Question 240
A model answer is truncated even at low temperature because the API has a small `max_output_tokens` value. Which parameter should be adjusted?
Explanation
Correct Answer
The completion length cap directly limits emitted tokens, unlike sampling parameters.
Incorrect Answers
Retrieval breadth does not enlarge the response budget.
Sampling variability cannot extend a hard output-token allowance.
Removing references does not eliminate the explicit token cap.