Few-Shot and Zero-Shot

Few-Shot and Zero-Shot Prompting

Zero-Shot: The Direct Request

Zero-Shot Prompting means asking the model to perform a task directly without any prior examples.

You trust that the model understands the task from its description alone.

Example: "Classify this review as Positive, Negative, or Neutral: 'The product was fine but delivery was late'"

This is zero-shot because you gave no classification examples.

Few-Shot: Teaching by Example

Few-Shot Prompting means providing several examples of the desired input/output format before the actual task.

The examples teach the model the "template" you want.

Example:

Classify the following reviews:

Review: "Excellent quality, fair price"
Classification: Positive

Review: "Order never arrived, terrible service"
Classification: Negative

Review: "Product is average, no strong feelings either way"
Classification: Neutral

Now classify:
Review: "The product was fine but delivery was late"
Classification:

When to Use Each?

| Situation | Best choice | Reason | |-----------|-------------|--------| | Common, clear task | Zero-Shot | Model understands without examples | | You want a specific style/format | Few-Shot | Examples teach the pattern | | Unexpected or wrong output | Few-Shot | Examples correct the direction | | Task unique to your project | Few-Shot | Model may not have learned this pattern |

Number of Examples: How Many is Enough?

• 1-2 examples: One-shot — useful for style adjustment
• 3-5 examples: Few-shot — good balance of cost and accuracy
• 10+ examples: Many-shot — useful for very complex tasks

More isn't always better — quality of examples matters more than quantity.

Pro Tip

Choose examples that cover edge cases, not just obvious ones. If your task is "sentiment classification," include a sarcastic or ambiguous example so the model knows how to handle it.