micahr234/play_ns_cartpole_long

这是一个强化学习环境的数据集，包含多个特征字段，如环境名称、全局步骤、回合步骤、动作、奖励、变换后的奖励、完成标志、观察值以及元数据（如质量杆标志、重力标志等）。数据集分为训练集和评估集，训练集包含1000万个示例，评估集包含20万个示例，总大小约为1.4 GB。该数据集适用于智能体训练和性能评估，可能基于物理模拟环境（如倒立摆）。This is a reinforcement learning environment dataset that includes multiple feature fields such as environment name, global step, episode step, action, reward, transformed reward, done flag, observation, and metadata (e.g., masspole flag, gravity flag). The dataset is split into train and eval sets, with the train set containing 10 million examples and the eval set containing 200,000 examples, totaling approximately 1.4 GB in size. It is suitable for agent training and performance evaluation, likely based on physical simulation environments (e.g., inverted pendulum).