AI-102 Question 478
Select 3You are designing an Azure OpenAI solution to generate text responses for a customer support chatbot. The business requires the responses to be concise, avoid unnecessary repetition, and align closely with the input context. Which parameters should you configure in the Azure OpenAI service to achieve this behavior?
- A
Temperature
- B
Top-p
- C
Frequency penalty
- D
Presence penalty
- E
Max tokens
Show answer and explanation
Correct answers: A, C, D
Explanation
To configure the Azure OpenAI service for generating concise and context-aligned responses without unnecessary repetition, you should focus on parameters that directly influence the behavior of the model. Temperature ensures deterministic outputs, frequency penalty avoids repetition, and presence penalty promotes variation. Other parameters like Top-p and Max tokens are less relevant to the specific requirements in this scenario.
- A. Correct.
Temperature controls the randomness of the output. Lowering this value makes responses more deterministic, which aligns with the need for concise and context-aligned responses.
- B. Incorrect.
Top-p (nucleus sampling) determines the diversity of the output by controlling the probability distribution. While useful for creative tasks, it is less critical when aligning responses closely to the input context.
- C. Correct.
Frequency penalty reduces the model's tendency to repeat words or phrases, making it essential for avoiding unnecessary repetition in responses.
- D. Correct.
Presence penalty encourages the generation of new content rather than repeating existing information, which is useful for ensuring variety in responses.
- E. Incorrect.
Max tokens limit the length of the output but do not directly address the requirements for conciseness, avoidance of repetition, or alignment with the input context.