewsAll IDCA News
9 Aug 2022
Meta's NLLB-200 AI Model Translates More Accurately than Its Predecessor by an Average of 44%
Meta, the translation technology company, has announced that its NLLB-200 Neural Machine Translation (NMT) model has exceeded the BLEU-6 score on the industry standard reference set of Chinese to English and Arabic to English machine translation language pairs by 44%. Several financial services companies are currently using this Model to improve communication across their businesses worldwide.
The Model is a statistical machine translation model that uses a neural network to improve translation quality. This is an important step forward in machine translation, and we are excited to see what the future holds for this technology. Even when the translation is off, it's close enough for a native speaker to comprehend.
A press release from Meta Reads:
"To help people connect better today and be part of the metaverse of tomorrow, our AI researchers created No Language Left Behind (NLLB), an effort to develop high-quality machine translation capabilities for most of the world's languages.
The metaverse aims to be borderless. The company’s translation service will have to offer correct translations quickly to achieve this goal.
"As the metaverse begins to take shape, the ability to build technologies that work well in a wider range of languages will help democratize access to immersive experiences in virtual worlds," the company explained.
In some African and Indian languages, NLLB-200's translations were over 70 percent more accurate than previous AI research.
FLORES-200 and NLLB-200 have been opened to developers who can use them to build on Meta's work and perfect the translation tools.
Victor Botev, the CTO at Iris.ai, commented, "Despite the hype, "It's worth bearing in mind that these models are not the cure-all they may first appear. The models that Meta uses are massive, unwieldy beasts. So, when you get into the minutiae of individualized use-cases, they can easily find themselves out of their depth – overgeneralized and incapable of performing the specific tasks required."
"Another point to note is that the validity of these measurements has yet to be scientifically proven and verified by their peers. The datasets for different languages are too small, as shown by the challenge in creating them in the first place, and the metric they're using, BLEU, is not particularly applicable."
Click this link to view the NLLB-200 demo video. As the video states, We've created a demo that uses the latest AI advancements from the No Language Left Behind project to translate books from their languages of origin, such as Indonesian, Burmese, and Somali, into more languages for readers – with hundreds available in the coming months. With this AI tool, families can now read stories together from around the world in a language that works for them.
Follow us on social media: