Facebook introduces its latest machine learning model that can directly translate between any pair of 100 languages.
Facebook has launched its revolutionary machine learning model capable of performing direct translations between any pair of 100 languages. The company points out that this new model offers enhanced accuracy compared to its predecessors, as it doesn't rely on English as an intermediate translation layer.
Designed as part of an open-source AI software, the tool aims to improve services across the social network's platform, which boasts over two billion users worldwide and supports up to 160 languages.
By executing more than 20 billion translations daily, Facebook hopes to achieve more precise outcomes by directly translating between any pair of 100 languages without the need for an initial translation into English. This model removes the reliance on English as a bridging layer, translating directly instead.
Angela Fan, the research assistant of the project, shared her insights on this groundbreaking initiative: "This milestone is a culmination of years of fundamental work in machine translation by Facebook's AI team."
Highlighting the uniqueness of their approach, Fan explained, "When translating from Chinese to French, most models typically first translate from Chinese to English and then from English to French." With their new model, she expressed confidence in capturing the essence of words more accurately by translating directly from Chinese to French.
