according toArtificial Intelligence Industry Development Alliance AIIA News, January 25, 2024,China Academy of Information and Communications Technology and Industrial and Commercial Bank of China jointly released theIntelligent software engineering technology and application requirements Part 1: CodeLarge Model》standard, which is applicable to enterprises in the process of R&D, evaluation and acceptance of code big models, and provides reference for the construction and improvement of code big model capabilities and guidance for the technical selection of code big models.
▲ Image source: Artificial Intelligence Industry Development Alliance AIIA, the same below
The standard is divided into three parts: general capabilities, dedicated scenario capabilities and application maturity.Includes 16 competency items and more than 100 competency requirements, mainly from the dimensions of input diversity, task diversity, language completeness, result acceptability, and result accuracy, it puts forward full-stack technical capability requirements for the code big model.
The core contributors to this standard include:Industrial and Commercial Bank of China, iFlytek, Huawei, Ping An Bank, Alibaba Cloud, 360, Ant, Tianyi Digital Life, Bank of Communications, NetEase Cloud Computing, Shenzhen JiyiIT Home has attached a complete picture of the participating units as follows:
AIIA said that in the field of intelligent software engineering, three types of large models continue to emerge, led by the code large model.The number of code-based models is the largest, followed by operation and maintenance and testing-based models..
-
The code big model helps improve software coding efficiency and quality with its powerful code understanding, code generation, code checking, code optimization, and R&D question-answering capabilities.
-
The test big model currently focuses on intelligent unit testing, test script generation, test data generation, etc. In the future, it will truly improve the testing capabilities from the perspectives of system testing and business testing;
-
The operation and maintenance big model currently focuses on operation and maintenance knowledge Q&A, work order processing and other aspects. In the future, it will promote the improvement of the intelligence level of the entire AIOps process from multiple dimensions such as fault identification, fault prediction, and operation and maintenance safety.
