Showing cs.CLShow all
2 papers · 1 filter
cs.CL2026
LOCUS: Task-Aware Low-Rank Post-Training for Token-Efficient Language Generation
Dongfang Zhao
Large language model serving costs scale directly with output sequence length, yet standard preference alignment often inflates response verbosity without improving utility. We stu…
cs.CL2026
DIVE: Embedding Compression via Self-Limiting Gradient Updates
Dongfang Zhao
High-dimensional language-model embeddings increase storage and search costs, while supervised compressors can overfit when relevance labels are scarce. We present DIVE (Dimensiona…