Vaswani

• Transformer

  Attention-based neural architecture without recurrence; enables efficient parallel training and strong performance on language tasks. Published by Vaswani et al., 2017.