NathanGavenski/Hopper-v4

该数据集由Imitation Learning Datasets项目创建，使用了来自HuggingFace的TD3策略的Stable Baselines权重。数据集包含1000个episode，每个episode的平均奖励为3531.1930。每个条目包括观察值（obs）、动作（action）、奖励（reward）和episode开始标志（episode_starts）。This dataset was created by the Imitation Learning Datasets project using Stable Baselines weights from a TD3 policy from HuggingFace. The dataset consists of 1,000 episodes with an average episodic reward of 3531.1930. Each entry includes observation (obs), action (action), reward (reward), and episode start flag (episode_starts).