Incorporating Structural Alignment Biases into an Attentional Neural Translation Model

23255

Knowledge Triples