Core Concepts
Few-Shot
Providing a small number of input-output examples in the prompt to guide the model.
Few-shot prompting includes 2–10 worked examples before the actual task, helping the model infer the expected format and reasoning pattern. This consumes more input tokens but often dramatically improves output quality and consistency, especially for structured tasks like JSON extraction or classification with custom labels.
Related Terms
Zero-Shot
Prompting a model to perform a task without providing any examples.
Prompt Engineering
The practice of designing inputs to elicit optimal outputs from language models.
Token
The basic unit of text that language models process and are billed by.
Chain of Thought
A prompting technique that instructs the model to reason step-by-step before answering.