One-shot Prompt

Definition

A prompting technique that provides the model with a single example of the task before asking it to perform a similar task. This approach gives the model context about the expected input-output format and style.

Characteristics

• Contains exactly one example
• Shows input-output relationship
• Demonstrates desired format and style
• Bridges zero-shot and few-shot approaches

Example

Prompt:

Translate the following English sentence to French:
English: "Hello, how are you?"
French: "Bonjour, comment allez-vous?"

Now translate this sentence:
English: "Good morning."
French:

Expected Response: "Bonjour."

Use Cases

• Translation tasks
• Format transformation
• Style demonstration
• Classification with specific output format
• Template-based responses

Benefits

• Provides clear example of expected output
• Better performance than zero-shot for many tasks
• Efficient use of context (minimal examples)
• Good balance between guidance and brevity
• Helps establish output format and style

Limitations

• Single example may not cover edge cases
• May lead to overfitting to the specific example
• Less robust than few-shot for complex tasks
• Example quality critically impacts performance
• May not represent full task complexity

Best Practices

• Choose representative, high-quality examples
• Ensure example demonstrates desired output format
• Select examples that cover common use cases
• Avoid overly complex or edge-case examples
• Make the pattern clear and generalizable

When to Use

• When you have limited context space
• For straightforward tasks with clear patterns
• When one good example sufficiently demonstrates the task
• As a starting point before scaling to few-shot