ZenGen：面向 LLM 的社会心智框架，含 SoMBench 基准与 Actio 推理架构
ZenGen: Social Mind for LLMs
研究团队提出 ZenGen 框架，用于测量、内化并落地大语言模型的社会智能。配套基准 SoMBench 覆盖 3 个一级维度、17 个二级维度和 71 种任务范式，共 284 个场景、3,481 个专家核验实例；20 个代表性 LLM 中最佳模型总体准确率仅 72.08%，17 个二级维度无一达到 90%。
Authors:ZenGen Team, Ao Xiang, Bi Jingping, Chen Jiahui, Chen Lehan, Chen Yilin, Cheng Xueqi, Fan Yixing, Gan Kairong, Gao Haowen, Gao Jinhua, Gao Shuxuan, Gong Chang, Guo Jiafeng, Guo Ruijie, Han Zhouyu, He Guangfu, He Yichun, Jiang Shuo, Jing Shaoling, Jing Ya, Lei Chenhao, Lei Yan, Li Anqi, Li Chengao, Li Haoyu, Li Shitian, Liang Xinjian, Liu Zhaoge, Lyu Xingyu, Nie Zhuwei, Pang Liang, Quan Zeping, Shan Shiguang, Shen Huawei, Tang Xinran, Tian Feng, Wang Qian, Wang Ruiping, Wang Xiaohong, Xia Zaiyu, Xiao Yi, Xu Jiayuan, Xu Kehan, Xu Qianqian, Xu Tianyu, Xu Yongjun, Xue Yuanhai, Yang Haoming, Yang Jun, Yao Di, Yu Xiaoming, Zhang Futong, Zhang Jie, Zhang Shixuan, Zhang Yuxuan, Zhao Xinyu, Zhao Zhuoran, Zhong Yunfei, Zhu Shengyu
Abstract:As large language models move from isolated task solving toward long-term service in human environments, they require social intelligence: the ability to infer mental states, track social relations, reason over norms, and adapt behavior under context. This report presents ZenGen, an integrated framework for measuring, internalizing, and grounding social intelligence. For measurement, we introduce SoMBench, a psychology-grounded benchmark spanning 3 primary dimensions, 17 secondary dimensions, and 71 task paradigms. It controls question format, narrative perspective, and context length across 284 shared scenarios and 3,481 expert-verified instances. Evaluation of 20 representative LLMs reveals substantial headroom: the best model achieves only 72.08% overall accuracy, and none of the 17 secondary dimensions reaches the 90% near-ceiling band. For internalization, we develop ZenGen, a diagnosis-driven training recipe combining supervised fine-tuning, on-policy distillation, and rubric-based reinforcement learning. Across five social-cognition benchmarks, ZenGen consistently outperforms its base models, with ZenGen-27B-Stage2 achieving the best average score and ZenGen-32B-Stage2 remaining competitive with DeepSeek-V4-Pro. For deployment-time grounding, we build Actio, a harness-controlled inference architecture that routes four typed supports into reasoning: PRISM for procedural guidance, Starling for runtime mental-state representation, SAGE for reusable experience, and gated RAG for external social and normative knowledge. Across five base models and three benchmarks, the full harness improves 14 of 15 model-benchmark pairs and is best or tied for best in 8, demonstrating the effectiveness of typed runtime support. Together, these results show that socially intelligent LLMs require coordinated advances in evaluation, parametric internalization, and deployment-time grounding.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2607.23740 [cs.CL] |
| (or arXiv:2607.23740v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2607.23740 arXiv-issued DOI via DataCite |
Submission history
From: Chengao Li [view email]
[v1]
Sun, 26 Jul 2026 16:25:45 UTC (8,957 KB)
[v2]
Fri, 21 Aug 2026 07:25:22 UTC (8,911 KB)
[v3]
Fri, 9 Oct 2026 03:55:41 UTC (8,913 KB)
来源：arXiv cs.CL · arxiv.org