AI-Assisted Programming with Test-based Refinement
收藏DataCite Commons2024-09-29 更新2025-04-16 收录
下载链接：
http://dataverse.jpl.nasa.gov/citation?persistentId=doi:10.48577/jpl.U0VAL1
下载链接
链接失效反馈官方服务：
资源简介：
This work explores the utilization of a Large Language Model (LLM), specifically OpenAI's ChatGPT to develop a program as a sequence of refinements. Traditionally in formal methods literature such refinements are proven correct, which can be time consuming. In this work the refinements are tested using property-based testing. This approach addresses the problem of ensuring that the code generated by a LLM is correct, which is one of the main challenges of code generation with LLMs. Programs are developed in Scala and testing is performed with ScalaCheck. This approach is demonstrated through the development and testing of a classical bridge controller, originally presented in documentation for the refinement-based Event-B theorem prover.
提供机构：
Root
创建时间：
2024-09-29
