Investigation of Transformer-based Latent Attention Models for Neural Machine Translation

29156

Knowledge Triples