1 paper
Shuhei Kondo, Katsuhito Sudoh, Yuji Matsumoto
Most of modern neural machine translation (NMT) models are based on an encoder-decoder framework with an attention mechanism. While they perform well on standard datasets, they can…