First, request sentiment labels without examples.
Next, provide a few input-and-label examples in the prompt.
What is the second method called?
Providing a small number of examples in the prompt is few-shot prompting; instructions without examples are zero-shot.
Detailed explanation
A few input/output examples show the desired format or decision pattern.
Zero-shot prompting provides instructions without task examples.
Encryption protects data and is unrelated to example-based prompting.
Pre-training is broad model training, not placing examples in a prompt.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01ガイドのDomain 3.1でプロンプトによるモデル適応の用語を確認する。
Official references
What to confirm
例の有無でゼロショットと少数ショットを区別できる。
Key points
- No examples
- A few examples
- Prompt adaptation
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
Example quality
Few-shot examples should be accurate, representative, and safe because they influence the model's interpretation.
Context budget
Examples consume tokens and can reduce the room available for the actual request and response.