2.4 Leveraging Gemini Enterprise Agent Platform for End-To-End ML Workflows PROFESSIONAL-CLOUD-ARCHITECT Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   What is the role of AI Hypercomputer in ML/AI workloads?

   • ATo automatically generate ML models from raw data.
   • BTo provide a user interface for managing ML experiments.
   • CTo replace the need for data preprocessing in ML workflows.
   • DTo provide a scalable and optimized infrastructure for training and serving AI models.
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A company has data in a transactional database that is updated frequently. They want to use this data to train a model that is retrained daily. They are considering using a batch export to Cloud Storage or a change data capture (CDC) approach. What is the primary trade-off they need to consider?

   • AThe need to use a specific data format (e.g., Parquet vs. Avro).
   • BThe cost of storage in Cloud Storage versus the cost of the database.
   • CThe choice between using a GPU or a TPU for training.
   • DThe complexity of setting up CDC versus the freshness of the data in the training pipeline.
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   What is a typical use case for Cloud Run functions in an ML workflow with Agent Platform?

   • AStoring and versioning ML models.
   • BTraining a deep neural network on a large dataset.
   • CRunning a lightweight data validation or transformation step when new data arrives.
   • DServing a large-scale model with low latency requirements.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A company is about to run a large-scale training job on AI Hypercomputer. They are concerned about the cost of the job and want to ensure they can stop it if the model's performance does not improve. What is the best practice to manage this?

   • AImplement early stopping in the training script and use a custom job service that can terminate the job based on a metric.
   • BUse a smaller model to reduce the training time and cost.
   • CRun the job on spot instances and rely on the preemption to stop the job if it runs too long.
   • DManually monitor the training job and stop it when the loss curve flattens.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   Which benefit does Agent Platform Pipelines provide for ML workflows?

   • AElimination of the need for data validation and preprocessing in ML projects.
   • BGuaranteed improvement in model accuracy for all ML tasks.
   • CAutomated execution and tracking of the ML workflow, reducing manual effort and errors.
   • DManual execution of each step to allow for human intervention at every stage.
   Reveal Solution Discussion Select an answer first