Prompting & Inference

Zero-Shot Learning

Asking a model to do a task with no examples, relying only on instructions and what it already knows.

Zero-shot means the model tackles a task it was given no worked examples for, working purely from the instruction and its general training. Modern large language models are surprisingly good at this, which is why you can often just ask plainly and get a usable answer. When zero-shot falls short, adding a few examples, called few-shot, usually helps.

Reference: Wikipedia: Zero-Shot Learning