Dialogue model training method and device, equipment, storage medium and program product

Through adversarial training and end-to-end dialogue training, multilingual representations between the source language and the target language are determined, and dialogue content is generated and matched, which solves the problems of insufficient information processing fusion capabilities and error accumulation of intelligent dialogue systems, and improves the overall performance of the system.

CN119831049APending Publication Date: 2025-04-15CHINA SOUTHERN POWER GRID ARTIFICIAL INTELLIGENCE TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411954061.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-27
Publication Date
2025-04-15

Smart Images

  • Figure CN119831049A_ABST
    Figure CN119831049A_ABST
Patent Text Reader

Abstract

The invention discloses a dialogue model training method and device, equipment, a storage medium and a program product. The method comprises the steps that multilingual representation between a source language and a target language is determined; carrying out adversarial training on the large language model by utilizing multilingual representation to obtain a trained large language model; inputting a plurality of source languages into a generator of the trained large language model to generate a plurality of pieces of text data; on the basis of the multiple pieces of text data, dialogue content matched with each piece of text data is obtained through a preset method, and multiple pieces of multilingual dialogue data are obtained on the basis of each piece of text data and the dialogue content corresponding to the text data; performing end-to-end dialogue training on the dialogue model by using the plurality of multilingual dialogue data to obtain a trained dialogue model; the trained dialogue model is used for being deployed in the intelligent dialogue system to generate dialogue content corresponding to the input voice data. By adopting the method, the overall performance of the intelligent dialogue system can be improved.
Need to check novelty before this filing date? Find Prior Art