Year: 2026
Type: conference-paper
Abstract: LLM role-playing seeks to portray arbitrary characters in interactive narratives, yet existing systems often lack immersion and adaptability.They typically under-model dynamic environment information and assume a largely static scene/cast, offering limited support for multi-character orchestration, scene transitions, and on-the-fly character introduction.We propose an adaptive multi-agent interaction framework dubbed AdaMARP, which featuring an immersive message format that interleaves [Thought], (Action), <Environment>, and Speech, and an explicit Scene Manager that controls role-playing via discrete actions (init_scene, pick_speaker, switch_scene, add_role, end) with rationales.To train these abilities, we construct AdaRPSet for the Actor Model and AdaSMSet for supervising orchestration decisions, and introduce Adap-tiveBench for trajectory-level evaluation.Experiments across multiple backbones and scales show consistent gains: AdaRPSet improves character consistency, environment grounding, and narrative coherence-an 8B actor outperforming several commercial LLMs, while AdaSMSet enables smoother scene transitions and more natural role introductions, surpassing Claude Sonnet 4.5 with only 14B LLMs.Our code and datasets are publicly available at ht tps://xuzhenhua55.github.io/
Authors: Zhenhua Xu, Dongsheng Chen, Shuo Wang, Jian Li, Chengjie Wang, Meng Han, Yabiao Wang
Institutions: Zhejiang University, Tencent (China)
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Funder: Tencent
OpenAlex ID: W7166852511