ChatPromptTemplate for conversations in LangChain - Mini Project: Build & Apply

Create a ChatPromptTemplate for Conversations with Langchain

📖 Scenario: You are building a chatbot that uses Langchain to manage conversations. You want to create a prompt template that formats the chat messages properly before sending them to the language model.

🎯 Goal: Build a ChatPromptTemplate that takes a list of messages and formats them for a conversation with a user and an assistant.

📋 What You'll Learn

Create a list of messages with exact roles and content

Define a variable for the template format

Use ChatPromptTemplate.from_messages with the correct message classes

Add the final call to format the messages with input variables

💡 Why This Matters

🌍 Real World

Chatbots and virtual assistants use prompt templates to organize conversations before sending them to AI models.

💼 Career

Understanding how to build and format chat prompts is essential for AI developers and software engineers working with conversational AI.

1

Create the initial list of messages

Create a list called messages with two dictionaries: the first with 'role': 'system' and 'content': 'You are a helpful assistant.', the second with 'role': 'user' and 'content': 'Hello, who are you?'.

LangChain

# Create the messages list with system and user roles
# Your code here

Need a hint?

Use a list of dictionaries with keys 'role' and 'content'.

2

Define the template format string

Create a variable called template and set it to the string '{role}: {content}' to format each message.

LangChain

messages = [
    {'role': 'system', 'content': 'You are a helpful assistant.'},
    {'role': 'user', 'content': 'Hello, who are you?'}
]
# Define the template string
# Your code here

Need a hint?

Use a simple string with placeholders {role} and {content}.

3

Create the ChatPromptTemplate from messages

Import ChatPromptTemplate and SystemMessagePromptTemplate, HumanMessagePromptTemplate from langchain.prompts.chat. Then create a variable called chat_prompt using ChatPromptTemplate.from_messages with a list containing SystemMessagePromptTemplate.from_template(template) and HumanMessagePromptTemplate.from_template(template).

LangChain

messages = [
    {'role': 'system', 'content': 'You are a helpful assistant.'},
    {'role': 'user', 'content': 'Hello, who are you?'}
]
template = '{role}: {content}'

# Import classes and create chat_prompt from messages
# Your code here

Need a hint?

Use the from langchain.prompts.chat import ... statement and call ChatPromptTemplate.from_messages with the correct message templates.

4

Format the chat prompt with input variables

Call chat_prompt.format_prompt with keyword arguments role='user' and content='What can you do?' and assign the result to a variable called formatted_prompt.

LangChain

from langchain.prompts.chat import ChatPromptTemplate, SystemMessagePromptTemplate, HumanMessagePromptTemplate

messages = [
    {'role': 'system', 'content': 'You are a helpful assistant.'},
    {'role': 'user', 'content': 'Hello, who are you?'}
]
template = '{role}: {content}'

chat_prompt = ChatPromptTemplate.from_messages([
    SystemMessagePromptTemplate.from_template(template),
    HumanMessagePromptTemplate.from_template(template)
])

# Format the prompt with role and content
# Your code here

Need a hint?

Call format_prompt on chat_prompt with the exact keyword arguments.