Constrained CartPole-v1

该数据集是对经典CartPole-v1环境的变种，其中增加了速度限制，旨在评估强化学习代理在安全性能和可解释性方面的表现。该数据集用于衡量约束违反情况以及通过SHAP归因来评估可解释性，重点关注安全意识训练。规模为单代理训练，任务是在安全约束下的强化学习。

This dataset is a modified variant of the classic CartPole-v1 environment that incorporates velocity constraints, designed to evaluate the performance of reinforcement learning agents in terms of safety capabilities and interpretability. It is used to quantify constraint violations and assess interpretability via SHAP attribution, with a focus on safety-aware training. The dataset is tailored for single-agent training, with the core task focusing on reinforcement learning under safety constraints.