Implement Custom Language Models AI-102 Practice Questions (Page 7)

Questions 31–35

1. 31foundation · easy

   Which of the following is a common type of external source that can be imported into a question answering knowledge base?

   • AAn FAQ page
   • BA trained language model
   • CAn Azure Functions app
   • DA virtual network configuration
   Reveal Solution Discussion Select an answer first
2. 32foundation · easy

   Why would you create a backup of a custom language model?

   • ATo prevent data loss and enable recovery in case of accidental deletion or corruption
   • BTo test the model's performance
   • CTo improve the model's accuracy
   • DTo deploy the model to a new region
   Reveal Solution Discussion Select an answer first
4. 33application · medium

   A multinational company wants to provide customer support in English, Spanish, and French using a single custom question answering solution. They have FAQ documents in all three languages. What is the recommended approach?

   • ACreate a separate question answering project for each language and import the corresponding FAQ document.
   • BCreate one project and set the language to 'Multiple' to handle all languages.
   • CImport all FAQ documents into a single project and rely on the service to detect the language.
   • DUse Azure Translator to translate all FAQs to English and create one English project.
   Reveal Solution Discussion Select an answer first
5. 34application · medium

   A company is building a customer support chatbot using custom question answering. Users often need to provide additional details after their initial question, such as their account number or issue category. The team wants the chatbot to ask follow-up questions based on the user's previous response. What should they implement?

   • ACreate multiple intents in the language model to handle each follow-up question.
   • BPublish the knowledge base and use a separate bot framework to handle follow-ups.
   • CAdd multi-turn conversation prompts to the relevant Q&A pairs to guide users through a sequence of questions.
   • DUse alternate phrasing to cover all possible follow-up questions.
   Reveal Solution Discussion Select an answer first
6. 35expert · medium

   A company has deployed a custom language model for a customer service chatbot. The chatbot is receiving high traffic, and the team notices that some requests are timing out. The model is deployed on a single endpoint with a standard tier. The team needs to improve reliability without changing the model's code. What should they do?

   • AIncrease the confidence threshold to filter out low-confidence requests.
   • BMove the model to a different region to improve latency.
   • CScale up the endpoint to a higher tier or enable autoscaling to handle the increased load.
   • DReduce the number of intents in the model to simplify inference.
   Reveal Solution Discussion Select an answer first