作者位次 #2/7ACL 2026Poster· Natural Language Processing
C ontrast KV : Robust KV Cache Eviction via Contrastive Signal Fusion for Multi-Query Generalization
Xingchi Chen · Peiyuan Zong · Ziqiang Gao · Qing Li · Yong Jiang · Fa Zhu · Hui Li
摘要
Large Language Models (LLMs) face significant memory and latency overheads during long-context inference due to the growing KV cache, especially in Knowledge Base Question Answering (KBQA) settings that require support for multiple downstream queries. Query-aware eviction methods do not generaliz...