In-context learning (zero-shot and few-shot)

In-context learning is a model's ability to pick up a task from the prompt itself, without any change to its internal settings. You demonstrate or describe what you want, and the model adapts on the spot.

Two related terms appear constantly. "Zero-shot" means you give the model only an instruction, with no examples ("Translate this to French."). "Few-shot" means you include a handful of worked examples first, and the model imitates the pattern. Chain-of-thought prompting (Chapter 1) comes in both flavors: few-shot, where you show example problems with their reasoning, and zero-shot, where you simply add a phrase like "let's think step by step."