Neural machine translation

Abstract

Abstract In this paper I would like to present the idea behind neural machine translation starting from the sequence to sequence models based on recurrent neural networks and long short memory term networks and finishing with the state of the art Transformer model. Also, I will explain the complete process of preparing ready to used translation models, which consists of language pair selection, corpus processing and model training. The main idea is to show the growth of machine translation architectures and how they improve the translation quality.