Unverified50% confidenceSolutionExact time
生产中典型做法是先用向量检索和BM25快速召回50-100个候选结果，再用reranker精排取top-k送入LLM，常用开源reranker包括Cohere Rerank、bge-reranker和cross-encoder/ms-marco系列
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/28/2026
First Seen
Valid until: 11/26/2026
Sources
Related Entities
Related Claims
UnverifiedHybrid retrieval combines BM25 sparse retrieval with dense vector retrieval, merging ranking results via RRF or linear weighting.67% similarUnverifiedHybrid retrieval combining vector similarity and keyword-based BM25 can significantly improve recall accuracy.66% similarUnverifiedProduction-grade RAG systems commonly adopt a two-stage architecture of hybrid retrieval plus reranking, first using sparse retrieval (such as BM25) to recall candidate documents, then using a Cross-Encoder to rerank and select passages.65% similarUnverifiedDify 在 RAG 管道中支持混合检索（向量检索 + BM25 关键词检索，通过 RRF 算法融合）以及 Reranker 重排序64% similarUnverifiedReranking 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.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/813713API
curl https://kongchang.com/api/v1/knowledge/claims/813713MCP
get_claim(id=813713)