RoleMRC: A Fine-Grained Composite Benchmark for Role-Playing and Instruction-Following

TL;DR

RoleMRC is a comprehensive benchmark designed to evaluate and enhance large language models' abilities in fine-grained role-playing and instruction-following across diverse, complex multi-turn scenarios.

Contribution

We introduce RoleMRC, a novel benchmark with extensive role profiles and instructions, enabling detailed assessment and improvement of LLMs' role-playing and instruction-following skills.

Findings

Models fine-tuned on RoleMRC show improved instruction-following.

RoleMRC enhances models' ability to handle complex, multi-turn role-playing tasks.

Cross-evaluation confirms the effectiveness of RoleMRC in boosting LLM capabilities.

Abstract

Role-playing is important for Large Language Models (LLMs) to follow diverse instructions while maintaining role identity and the role's pre-defined ability limits. Existing role-playing datasets mostly contribute to controlling role style and knowledge boundaries, but overlook role-playing in instruction-following scenarios. We introduce a fine-grained role-playing and instruction-following composite benchmark, named RoleMRC, including: (1) Multi-turn dialogues between ideal roles and humans, including free chats or discussions upon given passages; (2) Role-playing machine reading comprehension, involving response, refusal, and attempts according to passage answerability and role ability; (3) More complex scenarios with nested, multi-turn and prioritized instructions. The final RoleMRC features a 10.2k role profile meta-pool, 37.9k well-synthesized role-playing instructions, and 1.4k testing samples. We develop a pipeline to quantitatively evaluate the fine-grained role-playing and instruction-following capabilities of several mainstream LLMs, as well as models that are fine-tuned on our data. Moreover, cross-evaluation on external role-playing datasets confirms that models fine-tuned on RoleMRC enhances instruction-following without compromising general role-playing and reasoning capabilities. We also probe the neural-level activation maps of different capabilities over post-tuned LLMs. Access to our RoleMRC, RoleMRC-mix and Codes: https://github.com/LuJunru/RoleMRC.