Facebook has developed First artificial intelligence Model That Translates 100 Languages Without Relying on English, as many existing systems do. The social network model outperforms these systems by 10 points on a 100-point scale used by academics to automatically rate the quality of machine translations, according to Facebook. Dubbed M2M-100, the system is currently only a research project, but could eventually be used to translate messages from Facebook users, nearly two-thirds of whom use a language other than English.
Traditional machine translation systems require the creation of separate artificial intelligence models for each language and each task, but, according to Facebook, this approach is not effective on its social network, where people post content in more than 160 languages on billions of messages. Also, advanced multilingual systems can process multiple languages at once, but compromise accuracy by relying on data in English to bridge the gap between the source language and the target language.
To better serve its community, Facebook last week introduced the first multilingual machine translation (MMT) model that can translate between any pair of 100 languages without relying on English data. The model was developed in open source, according to a blog post by Angela Fan, a research assistant at Facebook.
“For years, artificial intelligence researchers have been working to build a single universal model that can understand all languages through different tasks. A single model that supports all languages, dialects and modalities will help us better serve more people, keep translations up to date, and create new experiences for billions of people equally. This work brings us closer to that goal ”. Fan said in the post.