研究人员开发了Code-MUE，一个旨在衡量代码大语言模型（LLMs）不确定性的新框架。这个纯粹的黑盒系统利用基于执行的语义交互图，通过分析运行时行为和计算解空间的冯·诺依曼熵来评估不确定性。涉及八个最先进大语言模型的实证研究表明，Code-MUE能有效关联功能正确性，在软件工程工作流的风险检测方面优于传统的词汇和基于嵌入的方法。 AI
影响 该框架可以通过更好地量化模型的不确定性来提高代码生成工具的可靠性和安全性。
排序理由 该集群描述了一篇介绍用于评估代码大语言模型的新颖框架的研究论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2607.12273v1 Announce Type: cross Abstract: As Code Large Language Models (LLMs) become central to modern software engineering, their inherent stochasticity poses significant real-world risks, where even minor errors can lead to severe functional, security, or safety conseq…
As Code Large Language Models (LLMs) become central to modern software engineering, their inherent stochasticity poses significant real-world risks, where even minor errors can lead to severe functional, security, or safety consequences. Reliable automation, therefore, demands th…
As Code Large Language Models (LLMs) become central to modern software engineering, their inherent stochasticity poses significant real-world risks, where even minor errors can lead to severe functional, security, or safety consequences. Reliable automation, therefore, demands th…