CharacterBox: Evaluating the Role-Playing Capabilities of LLMs in Text-Based Virtual Worlds
This addresses the problem of inadequate evaluation methods for role-playing in LLMs, which is crucial for applications like intelligent NPCs and digital twins, though it appears incremental as it builds on existing simulation and fine-tuning approaches.
The paper tackles the challenge of evaluating role-playing capabilities in LLMs by proposing CharacterBox, a simulation sandbox that generates fine-grained character behavior trajectories, and demonstrates competitive performance with fine-tuned smaller models compared to advanced GPT APIs.
Role-playing is a crucial capability of Large Language Models (LLMs), enabling a wide range of practical applications, including intelligent non-player characters, digital twins, and emotional companions. Evaluating this capability in LLMs is challenging due to the complex dynamics involved in role-playing, such as maintaining character fidelity throughout a storyline and navigating open-ended narratives without a definitive ground truth. Current evaluation methods, which primarily focus on question-answering or conversational snapshots, fall short of adequately capturing the nuanced character traits and behaviors essential for authentic role-playing. In this paper, we propose CharacterBox, which is a simulation sandbox designed to generate situational fine-grained character behavior trajectories. These behavior trajectories enable a more comprehensive and in-depth evaluation of role-playing capabilities. CharacterBox consists of two main components: the character agent and the narrator agent. The character agent, grounded in psychological and behavioral science, exhibits human-like behaviors, while the narrator agent coordinates interactions between character agents and environmental changes. Additionally, we introduce two trajectory-based methods that leverage CharacterBox to enhance LLM performance. To reduce costs and facilitate the adoption of CharacterBox by public communities, we fine-tune two smaller models, CharacterNR and CharacterRM, as substitutes for GPT API calls, and demonstrate their competitive performance compared to advanced GPT APIs.