RoboArena/DataDump_08-05-2025

RoboArena数据集快照包含了从RoboArena基准测试开始到2025年8月5日的自主策略回放、任务成功率、策略偏好标注及相关元数据。数据集按照特定的目录结构组织，包括全局元数据文件和评估会话目录，每个会话目录下有各自的元数据文件和不同策略的数据。

This dataset snapshot includes autonomous policy rollouts, task success scores, policy preference annotations, and related metadata collected from the RoboArena benchmark from its inception through August 5, 2025. The dataset is structured with global metadata and evaluation sessions directories, each containing metadata files and data for different policies.