Extract an invoice number with a stable format.
The same input should produce nearly the same output format each time.
Which direction for temperature is generally appropriate?
A lower temperature can reduce variation for a deterministic extraction task, although the application must still validate the output.
Detailed explanation
Lower randomness is generally preferable for a stable, structured extraction result.
Higher randomness can increase variation and format deviations.
Temperature is an inference-time setting and does not update model parameters.
Temperature affects generation, not authentication.
Check official documentation
Compare the following points with the official documentation.
対象モデルのAmazon Bedrock推論パラメータ仕様を確認し、代表入力で再現性を測定する。
Official references
What to confirm
温度が推論時のランダム性に関する設定で、訓練や認証とは別だと説明できる。
Key points
- Inference setting
- Randomness
- Structured extraction
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
Variability trade-off
More sampling variation can help creative generation but makes exact repetition less likely.
Evaluation repeatability
Record model, prompt, parameters, and seed or service conditions when comparing outputs.