Variables and dynamic content in LangChain - Step-by-Step Execution

Concept Flow - Variables and dynamic content

Define Variables

↓

Insert Variables into Template

↓

Generate Dynamic Content

↓

Use Output in Chain or App

↓

Update Variables if Needed

↩Back to Insert Variables

This flow shows how variables are defined, inserted into templates, and used to generate dynamic content in LangChain.

Execution Sample

LangChain

from langchain import PromptTemplate

template = "Hello, {name}!"
prompt = PromptTemplate(template=template, input_variables=["name"])
output = prompt.format(name="Alice")

This code defines a template with a variable and generates a greeting by filling in the variable.

Execution Table

Step	Action	Variable State	Output
1	Define template with variable {name}	template='Hello, {name}!'
2	Create PromptTemplate with input_variables=['name']	input_variables=['name']
3	Call prompt.format(name='Alice')	name='Alice'	Hello, Alice!
4	Output generated	name='Alice'	Hello, Alice!

💡 Variables inserted and dynamic content generated successfully.

Variable Tracker

Variable	Start	After Step 2	After Step 3	Final
template	undefined	Hello, {name}!	Hello, {name}!	Hello, {name}!
input_variables	undefined	['name']	['name']	['name']
name	undefined	undefined	Alice	Alice
output	undefined	undefined	Hello, Alice!	Hello, Alice!

Key Moments - 2 Insights

Why do we need to specify input_variables when creating a PromptTemplate?

Specifying input_variables tells LangChain which parts of the template are dynamic placeholders to replace. See execution_table step 2 where input_variables=['name'] enables substitution.

What happens if we call prompt.format() without providing a variable value?

LangChain will raise an error because the template expects a value for 'name'. This is shown in execution_table step 3 where 'name' must be provided to generate output.

Visual Quiz - 3 Questions

Look at the execution_table at step 3, what is the value of 'output'?

A Hello, {name}!

B Hello, Alice!

C name='Alice'

D undefined

💡 Hint

Check the 'Output' column in execution_table row with Step 3.

At which step is the variable 'name' assigned the value 'Alice'?

A Step 3

B Step 2

C Step 1

D Step 4

💡 Hint

Look at the 'Variable State' column in execution_table to see when 'name' changes.

If we change the template to "Hi, {user}!", what else must we update to generate output?

A No changes needed, just call format(name='Alice')

B Remove input_variables list

C Change input_variables to ['user'] and provide user='value' in format()

D Change output variable name

💡 Hint

Refer to key_moments about input_variables and variable substitution.

Concept Snapshot

Variables and dynamic content in LangChain:
- Define a template string with placeholders like {name}
- Specify input_variables list matching placeholders
- Use PromptTemplate.format() with variable values
- Output is the template with variables replaced
- Missing variables cause errors
- Update variables and template together for dynamic content

Full Transcript

In LangChain, variables let you create templates with placeholders that you fill dynamically. First, you define a template string with placeholders inside curly braces, like {name}. Then, you tell LangChain which variables to expect by listing them in input_variables. When you call the format method on the PromptTemplate, you provide actual values for these variables. LangChain replaces the placeholders with the values and generates the final output string. If you forget to provide a variable, LangChain will raise an error. This process allows you to create flexible prompts and content that change based on input values.