提出统一框架，让大模型选择更智能高效。
A Unified Approach to Routing and Cascading for LLMs
- 将路由与级联结合，理论证明最优性
- 实验显示性能显著优于单独使用任一方法
- 适合需要平衡成本与效果的模型部署场景
大量嵌入代理系统的大型语言模型（LLMs）为模型选择策略优化成本-性能权衡带来了新可能。现有策略分为路由（每查询选一个模型）和级联（依次运行更大模型直至满意结果）。但当前方法存在三大局限：缺乏最优性的严格证明、无法确定何时最有效、难以融合两种范式提升表现。本文首先推导出一种新型最优级联策略，并证明已有路由策略的最优性；进而提出级联路由（cascade routing），一个统一整合路由与级联的理论最优框架。分析表明，高质量的估计器是模型选择成功的关键因素。实验显示，级联路由在各项指标上均显著优于单一策略，并通过质量估计器分析明确了路由与级联适用的条件。
原文摘要 · Abstract (English)
The availability of a wide range of large language models (LLMs) embedded in various agentic systems has significantly increased the potential of model selection strategies to improve the cost-performance tradeoff. Existing strategies involve either routing, where a single model is chosen per query, or cascading, which sequentially runs increasingly larger models until a satisfactory answer is found. However, current approaches face three key limitations: they (1) lack formal proofs of optimality, (2) fail to identify the conditions under which these strategies are most effective to improve the cost-performance tradeoff, and (3) are unable to combine both paradigms for further improvements. To address these issues, we first derive a novel optimal strategy for cascading and prove the optimality of an existing routing strategy. Further, we propose cascade routing, a unified framework that integrates routing and cascading into a theoretically optimal strategy. Through our analysis, we identify good quality estimators as the critical factor for the success of model selection paradigms. Finally, in our experiments, we show that cascade routing consistently outperforms the individual approaches by a large margin and we analyze quality estimators to determine when routing and/or cascading are useful paradigms for model selection.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。