DL bug benchmark dataset
收藏arXiv2023-10-11 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2310.06912v1
下载链接
链接失效反馈官方服务：
资源简介：
DL bug benchmark dataset是由约克大学拉松德工程学院的研究人员手动创建的数据集，专门用于评估深度学习（DL）模糊测试工具。该数据集包含2020年至2022年间从TensorFlow和PyTorch库中用户报告的627个真实世界DL bug。数据集的创建过程涉及自动过滤和手动分析，确保了数据的质量和相关性。该数据集主要用于评估DL模糊测试工具的性能，特别是在检测真实世界bug方面的有效性。通过此数据集，研究人员能够更深入地了解DL库中的bug类型和分布，从而推动DL模糊测试工具的改进和发展。
The DL bug benchmark dataset was manually created by researchers from the Lassonde School of Engineering, York University, and is specifically designed to evaluate deep learning (DL) fuzz testing tools. This dataset contains 627 real-world DL bugs reported by users from the TensorFlow and PyTorch libraries between 2020 and 2022. The dataset creation process involves automated filtering and manual analysis to ensure the quality and relevance of the data. It is primarily used to assess the performance of DL fuzz testing tools, particularly their effectiveness in detecting real-world bugs. Through this dataset, researchers can gain a deeper understanding of the types and distribution of bugs in DL libraries, thereby promoting the improvement and development of DL fuzz testing tools.
提供机构：
约克大学拉松德工程学院
创建时间：
2023-10-11
