Articles → LANGCHAIN → Retrieval Augmented Generation

Retrieval Augmented Generation

This article describes Retrieval-Augmented Generation (RAG).

Purpose

RAG is a way for an AI system to search for relevant information first and then use that information to generate an answer.

Augmentation means to make something larger or better by adding to it. The definition is.

Augmentation means adding relevant information retrieved from an external knowledge source to the user's question before sending it to the LLM.

Here is the RAG Pipeline.

Documents → Chunking → Embeddings → Vector Database → Retrieval → LLM → Answer

Component	Description
Data Sources	Your knowledge comes from PDFs, websites, databases, Word files, FAQs, etc.
Document Processing / Chunking	Large documents are split into smaller pieces called chunks so they can be searched efficiently.
Embedding Model	Each chunk is converted into a numerical representation called an embedding.This captures the semantic meaning of the text.
Vector Database / Vector Store	The embeddings are stored here.
Retriever	When the user asks a question, the retriever searches the vector database and finds the chunks most relevant to the question.
Prompt / Context Builder	The retrieved information is added to the user's question to create a prompt for the LLM.
LLM (Large Language Model)	The LLM uses the question plus the retrieved context to generate the final answer.
Response / Generation	The final answer is returned to the user, sometimes with citations or references showing where the information came from.

Query/Feedback