Neural machine translation (NMT) uses a neural network to learn to translate text from a source language into a target language. Unlike SMT, one key advantage of NMT is that it needs only one model to translate end-to-end from a source language into a target language. More importantly, NMT works on whole segments of the source text rather than chunks or phrases as in SMT. This is achieved by learning context through word embeddings. Therefore, NMT performs translation while preserving the context of the original text, as well. We will now look at some of the common deep learning architectures used in NMT.