Deep Learning and NLP: Project experiences.

Transformer architecture is a breakthrough in the use of Machine Learning for Natural Language Processing. New models set new state-of-the-art results in many practical tasks. However new models are bigger and require much more computational resources for both training and inference.

In my speech I will share my experiences in deploying Transformer models for NLP tasks. I will cover custom dataset creation process, choosing optimal model and increasing reliability of the solution.