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

2025-01-26

AI Summary

This work addresses the weak generalization capability and high customization cost of large language models (LLMs) in role-playing tasks. To this end, we propose an efficient role-generalization modeling framework. First, we construct a synthetic persona dataset comprising over 10,000 diverse roles, leveraging a novel Persona Hub–driven persona profiling approach. Second, we introduce a dual-path data construction paradigm—role-aware response generation and rewriting—to enable multi-stage role alignment via data distillation. Finally, we perform supervised fine-tuning on LLaMA-3 8B. Experiments demonstrate substantial improvements in zero-shot role transfer, with role-playing dialogue quality approaching that of GPT-4o. We publicly release the entire synthetic persona dataset and instruction-following dialogue corpus, establishing critical infrastructure and a methodological foundation for open research in role-aware AI.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

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.

Problem

• Role Generalization
• Language Models
• Training Methods

Innovation

• Role Generalization
• Customizable Dialogue Data
• Persona-based Fine-tuning