Adaptive ModernMT

Adaptive ModernMT is a state-of-the-art machine translation system that uses deep learning to automatically translate text from one language to another. This innovative system is the product of a project called ModernMT, which was led by Translated, in collaboration with FBK Research Center, the University of Edinburgh, and the European Commission.

The system brings two significant layers of innovation to NMT:

• Contextual translation: Unlike traditional NMT systems which often translate each sentence in isolation, ModernMT uses the context of the entire document to improve the translation of each individual sentence. This enables it to generate more accurate and coherent translations, especially for complex texts where the meaning of a sentence can be highly dependent on the surrounding content.

• Real-time learning: ModernMT has the ability to learn from each correction made by human translators in real time. This continuous learning process allows the system to constantly improve its performance and adapt to new languages, dialects, or stylistic preferences of the translators.

Demo

Screenshots