A company uses OCI Generative AI Service to build a chatbot for customer support. They notice that the model sometimes generates inappropriate responses. What is the MOST effective way to mitigate this without retraining the model?
Trap 1: Fine-tune the model with curated safe examples
Fine-tuning requires additional data and resources, and is not a quick mitigation.
Trap 2: Reduce the temperature parameter to 0
Lower temperature reduces randomness but does not prevent inappropriate content.
Trap 3: Use the moderation API to filter responses
The moderation API filters after generation, which is less proactive than setting instructions upfront.
- A
Fine-tune the model with curated safe examples
Why wrong: Fine-tuning requires additional data and resources, and is not a quick mitigation.
- B
Configure system instructions to define acceptable behavior
System instructions constrain the model's output at inference time without retraining.
- C
Reduce the temperature parameter to 0
Why wrong: Lower temperature reduces randomness but does not prevent inappropriate content.
- D
Use the moderation API to filter responses
Why wrong: The moderation API filters after generation, which is less proactive than setting instructions upfront.