Unverified50% confidenceFactExact time
The RAG retrieval process is: question → rewriting → embedding → top-k → reranking → context → LLM; the model does not perform the search—it is the application code that searches and puts the results into the context; embedding uses a specially trained model to convert text into fixed-length vectors, and cosine similarity measures semantic distance.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedRAG工作流通常涉及两类模型：嵌入模型（将文本转换为向量用于相似度检索）和对话大模型（综合检索结果生成答案）71% similarUnverifiedA practical RAG technique is to first have the model write out a plausible answer and then search in the vicinity of that hypothetical answer to improve recall quality67% similarUnverifiedRAG changes the model's input by retrieving external documents and inserting them into the prompt, but does not touch the model's own reasoning logic or weights, constituting guidance rather than setting behavior.66% similarUnverifiedReranking models are a second-stage optimization in RAG pipelines that perform fine-grained relevance scoring and reordering of candidate documents against a query after vector retrieval returns candidate results66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1048077API
curl https://kongchang.com/api/v1/knowledge/claims/1048077MCP
get_claim(id=1048077)