Home / AWS Certified AI Practitioner (AIF-C01) / 2 Practice / AWS-AIF-241
Make FAQ responses stable but brainstorm outputs diverse.
Understand temperature and top-p.
Choose two correct statements.
Hint 1: Focus on the requirement Change generation controls one at a time, then compare the same quality, safety, format, and cost measures before choosing a setting.
Hint 2: Key terms Review generation sampling parameters.
TemperatureA parameter that changes how broadly a model samples the next token; higher values tend to add variation.
Top-pA sampling rule that limits candidates to a smallest set whose cumulative probability reaches a chosen value.
DeterminismThe property that the same input and settings produce the same or nearly the same output.
Evaluation setA fixed collection of inputs used to compare model settings under consistent conditions. × Incorrect Temperature and top-p affect sampling diversity; settings should be fixed per use case and compared on the same quality, safety, format, and cost set.
Detailed explanation A Incorrect Change multiple parameters without recording them
The cause of a difference is unclear.
B Incorrect Higher temperature always improves factuality
More variation can also increase errors.
C Correct Increasing temperature can broaden choices and increase variation
It tunes a creativity and consistency trade-off.
D Correct Compare factuality, safety, format, and cost on the same evaluation set after changes
Parameter impact can be measured.
Check official documentation Compare the following points with the official documentation.
Amazon Bedrock公式推論パラメータとAIF-C01 Domain 2.3を確認する。
Official references Official documentation 1 Official documentation 2 What to confirm 生成パラメータを品質保証ではなく、評価対象の設定として扱える。
Key points Temperature Top-p 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 Temperature A parameter that changes how broadly a model samples the next token; higher values tend to add variation.
Top-p A sampling rule that limits candidates to a smallest set whose cumulative probability reaches a chosen value.
Path & Pass is an independent learning site and is not affiliated with any certification body. Questions are original practice content based on published objectives.
