演化模型库中的持续模型路由
Continual Model Routing in Evolving Model Hubs
AI总结：
针对模型库快速扩展带来的模型选择和路由更新挑战，提出持续模型路由（CMR）问题，构建大规模基准CMRBench，并设计基于对比嵌入的CARvE方法，通过检查点锚定和结构化重放实现高效路由，显著优于多种基线。
摘要：
AI模型库提供了对快速增长的大量预训练模型的访问，使得具有不同路由策略的现成混合专家系统成为可能。然而，这种快速增长带来了两个基本挑战：跨数千个专家进行模型选择的扩展，以及随着新模型和任务的引入持续更新路由机制。在本文中，我们将这一设置形式化为持续模型路由（CMR），并提出了CMRBench，这是一个新的大规模基准，模拟现实的模型库扩展，包括超过2000个候选模型。最后，我们介绍了CARvE，一种对比嵌入方法，通过基于检查点的锚定和结构化重放实现高效的持续模型路由。大量的实验结果和消融研究表明，CARvE在模型、家族和领域级别的准确性上显著优于零样本检索、微调和适配器合并基线。
英文摘要：
AI model hubs provide access to a rapidly growing collection of powerful pre-trained models, enabling off-the-shelf mixture-of-experts systems with different routing strategies. However, this rapid growth poses two fundamental challenges: scaling model selection across thousands of experts and continually updating routing mechanisms as new models and tasks are introduced. In this paper, we formalise this setting as Continual Model Routing (CMR) and propose CMRBench, a new large-scale benchmark simulating realistic hub expansion and including over 2,000 candidate models. Finally, we introduce CARvE, a contrastive embedding approach for efficient continual model routing via checkpoint-based anchoring and structured replay. Extensive empirical results and ablations show that CARvE significantly outperforms zero-shot retrieval, fine-tuning, and adapter-merging baselines in model, family, and domain-level accuracy.
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。
