HER: Human-like Reasoning and Reinforcement Learning for LLM Role-playing
This addresses the problem of cognitive simulation in LLM role-playing for applications like companionship and gaming, representing a novel method for a known bottleneck.
The paper tackles the challenge of simulating inner thoughts in LLM role-playing by proposing HER, a framework that uses dual-layer thinking and reinforcement learning, resulting in a 30.26 improvement on the CoSER benchmark and a 14.97 gain on the Minimax Role-Play Bench compared to the baseline.
LLM role-playing, i.e., using LLMs to simulate specific personas, has emerged as a key capability in various applications, such as companionship, content creation, and digital games. While current models effectively capture character tones and knowledge, simulating the inner thoughts behind their behaviors remains a challenge. Towards cognitive simulation in LLM role-play, previous efforts mainly suffer from two deficiencies: data with high-quality reasoning traces, and reliable reward signals aligned with human preferences. In this paper, we propose HER, a unified framework for cognitive-level persona simulation. HER introduces dual-layer thinking, which distinguishes characters' first-person thinking from LLMs' third-person thinking. To bridge these gaps, we curate reasoning-augmented role-playing data via reverse engineering and construct human-aligned principles and reward models. Leveraging these resources, we train HER models based on Qwen3-32B via supervised and reinforcement learning. Extensive experiments validate the effectiveness of our approach. Notably, our models significantly outperform the Qwen3-32B baseline, achieving a 30.26 improvement on the CoSER benchmark and a 14.97 gain on the Minimax Role-Play Bench. Our datasets, principles, and models will be released to facilitate future research.