Unverified50% confidenceFactExact time
RadixAttention manages shared prefixes through a Radix Tree structure to reduce redundant computation.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedRadixAttention 引入全局共享 KV Cache 池，通过引用计数和 LRU 淘汰策略管理缓存生命周期，并在 Radix Tree 中执行最长前缀匹配72% similarUnverifiedRadixAttention 通过将 KV Cache 组织为前缀树结构，使具有相同前缀的多个请求共享同一份缓存69% similarUnverified全局上下文文件可以结合检索增强生成（RAG）机制，在项目规模扩大后按需注入相关规则片段，避免超出模型的上下文窗口限制63% similarUnverified工程界普遍采用RAG架构：将历史信息切分成语义块，通过嵌入模型转化为向量存入向量数据库（如Faiss、Pinecone、Weaviate），用户提问时计算余弦相似度检索最相关片段60% similarPartially VerifiedSGLang's core innovation is the RadixAttention technology, which achieves automatic KV Cache reuse through a prefix tree structure.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/893348API
curl https://kongchang.com/api/v1/knowledge/claims/893348MCP
get_claim(id=893348)