FrozenLake8x8-v1
收藏arXiv2025-09-30 收录
下载链接：
https://gymnasium.farama.org/environments/toy_text/frozen_lake/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集为一个具有64个状态和4个动作的随机环境。在其中，智能体以1/3的概率向预期方向移动，并有相等的概率向垂直方向移动。当智能体达到目标时，它会获得1的奖励。此外，当智能体掉入陷阱或达到目标时，剧集将终止。该数据集的规模为64个状态和4个动作，任务类型为强化学习。
This dataset is a stochastic reinforcement learning environment with 64 states and 4 actions. In this environment, the agent has a 1/3 probability of moving toward the intended direction, and equal probabilities of moving in each perpendicular direction. When the agent reaches the goal, it receives a reward of 1. Additionally, episodes terminate when the agent falls into a trap or reaches the goal.
提供机构：
Gymnasium
