扩散语言模型（Diffusion Large Language Models，dLLMs）凭借并行生成多个 token 的能力，为高效代码生成提供了一条新的技术路线，尤其可以用于高效生成测试用例，快速提升代码覆盖率。然而，扩散语言模型仍面临一个关键问题：生成得更快，往往会导致生成质量下降。 针对这一问题，本文提出了面向扩散语言模型生成单元测试的加速框架 DiffuTester。其核心思想来自一个简 ...
Like many of us, [Austin] finds writing code satisfying — especially if it is challenging. His latest challenge: shoehorn ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results