Expired50% confidenceFactExact time
The recall stage typically uses Bi-Encoder models (such as text-embedding-ada-002 or the BGE series), while the reranking stage uses Cross-Encoder models (such as Cohere Rerank or BGE-Reranker).
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/20/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 10/18/2026(expired)
Sources
Related Claims
UnverifiedThe retrieval stage uses a Bi-Encoder dual-tower model (such as text-embedding-ada-002 or the BGE series), with vectors precomputed offline and stored in a vector database (such as FAISS or Milvus).78% similarVerifiedRerankers typically adopt a Cross-Encoder architecture, in contrast to the Bi-Encoder architecture used in vector retrieval.75% similarUnverified现代Sentence Embedding模型通常采用双塔（Bi-Encoder）结构，查询和文档分别独立编码为向量70% similarUnverifiedCross-Encoders concatenate the query and document into a single input sequence and perform deep interaction modeling through Transformer's full attention mechanism; representative models include Cohere Rerank, BGE-Reranker, and ms-marco-MiniLM.69% similarUnverified主流Embedding模型（如BGE、text-embedding系列）通常基于BERT架构的编码器（Encoder-only），通过对比学习让语义相近的句子在向量空间中靠拢；而GPT类生成模型基于解码器（Decoder-only）69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/565221API
curl https://kongchang.com/api/v1/knowledge/claims/565221MCP
get_claim(id=565221)