2.3Standard
Generation diversity and stability must be adjusted.
Evaluate sampling.
Choose two descriptions.
× Incorrect
Top-k limits the number of candidate tokens, while top-p limits cumulative probability; both should be tested with temperature and quality checks.
Detailed explanation
CorrectTop-k limits selection to the k highest-probability candidates
Candidate count is direct.
CorrectTop-p selects from candidates reaching cumulative probability p and should be evaluated with temperature
The set adapts to probability.
IncorrectChanging top-k or top-p updates training data
They are generation settings.
IncorrectEither setting guarantees factual and safe output
Separate evaluation is needed.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01 Domain 2.3のtop-k、top-p、temperature、サンプリングを確認する。
Official references
What to confirm
サンプリング設定の役割と、出力の事実性を別途評価する必要性を説明できる。
Key points
- Top-k
- Top-p
- Sampling
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
Top-k
A sampling setting that limits candidates to the k highest-probability tokens.
Top-p
A sampling setting that selects a variable candidate set within a cumulative probability mass.
Question IDAWS-AIF-482
EnvironmentAWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
Last technical review2026-08-20
Report an error or rights issue