1 paper · 2 filters
Daming Luo, Christy Liang, Junyu Xuan
KV-cache compression methods are predominantly evaluated with the query appended to the context before compression -- a query-aware protocol. Yet the economic case for a compressed…