Part-of-speech tagging (POS) is one of the most important classification task in natural language processing.
In this assignment we experimented a bidirectional model, along three variations, implemented with Tensorflow API, based on Keras backend, and tested the performances with different input data strategies and hyperparameters.