Transformer structure, stacked by a sequence of encoder and decoder network layers,
achieves significant development in neural machine translation. However, vanilla …