PsyMem: Fine-grained psychological alignment and Explicit Memory Control for Advanced Role-Playing LLMs
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912660413480960 |
|---|---|
| author | Cheng, Xilong Qin, Yunxiao Tan, Yuting Li, Zhengnan Wang, Ye Xiao, Hongjiang Zhang, Yuan |
| author_facet | Cheng, Xilong Qin, Yunxiao Tan, Yuting Li, Zhengnan Wang, Ye Xiao, Hongjiang Zhang, Yuan |
| contents | Existing LLM-based role-playing methods often rely on superficial textual descriptions or simplistic metrics, inadequately modeling both intrinsic and extrinsic character dimensions. Additionally, they typically simulate character memory with implicit model knowledge or basic retrieval augment generation without explicit memory alignment, compromising memory consistency. The two issues weaken reliability of role-playing LLMs in several applications, such as trustworthy social simulation. To address these limitations, we propose PsyMem, a novel framework integrating fine-grained psychological attributes and explicit memory control for role-playing. PsyMem supplements textual descriptions with 26 psychological indicators to detailed model character. Additionally, PsyMem implements memory alignment training, explicitly trains the model to align character's response with memory, thereby enabling dynamic memory-controlled responding during inference. By training Qwen2.5-7B-Instruct on our specially designed dataset (including 5,414 characters and 38,962 dialogues extracted from novels), the resulting model, termed as PsyMem-Qwen, outperforms baseline models in role-playing, achieving the best performance in human-likeness and character fidelity. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_12814 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | PsyMem: Fine-grained psychological alignment and Explicit Memory Control for Advanced Role-Playing LLMs Cheng, Xilong Qin, Yunxiao Tan, Yuting Li, Zhengnan Wang, Ye Xiao, Hongjiang Zhang, Yuan Computation and Language Artificial Intelligence Existing LLM-based role-playing methods often rely on superficial textual descriptions or simplistic metrics, inadequately modeling both intrinsic and extrinsic character dimensions. Additionally, they typically simulate character memory with implicit model knowledge or basic retrieval augment generation without explicit memory alignment, compromising memory consistency. The two issues weaken reliability of role-playing LLMs in several applications, such as trustworthy social simulation. To address these limitations, we propose PsyMem, a novel framework integrating fine-grained psychological attributes and explicit memory control for role-playing. PsyMem supplements textual descriptions with 26 psychological indicators to detailed model character. Additionally, PsyMem implements memory alignment training, explicitly trains the model to align character's response with memory, thereby enabling dynamic memory-controlled responding during inference. By training Qwen2.5-7B-Instruct on our specially designed dataset (including 5,414 characters and 38,962 dialogues extracted from novels), the resulting model, termed as PsyMem-Qwen, outperforms baseline models in role-playing, achieving the best performance in human-likeness and character fidelity. |
| title | PsyMem: Fine-grained psychological alignment and Explicit Memory Control for Advanced Role-Playing LLMs |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2505.12814 |