# CharacterGLM: Customizing Chinese Conversational AI Characters with Large Language Models

**Authors:** Jinfeng Zhou, Zhuang Chen, Dazhen Wan, Bosi Wen, Yi Song, Jifan Yu, Yongkang Huang, Libiao Peng, Jiaming Yang, Xiyao Xiao, Sahand Sabour, Xiaohan Zhang, Wenjing Hou, Yijia Zhang, Yuxiao Dong, Jie Tang, Minlie Huang  
**arXiv:** [2311.16832](https://arxiv.org/abs/2311.16832)  
**Date:** 2023-11-28  
**Summary page:** https://www.summarizepaper.com/en/paper/2311.16832v1/

## Key points

- Introduction of CharacterGLM, an extension of ChatGLM with models ranging from 6B to 66B parameters
- Specifically designed for generating Character-based Dialogues (CharacterDial) to enhance conversational AI systems with character customization capabilities
- Surpasses mainstream large language models like the GPT series in terms of consistency, human-likeness, and engagement based on manual evaluations
- Plan to release the 6B version of CharacterGLM along with a subset of training data for further research in character-based dialogue generation

## Summary

, , , , 


In this paper, we introduce CharacterGLM, an extension of ChatGLM that consists of models ranging from 6B to 66B parameters. CharacterGLM is specifically designed for generating Character-based Dialogues (CharacterDial) to enhance conversational AI systems with character customization capabilities. Our model surpasses mainstream large language models like the GPT series in terms of consistency, human-likeness, and engagement based on manual evaluations. To facilitate further research in character-based dialogue generation, we plan to release our 6B version of CharacterGLM along with a subset of training data. An innovative approach to character-based dialogue generation. The foundation for our extended model. The target for enhancing with personalized characters. A key feature of our model for fulfilling social desires and needs. The focus of our research and development efforts.