Few-Shot Learning
Teaching a model to perform a task by providing a small number of examples (usually 2–10) in the prompt.
Few-shot learning is teaching without retraining. Instead of fine-tuning a model on hundreds of examples, you provide a handful of examples in the prompt and ask the model to learn the pattern on the fly.
For example: - One-shot: "Classify as positive or negative: "I love this product!" Positive. Classify: "This is awful."" - Few-shot: Add 2–5 more labeled examples before your unlabeled input.
Few-shot is powerful for quick prototyping but less stable than proper fine-tuning. Performance improves with better examples and clearer instructions. Modern models are better at few-shot learning because of scale and instruction-following improvements.
Few-shot is the default approach for most AI tool users because it's free (no fine-tuning setup) and fast.
Example
Few-shot sentiment classification: Show 3 examples of positive reviews and 3 negative reviews, then ask the model to classify your unlabeled review.
Related terms
Zero-Shot Learning
Asking a model to perform a task without providing examples—just an instruction.
Fine-Tuning
The process of updating a pre-trained model with task-specific or domain-specific data to improve performance.
Prompt
The text instruction you send to an AI model asking it to do something.