Hanabi Learning Environment

该数据集是一个开源的学习环境，专为纸牌游戏汉纳比设计，用于测试多代理强化学习算法。该环境允许代理在共享策略和经验的同时进行训练，同时保持各自的学习限制。它支持2至5名玩家的游戏规模，并在汉纳比纸牌游戏的背景下执行多代理强化学习任务。

This dataset is an open-source learning environment purpose-built for the card game Hanabi, designed to test multi-agent reinforcement learning algorithms. This environment enables agents to train while sharing policies and experiences, while maintaining their respective learning constraints. It supports game sessions with 2 to 5 players, and conducts multi-agent reinforcement learning tasks within the context of the Hanabi card game.

搜集汇总

背景与挑战

背景概述

Hanabi Learning Environment是一个用于Hanabi实验的研究平台，提供类似OpenAI Gym的RL环境API和低级游戏接口。该平台支持通过pip安装，并包含示例代码以帮助用户快速上手。