Objective 4.1 Deploy a Model AIP-210 Practice Questions (Page 8)

Questions 36–39

1. 36foundation · easy

   Which of the following is an example of a deployable model artifact that includes the model and its dependencies?

   • AA model training script
   • BA CSV file of predictions
   • CA Jupyter notebook
   • DA container image
   Reveal Solution Discussion Select an answer first
2. 37foundation · easy

   What is the purpose of a rollback procedure in model deployment?

   • ATo permanently delete the current model
   • BTo retrain the model on new data
   • CTo revert to a previous model version if the new one causes issues
   • DTo increase the model's inference speed
   Reveal Solution Discussion Select an answer first
4. 38expert · hard

   A team has a model that uses a custom tokenizer written in C++. They need to deploy it to a production environment that does not have a C++ compiler. What is the best way to package the model?

   • ACompile the tokenizer into a shared library and include it in the container image
   • BShip the C++ source code and compile it during deployment
   • CRewrite the tokenizer in Python and include it in the model package
   • DUse a pre-built binary from the internet without verification
   Reveal Solution Discussion Select an answer first
5. 39expert · hard

   A company has a model that predicts equipment failure. The model is deployed on edge devices in factories. The ML team wants to monitor the model's performance and receive alerts when it degrades. What is the best way to set up monitoring?

   • ACollect prediction logs from edge devices and send them to a central monitoring system for analysis
   • BHave each edge device send an alert if the model's confidence is low
   • CDo not monitor edge devices because they are offline most of the time
   • DMonitor each edge device's CPU and memory usage locally
   Reveal Solution Discussion Select an answer first