While running a model training pipeline on Vertex Al, you discover that the evaluation step is failing because of an out-of-memory error. You are currently using TensorFlow Model Analysis (TFMA)…

Question

Options

• AInclude the flag -runner=DataflowRunner in beam_pipeline_args to run the evaluation step on
• BMove the evaluation step out of your pipeline and run it on custom Compute Engine VMs with
• CMigrate your pipeline to Kubeflow hosted on Google Kubernetes Engine, and specify the
• DAdd tfma.MetricsSpec () to limit the number of metrics in the evaluation step.

• A
  75% (24)
• B
  13% (4)
• C
  9% (3)
• D
  3% (1)

Explanation

The evaluation step runs as an Apache Beam pipeline under the hood (TFMA uses Beam). Out-of-memory errors occur because the default runner (DirectRunner) processes everything on a single machine. Option D (limiting metrics) reduces evaluation quality, violating the requirement. Options B and C both introduce significant infrastructure overhead (custom VMs and GKE cluster management, respectively). Option A is correct: adding '--runner=DataflowRunner' to beam_pipeline_args redirects the evaluation Beam pipeline to Google Cloud Dataflow, which automatically distributes the workload across managed workers, resolving memory pressure with no manual infrastructure management.

No community discussion yet for this question.