一句话：提出 Kitten，一个简单有效的基线，用于评估基于 LLM 的编译器测试技术。
编译器测试对提升编译器正确性至关重要。近年来，基于大语言模型（LLM）的编译器测试技术（如 Fuzz4All）在发现真实缺陷、降低程序生成器设计工程量方面展现出潜力。然而，随着 LLM 持续演进和新方法不断出现，领域亟需稳健基线以进行严格评估并推动后续研究。本文提出 Kitten：一个简单而有效的基线，用于评估基于 LLM 的编译器测试技术。其核心目标是提供低复杂度、易复现的参照，使研究者能够判断新方法的增益是否来自真正的创新，而非仅由模型能力或实现复杂度带来。该工作强调基线在 LLM 编译器测试研究中的必要性，并为未来技术的系统比较与进步奠定基础。
Compiler testing is critical and indispensable to improve the correctness of compilers. Spurred by recent advancements in Large Language Models (LLMs), LLM-based compiler testing techniques such as Fuzz4All, have demonstrated their potential in uncovering real bugs in diverse compilers and reducing the required engineering efforts in designing program generators. Given the continuous evolution of LLMs and the emergence of new LLM-based approaches, establishing robust baselines is crucial for rigorous evaluation and driving future advancements in this promising research direction.
@inproceedings{paperbot9588,
title = {Kitten: A Simple Yet Effective Baseline for Evaluating LLM-Based Compiler Testing Techniques},
author = {Yuanmin Xie and Zhenyang Xu and Yongqiang Tian and Min Zhou and Xintong Zhou and Chengnian Sun},
booktitle = {ISSTA},
year = {2025},
doi = {10.1145/3713081.3731731}
}