The Few Shot Prompt Template node enables the creation of prompts with multiple examples, allowing Language Models to learn from demonstrations before processing the main input.
Name: fewShotPromptTemplate
Type: FewShotPromptTemplate
Version: 1.0
Category: Prompts
FewShotPromptTemplate
BasePromptTemplate
Examples (Required)
Type: string (JSON)
Description: Array of example objects with key-value pairs
Example:
Example Prompt (Required)
Type: PromptTemplate
Description: Template for formatting individual examples
Prefix (Optional)
Type: string
Description: Text appearing before the examples
Example: “Give the antonym of every input”
Suffix (Required)
Type: string
Description: Text after examples, containing input variable
Example: “Word:
\nAntonym:”
Example Separator (Optional)
Type: string
Default: “\n\n”
Description: String used to separate examples
Template Format (Optional)
Type: options
Options: [“f-string”, “jinja-2”]
Default: “f-string”
Description: Format style for template strings
Example Processing
JSON parsing
Example formatting
Template application
Separator handling
Template Management
Format selection
Variable extraction
Context building
Structure validation
Pattern Teaching
Response formatting
Style demonstration
Behavior guidance
Context setting
Complex Tasks
Multi-step processes
Format standardization
Response structuring
Pattern recognition
Uses @langchain/core/prompts
Supports multiple example formats
Handles variable extraction
Manages template formatting
Example Design
Clear demonstrations
Consistent formatting
Diverse examples
Progressive complexity
Template Structure
Compatible formats
Clear separators
Proper variables
Consistent style
Performance Optimization
Efficient examples
Minimal redundancy
Clear instructions
Optimal spacing