Your prompt is sent, but the reply stops mid-sentence. What setting should you check?
Why A is the answer
The maximum number of output tokens
Max tokens limits the length of the response. A low value cuts it off.
In detail
- A reply that stops mid-sentence often hit the output token limit.
- Very long replies also cost more.
Use case
An illustrative scenario, not a real incident.
A reply stops mid-sentence. The maximum output tokens is too low, so raising it lets the reply finish.
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 LR S0["Reply stops early"] --> S1["Check max tokens"] --> S2["Raise it"] --> S3["Reply completes"]
Key point: Cut-off replies often mean a low token limit.
Why the other options are wrong
- B. The name of the Foundry project
- A project name has no effect.
- C. The temperature
- Temperature changes variety, not length.
- D. The top-p value
- Top-p changes sampling, not length.