A summarization app needs MORE creative, varied outputs from the same model. Which inference parameter should be increased?
A.Temperature
Correct. Higher temperature = more creative sampling.
B.Top-K (lowering it)
Lowering Top-K restricts choices — less variety (raising it helps variety).
C.Stop sequences
Stop sequences end generation — unrelated to creativity.
D.Max tokens
Max tokens controls length, not randomness.
Why this answer
Temperature controls randomness: higher = more varied/creative sampling, lower = more deterministic. Raising Top-K widens the candidate pool (also more variety); lowering it, increasing stop sequences, or raising max tokens don't add creativity.
Correct answer: A · Last reviewed 2026-10-28
Official AWS references
Practice this interactively
Answer without seeing the solution first in the Generative AI & Bedrock practice session.
Practice Generative AI & Bedrock