Trigger Environment
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/Information-Fusion-Lab-Umass/causal_transfer_learning
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了不同设置的网格世界环境，其中学习了对象之间的因果关系。这些设置包括网格大小、钥匙和锁的数量变化。数据集还包含一个19维向量，它代表了代理属性和环境特征，这有助于研究强化学习中的因果关系。规模上，网格大小从5x5到75x75不等。该数据集的任务是进行因果动态建模和迁移学习。
This dataset includes grid-world environments with various configurations for learning causal relationships between objects. These configurations cover variations in grid size, as well as the number of keys and locks. The dataset also contains a 19-dimensional vector that represents agent attributes and environmental characteristics, which facilitates research on causal relationships in reinforcement learning. In terms of scale, the grid size ranges from 5×5 to 75×75. The tasks supported by this dataset focus on causal dynamic modeling and transfer learning.
