Few-Shot Learning
Few-shot learning, unpacked: how a large language model can generalize from just a handful of worked examples in the prompt — no retraining needed — where this ability comes from, when it shines, and when plain rules work better.
Few-shot learning lets a model master a new task from just a few examples at inference time, with no parameter changes: two or three input-output pairs in the prompt, and it induces the underlying pattern and applies it to fresh questions. This topic explains why examples can teach a model, how their count and quality shape results, the boundary with zero-shot prompting and fine-tuning, and the classic few-shot Chain-of-Thought style.
