What does a higher temperature do to a foundation model's output?
Why C is the answer
Makes it more varied
Temperature affects sampling randomness. Length and price are controlled elsewhere.
In detail
- A higher temperature makes output more varied.
- Temperature changes randomness, not length or price.
Use case
An illustrative scenario, not a real incident.
A marketing team raises temperature for varied slogans, and lowers it for a product specification.
See it in different ways
mindmap
root((Generative AI))
Token
Unit of text
Drives cost
Embedding
Meaning as numbers
Chunking
Focused passages
Hallucination
Confident and false
Temperature
Varietymindmap
root(("Temperature"))
n1["Lower"]
n2["Predictable"]
n3["Factual tasks"]
n4["Higher"]
n5["Varied"]
n6["Creative tasks"]Key point: Higher temperature gives more varied output.
Why the other options are wrong
- A. Makes it cheaper per token
- It does not change the token price.
- B. Makes it longer on average
- Length is set by other limits.
- D. Makes it faster to generate
- It does not change speed.