Design and Implement Models and Embeddings DP-800 Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A multinational bank stores customer communication logs in a SQL database. They want to use an external AI model to generate embeddings for these logs. The logs contain text, images, and audio files. The team needs to search across all modalities. Which capability must the external model have?

   • AStructured output
   • BMultilingual support
   • CLarge model size
   • DMultimodal support
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A company uses an external embedding model to generate vectors for a product search feature. The model vendor releases a new version that improves accuracy but changes the vector dimensionality from 384 to 768. The team must update the system without breaking existing search functionality. What should they do first?

   • AUpdate the search index schema first, then regenerate embeddings incrementally over time.
   • BKeep the old model running and only use the new model for new products.
   • CRegenerate all embeddings using the new model and update the search index schema to support the new dimensionality.
   • DContinue using the old model because changing dimensionality is too risky.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A logistics company uses Azure SQL Database to store shipment records. They generate embeddings for the 'Notes' column to enable semantic search. The Notes column is updated frequently, and the team needs to maintain embeddings with minimal latency. They decide to use Change Data Capture (CDC). What is the next step to generate embeddings for changed rows?

   • ACreate an Azure Function that reads CDC changes and calls the embedding API for each changed row.
   • BEnable Change Tracking instead of CDC, because CDC is deprecated.
   • CCreate a database trigger that updates the embedding column directly.
   • DUse a SQL Server Agent job to rebuild all embeddings nightly.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which component is required in the CREATE EXTERNAL MODEL statement in Azure SQL Database to specify the location of the external model?

   • ALOCATION
   • BENDPOINT
   • CDATA_SOURCE
   • DMODEL_TYPE
   Reveal Solution Discussion Select an answer first
6. 20foundation · easy

   When designing chunks for embeddings, what is a key consideration to ensure the chunks are meaningful?

   • AChunks should be as large as possible
   • BChunks should be random pieces of text
   • CChunks should be exactly 100 tokens
   • DChunks should preserve semantic boundaries
   Reveal Solution Discussion Select an answer first