Exams / AIF-C01 / Choosing FMs and inference parameters / Question 2
AIF-C01 practice question: Choosing FMs and inference parameters (2 of 4)
Applications of Foundation Models A free question from our original AIF-C01 practice exam. Try it, then open the answer to see why each option is right or wrong.
A developer sets Top P to 0.8 on a text model in Amazon Bedrock. What effect does this have when the model chooses each next token?
Choose one answer.
- AA. It caps the response at 80 per cent of the maximum allowed length
- BB. It considers only the 80 most likely candidate tokens at each step
- CC. It stops generating once the model is 80 per cent confident
- DD. It samples from the tokens making up the top 80% of probability
Show the answer and explanation
Answer: D
- A. Wrong: Response length is set by a separate parameter. Top P does not limit the number of tokens returned.
- B. Wrong: That describes Top K, which sets a fixed number of candidates. Top P sets a share of the probability distribution.
- C. Wrong: Generation ends at a stop sequence, the length limit or the natural end. Top P has no confidence cut-off.
- D. Right: Correct. Top P limits choices to the smallest set of likely tokens whose combined probability reaches the chosen share.
Explanation
Top P (nucleus sampling) restricts the next-token choice to the most likely tokens that together make up the chosen share of probability. Lower values make output more focused; higher values allow less likely tokens. Top K instead sets a fixed number of candidates, and temperature reshapes the whole distribution.
Study
Keep practising
Answer questions with a score at the end, or practise just this topic with the explanation after each one. No sign-up needed.