Neural machine translation model training methods, devices and storage media

摘要

PROBLEM TO BE SOLVED: To provide a training method, an apparatus and a storage medium for a neural machine translation model. SOLUTION: A training method, an apparatus and a storage medium of a neural machine translation model according to an embodiment of the present invention have obtained a predicted sentence of a source sentence translated by using the trained neural machine translation model. The original neural by synthesizing a new parallel corpus based on the degree of N-gram matching between the predicted sentence and the target sentence and training the neural machine translation model with the original corpus and the added new corpus. The features of the poorly trained part of the machine translation model are enhanced, and the translation performance of the finally trained neural machine translation model is improved. [Selection diagram] Fig. 1