Question 29
Domain 3: Implement Generative AI SolutionsYou are configuring an Azure OpenAI deployment to generate product descriptions. The requirements are: - Responses must be creative and varied - Responses must not repeat product features already mentioned - Maximum 200 words per description Which parameter combination is most appropriate?
Correct answer: B
Explanation
A high "temperature" such as 0.9 increases creativity and variation in generated text. A "frequency_penalty" of 0.8 reduces repetition of already used terms, matching the requirement that responses must not repeat product features. Setting "max_tokens=200" limits the description length to the required maximum.
Why each option is right or wrong
A. temperature=0, max_tokens=200, frequency_penalty=0
B. temperature=0.9, max_tokens=200, frequency_penalty=0.8
Azure OpenAI generation controls map directly to the stated constraints: the `temperature` setting governs randomness, and a value of `0.9` is near the upper end of the typical `0.0–1.0` range, which is appropriate when the output must be creative and varied. The `frequency_penalty` parameter is designed to discourage repeated tokens/phrases, and `0.8` is a strong penalty that helps avoid restating product features already mentioned, while `max_tokens=200` caps the completion at 200 tokens, aligning with the maximum-length requirement.
C. temperature=2.0, max_tokens=200, presence_penalty=-1.0
D. top_p=0.1, max_tokens=200, frequency_penalty=0