RMTBench: Benchmarking LLMs Through Multi-Turn User-Centric Role-Playing

TL;DR

RMTBench is a new bilingual, user-centric role-playing benchmark for LLMs that evaluates multi-turn dialogues based on user motivations, aiming to better reflect real-world applications and improve assessment accuracy.

Contribution

It introduces a comprehensive, user-focused role-playing benchmark with diverse characters and dialogue simulation, addressing limitations of previous character-centric evaluations.

Findings

Benchmark includes 80 characters and 8,000 dialogue rounds.

Evaluation based on user motivation alignment improves assessment relevance.

Provides a multi-turn dialogue simulation mechanism for realistic testing.

Abstract

Recent advancements in Large Language Models (LLMs) have shown outstanding potential for role-playing applications. Evaluating these capabilities is becoming crucial yet remains challenging. Existing benchmarks mostly adopt a \textbf{character-centric} approach, simplify user-character interactions to isolated Q&A tasks, and fail to reflect real-world applications. To address this limitation, we introduce RMTBench, a comprehensive \textbf{user-centric} bilingual role-playing benchmark featuring 80 diverse characters and over 8,000 dialogue rounds. RMTBench includes custom characters with detailed backgrounds and abstract characters defined by simple traits, enabling evaluation across various user scenarios. Our benchmark constructs dialogues based on explicit user motivations rather than character descriptions, ensuring alignment with practical user applications. Furthermore, we construct an authentic multi-turn dialogue simulation mechanism. With carefully selected evaluation dimensions and LLM-based scoring, this mechanism captures the complex intention of conversations between the user and the character. By shifting focus from character background to user intention fulfillment, RMTBench bridges the gap between academic evaluation and practical deployment requirements, offering a more effective framework for assessing role-playing capabilities in LLMs. All code and datasets will be released soon. We release the datasets at https://huggingface.co/datasets/xiangh/RMTBENCH.