Google Professional Machine Learning Engineer
Get started today
Ultimate access to all questions.
You have recently developed a wide and deep model using TensorFlow for generating daily recommendations. To prepare your training datasets, you employed a SQL script to preprocess raw data from BigQuery by performing instance-level transformations. You now need to create a robust training pipeline that will automatically retrain the model on a weekly basis. Considering the need to minimize model development and training time, what approach should you take to develop the training pipeline?
You have recently developed a wide and deep model using TensorFlow for generating daily recommendations. To prepare your training datasets, you employed a SQL script to preprocess raw data from BigQuery by performing instance-level transformations. You now need to create a robust training pipeline that will automatically retrain the model on a weekly basis. Considering the need to minimize model development and training time, what approach should you take to develop the training pipeline?
Explanation:
The correct answer is A: 'Use the KubeFlow Pipelines SDK to implement the pipeline. Use the BigQueryJobOp component to run the preprocessing script and the CustomTrainingJobOp component to launch a Vertex AI training job.' This approach leverages both KubeFlow Pipelines for orchestrating steps and existing BigQuery jobs to handle preprocessing, minimizing complexity and ensuring efficiency. BigQueryJobOp allows you to directly use the SQL script for preprocessing without additional overhead, and CustomTrainingJobOp will facilitate launching training tasks on Vertex AI, thereby streamlining the process.