arXiv cs.CL· Minwoo Kang, T\'ea Wright, Seun Eisape, Ayush Raj, Suhong Moon, Joseph Suh, Alane Suhr, David M. Chan, John Canny·· 9 天前AI 评分30
预训练 Persona Mixture Models 与串联模型用于人类模拟
Pretrained Persona Mixture Models and Tandem Models for Human Simulation
AI 导读
研究提出 Persona Mixture Models（PMMs），用特定人物的短对话样本将 LLM 绑定到人格，比指令微调模型的角色扮演预测更准确，并保留更多人类对话中的词汇、语义和语用多样性。PMMs 目前以预训练基础模型实现，但可能产生域外对话并在长上下文中丢失人物内部状态。为此提出的串联模型将预训练模型与指令微调监督器结合，在实验中取得最佳整体准确率与多样性。
正文 · 原文
Abstract:We argue here that the current dominant practice in LLM human simulation: prompting instruction-tuned assistant language models to role-play personas, is inaccurate and produces stereotyped predictions (lacking natural diversity). It has previously been shown that LLMs can be bound to personas using naturalistic, freetext dialog avoiding stereotyping. Here we show that binding can also be achieved using short, individual samples of dialog from specific people. Demographics can be added later without negative effects by simply querying the model. We use the term Persona Mixture Models (PMMs) for well-calibrated human models, currently realized as pretrained base models. We show that PMMs produce more accurate predictions than instruction-tuned models and retain more of the lexical, semantic, and pragmatic diversity found in human dialog. We measure realism and diversity of LLMs simulating human interlocutors across a diverse set of corpora spanning open-domain text, human-AI chat, and task-oriented dialogue between human speakers. However, base pretrained models can produce out-of-domain dialog and may lose some of the human's internal state over long contexts. We propose and explore tandem models which combine a pre-trained model with an instruction-tuned supervisor. Tandem models achieve the best overall accuracy and diversity in our experiments.
| Comments: | 11 pages in body, 36 with appendices. 7 figures. 8 tables |
| Subjects: | Computation and Language (cs.CL) |
| ACM classes: | I.2.7 |
| Cite as: | arXiv:2609.22607 [cs.CL] |
| (or arXiv:2609.22607v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2609.22607 arXiv-issued DOI via DataCite |
Submission history
From: John Canny [view email]
[v1]
Fri, 18 Sep 2026 21:50:35 UTC (7,604 KB)
[v2]
Wed, 7 Oct 2026 22:28:28 UTC (7,612 KB)
来源：arXiv cs.CL · arxiv.org