開放角色：使用大規模合成人物形象訓練可定制角色扮演LLM

OpenCharacter: Training Customizable Role-Playing LLMs with Large-Scale Synthetic Personas

January 26, 2025

作者: Xiaoyang Wang, Hongming Zhang, Tao Ge, Wenhao Yu, Dian Yu, Dong Yu

cs.AI

摘要

在大型語言模型（LLMs）中可定制的角色扮演，也被稱為角色泛化，因其在開發和部署角色扮演對話代理時的多功能性和成本效益而受到越來越多的關注。本研究探索了一種大規模數據合成方法，以賦予LLMs角色泛化能力。我們首先使用Persona Hub 中的角色概要合成大規模角色概要，然後探索兩種策略：回應重寫和回應生成，以創建與角色對齊的指導回應。為了驗證我們的合成指導數據對角色泛化的有效性，我們使用LLaMA-3 8B模型進行監督微調（SFT）。我們表現最佳的模型加強了原始的LLaMA-3 8B Instruct 模型，在角色扮演對話方面實現了與GPT-4o模型相當的性能。我們釋放我們的合成角色和指導調整對話，以支持公共研究。

English

Customizable role-playing in large language models (LLMs), also known as character generalization, is gaining increasing attention for its versatility and cost-efficiency in developing and deploying role-playing dialogue agents. This study explores a large-scale data synthesis approach to equip LLMs with character generalization capabilities. We begin by synthesizing large-scale character profiles using personas from Persona Hub and then explore two strategies: response rewriting and response generation, to create character-aligned instructional responses. To validate the effectiveness of our synthetic instruction tuning data for character generalization, we perform supervised fine-tuning (SFT) using the LLaMA-3 8B model. Our best-performing model strengthens the original LLaMA-3 8B Instruct model and achieves performance comparable to GPT-4o models on role-playing dialogue. We release our synthetic characters and instruction-tuning dialogues to support public research.