Transformer

A deep learning architecture based on self-attention, widely used for sequence modeling.