Decoder. plain-English AI glossary

Exemplar selection

▲ Rising

Choosing which examples to include in a few-shot prompt — diversity, relevance, and order all matter.

Think of it like

Picking which photos to show in a portfolio; you choose diverse, impressive shots, not random ones.

Example

For a task, select examples that are similar to the test input (relevant) but diverse in edge cases (coverage).

How it actually works

Naive few-shot uses random examples; smarter selection uses similarity to the test input (semantic matching) or active learning principles (uncertain cases). Optimal strategies are task-dependent. Research shows careful selection can boost performance 5–15% over random.

For product teams

Squeeze more value from few-shot prompts with minimal cost.

For engineers

Compute embeddings of examples and test input. Select k-nearest neighbors or maximize diversity. Experiment with strategies.

Few-shot examples — Improves few-shot prompting.

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open Decoder Add to Chrome