We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models Peters et al. (2018); Radford et al. …
0 minutes · Arxiv