Free AWS AIP-C01 Actual Exam Questions - Question 8 Discussion
Scenario: A publishing company uses a text-to-text foundation model (FM) on Amazon Bedrock for summarization. The model misinterprets casual language, local expressions, and abbreviations in customer feedback, leading to inaccurate summaries. Question- Which solution provides the most efficient and cost-effective approach to improve the model's understanding of customer feedback? Options:
B/D? Fine-tuning (B) helps the model learn the exact language style, but adding metadata with CER (D) could boost context understanding without full retraining. Both seem practical compared to heavy preprocessing or training.
It’s C since cleaning data upfront avoids expensive retraining and keeps summaries consistent.
Option C feels risky since removing slang could lose important context needed for accurate summaries.
Maybe D, since adding slang as metadata could help the model without costly retraining or losing context.
Could C be risky if it strips out useful context from slang or local expressions? Sometimes standardizing text loses nuance that the model actually needs to summarize properly.
B/C? Fine-tuning (B) seems more targeted for slang and informal language, but preprocessing (C) avoids retraining costs and can quickly standardize input. Depends if the company wants ongoing updates or a quick fix.
Maybe E and D. The model needs to be registered so SageMaker Canvas can recognize it, and the data specialists definitely need permissions to access the S3 bucket where the artifacts are stored. I don’t think the endpoint is mandatory just to get Canvas access; it’s more about having the model available and accessible. Also, converting the format (C) isn’t necessarily required if the model’s already compatible with SageMaker.
I think D and E make the most sense here. Registering the model in the registry and giving the data team S3 access should let Canvas use the model without extra setup.