发表机构
University of California, Berkeley; Google（加州大学伯克利分校; 谷歌）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出人格混合模型（PMMs）和串联模型以改进LLM人类模拟，前者基于预训练模型绑定人格，后者结合预训练与指令调优监督者，实验表明串联模型在准确性和多样性上最优。
AI 中文摘要
我们在此论证，当前LLM人类模拟的主流做法——提示指令调优的助手语言模型进行角色扮演——是不准确的，并且会产生刻板化的预测（缺乏自然多样性）。先前已有研究表明，使用自然主义的、自由文本的对话可以将LLM绑定到特定人格，从而避免刻板印象。在此我们展示，绑定也可以通过使用特定人物的简短、个体对话样本实现。人口统计信息可以在之后通过简单查询模型来添加，而不会产生负面影响。我们使用术语“人格混合模型”（PMMs）来指代校准良好的人类模型，目前实现为预训练的基础模型。我们证明，PMMs比指令调优模型产生更准确的预测，并保留更多人类对话中发现的词汇、语义和语用多样性。我们测量了LLM在模拟人类对话者时的真实性和多样性，跨越了涵盖开放域文本、人机对话以及人类说话者之间的任务导向对话的多种语料库。然而，基础预训练模型可能产生域外对话，并且在长上下文中可能丢失人类的一些内部状态。我们提出并探索了串联模型，它将预训练模型与指令调优的监督者相结合。在我们的实验中，串联模型实现了最佳的整体准确性和多样性。
英文摘要
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.
Comments11 pages in body, 36 with appendices. 7 figures. 8 tables
