A city taxi-medallion desk needs stable automation locators from an LLM, not brainstormed edge-case names. Which temperature choice best fits that reproducibility-first task?
Select an answer to reveal the explanation.
Short Explanation
Think of temperature as matching the job, not a universal “always zero” rule. When you need the same locator style run after run, dial it down. High temperature is the poorer fit when variety matters less than reproducibility.
Full Explanation
City taxi-medallion automation needs stable locators, not creative synonyms each run. The temperature choice that best fits a reproducibility-first task is a low temperature so generated scripts stay more consistent across runs. Match the dial to the job: locator stability over brainstorming.
Raising temperature so every locator draft invents fresh synonyms fails by concept for this goal—variety fights reproducibility. Ignoring temperature and relying only on longer prompts fails because prompt length does not replace sampling control when consistency is the requirement. Setting temperature to maximum fails for the same reason: maximum exploration is the wrong fit when identical locator style matters.
Exam caveat: low temperature is preferred for reproducibility-first generation; it is not a universal “always zero” rule for every municipal GenAI task.
Operational check: define temperature profiles by task class (locators/scripts vs. idea generation), default medallion automation jobs to low temperature, sample multiple runs for locator drift, and reject packs that rename controls without basis justification. Medallion desks should version-control the temperature profile with the automation pack so audits can show why locator wording stayed stable across the week’s regenerations.