arXiv cs.AI· Kounianhua Du, Jianxing Liu, Kangning Zhang, Wenxiang Jiao, Yuan Lu, Jiarui Jin, Jianghao Lin, Weiwen Liu, Yong Yu, Weinan Zhang·· 3 天前AI 评分31
FleetPilot：细粒度引导实现真实场景下的 LLM 个性化
FleetPilot: Fine-Grained Steering Empowers Real-World LLM Personalization
AI 导读
FleetPilot 是一个细粒度、按实例定制的大语言模型个性化引导框架，可从用户数据动态生成样本级干预向量并注入模型前向传播。它通过 match contrast 提取用户相关信号、细粒度引导组件挂钩 attention 与 MLP 层激活、查询条件聚合模块合成上下文相关增强。
正文 · 原文
Abstract:With the rapid evolution of large language models (LLMs), there is an increasing demand for effective personalization techniques that can adapt model behavior to individual user preferences. Personalized LLMs must infer a user's query-dependent preferences from limited, heterogeneous, and evolving interaction histories. Existing approaches rely primarily on in-context personalization, parametric adaptation, or personalized reward modeling, but respectively incur recurring costs in context management, model optimization and deployment, or multi-candidate generation and scoring. These limitations become particularly restrictive when user observations are sparse and the preference evidence relevant to each query varies across contexts and over time. To address these challenges, we propose \modelname, a fine-grained and instance-tailored steering framework that dynamically generates sample-level interference vectors from user data and injects them into the model's forward pass for personalized adaptation. Our approach introduces: match contrast that produces user-relevant signals, fine-grained steering component that captures nuanced signals by hooking activations from attention and MLP layers, and query-conditioned aggregation module that synthesizes these signals into contextually relevant enhancements. Extensive experiments across diverse scenarios, including short-to-long text generation, and web function calling, validate the effectiveness and compatibility of \modelname. Results show that our method consistently brings significant performance gains, and show extreme superiority in fast-shifting and data-sparse environments. Additionally, \modelname rids us of training cycles and incurs light overhead of storage or inference latency. These make \modelname a strong and real-world deployable method. Implementation is available \footnote{this https URL}.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.14970 [cs.AI] |
| (or arXiv:2610.14970v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.14970 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Kounianhua Du [view email]
[v1]
Sat, 10 Oct 2026 14:44:50 UTC (787 KB)
来源：arXiv cs.AI · arxiv.org