In 2017, the introduction of transformers into feed-forward neural networks enabled computers to achieve human-level performance on a range of language tasks.
In this talk, we will describe some of the theory behind these systems, and explain how it can be applied to search and information retrieval problems.