Training methods, devices, electronic devices and storage media for machine translation models

摘要

PROBLEM TO BE SOLVED: To provide a training method, an apparatus, an electronic device and a storage medium of a machine translation model capable of saving time and labor and efficiently improving the training efficiency of the machine translation model in a target field. SOLUTION: The method is to select a set of samples from a parallel corpus that satisfy preset requirements for translation quality and that have features of a common field and / or features of a target field, and a first training sample set. Select a set of samples that meet the preset requirements of translation quality from the parallel corpus and do not have the characteristics of the common field and the characteristics of the target field to form the second training sample set, and the first training. Using the sample set and the second training sample set, respectively, the discriminator placed in each coding layer of the machine translation model encoder and encoder in the target field, and the encoder and decoder of the machine translation model in the target field are sequentially used. train. [Selection diagram] Fig. 1