Summary

In this chapter, we first examined several state-of-the-art NLP language models. BERT, in particular, seems to have been widely accepted as the industry standard state-of-the-art language model, and BERT and its variants are widely used by businesses in their own NLP applications.

Next, we examined several areas of focus for machine learning moving forward; namely semantic role labeling, constituency parsing, textual entailment, and machine comprehension. These areas will likely make up a large percentage of the current research being conducted in NLP moving forward.

Now that you have a well-rounded ability and understanding when it comes to NLP deep learning models and how to implement them in PyTorch, perhaps you'll feel inclined to be a part of this research moving forward. Whether this is in an academic or business context, you now hopefully know enough to create your own deep NLP projects from scratch and can use PyTorch to create the models you need to solve...