Method, device and equipment for improving large model code capability and storage medium

The invention provides a method, device and equipment for improving large model code capability and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the following steps: crawling and cleaning code corpora, and storing the cleaned code corpora and text data corresponding to the cleaned code corpora into a search engine retrieval library; constructing a query statement corresponding to a user question, and obtaining a query result of the query statement through the search engine retrieval library; and based on a query result of the query statement, the user question and an output result of the manual annotation, performing fine adjustment on the large model. Therefore, the fine-tuned large model can output the code content which is more related to the user question and is accurate, and the code generation capability of the large model is improved. 本发明提供一种提升大模型代码能力的方法、装置、设备及存储介质，涉及人工智能技术领域，其中方法包括：爬取并清洗代码语料，将清洗后的代码语料和所述清洗后的代码语料对应的文本数据存入搜索引擎检索库；构建用户问题对应的查询语句