Generative natural language model training method, system and device and storage medium

The invention relates to the technical field of artificial intelligence, particularly provides a generative natural language model training method, system and device and a storage medium, and aims to solve the technical problems of long training time and large resource consumption of a generative natural language model. In order to achieve the purpose, the method comprises the steps that coded sample data corresponding to all text data samples are obtained and spliced, and spliced sample data are obtained; acquiring an attention mask corresponding to the spliced sample data; and training the generative natural language model based on the spliced sample data and the attention mask. Through data splicing, the training iteration process is greatly shortened, and the model training efficiency is greatly improved; in addition, by introducing an attention mask mechanism for shielding different samples, the final training effect is not affected by sample splicing. 本发明涉及人工智能技术领域，具体提供一种生成式自然语言模型训练方法、系统、设备及存储介质，旨在解决生成式