Unverified50% confidenceFactExact time
缓存机制在技术上通常被称为 KV Cache 或 Prompt Caching
1
Sources
50%
Confidence
Long-term
Relevance
9/16/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedKV Cache通过将已计算的键值对持久化存储,使相同Prompt前缀的后续请求可直接复用,跳过重复的矩阵运算76% similarUnverifiedAnthropic的提示缓存(Prompt Caching)利用Transformer架构中的KV Cache机制,通过缓存共享提示前缀对应的键值状态来降低延迟和成本72% similarUnverifiedKV Cache(键值缓存)机制使得共享相同前缀的API请求可以复用已计算的Key和Value向量,从而大幅降低推理成本72% similarUnverifiedPrompt Cache本质上是大模型推理架构中KV Cache机制的服务端延伸,缓存Transformer中每个token生成的Key和Value矩阵72% similarUnverifiedvLLM的前缀缓存(Prefix Caching)对KV Cache块计算哈希值,当新请求输入前缀与已缓存请求重叠时直接复用物理缓存块,跳过该部分prefill计算71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/927829API
curl https://kongchang.com/api/v1/knowledge/claims/927829MCP
get_claim(id=927829)