让大模型游戏角色按设定难度智能调整表现，提升可控性与公平性。
DVM: Towards Controllable LLM Agents in Social Deduction Games
- 通过预测-决策-讨论三模块结合强化学习，实现动态性能调节。
- 在狼人杀中达成预设胜率目标，表现优于现有方法。
- 适合研究可控智能体、游戏难度调节及大模型安全性评估者。
大型语言模型（LLMs）提升了社交推理游戏（SDGs）中游戏角色的能力。这类游戏高度依赖对话交互，要求角色基于信息进行推断、决策和表达。尽管当前进展使非玩家角色（NPCs）更智能且更具策略性，但仍需对其能力水平进行控制。这种控制不仅可使NPC适应不同游戏难度，还能为大模型代理的安全性与公平性提供洞察。本文提出DVM框架，用于开发可调控的SDG大模型代理，并在最受欢迎的社交推理游戏之一——狼人杀中验证。DVM包含三个核心组件：预测器、决策器和讨论器。通过将强化学习与胜率约束的决策链奖励机制相结合，使代理能动态调整其游戏表现以达到指定胜率。实验表明，DVM不仅在狼人杀中超越现有方法，还能成功调节性能以满足预设胜率目标。这些结果为大模型代理在社交推理游戏中的自适应与平衡表现铺平道路，开辟了可控游戏智能体研究的新方向。
原文摘要 · Abstract (English)
Large Language Models (LLMs) have advanced the capability of game agents in social deduction games (SDGs). These games rely heavily on conversation-driven interactions and require agents to infer, make decisions, and express based on such information. While this progress leads to more sophisticated and strategic non-player characters (NPCs) in SDGs, there exists a need to control the proficiency of these agents. This control not only ensures that NPCs can adapt to varying difficulty levels during gameplay, but also provides insights into the safety and fairness of LLM agents. In this paper, we present DVM, a novel framework for developing controllable LLM agents for SDGs, and demonstrate its implementation on one of the most popular SDGs, Werewolf. DVM comprises three main components: Predictor, Decider, and Discussor. By integrating reinforcement learning with a win rate-constrained decision chain reward mechanism, we enable agents to dynamically adjust their gameplay proficiency to achieve specified win rates. Experiments show that DVM not only outperforms existing methods in the Werewolf game, but also successfully modulates its performance levels to meet predefined win rate targets. These results pave the way for LLM agents' adaptive and balanced gameplay in SDGs, opening new avenues for research in controllable game agents.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。