conftrace_

kv cache reuse

2 papers

Explore in graph

Also known as

KV

Co-occurring keywords · papers mentioning both

judge consistency rate (1) cross-candidate attention (1) prompt scheduling (1) distributed llm serving (1) multi-agent system (1) large language model (1) load balancing (1) inference optimization (1)

Papers

When KV Cache Reuse Fails in Multi-Agent Systems: Cross-Candidate Interaction is Crucial for LLM Judges ACL 2026 Preble: Efficient Distributed Prompt Scheduling for LLM Serving ICLR 2025