Use the Snowflake Model Registry SNOWPRO-ADVANCED-DATA-SCIENTIST Practice Questions (Page 5)

Questions 21–25

1. 21expert · medium

   A team has a model in the Snowflake Model Registry with version 1 in production and version 2 in staging. They want to run a canary deployment where a small percentage of inference requests use version 2, while the majority use version 1. They plan to do this by running two separate inference jobs and combining the results. What is the main drawback of this approach?

   • AIt is not possible to run two model versions in the same query.
   • BIt will not allow comparing the results of the two versions.
   • CIt requires manual splitting of the data and combining results, which is error-prone.
   • DIt will cause performance degradation because two queries are run.
   Reveal Solution Discussion Select an answer first
2. 22application · medium

   A model in the Snowflake Model Registry has been in production for several months. A new version has been thoroughly tested and is ready to replace it. The team wants to ensure that all new inference calls use the new version, but they also want to keep the old version available for rollback. What should they do?

   • ARegister the new version and set it as the default version, while keeping the old version available.
   • BSet the old version to deprecated and rely on the registry to automatically route to the new version.
   • CDelete the old version and register the new version with the same name.
   • DCreate a new model with a different name for the new version and update all references.
   Reveal Solution Discussion Select an answer first
4. 23foundation · easy

   Which of the following is a key feature of the Snowflake Model Registry?

   • ABuilt-in hyperparameter tuning for all registered models
   • BDirect integration with external model registries like MLflow
   • CAutomatic model retraining based on data drift
   • DVersioning of models with the ability to manage multiple versions
   Reveal Solution Discussion Select an answer first
5. 24application · medium

   A team has a model registered in the Snowflake Model Registry with multiple versions. They want to run batch inference on a large table using a specific version of the model, but they are concerned about performance. What is the best practice for generating predictions on a large dataset?

   • AUse a SQL SELECT statement with the model function, which leverages Snowflake's parallel processing for batch inference.
   • BUse the model's predict method in a Python UDF, which processes rows sequentially.
   • CUse a Python stored procedure to iterate over the table row by row and call the model for each row.
   • DExport the data to a separate system and run inference there, then load the results back.
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   Which command is used to set a model version as the default for inference in Snowflake?

   • ASET MODEL ... DEFAULT
   • BUPDATE MODEL ... SET VERSION
   • CMODIFY MODEL ... DEFAULT VERSION
   • DALTER MODEL ... SET DEFAULT VERSION
   Reveal Solution Discussion Select an answer first