Comparing prompt versions in LangChain - Mini Project: Build & Apply

Comparing prompt versions

📖 Scenario: You are building a simple tool to compare two versions of text prompts used in a language model application. This helps you see what changed between versions.

🎯 Goal: Create a Python script that stores two prompt versions, sets a threshold for similarity, compares the prompts, and outputs whether they are similar or different.

📋 What You'll Learn

Create two string variables named prompt_v1 and prompt_v2 with exact given texts

Create a float variable similarity_threshold set to 0.8

Write a function compare_prompts that takes two prompts and returns True if their similarity is above the threshold

Add a final line that calls compare_prompts with prompt_v1 and prompt_v2

💡 Why This Matters

🌍 Real World

Comparing prompt versions helps developers track changes and improvements in language model inputs.

💼 Career

Skills in text processing and comparison are useful for AI developers, data scientists, and software engineers working with natural language.

1

DATA SETUP: Define two prompt versions

Create two string variables called prompt_v1 and prompt_v2 with these exact values:
prompt_v1 = "What is the weather like today?"
prompt_v2 = "Can you tell me the weather forecast for today?"

LangChain

# Your code here

Need a hint?

Use exact variable names and string values as shown.

2

CONFIGURATION: Set similarity threshold

Create a float variable called similarity_threshold and set it to 0.8

LangChain

prompt_v1 = "What is the weather like today?"
prompt_v2 = "Can you tell me the weather forecast for today?"
# Your code here

Need a hint?

Use the exact variable name and value.

3

CORE LOGIC: Write a function to compare prompts

Write a function called compare_prompts that takes two parameters p1 and p2. Inside, calculate a simple similarity score by dividing the number of common words by the total unique words in both prompts. Return True if the similarity is greater than or equal to similarity_threshold, otherwise False.

LangChain

prompt_v1 = "What is the weather like today?"
prompt_v2 = "Can you tell me the weather forecast for today?"
similarity_threshold = 0.8
# Your code here

Need a hint?

Use sets to find common and unique words. Compare similarity to similarity_threshold.

4

COMPLETION: Call the comparison function

Add a line that calls compare_prompts with prompt_v1 and prompt_v2 as arguments and assigns the result to a variable called are_similar.

LangChain

prompt_v1 = "What is the weather like today?"
prompt_v2 = "Can you tell me the weather forecast for today?"
similarity_threshold = 0.8

def compare_prompts(p1, p2):
    words_p1 = set(p1.lower().split())
    words_p2 = set(p2.lower().split())
    common_words = words_p1.intersection(words_p2)
    total_unique_words = words_p1.union(words_p2)
    similarity = len(common_words) / len(total_unique_words)
    return similarity >= similarity_threshold

# Your code here

Need a hint?

Assign the function call result to are_similar.