1.3Standard
Repeated training produces different scores.
Make selection reproducible.
Choose two practices.
× Incorrect
Record seeds, splits, environments, code, and settings, and evaluate several seeds for mean, variation, and safety rather than one lucky result.
Detailed explanation
CorrectEvaluate several seeds for average, variation, and safety
A single lucky result is avoided.
CorrectRecord random seed, data split, environment, code, and model settings
Differences can be reproduced and compared.
IncorrectChoose from the highest single score
Random variation is ignored.
IncorrectChange splits every time without saving settings
Comparisons are impossible.
Check official documentation
Compare the following points with the official documentation.
AWS公式AIF-C01 Domain 1.3の実験管理、再現性、モデル評価を確認する。
Official references
What to confirm
確率的な学習結果を再現可能な条件で比較し、ばらつきも判断できる。
Key points
- Seed
- Variation
- Reproducibility
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
Random seed
A value that controls pseudo-random operations so an experiment can be repeated.
Experimental variability
The spread in results caused by different seeds, splits, or stochastic processes.
Question IDAWS-AIF-394
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