Showing 2026Show all
3 papers · 1 filter
cs.LG2026
Elastic Selective Spectral Hybrids for Train-Once, Export-Many Budgeted Inference
Dachuan Song, Chuchu Chen, Xuan Wang
Deploying a language model under different computing and latency budgets calls for compact models with different quality-cost trade-offs. Towards this end, elastic spectral state s…
cs.CL2026
Mergeable Model-Side Aggregation States for Long-Context Language Models
Dachuan Song, Junyu Yin, Zechen Hu +1
A known limitation of long-context language models is their increasingly unreliable performance in non-additive, set-based aggregation as context length grows. Examples include car…
cs.LG2026
Elastic Spectral State Space Models for Train-Once Budgeted Inference
Dachuan Song, Xuan Wang, Junyu Yin +1
Modern sequence models are typically trained at a fixed computational capacity, while real-world applications require deployment across platforms with different resource constraint…