most citedUnderstanding Large Language Models in Your Pockets: Performance Study on COTS Mobile Devices

3 citations · 3 across the 10 of their papers we have counts for

collaborators
Showing cs.CLShow all

1 paper · 1 filter

cs.CL2026

SpecLA: Efficient Speculative Decoding for Linear-Attention Models

Zhibin Wang, Xuying Han, Zhaohua Yang +5

Linear-attention models replace the growing KV cache with recurrent states, but autoregressive decoding still reads, updates, and writes these states one token at a time. Speculati…