A summarizer returns different wording each time and you want more consistent output. Which parameter do you lower first?
Why D is the answer
Temperature
Temperature controls randomness in sampling. Max tokens limits length, and quota is about capacity, so neither affects consistency.
In detail
- Temperature controls how varied the output is.
- A lower value gives more consistent wording.
- Keep a higher temperature for creative writing.
Use case
An illustrative scenario, not a real incident.
A summarizer gives different wording each run, and reviewers object. Lowering temperature makes the output steadier, without changing the model.
See it in different ways
mindmap
root((Settings))
Temperature
Lower is more consistent
Higher is more varied
Max tokens
Limits reply length
System prompt
Role and rulesflowchart TD
Q{"What do you want?"}
Q -->|"Consistency"| A0["Lower temperature"]
Q -->|"Variety"| A1["Higher temperature"]Key point: Lower temperature, steadier output.
Why the other options are wrong
- A. Quota
- Quota affects how much you can use.
- B. Max tokens
- Max tokens limits length, not consistency.
- C. Number of deployments
- Deployment count affects capacity.