Elamur Research⌕
☰
home / papers / 2025 / Beyond Dialogue: A Profile-Dialogue Alignment Framework Towards General Role-Playing Language Model

Beyond Dialogue: A Profile-Dialogue Alignment Framework Towards General Role-Playing Language Model

Paper · 2025 · Yeyong Yu, Runsheng Yu, Haojie Wei et al.

Summary

Role-playing LLMs are usually fine-tuned on scene dialogues paired with a character profile written for an entire novel or script. The authors show this causes a mismatch: a single scene rarely displays every listed trait. In a sample of 800 HPD scenarios, 83.2% had at least one unaligned dimension. Such training also never teaches the model which sentences express which trait.

BEYOND DIALOGUE addresses both problems automatically and at low cost. For each scenario, an LLM-based prompting mechanism adjusts the profile to match what the dialogue actually shows, which yields cleaner training pairs. The same mechanism produces reasoning data linking individual sentences to profile attributes. These "beyond dialogue" alignment tasks are combined with ordinary dialogue data during supervised fine-tuning of Qwen2 and Mistral-Nemo.

Evaluation is recast as objective questions, such as multiple-choice and true/false items. LLM judges answer them over automatically generated multi-turn dialogues. The tuned models follow profiles better than most proprietary general and specialized baselines, including GPT-4o and Baichuan-NPC-Turbo. This offers an open route to customizable role-playing models.

Key points

  • Bias analysis: Among 800 sampled HPD scenarios, 83.2% displayed a profile–dialogue mismatch on one or more dimensions, for example speaking style or character.
  • Five profile dimensions (CSERP): the framework aligns Character, Style, Emotion, Relationship and Personality at the sentence level through generated reasoning tasks used in training.
  • Base models and baselines: the method was applied to Qwen2 and Mistral-Nemo, and the tuned models beat GPT-4o and Baichuan-NPC-Turbo on key role-following aspects.
  • Ablation: the full task set reached a Qualification Rate of 56.33. Removing the Personality task dropped it to 44.67, and removing the Character task dropped it to 45.0, which were the largest losses.
  • Resources: code and data are available at github.com/yuyouyu32/BeyondDialogue. The processed dialogue dataset is released under CC BY-NC 4.0 for non-commercial research only.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 3,320 worksDialogue systems · 624 works

Cite

Cite as Yeyong Yu, Runsheng Yu, Haojie Wei, Zhanqiu Zhang and Quan Qian. 2025. Beyond Dialogue: A Profile-Dialogue Alignment Framework Towards General Role-Playing Language Model. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.acl-long.586

@inproceedings{yu2025beyond,
  title = {{Beyond Dialogue: A Profile-Dialogue Alignment Framework Towards General Role-Playing Language Model}},
  author = {Yeyong Yu, Runsheng Yu, Haojie Wei, Zhanqiu Zhang and Quan Qian},
  booktitle = {Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2025},
  doi = {10.18653/v1/2025.acl-long.586},
  url = {https://doi.org/10.18653/v1/2025.acl-long.586}
}

References

34 references, 18 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Seungone Kim, Jamin Shin, Yejin Cho, Joel Jang, Shayne Longpre, Hwaran Lee, Sangdoo Yun, Seongjin Shin, Sungdong Kim, James Thorne, et al. 2023. Prometheus: Inducing fine-grained evaluation capability in language models. In The Twelfth International Conference on Learning Representations.
  2. Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. 2023. Gpt-4 technical report. Preprint, arXiv:2303.08774.
  3. Anthropic. 2024. The claude 3 model family: Opus, sonnet, haiku. Online technical report. Accessed: 2025-03-01. find
  4. Xiao Bi, Deli Chen, Guanting Chen, Shanhuang Chen, Damai Dai, Chengqi Deng, Honghui Ding, Kai Dong, Qiushi Du, Zhe Fu, et al. 2024. Deepseek llm: Scaling open-source language models with longtermism. arXiv preprint arXiv:2401.02954, arXiv:2401.02954. bilibili. 2024. Index1.9b technical report. Onlin… find
  5. Jiangjie Chen, Xintao Wang, Rui Xu, Siyu Yuan, Yikai Zhang, Wei Shi, Jian Xie, Shuang Li, Ruihan Yang, Tinghui Zhu, et al. 2024a. From persona to personalization: A survey on role-playing language agents. Preprint, arXiv:2404.18231.
  6. Nuo Chen, Y Wang, Yang Deng, and Jia Li. 2024b. The oscars of ai theater: A survey on role-playing with language models. Preprint, arXiv:2407.11484. — [arXiv:2407.11484](https://arxiv.org/abs/2407.11484) find
  7. Nuo Chen, Yan Wang, Haiyun Jiang, Deng Cai, Yuhan Li, Ziyang Chen, Longyue Wang, and Jia Li. 2023. Large language models meet harry potter: A dataset for aligning dialogue agents with characters. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 8506–8520, Singapore. As…
  8. Daixuan Cheng, Shaohan Huang, and Furu Wei. 2023. Adapting large language models via reading comprehension. Preprint, arXiv:2309.09530. find
  9. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. Preprint, arXiv:2407.21783.
  10. Paul Ekman. 1992. Are there basic emotions? Psychological Review, 99(3):550–553. find
  11. Seungju Han, Beomsu Kim, Jin Yong Yoo, Seokjun Seo, Sangbum Kim, Enkhbayar Erdenee, and Buru Chang. 2022. Meet your favorite character: Open-domain chatbot mimicking fictional characters with only a few utterances. In Proceedings of the 2022 Conference of the North American Chapter of the Associatio… find
  12. Cheng Li, Ziang Leng, Chenxi Yan, Junyi Shen, Hao Wang, Weishi Mi, Yaying Fei, Xiaoyang Feng, Song Yan, HaoSheng Wang, et al. 2023. Chatharuhi: Reviving anime character in reality via large language model. Preprint, arXiv:2308.09597.
  13. Yanran Li, Hui Su, Xiaoyu Shen, Wenjie Li, Ziqiang Cao, and Shuzi Niu. 2017. Dailydialog: A manually labelled multi-turn dialogue dataset. In Proceedings of The 8th International Joint Conference on Natural Language Processing (IJCNLP 2017).
  14. Keming Lu, Bowen Yu, Chang Zhou, and Jingren Zhou. 2024. Large language models are superpositions of all characters: Attaining arbitrary role-play via self-alignment. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7828–7840,…
  15. MistralAI. 2024. Mistral nemo. Online technical report. Accessed: 2025-03-01. find
  16. Yunfan Shao, Linyang Li, Junqi Dai, and Xipeng Qiu. 2023. Character-LLM: A trainable agent for role-playing. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 13153–13187, Singapore. Association for Computational Linguistics.
  17. Xingwu Sun, Yanfeng Chen, Yiqing Huang, Ruobing Xie, Jiaqi Zhu, Kai Zhang, Shuaipeng Li, Zhen Yang, Jonny Han, Xiaobo Shu, et al. 2024. Hunyuan-large: An open-source moe model with 52 billion activated parameters by tencent. Preprint, arXiv:2411.02265.
  18. Yu Sun, Shuohuan Wang, Shikun Feng, Siyu Ding, Chao Pang, Junyuan Shang, Jiaxiang Liu, Xuyi Chen, Yanbin Zhao, Yuxiang Lu, et al. 2021. Ernie 3.0: Large-scale knowledge enhanced pre-training for language understanding and generation. Preprint, arXiv:2107.02137.
  19. Yihong Tang, Jiao Ou, Che Liu, Fuzheng Zhang, Di Zhang, and Kun Gai. 2024. Erabal: Enhancing role-playing agents through boundary-aware learning. Preprint, arXiv:2409.14710. — [arXiv:2409.14710](https://arxiv.org/abs/2409.14710) find
  20. Meiling Tao, Liang Xuechen, Tianyu Shi, Lei Yu, and Yiting Xie. 2024. RoleCraft-GLM: Advancing personalized role-playing in large language models. In Proceedings of the 1st Workshop on Personalization of Generative AI Systems (PERSONALIZE 2024), pages 1–9, St. Julians, Malta. Association for Computa… find
  21. Linguistics (Volume 1: Long Papers), pages 1840– 1873, Bangkok, Thailand. Association for Computational Linguistics. find
  22. Kimi Team, Angang Du, Bofei Gao, Bowei Xing, Changjiu Jiang, Cheng Chen, Cheng Li, Chenjun Xiao, Chenzhuang Du, Chonghua Liao, et al. 2025. Kimi k1. 5: Scaling reinforcement learning with llms. Preprint, arXiv:2501.12599. — Kimi k1.5: Scaling Reinforcement Learning with LLMs
  23. Aiyuan Yang, Bin Xiao, Bingning Wang, Borong Zhang, Ce Bian, Chao Yin, Chenxu Lv, Da Pan, Dian Wang, Dong Yan, et al. 2023. Baichuan 2: Open large-scale language models. Preprint, arXiv:2309.10305.
  24. Quan Tu, Shilong Fan, Zihang Tian, Tianhao Shen, Shuo Shang, Xin Gao, and Rui Yan. 2024. CharacterEval: A Chinese benchmark for role-playing conversational agent evaluation. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1183…
  25. Jack Urbanek, Angela Fan, Siddharth Karamcheti, Saachi Jain, Samuel Humeau, Emily Dinan, Tim Rocktäschel, Douwe Kiela, Arthur Szlam, and Jason Weston. 2019. Learning to speak and act in a fantasy text adventure game. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Proc… find
  26. An Yang, Baosong Yang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Zhou, Chengpeng Li, Chengyuan Li, Dayiheng Liu, et al. 2024. Qwen2 technical report. Preprint, arXiv:2407.10671. find
  27. Alex Young, Bei Chen, Chao Li, Chengen Huang, Ge Zhang, Guanwei Zhang, Guoyin Wang, Heng Li, Jiangcheng Zhu, Jianqun Chen, et al. 2024. Yi: Open foundation models by 01. ai. arXiv preprint arXiv:2403.04652, arXiv:2403.04652.
  28. Phillip B Zarrilli. 2005. Acting (re) considered: a theoretical and practical guide. Routledge. find
  29. Jinfeng Zhou, Zhuang Chen, Dazhen Wan, Bosi Wen, Yi Song, Jifan Yu, et al. 2024. CharacterGLM: Customizing social characters with large language models. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 1457–1476, Miami, Florida, US. Ass… find
  30. Noah Wang, Z.y. Peng, Haoran Que, Jiaheng Liu, Wangchunshu Zhou, Yuhan Wu, Hongcheng Guo, Ruitong Gan, Zehao Ni, Jian Yang, Man Zhang, Zhaoxiang Zhang, Wanli Ouyang, Ke Xu, Wenhao Huang, Jie Fu, and Junran Peng. 2024a. RoleLLM: Benchmarking, eliciting, and enhancing role-playing abilities of large l…
  31. Xiao Wang, Weikang Zhou, Can Zu, Han Xia, Tianze Chen, Yuansen Zhang, Rui Zheng, Junjie Ye, Qi Zhang, Tao Gui, et al. 2023. Instructuie: Multi-task instruction tuning for unified information extraction. Preprint, arXiv:2304.08085.
  32. Xiaoyang Wang, Chen Li, Jianqiao Zhao, and Dong Yu. 2021. Naturalconv: A chinese dialogue dataset towards multi-turn topic-driven conversation. In Proceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI-21). find
  33. Xiaoyang Wang, Hongming Zhang, Tao Ge, Wenhao Yu, Dian Yu, and Dong Yu. 2025. Opencharacter: Training customizable role-playing llms with large-scale synthetic personas. Preprint, arXiv:2501.15427. — [arXiv:2501.15427](https://arxiv.org/abs/2501.15427) find
  34. Xintao Wang, Yunze Xiao, Jen-tse Huang, Siyu Yuan, Rui Xu, Haoran Guo, Quan Tu, Yaying Fei, Ziang Leng, Wei Wang, Jiangjie Chen, Cheng Li, and Yanghua Xiao. 2024b. InCharacter: Evaluating personality fidelity in role-playing agents through psychological interviews. In Proceedings of the 62nd Annual…

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yu-2025-beyond-dialogue-profile-dialogue-alignment/