Unverified50% confidenceFactExact time
大模型二次开发的核心技术路径有两条：RAG通过向量数据库（如Faiss、Milvus、Chroma）注入私域知识无需修改权重，Fine-tuning用行业数据微调，常用LoRA、QLoRA等参数高效微调技术
1
Sources
50%
Confidence
Long-term
Relevance
7/5/2026
First Seen
Sources
Related Claims
UnverifiedRAG (Retrieval-Augmented Generation) systems can be built on top of vector databases such as Milvus, Weaviate, and Qdrant to supplement large models with private domain knowledge.85% similarUnverified检索增强生成（RAG）技术通过将向量数据库（如Pinecone、Chroma）与大模型结合，使Agent能够实时检索PubMed、arXiv等学术数据库的最新文献73% similarUnverifiedRAG通过检索+生成两阶段架构，先将问题转化为向量表示在向量数据库（如Faiss、Milvus、Chroma）中进行语义检索，弥补大模型知识固化和无法感知私有信息的缺陷73% similarUnverifiedBuilding a RAG system requires three core components working in concert: a vector database (such as Pinecone or Chroma), an embedding model, and an LLM.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/112316API
curl https://kongchang.com/api/v1/knowledge/claims/112316MCP
get_claim(id=112316)