A model evaluation method, terminal, system and storage medium

Abstract

The invention relates to a model evaluation method, a terminal, a system and a storage medium. The method includes obtaining query keywords of an algorithm model to be evaluated and data to be evaluated; Word query; evaluate the obtained query results, and generate model evaluation records, based on BERT technology, combined with the context, and then can be flexible, efficient and scalable for different models of the model warehouse, and different training sets of the data warehouse Data can be retrieved more accurately to solve the problems existing in existing retrieval technologies, achieve more accurate query results, improve evaluation efficiency, and optimize user experience.

Description

technical field [0001] The present invention relates to the communication field, in particular to a model evaluation method, terminal, system and storage medium. Background technique [0002] City Brain is the product of the combination of artificial intelligence, big data, blockchain, 5G, Internet of Things and other new generation information technologies and smart city construction. For application scenarios such as grid prevention and control, medical and health care, tourism, environmental protection, and urban refined management, build a city-level neural perception network to realize refined and dynamic urban management, improve urban management effectiveness, alleviate "big city diseases", and improve Improve the quality of life of citizens and push urban informatization to a higher stage. [0003] The artificial intelligence computing and processing center platform is an important part of the city brain, the basis for supporting the long-term efficient operation of...

AI Technical Summary

Problems solved by technology

[0003] The artificial intelligence computing and processing center platform is an important part of the city brain, the basis for supporting the long-term efficient operation of the "city brain", the basic algorithm analysis platform for the "city brain", and the core engine for the continuous evolution of the "city brain"; The brain needs numerous AI algorithms as business support. For different models accessed by different manufacturers, it is necessary to provide a unified AI evaluation system, which also includes the management of model warehouses and data warehouses, so that the models of various manufacturers and different types Algorithm evaluation data for effective management and scheduling; in AI algorithm evaluation, traditional fuzzy query technology is mainly fuzzy string matching method, which calculates the difference between two sequences based on Levenshtein Distance algorithm, Levenshtein Distance algorithm, and Called the Edit Distance algorithm, it refers to the minimum number of editing operations required to convert one character string into another between two character strings; permitted editing operations include replacing one character with another character, inserting a character, and deleting a character; The smaller the edit distance, the greater the similarity between the two strings; but if the names of the models or data sets are similar, the final matching effect of the Levenshtein Distance algorithm is not good, and retrieval errors are prone to occur

Images