Implement Custom Language Models AI-102 Practice Questions (Page 4)

Questions 16–20

1. 16foundation · easy

   What does it mean to deploy a custom language model in Azure AI Language?

   • ATo evaluate the model's performance on test data
   • BTo make the trained model available as a runtime endpoint for applications to call
   • CTo add more utterances to the model's training set
   • DTo save the model's training data to a backup location
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A company has finished testing their custom question answering knowledge base and is satisfied with the accuracy. They now want to make it available to their customers through a web app. What should they do?

   • AExport the knowledge base and embed it into the web app.
   • BPublish the knowledge base to obtain an endpoint and then integrate it into the web app.
   • CTrain the knowledge base again to ensure it is ready.
   • DCreate a new project and copy the QnA pairs into it.
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   What is the purpose of evaluating a custom language model with test data?

   • ATo create a backup of the model's configuration
   • BTo measure the model's accuracy and identify areas for improvement
   • CTo deploy the model to a production endpoint
   • DTo add more sample utterances to the training set
   Reveal Solution Discussion Select an answer first
5. 19expert · medium

   A custom question answering knowledge base has been tested and is ready for production. The team wants to integrate it into an existing customer portal that uses Azure Active Directory for authentication. The portal will call the knowledge base's REST API. What is the most secure way to authenticate the portal's requests?

   • AUse the knowledge base's primary key in the request header to authenticate.
   • BUse a shared access signature (SAS) token generated for the knowledge base.
   • CUse Azure AD authentication with a managed identity for the portal.
   • DEmbed the endpoint URL in the client-side code and call it directly.
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A company has invested significant effort in building a custom language model with hundreds of intents and thousands of utterances. They want to ensure they can recover the model if the Azure region where it is deployed experiences an outage. What should they do to protect against data loss?

   • ACopy the model's endpoint URL to a text file and save it locally.
   • BTake a snapshot of the virtual machine running the model.
   • CRely on Azure's automatic backups for the Language service.
   • DExport the model's project assets (intents, entities, utterances) to a JSON file and store it in a different region.
   Reveal Solution Discussion Select an answer first