Basic RAG chain with LCEL in LangChain - Step-by-Step Execution

Concept Flow - Basic RAG chain with LCEL

User Query Input

↓

Retrieve Relevant Docs

↓

Pass Docs + Query to LCEL

↓

LCEL Generates Answer

↓

Return Answer to User

The flow starts with a user query, retrieves documents, then uses LCEL to generate an answer based on those docs and the query.

Execution Sample

LangChain

from langchain.chains import RetrievalQA
from langchain.llms import OpenAI
from langchain.document_loaders import SimpleDirectoryLoader
from langchain.vectorstores import FAISS

# Setup retriever and LCEL chain
retriever = FAISS.load_local('faiss_index', OpenAI().embedding)
qa_chain = RetrievalQA.from_chain_type(llm=OpenAI(), retriever=retriever)

# Run query
answer = qa_chain.run('What is LangChain?')

This code sets up a retrieval-based QA chain using LCEL and runs a query to get an answer.

Execution Table

Step	Action	Input	Output	Notes
1	Receive user query	'What is LangChain?'	Query stored	Start with user input
2	Retrieve documents	Query + FAISS index	Top relevant docs	Retriever finds related info
3	Pass to LCEL	Docs + Query	Prompt for LCEL	Prepare input for language model
4	LCEL generates answer	Prompt	Answer text	Language model creates response
5	Return answer	Answer text	Displayed to user	Final output
6	End	N/A	Process complete	No more steps

💡 Process ends after answer is returned to user.

Variable Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	After Step 4	Final
query	None	'What is LangChain?'	'What is LangChain?'	'What is LangChain?'	'What is LangChain?'	'What is LangChain?'
retrieved_docs	None	None	[Doc1, Doc2, ...]	[Doc1, Doc2, ...]	[Doc1, Doc2, ...]	[Doc1, Doc2, ...]
lc_el_prompt	None	None	None	Prompt with docs + query	Prompt with docs + query	Prompt with docs + query
answer	None	None	None	None	'LangChain is a framework...'	'LangChain is a framework...'

Key Moments - 3 Insights

Why do we need to retrieve documents before asking LCEL?

The execution_table row 2 shows retrieval of relevant documents. LCEL uses these docs to give accurate answers instead of guessing.

What exactly does LCEL receive as input?

Row 3 in execution_table shows LCEL gets a prompt combining the user query and retrieved documents to generate a precise answer.

When does the process stop?

Row 6 marks the end after the answer is returned, so no further steps happen beyond that.

Visual Quiz - 3 Questions

Look at the execution_table, what is the output after Step 2?

A Top relevant documents

B User query stored

C Answer text

D Process complete

💡 Hint

Check the 'Output' column for Step 2 in the execution_table.

At which step does LCEL generate the answer?

A Step 1

B Step 3

C Step 4

D Step 5

💡 Hint

Look for 'LCEL generates answer' in the 'Action' column.

If the retriever returns no documents, what changes in the execution_table?

A Step 2 output would be empty or None

B Step 4 would generate answer normally

C Step 1 would change

D Process would skip Step 3

💡 Hint

Focus on the 'retrieved_docs' variable in variable_tracker after Step 2.

Concept Snapshot

Basic RAG chain with LCEL:
1. Input user query.
2. Retrieve relevant documents using vector search.
3. Combine docs + query into prompt.
4. LCEL (language model) generates answer.
5. Return answer to user.
This ensures answers are based on real info, not guesswork.

Full Transcript

This visual execution trace shows how a Basic RAG chain with LCEL works. First, the user inputs a query. Then the system retrieves relevant documents from a vector store. These documents plus the query form a prompt for the LCEL language model. LCEL generates an answer based on this prompt. Finally, the answer is returned to the user. Variables like 'query', 'retrieved_docs', 'lc_el_prompt', and 'answer' change step-by-step. Key moments clarify why retrieval is needed before LCEL and when the process ends. The quiz tests understanding of each step's output and flow.