Unverified90% confidenceFactTime unknown
RAG检索阶段采用两阶段策略：第一阶段向量检索采用双编码器架构进行粗排，第二阶段Rerank模型采用交叉编码器架构进行精排
1
Sources
90%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe two-stage retrieval strategy first uses fast vector retrieval to recall Top-N candidates, then uses a cross-encoder to precisely score those N candidates and select the final Top-K.76% similarUnverifiedThe reranking module typically uses a cross-encoder model to compute relevance scores between queries and documents, performing a secondary ranking of the candidate documents from the initial retrieval.63% similarUnverified「Bi-Encoder粗召回 + Cross-Encoder精排」的两阶段级联架构已成为行业标准配置，MS MARCO等基准测试上的主流方案均采用此范式62% similarUnverifiedThe test contains two programming exam questions covering dimensions such as cross-project migration, architecture adaptation, full-stack development, and streaming output61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/18816API
curl https://kongchang.com/api/v1/knowledge/claims/18816MCP
get_claim(id=18816)