Basic RAG chain with LCEL in LangChain - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

What is the main purpose of a Retrieval-Augmented Generation (RAG) chain?

In simple terms, what does a RAG chain do in a language model system?

A It compresses large documents into smaller summaries without generating new text.

B It only generates text without using any external information.

C It translates text from one language to another without retrieving documents.

D It retrieves relevant documents and uses them to generate more accurate answers.

💡 Hint

Think about how RAG combines searching and writing.

A RAG chain first finds useful information from documents, then uses that info to create better answers.

❓ Predict Output

intermediate

Output of a simple RAG chain

Given the following code snippet using LangChain, what will be the output?

LangChain

from langchain.chains import RetrievalQA
from langchain.llms import OpenAI

retriever = DummyRetriever()
qa = RetrievalQA.from_chain_type(llm=OpenAI(temperature=0), chain_type="stuff", retriever=retriever)

query = "What is AI?"
result = qa.run(query)
print(result)

A SyntaxError: invalid syntax in chain_type argument

B "AI is a type of hardware used in computers."

C "AI stands for Artificial Intelligence, which is the simulation of human intelligence by machines."

D TypeError: 'DummyRetriever' object is not callable

💡 Hint

Assume DummyRetriever returns a document defining AI correctly.

The retriever provides a document defining AI, and the LLM generates the answer based on it.

❓ Hyperparameter

advanced

Choosing the best temperature setting for a RAG chain

In a RAG chain using LCEL, which temperature setting is best for producing consistent factual answers?

A Temperature = 0 (lowest randomness)

B Temperature = 1 (highest randomness)

C Temperature = 0.7 (medium randomness)

D Temperature = -1 (invalid value)

💡 Hint

Lower temperature means less creativity, more factual.

Setting temperature to 0 makes the model deterministic and factual, ideal for RAG chains.

❓ Metrics

advanced

Evaluating RAG chain performance with accuracy

You run a RAG chain on 100 questions. It answers 85 correctly. What is the accuracy?

A 15%

B 85%

C 100%

D Cannot be determined without loss value

💡 Hint

Accuracy = (correct answers / total questions) * 100

85 correct out of 100 means accuracy is 85%.

🔧 Debug

expert

Debugging a RAG chain that returns empty answers

Given this snippet, why does the RAG chain return empty answers?

retriever = SomeRetriever()
qa_chain = RetrievalQA.from_chain_type(llm=OpenAI(), retriever=retriever)
answer = qa_chain.run("")
print(answer)

A The query string is empty, so no documents are retrieved or answers generated.

B The retriever is not initialized properly, causing a runtime error.

C The OpenAI model is not connected to the internet.

D The chain_type argument is missing, causing a syntax error.

💡 Hint

Check the input query string.

An empty query means no search happens, so no answer is generated.