[论文解读] DAK-UCB: Diversity-Aware Prompt Routing for LLMs and Generative Models

Donya Jafari, Farzan Farnia|arXiv (Cornell University)|Mar 24, 2026

Advanced Bandit Algorithms Research被引用 0

一句话总结

DAK-UCB 引入一种多样性感知的上下文带权随机臂方法，用于在线选择生成模型，在保留保真度的同时通过联合核距离分衡输出多样性。它还扩展到条件提示的混合策略以提升多样性。

ABSTRACT

The expansion of generative AI and LLM services underscores the growing need for adaptive mechanisms to select an appropriate available model to respond to a user's prompts. Recent works have proposed offline and online learning formulations to identify the optimal generative AI model for an input prompt, based solely on maximizing prompt-based fidelity evaluation scores, e.g., CLIP-Score in text-to-image generation. However, such fidelity-based selection methods overlook the diversity of generated outputs, and hence, they can fail to address potential diversity shortcomings in the generated responses. In this paper, we introduce the Diversity-Aware Kernelized Upper Confidence Bound (DAK-UCB) method as a contextual bandit algorithm for the online selection of generative models with diversity considerations. The proposed DAK-UCB method incorporates both fidelity and diversity-related metrics into the selection process. We design this framework based on prompt-aware diversity score functions that decompose to a two-sample-based expectation over prompt-output pairs in the previous generation rounds. Specifically, we illustrate the application of our framework using joint kernel distance and kernel entropy measures. Our experimental results demonstrate the effectiveness of DAK-UCB in promoting diversity-aware model selection while maintaining fidelity in the generations for a sequence of prompts. The code is available at https://github.com/Donya-Jafari/DAK-UCB.

研究动机与目标

• 强调需要考虑多样性而非仅保真度的提示感知模型选择.
• 提出将多样性度量整合到在线模型选择中的 Diversity-Aware Kernelized UCB (DAK-UCB) 框架.
• 使提示条件混合策略进一步提升生成输出的多样性.
• 为所提出的多样性感知方法提供理论保证（后悔界）。
• 在文本到图像和语言模型任务中展示在多样性和保真度方面的经验提升。

提出的方法

• 用联合乘积核将 KD 和 RKE 扩展到条件提示–输出设定。
• 将 Joint Kernel Distance (JKD) 和 Joint-RKE (JRKE) 定义为对提示与输出的两样本期望。
• 将 JKD/JRKE 与保真度分数（如 CLIP-Score）整合进带有多样性项的逐臂核化 UCB 目标。
• 通过对每个提示仅用单个样本进行核岭回归来估计提示级目标，从而实现在线置信界。
• 提供 Mixture-DAK-UCB 变体，通过求解带约束的二次规划来优化提示相关混合。
• 在标准假设下给出阶段性后悔界，证明该多样性感知扩展保留核化 UCB 的保证。

Figure 1: Comparison of baseline Kernelized-UCB model selection (CLIP-Score fidelity metric) (Hu et al. , 2025a ) vs. our proposed diversity-aware DAK-UCB over $T=500$ rounds. While the baseline Kernelized-UCB does not favor model $G_{2}$ with higher diversity over model $G_{1}$ , DAK-UCB selected t

实验结果

研究问题

• RQ1 如何将多样性纳入由提示驱动的生成模型的在线选择？
• RQ2 联合核多样性分数（JKD、JRKE）在平衡保真度与多样性时是否能改善模型选择？
• RQ3 提示条件的混合是否能在单臂选择之外进一步提升多样性？
• RQ4 对于像 DAK-UCB 这样的多样性感知上下文带权臂方法，存在哪些理论保证（后悔界）？
• RQ5 多样性感知方法是否能提升实际任务如文本到图像生成和提示感知的 LLM 路由？

主要发现

• 与仅保真度基线相比，DAK-UCB 推动了面向多样性的模型选择，同时保持保真度。
• 联合 JKD 与 Joint-RKE 能以两样本、提示感知形式实现在线估计，每个提示仅需一个样本。
• 在基于 MS-COCO 的实验中，DAK-UCB 实现了更高的多样性指标（Joint-RKE），并具有具有竞争力的保真度（如 CLIP-Score）。
• Mixture-DAK-UCB 使提示相关混合成为可能，进一步提升了相对于单臂选择的多样性。
• 该方法在标准假设下提供了类似于核化 UCB 的形式化后悔界。

Figure 2: Performance comparison on JKD score and Joint-RKE for MS-COCO prompt clusters using Kandinsky, SDXL, and GigaGAN.