PointMass

该数据集对控制环境中的奖励函数进行评估，包括稀疏奖励和密集奖励。在这个环境中，智能体可以在一条直线上向左或向右加速。数据集包含了多种奖励函数，如稀疏奖励（S）、密集奖励（D）以及幅度奖励（M），并用于评估EPIC距离方法。该数据集的规模为单任务评估，任务类型为连续控制任务。

This dataset evaluates reward functions within a control environment, covering both sparse and dense rewards. In this environment, the AI Agent can accelerate leftward or rightward along a straight line. The dataset encompasses multiple types of reward functions, including sparse reward (S), dense reward (D), and magnitude reward (M), and is used to assess the EPIC distance method. This dataset is designed for single-task evaluation, and the task type falls under continuous control tasks.