A company has a composed model in production. They need to add a new document type to the composed model. They have trained a new custom model for the new type. What is the correct deployment process?

Create a new composed model that includes the existing custom models and the new model, then update the application to use the new composed model's endpoint.

Delete the existing composed model and recreate it with the new model.

Use the new model separately and keep the old composed model for the existing types.

Add the new model to the existing composed model's configuration.

You have a set of documents that are unique to your business and do not match any prebuilt model. You need to extract specific fields from these documents. What should you create?

A prebuilt general document model

A custom document intelligence model

After training a custom document intelligence model, you want to evaluate its performance. What should you use?

A separate set of labeled documents not used in training

A company has a custom model that achieves 95% accuracy on a held-out test set. However, in production, the accuracy is only 80%. The production documents have more variations in layout and image quality than the training set. What should they do?

Deploy the model to a different region to improve performance.

Use the prebuilt general document model instead.

Increase the confidence threshold for the model's predictions.

Collect production-like documents, label them, and retrain the model with a more representative training set.

Microsoft Certified:Azure AI Engineer Associate

Domain 6Objective 2

Implement an Azure Document Intelligence in Foundry Tools Solution AI-102 Practice Questions (Page 2)

Part of the Implement knowledge mining and information extraction solutions domain, which accounts for 15-20% of the AI-102 exam. Microsoft does not publish an official question count, but from its 100-minute exam (~40–65 total, ~6–13 in this domain), expect 2–4 from this objective — we provide 35 practice questions to prepare you well beyond it. (estimate)

35questions here

7free pages

7concepts

15-20%of the exam

Questions 6–10

6expert · hard
A company has a composed model in production. They need to add a new document type to the composed model. They have trained a new custom model for the new type. What is the correct deployment process?
Select an answer first
7foundation · easy
You have a set of documents that are unique to your business and do not match any prebuilt model. You need to extract specific fields from these documents. What should you create?
Select an answer first
8foundation · easy
After training a custom document intelligence model, you want to evaluate its performance. What should you use?
Select an answer first
9foundation · easy
You are evaluating a custom document intelligence model. Which metric is commonly used to report the model's extraction accuracy?
Select an answer first
10expert · hard
A company has a custom model that achieves 95% accuracy on a held-out test set. However, in production, the accuracy is only 80%. The production documents have more variations in layout and image quality than the training set. What should they do?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to AI-102

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Microsoft. “AI-102” is a trademark of its owner, used for identification only.