MuJoCo Continuous Control Datasets
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/openai/spinningup
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了使用MuJoCo模拟器评估的各种连续控制任务，如HalfCheetah-v2、Ant-v2、Pusher-v2、Humanoid-v2、Hopper-v2和Swimmer-v2。此外，Hopper-v2和Swimmer-v2的结果已包含在补充材料中。实验通过改变熵指数来研究它们对探索的影响。该数据集规模涉及多个复杂度不同的任务，其研究主题是强化学习中的连续控制问题。
This dataset contains various continuous control tasks evaluated using the MuJoCo simulator, including HalfCheetah-v2, Ant-v2, Pusher-v2, Humanoid-v2, Hopper-v2, and Swimmer-v2. Additionally, the results for Hopper-v2 and Swimmer-v2 are included in the supplementary materials. Experiments study the effect of exploration by adjusting the entropy exponent. Covering multiple tasks with varying levels of complexity, this dataset focuses on the continuous control problem in reinforcement learning.
提供机构：
OpenAI
