Cheetah and Walker
收藏arXiv2025-09-30 收录
下载链接：
https://arnaudfickinger.github.io/gwil/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集对DeepMind控制套件中两种形态迥异的智能体之间的模仿学习进行了评估。它测试了GWIL（一种学习算法）在不同形态智能体之间模仿行为的能力，尤其是在具有非平凡转换的连续控制任务中。
This dataset evaluates imitation learning between two morphologically distinct agents within the DeepMind Control Suite. It tests the ability of GWIL, a learning algorithm, to imitate behaviors across agents with different morphologies, particularly in continuous control tasks with non-trivial transitions.
提供机构：
DeepMind Control Suite
