Regression testing for chains in LangChain - Mini Project: Build & Apply

Regression Testing for Chains in Langchain

📖 Scenario: You are working on a Langchain project that uses chains to process text inputs and generate outputs. To keep your project reliable, you want to add regression tests that check if your chains produce the expected results after changes.

🎯 Goal: Build a simple regression test setup for a Langchain chain that verifies the output for a given input text.

📋 What You'll Learn

Create a Langchain chain with a simple prompt template

Set up a test input string

Run the chain with the test input and capture the output

Write an assertion to check the output matches the expected result

💡 Why This Matters

🌍 Real World

Regression testing helps keep your Langchain chains reliable when you update or refactor your code. It ensures your chains still produce the expected outputs for known inputs.

💼 Career

Many jobs in AI and software development require writing tests to maintain code quality. Knowing how to test Langchain chains is useful for roles involving AI workflows and automation.

1

Create a Langchain chain with a prompt template

Import LLMChain and PromptTemplate from langchain. Create a prompt template called prompt with the template string 'Say hello to {name}.'. Then create a chain called chain using LLMChain with the prompt you created.

LangChain

# Your code here to import and create prompt and chain

Need a hint?

Use PromptTemplate to define how the input variable name is used in the prompt string.

2

Set up a test input string

Create a variable called test_input and set it to a dictionary with the key 'name' and the value 'Alice'.

LangChain

from langchain import LLMChain, PromptTemplate

prompt = PromptTemplate(template='Say hello to {name}.', input_variables=['name'])
chain = LLMChain(prompt=prompt)

# Create test_input dictionary with name 'Alice'
# Your code here

Need a hint?

Use a dictionary with the key exactly 'name' and value 'Alice'.

3

Run the chain with the test input and capture the output

Call chain.run() with the argument test_input['name'] and assign the result to a variable called output.

LangChain

from langchain import LLMChain, PromptTemplate

prompt = PromptTemplate(template='Say hello to {name}.', input_variables=['name'])
chain = LLMChain(prompt=prompt)

test_input = {'name': 'Alice'}

# Run the chain with test_input['name'] and save to output
# Your code here

Need a hint?

Use chain.run() with the string value from test_input.

4

Write an assertion to check the output matches the expected result

Write an assert statement that checks if output is equal to the string 'Say hello to Alice.'.

LangChain

from langchain import LLMChain, PromptTemplate

prompt = PromptTemplate(template='Say hello to {name}.', input_variables=['name'])
chain = LLMChain(prompt=prompt)

test_input = {'name': 'Alice'}

output = chain.run(test_input['name'])

# Assert output equals 'Say hello to Alice.'
# Your code here

Need a hint?

Use assert to compare output with the exact expected string.