Speeding up Transformer Decoding via an Attention Refinement Network.

Kaixin Wu, Yue Zhang, Bojie Hu, Tong Zhang

Browse the full COLING paper archive.