Постредакции

AbstractSupervised learning in Neural Machine Translation (NMT) standardly follows a teacher forcing paradigm where the conditioning context in the model’s prediction is constituted by reference tokens, instead of its own …...