IBM Project CodeNet
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/manishacharya60/rag-optimization
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了1200对C++源代码及其优化后的程序，其中1000对用于训练，200对用于测试。除了原始代码和优化代码版本外，数据集还包含了问题类型、复杂性的元数据以及用于检索的代码嵌入信息。该数据集的规模为1200对，任务旨在进行代码优化和性能评估。
This dataset comprises 1200 pairs of C++ source code and their optimized counterparts, where 1000 pairs are designated for training and the remaining 200 pairs for testing. In addition to the original and optimized code versions, the dataset encompasses metadata such as problem type and complexity, along with code embedding information tailored for retrieval purposes. With a total size of 1200 code pairs, this dataset is targeted at code optimization and performance evaluation tasks.
提供机构：
IBM
