Counterfactual Generation (CF)
收藏arXiv2025-09-30 收录
下载链接：
https://pradippramanick.github.io/coherent-explain/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是从RoboFail数据集中生成的，旨在创建一个更大且平衡的反事实示例数据集，其中包含1240个自动标注的例子。该数据集专注于根据失败类型修改计划、观察和行动来生成反事实示例。其任务是对一致性分类模型进行训练和评估。
This dataset is derived from the RoboFail dataset, and is designed to construct a larger and balanced counterfactual example dataset containing 1240 automatically annotated instances. This dataset focuses on generating counterfactual examples by revising plans, observations, and actions based on failure types. Its core task is to train and evaluate consistency classification models.
提供机构：
Authors' research team
