Intermediate · Fundamentals

Few-Shot Learning

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. Training a model to recognize new patterns from just a handful of labeled examples.

Technical Definition

Training a model to recognize new patterns from just a handful of labeled examples.

How it works

Few-shot learning enables learning from 1-10 examples per class. Meta-learning approaches like MAML train models to learn quickly. Prototypical Networks use nearest-neighbor in embedding space. In NLP, few-shot prompting provides examples directly in the prompt.

Related Concepts

• Transfer Learning — Leveraging knowledge from a model trained on one task to improve performance on a different but related task.
• Prompt Engineering — The art of crafting effective input instructions to guide LLM behavior without changing model weights.
• Zero-Shot Learning — The ability of a model to perform tasks or classify categories it has never explicitly been trained on.