Training-Free Token-Level Steering for LLM Personalized Co-Writing

2026-08-06

AI Summary

This work addresses the limitations of current large language models in personalized collaborative writing, which stem from insufficient fine-grained domain knowledge, the high cost of conventional fine-tuning, and the inability of retrieval-augmented generation to enable token-level control. To overcome these challenges, the authors propose SteerWrite, a novel framework that introduces, for the first time, a training-free, token-level personalization mechanism. By leveraging context-aware dynamic intervention strategies, SteerWrite efficiently adapts general-purpose large models to specific domains using minimal data. This approach transcends the constraints of both fine-tuning and retrieval-based paradigms, establishing a new non-chat-based collaborative writing framework. Extensive experiments across multiple datasets, base models, and evaluation metrics demonstrate state-of-the-art performance and a significant reduction in manual editing effort.

Abstract

While Large Language Models (LLMs) show great promise for personalization, they often lack specialized domain knowledge. Conventional solutions like fine-tuning struggle with high computational costs and rapid data updates, while Retrieval-Augmented Generation fails to provide fine-grained, token-level steering. Furthermore, chat-based interfaces remain dominant, whereas productive co-writing paradigms have not yet been well exploited beyond the coding domain. To this end, we introduce SteerWrite, a training-free framework designed for personalized co-writing. Our method effectively adapts the base model to specialized domains without gradient updates, with specific designs tailored to small datasets. Experiments demonstrate that SteerWrite achieves state-of-the-art performance across diverse datasets, metrics, and models, significantly reducing human editing effort.

Problem

• personalized co-writing
• domain adaptation
• token-level steering
• training-free
• retrieval-augmented generation

Innovation

• training-free
• token-level steering
• personalized co-writing
• domain adaptation
• SteerWrite