SPIN-Bench

该数据集名为SPIN-Bench，是一个多领域评估框架，旨在衡量人工智能代理在战略规划和社交推理方面的智能水平。它融合了经典的PDDL任务、竞技棋盘游戏、合作卡牌游戏以及多代理谈判场景。SPIN-Bench系统地调整了动作空间、状态复杂度以及交互代理的数量，以模拟各种社交环境。该框架包含了多个具有不同复杂性和代理互动的场景，其任务旨在评估战略规划和社交推理能力。

This dataset, named SPIN-Bench, is a multi-domain evaluation framework designed to measure the intelligence of AI Agents in strategic planning and social reasoning. It integrates classical PDDL tasks, competitive board games, cooperative card games, and multi-agent negotiation scenarios. SPIN-Bench systematically adjusts the action space, state complexity, and the number of interacting agents to simulate diverse social environments. This framework includes multiple scenarios with varying complexities and agent interaction patterns, where the tasks are intended to evaluate strategic planning and social reasoning capabilities.