langchain-rag

1<overview>
2Retrieval Augmented Generation (RAG) enhances LLM responses by fetching relevant context from external knowledge sources.
3 
4**Pipeline:**
51. **Index**: Load → Split → Embed → Store
62. **Retrieve**: Query → Embed → Search → Return docs
73. **Generate**: Docs + Query → LLM → Response
8 
9**Key Components:**
10- **Document Loaders**: Ingest data from files, web, databases
11- **Text Splitters**: Break documents into chunks
12- **Embeddings**: Convert text to vectors
13- **Vector Stores**: Store and search embeddings
14</overview>
15 
16<vectorstore-selection>
17 
18| Vector Store | Use Case | Persistence |
19|--------------|----------|-------------|
20| **InMemory** | Testing | Memory only |
21| **FAISS** | Local, high performance | Disk |
22| **Chroma** | Development | Disk |
23| **Pinecone** | Production, managed | Cloud |
24 
25</vectorstore-selection>
26 
27---
28 
29## Complete RAG Pipeline
30 
31<ex-basic-rag-setup>
32<python>
33End-to-end RAG pipeline: load documents, split into chunks, embed, store, retrieve, and generate a response.
34 
35```python
36from langchain_openai import ChatOpenAI, OpenAIEmbeddings
37from langchain_community.vectorstores import InMemoryVectorStore
38from langchain_text_splitters import RecursiveCharacterTextSplitter
39from langchain_core.documents import Document
40 
41# 1. Load documents
42docs = [
43    Document(page_content="LangChain is a framework for LLM apps.", metadata={}),
44    Document(page_content="RAG = Retrieval Augmented Generation.", metadata={}),
45]
46 
47# 2. Split documents
48…