What does top-p control?
Why D is the answer
How wide a range of likely tokens the model picks from
Top-p, or nucleus sampling, is a sampling setting.
In detail
- Top-p sets how wide a range of likely tokens the model picks from.
- Temperature is a related setting.
Use case
An illustrative scenario, not a real incident.
A low top-p keeps a model to its most likely words, while a high value allows more variety.
See it in different ways
mindmap
root((Prompting))
Examples
Zero-shot
One-shot
Few-shot
Persona
Role prompting
Steps
Prompt chaining
Tools
ReActflowchart TD
Q{"What do you want?"}
Q -->|"More focused output"| A0["Lower top-p"]
Q -->|"More varied output"| A1["Higher top-p"]Key point: Top-p limits the range of likely tokens.
Why the other options are wrong
- A. How many users can call the model at once
- That is capacity.
- B. Which region stores your data
- That is data location.
- C. How many documents are in the index
- That is about search.