Improving LLM Reasoning through Interpretable Role-Playing Steering
This addresses the need for more stable and interpretable role-playing methods in LLMs, though it is incremental as it builds on existing role-playing techniques.
The paper tackled the problem of unstable and non-interpretable role-playing in LLMs by introducing SRPS, a framework that manipulates internal features to steer role-playing behavior, resulting in accuracy improvements such as Llama3.1-8B on CSQA increasing from 31.86% to 39.80%.
Role-playing has emerged as an effective technique for enhancing the reasoning capabilities of large language models (LLMs). However, existing methods primarily rely on prompt engineering, which often lacks stability and interpretability. In this paper, we introduce Sparse Autoencoder Role-Playing Steering (SRPS), a novel framework that identifies and manipulates internal model features associated with role-playing behavior. Our approach extracts latent representations from role-play prompts, selects the most relevant features based on activation patterns, and constructs a steering vector that can be injected into the model's residual stream with controllable intensity. Our method enables fine-grained control over role-specific behavior and offers insights into how role information influences internal model activations. Extensive experiments across various reasoning benchmarks and model sizes demonstrate consistent performance gains. Notably, in the zero-shot chain-of-thought (CoT) setting, the accuracy of Llama3.1-8B on CSQA improves from 31.86% to 39.80%, while Gemma2-9B on SVAMP increases from 37.50% to 45.10%. These results highlight the potential of SRPS to enhance reasoning ability in LLMs, providing better interpretability and stability compared to traditional prompt-based role-playing.