Databricks Generative AI Engineer Associate Question 225
Select 3You are building a recommendation system for a product catalog using Databricks. To efficiently retrieve similar items based on product descriptions, you decide to use a Vector Search index. After creating embeddings for the product descriptions using a pre-trained model, what is the correct sequence of steps to create and query a Vector Search index in Databricks?
- A
Store the embeddings as a table in Databricks.
- B
Use the Databricks MLflow API to directly query the embeddings without creating an index.
- C
Create a Vector Search index on the embeddings table.
- D
Use a Vector Search query to retrieve similar items by providing a query vector.
- E
Export the embeddings to an external database before querying for similar vectors.
Show answer and explanation
Correct answers: A, C, D
Explanation
To perform Vector Search in Databricks, embeddings must first be stored in a table, followed by the creation of a Vector Search index on this table. Once the index is created, the similarity search can be conducted using a Vector Search query. Exporting data to an external database or relying on unrelated APIs like MLflow is not required or supported for this task.
- A. Correct.
Correct: To perform Vector Search, the embeddings must first be stored in a table in Databricks so they can be indexed and queried.
- B. Incorrect.
Incorrect: While MLflow is useful for managing models, it does not provide functionality for querying embeddings as part of Vector Search.
- C. Correct.
Correct: Creating a Vector Search index on the embeddings table allows for efficient similarity search operations.
- D. Correct.
Correct: A Vector Search query is used to find similar items by comparing query vectors with the indexed embeddings.
- E. Incorrect.
Incorrect: It is not necessary to export the embeddings to an external database for Vector Search, as Databricks supports querying embeddings directly within its environment.