Neural Language Translation

The simple binary classifier described in the previous section is a basic use case for the area of natural language processing (NLP) and doesn't fully justify the use of any techniques that are more complex than using a simple RNN or even simpler techniques. However, there are many complex use cases for which it is imperative to use more complex units such as LSTMs. Neural language translation is one such application.

The goal of a neural language translation task is to build a model that can translate a piece of text from a source language to a target language. Before starting with the code, let's discuss the architecture of this system.

Neural language translation represents a many-to-many NLP application, which means that there are many inputs to the system and the system produces many outputs as well.

Additionally, the number of inputs and outputs could be different as the same text can have a different number of words in the source and target language...