arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2511.16883cs.LG
收藏

PersonalizedRouter: 通过基于图的用户偏好建模实现个性化LLM路由

PersonalizedRouter: Personalized LLM Routing via Graph-based User Preference Modeling

University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校)

Zhongjie Dai, Tao Feng, Jiaxuan You

AI总结:

PersonalizedRouter通过基于图的用户偏好建模,实现个性化LLM选择,显著优于现有方法,具有高效和强泛化能力。

摘要:

随着大型语言模型(LLM)数量的增长,这些模型具有多样化的功能和响应风格,为用户提供更广泛的选择,但如何选择合适的LLM成为挑战,因为用户在性能、成本和响应风格上存在差异。当前的LLM选择方法通常优化单一固定目标,如性能、成本或它们之间的权衡,而无法从交互数据中学习个体用户偏好。为了解决这些限制,我们提出了PersonalizedRouter,一个基于图的框架,通过利用包含任务上下文、查询、候选LLM和用户决策的交互数据,建模多样的用户资料并进行个性化LLM选择。为了捕捉用户查询和最优LLM之间的上下文信息,PersonalizedRouter将交互数据转换为异构图,其中不同类型的节点之间的关系由边表示。为了评估跨用户适应性,我们设计了两种策略:多成本效率模拟策略和LLM-as-a-Judge策略。此外,我们构建了PersonaRoute-Bench,一个包含1000个模拟用户和10个LLM的大规模基准。实验结果表明,PersonalizedRouter在现有LLM选择方法上显著优于,并在两种模拟策略下分别超越最强方法15.38%和9.83%。在具有1000个用户的PersonaRoute-Bench上,它进一步在保持更高效率的同时,超越最佳方法16.19%和59.69%。此外,PersonalizedRouter表现出强大的少样本泛化能力,当适应到新用户和新LLM时,其性能达到完全训练模型的64.81%和85.80%。

英文摘要:

The growing number of Large Language Models (LLMs) with diverse capabilities and response styles provides users with a wider range of choices, which presents challenges in selecting appropriate LLMs, as user preferences vary in terms of performance, cost, and response style. Current LLM selection methods typically optimize for a single fixed objective, such as performance, cost, or a trade-off between them, and fail to learn individual user preferences from interaction data. To address these limitations, we propose PersonalizedRouter, a graph-based framework that models diverse user profiles and performs personalized LLM selection by leveraging interaction data that includes task context, queries, candidate LLMs, and user decisions. To capture contextual information between user queries and optimal LLMs, PersonalizedRouter converts the interaction data into a heterogeneous graph, where the relationships between different types of nodes are represented by edges. To evaluate adaptability across users, we design two strategies: the multi-cost-efficiency simulation strategy and the LLM-as-a-Judge strategy. In addition, we construct PersonaRoute-Bench, a large-scale benchmark with 1,000 simulated users and 10 LLMs. Experimental results show that PersonalizedRouter significantly outperforms existing LLM selection methods and surpasses the strongest methods by a large margin of 15.38% and 9.83% under two simulation strategies. On the PersonaRoute-Bench with 1,000 users, it further surpasses the best methods by 16.19% and 59.69% while maintaining higher efficiency. Moreover, PersonalizedRouter demonstrates strong few-shot generalization, achieving 64.81% and 85.80% of the fully trained model's performance when adapting to new users and new LLMs.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑