AI for QA / Quiz

Temperature, Sampling, and Repeatability

The dial that trades consistency for variety, and what it means for anyone who asserts on model output.

Difficulty
Medium
Format
Quiz
Points
150
Estimate
8 min

// MISSION BRIEF

Your Mission

Temperature and top_p decide whether the model plays it safe or takes chances, and testers sit on both sides of that dial: wanting diverse test data one hour, reproducible pipeline output the next. This quiz covers how sampling actually works and the assertions you can (and cannot) write against it.

// FIRST CONTACT

Battle teaser

What does the temperature setting actually do?

  1. ASets the emotional tone of the response
  2. BLimits the maximum output length
  3. CControls how long the model thinks before answering
  4. DReshapes the token probability distribution: low values concentrate on the most likely tokens, high values flatten it so less likely tokens get picked more often
Answers, scoring, hints, and the full battle stay sealed.

// SKILL TAGS

llm-fundamentalstemperaturesamplingreproducibility