Partial prompt templates in LangChain - Mini Project: Build & Apply

Building a Partial Prompt Template with LangChain

📖 Scenario: You are creating a chatbot that answers questions about movies. You want to reuse parts of the prompt that describe the chatbot's role and style, while allowing the movie title to change dynamically.

🎯 Goal: Build a partial prompt template using LangChain that fixes the chatbot's role and style, and later add the movie title dynamically to complete the prompt.

📋 What You'll Learn

Create a partial prompt template with fixed role and style

Add a variable placeholder for the movie title

Use the partial prompt template to create a full prompt by filling in the movie title

Ensure the prompt template is valid and usable with LangChain's PromptTemplate

💡 Why This Matters

🌍 Real World

Partial prompt templates let you reuse fixed parts of prompts and fill in dynamic data later, which is common in chatbots and AI assistants.

💼 Career

Knowing how to build and use partial prompt templates is useful for AI developers and prompt engineers working with LangChain or similar frameworks.

1

Create a partial prompt template with fixed role and style

Import PromptTemplate from langchain.prompts and create a partial prompt template called partial_prompt with the template string: "You are a helpful assistant who provides detailed answers about movies. Your style is friendly and informative."

LangChain

# Your code here

Hint

Use PromptTemplate and set the template argument to the fixed text.

2

Add a variable placeholder for the movie title

Modify the partial_prompt template string to include a variable placeholder called {movie_title} at the end, so the template becomes: "You are a helpful assistant who provides detailed answers about movies. Your style is friendly and informative. Tell me about the movie {movie_title}."

LangChain

from langchain.prompts import PromptTemplate

# Add {movie_title} placeholder to the template string
partial_prompt = PromptTemplate(template="You are a helpful assistant who provides detailed answers about movies. Your style is friendly and informative. Tell me about the movie {movie_title}.")

Hint

Insert {movie_title} exactly as shown inside the template string.

3

Create a full prompt by filling in the movie title

Use the partial_prompt to create a full prompt by calling its format method with movie_title='Inception' and assign the result to a variable called full_prompt.

LangChain

from langchain.prompts import PromptTemplate

partial_prompt = PromptTemplate(template="You are a helpful assistant who provides detailed answers about movies. Your style is friendly and informative. Tell me about the movie {movie_title}.")

# Create full_prompt by formatting partial_prompt with movie_title='Inception'
full_prompt = partial_prompt.format(movie_title='Inception')

Hint

Call partial_prompt.format() with the argument movie_title='Inception'.

4

Use the full prompt with LangChain's LLMChain

Import LLMChain and OpenAI from langchain. Create an llm instance using OpenAI(). Then create an llm_chain using LLMChain with the llm and partial_prompt. Finally, call llm_chain.run with movie_title='Inception'.

LangChain

from langchain.prompts import PromptTemplate
from langchain import LLMChain, OpenAI

partial_prompt = PromptTemplate(template="You are a helpful assistant who provides detailed answers about movies. Your style is friendly and informative. Tell me about the movie {movie_title}.")

full_prompt = partial_prompt.format(movie_title='Inception')

# Create llm, llm_chain and run llm_chain with movie_title='Inception'
llm = OpenAI()
llm_chain = LLMChain(llm=llm, prompt=partial_prompt)
llm_chain.run(movie_title='Inception')

Hint

Import LLMChain and OpenAI, create instances, then run the chain with the variable.

Practice

1. What is the main purpose of using PartialPromptTemplate in Langchain?

easy

A. To create reusable parts of prompts that can be filled later

B. To execute a prompt directly without variables

C. To store the final output of a prompt

D. To connect multiple language models together

Solution

1. Step 1: Understand the role of PartialPromptTemplate

   PartialPromptTemplate is designed to hold parts of a prompt with placeholders for variables.

2. Step 2: Recognize its use for reusability

   This allows you to reuse prompt pieces and fill variables later to form a complete prompt.

3. Final Answer:

   To create reusable parts of prompts that can be filled later -> Option A

4. Quick Check:

   PartialPromptTemplate = reusable prompt parts [OK]

Hint: Think reusable prompt pieces filled later [OK]

Common Mistakes:

• Confusing it with final prompt execution
• Thinking it stores output instead of template
• Assuming it connects models directly

2. Which of the following is the correct way to create a PartialPromptTemplate with a variable named name?

easy

A. PartialPromptTemplate(template="Hello {name}")

B. PartialPromptTemplate(template="Hello {name}", variables=["name"])

C. PartialPromptTemplate(template="Hello {name}", inputs=["name"])

D. PartialPromptTemplate(template="Hello {name}", input_variables=["name"])

Solution

1. Step 1: Check the required parameters for PartialPromptTemplate

   It requires a template string and a list named input_variables specifying variable names.

2. Step 2: Match the correct syntax

   PartialPromptTemplate(template="Hello {name}", input_variables=["name"]) correctly uses input_variables=["name"] to declare the variable.

3. Final Answer:

   PartialPromptTemplate(template="Hello {name}", input_variables=["name"]) -> Option D

4. Quick Check:

   Use input_variables list to declare variables [OK]

Hint: Remember input_variables param holds variable names [OK]

Common Mistakes:

• Using wrong parameter names like variables or inputs
• Omitting input_variables list
• Not matching variable names in template and list

3. Given the following code, what will be the output of full_prompt.format(name="Alice")?

from langchain.prompts import PartialPromptTemplate, PromptTemplate
partial = PartialPromptTemplate(template="Hello {name}", input_variables=["name"])
full_prompt = PromptTemplate(template="{greeting}, welcome!", input_variables=["greeting"])
full_prompt = full_prompt.partial(greeting=partial)

medium

A. "{greeting}, welcome!"

B. "Hello Alice, welcome!"

C. "Hello {name}, welcome!"

D. Error: missing variable 'name'

Solution

1. Step 1: Understand partial prompt substitution

   The partial prompt replaces the greeting variable in full_prompt with the partial template.

2. Step 2: Format the full prompt with name="Alice"

   Calling full_prompt.format(name="Alice") fills {name} in partial, producing "Hello Alice", then inserts it into full prompt, resulting in "Hello Alice, welcome!".

3. Final Answer:

   "Hello Alice, welcome!" -> Option B

4. Quick Check:

   Partial fills greeting, then full prompt formats [OK]

Hint: Partial fills variables inside main prompt [OK]

Common Mistakes:

• Expecting raw template string without substitution
• Confusing variable names and placeholders
• Missing that partial is nested inside full prompt

4. What is the error in the following code snippet?

partial = PartialPromptTemplate(template="Hi {user}", input_variables=["name"])

medium

A. Variable name in template and input_variables do not match

B. Missing import statement for PartialPromptTemplate

C. Template string must not contain variables

D. input_variables should be a string, not a list

Solution

1. Step 1: Compare template variables and input_variables list

   The template uses {user} but input_variables list contains "name".

2. Step 2: Identify mismatch causes error

   Variables must match exactly; mismatch causes runtime error when formatting.

3. Final Answer:

   Variable name in template and input_variables do not match -> Option A

4. Quick Check:

   Variable names must match in template and input_variables [OK]

Hint: Check variable names match exactly in template and list [OK]

Common Mistakes:

• Assuming variable names can differ
• Ignoring case sensitivity
• Thinking input_variables can be a string

5. You want to build a prompt that greets a user and mentions their favorite color using partial prompt templates. Which approach correctly combines two partial templates greet and color into a full prompt?

hard

A. Create two PartialPromptTemplates but combine by concatenating their templates as strings manually

B. Create one PartialPromptTemplate with all variables: PartialPromptTemplate(template="Hello {name}. Your favorite color is {color}.", input_variables=["name", "color"])

C. Create greet = PartialPromptTemplate(template="Hello {name}", input_variables=["name"]) and color = PartialPromptTemplate(template="Your favorite color is {color}", input_variables=["color"]), then combine with full = PromptTemplate(template="{greeting}. {color_info}.", input_variables=["greeting", "color_info"]) and use full.partial(greeting=greet, color_info=color)

D. Use PromptTemplate only with variables name and color without partial templates

Solution

1. Step 1: Define two partial templates for greeting and color

   Each partial holds a reusable piece with its own variables.

2. Step 2: Combine partials into a full prompt using placeholders

   The full prompt uses placeholders for each partial, then partial() method fills them with partial templates.

3. Step 3: This approach keeps prompts modular and variables scoped

   It allows filling variables later and keeps code organized.

4. Final Answer:

   Create greet = PartialPromptTemplate(template="Hello {name}", input_variables=["name"]) and color = PartialPromptTemplate(template="Your favorite color is {color}", input_variables=["color"]), then combine with full = PromptTemplate(template="{greeting}. {color_info}.", input_variables=["greeting", "color_info"]) and use full.partial(greeting=greet, color_info=color) -> Option C

5. Quick Check:

   Combine partials via placeholders and partial() method [OK]

Hint: Use partial() to nest partial templates inside full prompt [OK]

Common Mistakes:

• Trying to concatenate templates as strings manually
• Using one partial for all variables losing modularity
• Ignoring partial() method for combining templates