Persona Switch: Mixing Distinct Perspectives in Decoding Time

2026-01-22

AI Summary

This work addresses the inconsistent effectiveness of role-playing prompts in enhancing zero-shot reasoning capabilities of large language models. To overcome this limitation, the authors propose a dynamic decoding method that adaptively selects between standard zero-shot and role-playing prompt outputs at each generation step, based on the confidence derived from the logit gap of the current token prediction. This approach enables, for the first time, real-time switching and synergistic integration of prompting strategies during the decoding phase. Evaluated across multiple mainstream large language models, the method achieves an average accuracy improvement of up to 5.13%, significantly outperforming existing baselines.

Technology Category

• Natural Language Processing: Prompt Engineering / Prompting
• Machine Learning: Large Multimodal Models (LMMs)
• Reasoning under Uncertainty: Sequential Decision Making

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Role-play prompting is known to steer the behavior of language models by injecting a persona into the prompt, improving their zero-shot reasoning capabilities. However, such improvements are inconsistent across different tasks or instances. This inconsistency suggests that zero-shot and role-play prompting may offer complementary strengths rather than one being universally superior. Building on this insight, we propose Persona Switch, a novel decoding method that dynamically combines the benefits of both prompting strategies. Our method proceeds step-by-step, selecting the better output between zero-shot and role-play prompting at each step by comparing their output confidence, as measured by the logit gap. Experiments with widely-used LLMs demonstrate that Persona Switch consistently outperforms competitive baselines, achieving up to 5.13% accuracy improvement. Furthermore, we show that output confidence serves as an informative measure for selecting the more reliable output.

Problem

• role-play prompting
• zero-shot prompting
• decoding strategy
• language models
• output consistency

Innovation

• Persona Switch
• role-play prompting
• zero-shot reasoning
• logit gap
• dynamic decoding