SEAC Training Data

该数据集是由使用软弹性演员-评论家（SEAC）算法进行的实验生成的，比较了该算法与原始软演员-评论家（SAC）算法和近端策略优化（PPO）算法的性能。数据集包含了SEAC算法相较于SAC和PPO算法的平均回报、每项任务的时间成本以及能源消耗等指标。训练规模涉及大约90万至120万步，任务是强化学习，用于控制模拟环境中的智能体运动。

This dataset is generated from experiments conducted with the Soft Elastic Actor-Critic (SEAC) algorithm, which compares the performance of SEAC against the original Soft Actor-Critic (SAC) and Proximal Policy Optimization (PPO) algorithms. The dataset includes metrics such as the average return of SEAC relative to SAC and PPO, the time cost per individual task, and energy consumption. The training scale ranges from approximately 900,000 to 1,200,000 steps, and the tasks are reinforcement learning tasks focused on controlling agent movement in simulated environments.