ServiceActive

FastGPT Knowledge Base

RAG-based knowledge base with vector search, full-text search, hybrid retrieval, and multimodal image search.

Field note

What it does

FastGPT knowledge base uses PostgreSQL PG Vector with HNSW indexing and MongoDB for metadata. Supports multi-vector mapping, semantic/full-text/hybrid search, reranking, question optimization, and image retrieval via description or vector matching.

Capabilities

Available capabilities

Question Optimization Result Reranking Vector Search Hybrid Search Full-Text Search Image Retrieval

1. 引言

FastGPT Knowledge Base

What it does

Available capabilities

Tags

Ways to use it

Product features

1. 引言

1.1 RAG的定义

2.1 检索器（Retriever）

2.2 生成器（Generator）

2.3 RAG的工作流程

2. RAG模型的核心机制

3.1 检索阶段

3.2 生成阶段

3.3 多轮交互与反馈机制

3. RAG模型的工作原理

4. RAG的优势与局限

知识库基础原理介绍