Implement Custom Language Models AI-102 Practice Questions (Page 8)

Questions 36–40

1. 36foundation · easy

   After training a custom language model, what is the next step to make it usable by a client application?

   • AAdd more utterances to the training set
   • BDeploy the model to a production endpoint
   • CEvaluate the model using test data
   • DExport the model to a JSON file
   Reveal Solution Discussion Select an answer first
2. 37foundation · easy

   What does it mean to consume a custom language model in a client application?

   • ATo call the model's endpoint via an API to get predictions
   • BTo export the model to a file
   • CTo evaluate the model's performance
   • DTo train the model with new data
   Reveal Solution Discussion Select an answer first
4. 38application · medium

   A custom question answering project currently has a knowledge base built from a single FAQ document. Users are not finding answers to questions about product specifications, which are available on the company's public website. How can the team enhance the knowledge base without manually adding every Q&A pair?

   • AManually add each product specification as a Q&A pair in the knowledge base.
   • BCreate a new project for the product specifications and link it to the existing project.
   • CAdd the public website URL as an additional source in the project and retrain the knowledge base.
   • DUse the Azure AI Search service to index the website and return search results.
   Reveal Solution Discussion Select an answer first
5. 39foundation · easy

   What is the purpose of creating a multi-language question answering solution?

   • ATo translate the knowledge base into a single language
   • BTo allow users to ask questions in different languages and receive answers in the corresponding language
   • CTo create multi-turn conversations
   • DTo improve the accuracy of intent recognition
   Reveal Solution Discussion Select an answer first
6. 40application · medium

   A developer has created a custom language model with intents, entities, and sample utterances. They have just finished adding a new set of utterances to improve coverage for a specific intent. What should the developer do before making the model available to the client application?

   • ADeploy the model immediately so the new utterances take effect in production.
   • BEvaluate the model using the training data, then deploy without retraining.
   • CBack up the model, then deploy the backup to production.
   • DTrain the model, evaluate it with test data, and then deploy it to a prediction endpoint.
   Reveal Solution Discussion Select an answer first