Implement Custom Language Models AI-102 Practice Questions (Page 9)

Questions 41–45

1. 41application · medium

   A company has a custom question answering knowledge base with manually added QnA pairs. They want to expand it using their existing online documentation. What should they do?

   • ATrain the knowledge base on the documentation without importing it.
   • BPublish the knowledge base and rely on the documentation being linked.
   • CAdd the documentation URL as a source to import the content.
   • DCopy and paste the entire documentation into a single QnA pair.
   Reveal Solution Discussion Select an answer first
2. 42foundation · easy

   When building a custom language model for a support bot, which of the following is an example of a sample utterance for the intent 'ResetPassword'?

   • APasswordReset
   • BPassword
   • CI forgot my password and need to reset it
   • DResetPassword
   Reveal Solution Discussion Select an answer first
4. 43foundation · easy

   What is the goal of optimizing a custom language model?

   • ATo improve its accuracy and efficiency while reducing unnecessary resource usage
   • BTo deploy the model to multiple regions
   • CTo create a backup of the model's configuration
   • DTo export the model to a different format
   Reveal Solution Discussion Select an answer first
5. 44foundation · easy

   What is the benefit of importing external sources into a question answering knowledge base?

   • AIt automatically adds question-answer pairs derived from the source content
   • BIt translates the knowledge base into multiple languages
   • CIt trains a custom language model for intent recognition
   • DIt deploys the knowledge base to a production endpoint
   Reveal Solution Discussion Select an answer first
6. 45application · medium

   A company is building a custom language model for a support chatbot. The model must recognize a user's request to reset a password. The team has added the intent 'ResetPassword' and the entity 'AccountType'. They have collected 50 sample utterances from real users, but the model still misclassifies some requests as 'GetHelp'. What should the team do to improve intent classification accuracy?

   • ADeploy the model to production and use real user feedback to manually correct misclassifications.
   • BRemove the 'GetHelp' intent from the model to eliminate confusion.
   • CIncrease the number of entities in the model to capture more details from each utterance.
   • DAdd more varied utterances for the 'ResetPassword' intent, including synonyms and different phrasings, then retrain the model.
   Reveal Solution Discussion Select an answer first