CoSteer: Collaborative Decoding-Time Personalization via Local Delta Steering

TL;DR

CoSteer introduces a decoding-time personalization framework that enables real-time, resource-efficient, and privacy-preserving customization of large language models by leveraging local logit differences to steer cloud-based models.

Contribution

It presents a novel, tuning-free approach for real-time personalization that operates during decoding, using local logit differences to steer large models without fine-tuning.

Findings

Achieves high-quality personalized content generation.

Maintains system efficiency and user privacy.

Demonstrates robustness across various models and tasks.

Abstract

Personalization has become crucial for adapting models to the diverse and evolving needs of users across cultural, temporal, and contextual dimensions. While existing methods often rely on centralized fine-tuning or static preference alignment within a single model, they struggle to achieve both real-time and high-quality personalization under the resource and privacy constraints of personal devices. To address this challenge, we propose CoSteer, a collaborative framework that enables tuning-free, real-time personalization via decoding-time adaptation. By leveraging logit differences between context-aware and context-agnostic local small models, CoSteer steers cloud-based large models, ensuring effective personalization while preserving the large model's capabilities. Personalization is handled locally, with only final tokens sent to the cloud, maintaining both user context and system efficiency. Through extensive experiments across a wide range of tasks, we demonstrate that CoSteer generates high-quality personalized content, ensuring both effectiveness and computational efficiency. Our results highlight its robustness across models and environments, confirming its practical applicability in real-world scenarios.