Skip to main content

Zero-Shot & Few-Shot Prompting

Zero-shot and few-shot prompting are the foundational techniques for controlling Large Language Models (LLMs).

Zero-Shot Prompting​

Zero-shot prompting means asking an LLM to perform a task without providing any examples. This relies entirely on the model's pre-existing knowledge.

When to use:

  • Simple translations
  • Basic summarization
  • General knowledge queries

Few-Shot Prompting​

Few-shot prompting involves providing the model with a few examples (shots) of the expected input-output behavior. This "conditions" the model to follow a specific pattern or style.

Best Practices:​

  1. Diverse Examples: Cover different scenarios.
  2. Consistent Formatting: Use the exact same structure for examples as you do for the target query.
  3. Quality over Quantity: 3-5 high-quality examples are usually better than 10 mediocre ones.