GluonNLP is one of the newest toolkits for natural language processing (NLP), providing building blocks for text data pipelines and neural models. The focus is on enabling fast prototyping. This tutorial will provide an introduction to GluonNLP with basic examples. There will be a quick refresher for deep learning in NLP before delving into the specifics of the toolkit itself.
GluonNLP is a deep learning toolkit for natural language processing. It was recently released and promises to be a one-stop shop for models and data pipelines for natural language. We provide a quick and dirty refresher for deep learning as applied to NLP and then introduce GluonNLP and its specific features and merits. This toolkit is based on Amazon AI's MXNet, adding the Gluon front end on top. The talk will cover the following with examples - + How to find and load text data sets + How to construct a vocabulary and work with word embeddings + How to build a neural model + Hyperparameters and initialization + How to work with pre-trained models
This talk is designed to appeal to both beginners, who would like to get a taste of what deep learning for natural language processing looks like and to somewhat experienced people who would like to get a taste of GluonNLP to do NLP tasks already familiar to them.