About this event

What if we did not use transformers? Can we still do NLP? Yes.

In this talk, Vincent will discuss some effective tricks in NLP that don’t require a transformer or a GPU. It will be a collection of techniques that he has seen work in the field including some tricks with embeddings, annotation techniques, tf-idf, subword embeddings, and multi-word embeddings. Many of these techniques can also be used outside of the realm of NLP and most of the topics will be discussed via a live demo.

If there is time, Vincent might also show a very cool fine-tuning trick!

Speaker

Vincent Warmerdam worked as an engineer, consultant, researcher, team lead, and educator in the past. He currently works as a Machine Learning Engineer at Explosion. He also maintains many small open source packages, including many scikit-learn related plugins. He blogs over at koaning.io and he also maintains an increasingly popular free learning resource over at calmcode.io