analiza aiqbench
This research proposes RouteFM, a foundation model designed to generalize LLM routing decisions across heterogeneous candidate pools and varying deployment conditions without requiring retraining. By learning to characterize anonymous models through behavioral context, the approach achieves significant performance gains on unseen benchmarks with minimal observation data. This work shifts the paradigm from local fitting toward a reusable framework that adapts dynamically to new environments via context alone.
Pretrain Once, Route Anywhere: Towards a Foundation Model for LLM Routing
arXiv:2609.37362v1 Announce Type: new Abstract: Large language model (LLM) routing aims to assign each query to the most suitable model from a heterogeneous candidate pool, improving the quality--efficiency trade-off of LLM inference. Existing routers are typically learned through local fitting: a router is optimized for a particular query workload and candidate pool, and often requires additional supervision or retraining as the routing environment changes. We ask whether LLM routing can instead be approached from a foundation-model perspective, learning a reusable routing capability that generalizes across tasks, candidate models, and deployment conditions. To this end, we introduce RouteFM, which learns to characterize anonymous candidate models from behavioral context and infer their target-specific capabilities, rather than binding routing decisions to fixed model identities or a single environment. Through episodic pretraining across heterogeneous routing environments, this capability can be reused by a frozen router and adapted to new environments through context alone. Experiments demonstrate transfer across changes in domains, modalities, candidate pools, and context budgets, with the largest gains when behavioral evidence is limited. On MMR-Bench, which is excluded from pretraining, RouteFM outperforms the strongest baseline by 2.23 quality points with only eight observations per candidate. These results support moving LLM routing from repeated local fitting toward a pretrain once, route anywhere paradigm. Our code is publicly available at https://github.com/LAMDA-Model-Reuse/RouteFM.
Abstract:Large language model (LLM) routing aims to assign each query to the most suitable model from a heterogeneous candidate pool, improving the quality--efficiency trade-off of LLM inference. Existing routers are typically learned through local fitting: a router is optimized for a particular query workload and candidate pool, and often requires additional supervision or retraining as the routing environment changes. We ask whether LLM routing can instead be approached from a foundation-model perspective, learning a reusable routing capability that generalizes across tasks, candidate models, and deployment conditions. To this end, we introduce RouteFM, which learns to characterize anonymous candidate models from behavioral context and infer their target-specific capabilities, rather than binding routing decisions to fixed model identities or a single environment. Through episodic pretraining across heterogeneous routing environments, this capability can be reused by a frozen router and adapted to new environments through context alone. Experiments demonstrate transfer across changes in domains, modalities, candidate pools, and context budgets, with the largest gains when behavioral evidence is limited. On MMR-Bench, which is excluded from pretraining, RouteFM outperforms the strongest baseline by 2.23 quality points with only eight observations per candidate. These results support moving LLM routing from repeated local fitting toward a pretrain once, route anywhere paradigm. Our code is publicly available at this https URL.
Artificial Intelligence (cs.AI)
arXiv:2609.37362 [cs.AI]
(or arXiv:2609.37362v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.37362 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 308 cuvinte
Subiecte conexe
Game-Theoretic Disempowerment: Loss of Control to Agencyless AI
LessWrong · 27 sept. 2026
Introducing Instella-MoE: A State-of-the-Art Fully Open Mixture-of-Experts Language Model
AMD · 24 iul. 2026
A continually expandable foundation model for brain MRI
arXiv · 20 aug. 2026
CAPT: A Multi-task Continuous Autoregressive Transformer enabling Cross-dataset and Cross-species Transfer for Calcium Population Dynamics
arXiv · 28 iul. 2026