PerfCurator
收藏资源简介:
PerfCurator是由密歇根大学迪尔伯恩分校的研究团队创建的一个大规模数据集,专注于收集与性能错误相关的提交记录。该数据集包含114K Python提交、217.9K C++提交和76.6K Java提交,总计约408.5K条记录。创建过程涉及使用PcBERT-KD模型从GitHub仓库中挖掘相关提交。PerfCurator的应用领域主要集中在软件工程研究,特别是性能错误的检测与优化,旨在通过数据驱动的方法提高软件性能和资源利用效率。
PerfCurator is a large-scale dataset created by a research team from the University of Michigan-Dearborn, dedicated to collecting commit records associated with performance bugs. This dataset encompasses 114K Python commits, 217.9K C++ commits, and 76.6K Java commits, amounting to approximately 408.5K records in total. The construction of PerfCurator involved mining relevant commits from GitHub repositories using the PcBERT-KD model. Its primary application domains focus on software engineering research, particularly the detection and optimization of performance bugs, aiming to improve software performance and resource utilization efficiency via data-driven approaches.

- 1PerfCurator: Curating a large-scale dataset of performance bug-related commits from public repositories密歇根大学迪尔伯恩分校 · 2024年



