easyMultiple Choice
AIF-C01 Practice Question: What does the temperature parameter control in a…
What does the temperature parameter control in a text generation model?
⚠ Common exam trap
The trap is conflating temperature with other sampling controls — candidates who don't distinguish temperature (probability sharpening) from top-k/top-p (candidate restriction) or max_tokens (length limit) pick A or D.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
The degree of randomness in the generated output
Temperature is a scaling factor applied to the logits before the softmax function, controlling the sharpness of the resulting probability distribution. A low temperature (e.g., 0.1) sharpens the distribution, making the highest-probability token overwhelmingly likely and producing deterministic, focused output. A high temperature (e.g., 1.5) flattens the distribution, giving lower-probability tokens more chance of being sampled, producing more diverse and creative — but potentially incoherent — output.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
The number of candidate tokens considered at each step
Why it's wrong here
Temperature scales the logits before the softmax, reshaping the probability distribution; it does not limit how many candidate tokens are evaluated. Top-k or nucleus sampling controls candidate count. Temperature is tempting here because both parameters shape sampling randomness, but they act on different mechanisms.
- ✓
The degree of randomness in the generated output
Why this is correct
Temperature scales the probability distribution over the model's vocabulary before sampling, so higher values flatten it and increase randomness while lower values sharpen it toward the most likely tokens. This directly governs output variability, satisfying the question's focus on how stochastic versus deterministic the generated text becomes.
- ✗
The similarity to the training data distribution
Why it's wrong here
Temperature rescales logits to flatten or sharpen the output probability distribution; it does not measure or control similarity to the training data. That similarity reflects pretraining and fine-tuning. Temperature is tempting because low values yield predictable, training-like text, but the mechanism is distribution shaping, not data matching.
- ✗
The maximum number of tokens to generate
Why it's wrong here
Temperature adjusts the sharpness of the token probability distribution; it never sets an output length. The max tokens parameter caps generation length. Temperature is tempting because both influence generated output, but length is governed by a separate decoding limit, not by logit scaling.
About these practice questions
Courseiva writes every AIF-C01 question from scratch — 862 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Amazon Web Services exam blueprint
This AIF-C01 practice question is part of Courseiva's free Amazon Web Services certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AIF-C01 exam.