r2e-edits/deepswe-verifier-merged-with-regression-with-filenames-and-rewards-v2
收藏Hugging Face2025-06-29 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/r2e-edits/deepswe-verifier-merged-with-regression-with-filenames-and-rewards-v2
下载链接
链接失效反馈官方服务：
资源简介：
这是一个包含多个特征的复杂数据集，特征包括但不限于docker镜像名称、奖励值、实验名称、步数、补丁大小、退出原因、P2P比率、平均是概率、是否正确标记的布尔值、消息内容和角色等。数据集分为训练集，其大小为约1.5GB，包含7997个示例。数据集还提供了默认配置信息，包括训练数据文件的路径。
This is a complex dataset with multiple features including but not limited to docker image names, reward values, experiment names, number of steps, patch sizes, exit reasons, P2P rates, average yes probability, boolean values indicating correctness, message content, and roles. The dataset is split into a training set, which is approximately 1.5GB in size and contains 7997 examples. The dataset also provides default configuration information, including the path to the training data files.
提供机构：
r2e-edits
