11 papers
Routing Should Pay for Itself: Sparse Supervision for Economical LLM Routing
Guannan Lai, Gelin Bian, Hao-Xuan Ma +5
Large language model (LLM) routing reduces serving cost by assigning each query to an appropriate model while preserving response quality. Learning such a router, however, often re…
Pretrain Once, Route Anywhere: Towards a Foundation Model for LLM Routing
Guannan Lai, Han-Jia Ye
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 infe…
RouteJudge: An Open Platform for Reproducible and Preference-Aware LLM Routing
Guannan Lai, Haoran Hu, Han-Jia Ye
We present RouteJudge, an online pairwise preference evaluation framework for LLM routing systems, with a public platform available at https://routejudge.cn. Different from model-l…
From Sampled Outcomes to Capability Distributions: Rethinking Supervision for LLM Routing
Guannan Lai, Haoran Hu, Long Chen +2
Existing LLM routing methods often construct supervision from a single sampled response for each query--model pair. Because LLM generation is stochastic, however, such an observati…
The Golden Subspace: Where Efficiency Meets Generalization in Continual Test-Time Adaptation
Guannan Lai, Da-Wei Zhou, Zhenguo Li +1
Continual Test-Time Adaptation (CTTA) aims to enable models to adapt online to unlabeled data streams under distribution shift without accessing source data. Existing CTTA methods…
When Routing Collapses: On the Degenerate Convergence of LLM Routers
Guannan Lai, Han-Jia Ye
LLM routing aims to achieve a favorable quality--cost trade-off by dynamically assigning easy queries to smaller models and harder queries to stronger ones. However, across both un…
MMR-Bench: A Comprehensive Benchmark for Multimodal LLM Routing
Haoxuan Ma, Guannan Lai, Han-Jia Ye
Multimodal large language models (MLLMs) have advanced rapidly, yet heterogeneity in architecture, alignment strategies, and efficiency means that no single model is uniformly supe…
The Lie of the Average: How Class Incremental Learning Evaluation Deceives You?
Guannan Lai, Da-Wei Zhou, Xin Yang +1
Class Incremental Learning (CIL) requires models to continuously learn new classes without forgetting previously learned ones, while maintaining stable performance across all possi…
Order-Robust Class Incremental Learning: Graph-Driven Dynamic Similarity Grouping
Guannan Lai, Yujie Li, Xiangkun Wang +3
Class Incremental Learning (CIL) aims to enable models to learn new classes sequentially while retaining knowledge of previous ones. Although current methods have alleviated catast…
Exploring Open-world Continual Learning with Knowns-Unknowns Knowledge Transfer
Yujie Li, Guannan Lai, Xin Yang +3
Open-World Continual Learning (OWCL) is a challenging paradigm where models must incrementally learn new knowledge without forgetting while operating under an open-world assumption…
A New Perspective on Privacy Protection in Federated Learning with Granular-Ball Computing
Guannan Lai, Yihui Feng, Xin Yang +5
Federated Learning (FL) facilitates collaborative model training while prioritizing privacy by avoiding direct data sharing. However, most existing articles attempt to address chal…