arXiv 2610.29934 · cs.AI · 2026-10-21
ToM2Vec: Continuous Theory-of-Mind Tokens for Multi-Agent Coordination
Nithin Sivakumaran, Ali Payani, Hyunji Lee, Mohit Bansal, Elias Stengel-Eskin
University of North Carolina at Chapel Hill
一句话要点
ToM2Vec通过学习连续的理论思维令牌,在多智能体协作中实现对伙伴状态的抽象推理,显著提升合作与协调能力,并在多种环境下的表现优于基线模型。
ToM2Vec enables abstract theory-of-mind reasoning in multi-agent systems, outperforming baselines across diverse environments.
要点
- 多智能体协作中,伙伴状态存在不确定性,传统语言模型难以表达多种可能状态,因此需要一种能抽象建模伙伴思维的机制。
- ToM2Vec通过在隐空间中学习连续的理论思维令牌,利用伙伴真实想法作为监督信号,训练令牌匹配伙伴的潜在心理状态。
- 在Overcooked-AI、谈判、Decrypto等多场景实验中,ToM2Vec显著优于基线模型,在合作与竞争任务中均表现更优,且泛化至未见过的伙伴模型。
- 分析表明这些隐空间令牌真正编码了理论思维内容,能有效帮助智能体理解伙伴的信念、欲望和行为意图,而非仅学习任务特定信息。
English overview
ToM2Vec introduces continuous theory-of-mind tokens to enable multi-agent systems to reason about partner mental states under uncertainty. By training in a latent space with fixed reasoning tokens supervised by actual partner thoughts, it captures abstract beliefs and intentions. Experiments across diverse environments show consistent performance gains over baselines, with strong generalization to unseen partner models. Analysis confirms the tokens encode genuine theory-of-mind capabilities, not task-specific knowledge.
只根据 arXiv 官方摘要,由开源模型整理,经数字/专名自动核对(必须在摘要中出现),以论文原文为准。
阅读 arXiv 原文 PDF Hugging Face 讨论 代码仓库
主题与学科
同主题:大语言模型
- Jeva: System One Models from First Principles ▲371
- Post-Training Leaves Behavioral Shadows on Unrelated Decisions ▲273
- OmniEdu: Open Foundation Models for Learning and Teaching ▲240
- PAR: Progress-Aligned Recurrence for Length Generalization in State Tracking ▲163
- MemT-1.0: Towards Memory-Native Evolution ▲143
同主题:智能体
- Raven: The Harness of Harnesses for Composable Agentic Intelligence 2026-09-27 · ▲748
- False Frontiers: Diagnosing and Mitigating Co-Cheating in Self-Evolving Search Agents 2026-09-30 · ▲642
- Omni-IO Skills: Harnessing Your Agent Omni-Native 2026-09-25 · ▲485
- BizAgentBench: Unveiling Hidden Challenges in Deploying Long-Horizon Agents for Real-World Business 2026-10-14 · ▲388
- Any Time, Any Space: A World-State-Centric Framework for Agentic Video Generation 2026-10-14 · ▲364
数据说明
元数据来自 arXiv API(已核实);点赞来自 Hugging Face 每日论文;引用来自 Semantic Scholar。最近刷新 2026-10-21。
要点由本站根据摘要自写,不转载摘要原文,请阅读原文。