AdaMARP: An Adaptive Multi-Agent Interaction Framework for General Immersive Role-Playing
This work addresses the challenge of creating more immersive and adaptable role-playing systems for interactive narratives, representing a novel method for a known bottleneck in multi-character orchestration.
The paper tackles the problem of limited immersion and adaptability in LLM role-playing by proposing AdaMARP, an adaptive multi-agent framework with an immersive message format and a Scene Manager, which improves character consistency, environment grounding, and narrative coherence, with an 8B actor outperforming commercial LLMs and a 14B model surpassing Claude Sonnet 4.5 in scene transitions and role introductions.
LLM role-playing aims to portray arbitrary characters in interactive narratives, yet existing systems often suffer from limited immersion and adaptability. They typically under-model dynamic environmental information and assume largely static scenes and casts, offering insufficient support for multi-character orchestration, scene transitions, and on-the-fly character introduction. We propose an adaptive multi-agent role-playing framework, AdaMARP, featuring an immersive message format that interleaves [Thought], (Action), <Environment>, and Speech, together with an explicit Scene Manager that governs role-playing through discrete actions (init_scene, pick_speaker, switch_scene, add_role, end) accompanied by rationales. To train these capabilities, we construct AdaRPSet for the Actor Model and AdaSMSet for supervising orchestration decisions, and introduce AdaptiveBench for trajectory-level evaluation. Experiments across multiple backbones and model scales demonstrate consistent improvements: AdaRPSet enhances character consistency, environment grounding, and narrative coherence, with an 8B actor outperforming several commercial LLMs, while AdaSMSet enables smoother scene transitions and more natural role introductions, surpassing Claude Sonnet 4.5 using only a 14B LLM.