OpenCharacter: Training Customizable Role-Playing LLMs with Large-Scale Synthetic Personas
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912235408850944 |
|---|---|
| author | Wang, Xiaoyang Zhang, Hongming Ge, Tao Yu, Wenhao Yu, Dian Yu, Dong |
| author_facet | Wang, Xiaoyang Zhang, Hongming Ge, Tao Yu, Wenhao Yu, Dian Yu, Dong |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_15427 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | OpenCharacter: Training Customizable Role-Playing LLMs with Large-Scale Synthetic Personas Wang, Xiaoyang Zhang, Hongming Ge, Tao Yu, Wenhao Yu, Dian Yu, Dong Computation and Language 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. |
| title | OpenCharacter: Training Customizable Role-Playing LLMs with Large-Scale Synthetic Personas |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2501.15427 |