Add examples to a classification prompt.
Avoid instability from inconsistent or biased examples.
Choose two example practices.
Use consistent formats, labels, and representative boundary cases, and review source, privacy, quality, and ordering effects.
Detailed explanation
Negative and boundary behavior is under-specified.
Prompts and logs can expose them.
Example quality and leakage risk are governed.
The intended rule and format are clearer.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01 Domain 2.2のプロンプト例示、入力管理、限界を確認する。
Official references
What to confirm
例示の内容と並び順が出力品質・安全性へ与える影響を評価できる。
Key points
- Few-shot
- Boundary case
- Label quality
Notes
- Environment: AWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
- Compare the definition with the AIF-C01 objective and the service documentation before applying it.
- Keep the concept separate from a particular answer choice or implementation step.
Foundation review
Few-shot prompting
Showing a small number of input and output examples to communicate a task pattern.
Boundary case
An example near a decision boundary where similar inputs may receive different labels.