Question reformulation with history in LangChain - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ component_behavior

intermediate

What is the output of this LangChain question reformulation code?

Given the following LangChain code snippet that reformulates a follow-up question using chat history, what will be the output when the input question is "What is the capital?" and the chat history contains one previous question-answer pair?

LangChain

from langchain.chains import TransformChain
from langchain.prompts import PromptTemplate

prompt_template = """
Given the following conversation and a follow up question, rephrase the follow up question to be a standalone question.

Chat History:
{chat_history}
Follow Up Input: {question}
Standalone question:"""
prompt = PromptTemplate(input_variables=["chat_history", "question"], template=prompt_template)

chain = TransformChain(input_variables=["chat_history", "question"], output_variables=["rephrased_question"], transform=lambda inputs: {"rephrased_question": f"Standalone: {inputs['question']}"})

inputs = {"chat_history": "Q: Who is the president? A: Joe Biden.", "question": "What is the capital?"}
output = chain.run(inputs)
print(output)

A {"rephrased_question": "What is the capital?"}

B {"rephrased_question": "Standalone: What is the capital?"}

C SyntaxError due to missing colon in lambda

D KeyError because 'question' key is missing in inputs

💡 Hint

Look at how the lambda function formats the output dictionary using the input question.

The TransformChain uses a lambda that returns a dictionary with key 'rephrased_question' and value 'Standalone: ' plus the input question. The input question is 'What is the capital?', so the output dictionary is {"rephrased_question": "Standalone: What is the capital?"}.

📝 Syntax

intermediate

Which option causes a syntax error in this LangChain prompt template?

Consider the following prompt template string for question reformulation. Which option contains a syntax error that will prevent the PromptTemplate from being created?

LangChain

prompt_template = """
Given the following conversation and a follow up question, rephrase the follow up question to be a standalone question.

Chat History:
{chat_history}
Follow Up Input: {question}
Standalone question:"""

A prompt = PromptTemplate(input_variables=["chat_history", "question"], template=prompt_template)

B prompt = PromptTemplate(input_variables=["chat_history", "question"], template=prompt_template + "\n")

C prompt = PromptTemplate(input_variables=["chat_history", "question"], template=prompt_template[:-1])

D prompt = PromptTemplate(input_variables=["chat_history", "question"], template=prompt_template.replace('{question}', '{follow_up}'))

💡 Hint

Check if the input variables match the placeholders in the template string.

Option D replaces '{question}' with '{follow_up}', but 'follow_up' is not listed in input_variables, causing a KeyError or template error when rendering.

❓ state_output

advanced

What is the value of 'rephrased_question' after running this LangChain chain?

Given a LangChain TransformChain that reformulates a question by appending chat history, what is the value of 'rephrased_question' after running with the inputs below?

LangChain

from langchain.chains import TransformChain

chain = TransformChain(
    input_variables=["chat_history", "question"],
    output_variables=["rephrased_question"],
    transform=lambda inputs: {"rephrased_question": f"{inputs['chat_history']} Then ask: {inputs['question']}"}
)

inputs = {"chat_history": "Q: Who won the game? A: Team A.", "question": "What was the score?"}
output = chain.run(inputs)
rephrased = output["rephrased_question"]

A "Q: Who won the game? A: Team A. Then ask: What was the score?"

B "Then ask: What was the score? Q: Who won the game? A: Team A."

C "Q: Who won the game? A: Team A. What was the score?"

D KeyError because 'rephrased_question' is not in output

💡 Hint

Look at how the lambda formats the string using chat_history and question keys.

The lambda returns a dictionary with key 'rephrased_question' and value combining chat_history, ' Then ask: ', and question. So the value is 'Q: Who won the game? A: Team A. Then ask: What was the score?'.

🔧 Debug

advanced

Which option causes a runtime error when running this LangChain chain?

Given this LangChain TransformChain code, which option will cause a runtime error when calling chain.run with the provided inputs?

LangChain

from langchain.chains import TransformChain

chain = TransformChain(
    input_variables=["chat_history", "question"],
    output_variables=["rephrased_question"],
    transform=lambda inputs: {"rephrased_question": inputs["question"].upper()}
)

inputs = {"chat_history": "Previous Q&A", "question": "hello"}

A chain.run(inputs)

B chain.run({"chat_history": "Previous Q&A", "question": None})

C chain.run({"chat_history": "Previous Q&A"})

D chain.run({"question": "hello"})

💡 Hint

Check which inputs dictionary is missing required keys.

Option C is missing the 'question' key, so accessing inputs["question"] raises a KeyError at runtime.

🧠 Conceptual

expert

How does LangChain's question reformulation with history improve conversational AI?

Why is it important to reformulate follow-up questions into standalone questions using chat history in LangChain conversational chains?

A It allows the language model to understand the context without needing the entire conversation history, improving accuracy and relevance.

B It reduces the number of tokens sent to the model by removing all previous conversation context.

C It enables the model to generate questions instead of answers, improving engagement.

D It stores the entire conversation history in a database for later retrieval.

💡 Hint

Think about how standalone questions help the model understand context better.

Reformulating follow-up questions into standalone questions provides the model with all necessary context in a single input, avoiding confusion and improving response quality.