Transformers in NLP

In part due to the scarcity of text datasets, NLP had a slow progression until the release of Bidirectional Encoder Representations from Transformers (BERT) in 2018 [1].

Read at towardsdatascience.com