Unverified50% confidenceFactExact time
重排阶段通常对10-50条候选文档进行精排，能够捕捉双编码器遗漏的细微语义关联从而提升检索相关性
1
Sources
50%
Confidence
Long-term
Relevance
8/27/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedReranking uses a Cross-Encoder model to score candidate documents retrieved in the initial retrieval pairwise, then fine-ranks the Top-K to feed into the LLM.73% similarUnverified重排模型会对查询和每个候选文档进行更精细的相关性打分，将最相关的文档排在前列，从而提升检索质量72% similarVerified重排序（Re-ranking）机制在初步检索后引入交叉编码器（Cross-Encoder）对候选文档重新打分，提升输入大模型的上下文质量71% similarUnverifiedVector databases such as Pinecone, Weaviate, and Chroma can complete semantic similarity code retrieval at the millisecond level via approximate nearest neighbor (ANN) algorithms.68% similarUnverified本地代码搜索层流程分五步：切分代码、双路搜索并行、进一步压缩、追踪调用关系、打分过滤67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/808067API
curl https://kongchang.com/api/v1/knowledge/claims/808067MCP
get_claim(id=808067)