Google Professional Machine Learning Engineer Question 172
Select 3Google Cloud PlatformYou are working as a Machine Learning Engineer at an e-commerce company and need to develop a recommendation model for predicting user preferences. You decide to use Vertex AI Workbench to build the model. The company wants to use TensorFlow for this project. Which steps should you take to successfully develop the model in Vertex AI Workbench?
- A
Set up a Vertex AI Workbench notebook instance with TensorFlow pre-installed.
- B
Train the model locally on your laptop to avoid incurring cloud costs.
- C
Use Vertex AI Workbench to write and run TensorFlow code for data preprocessing, model training, and evaluation.
- D
Export the trained model as a TensorFlow SavedModel format for deployment.
- E
Manually install TensorFlow dependencies on Vertex AI Workbench every time the notebook is started.
Show answer and explanation
Correct answers: A, C, D
Explanation
To develop models in Vertex AI Workbench using TensorFlow, you should leverage the built-in capabilities for environment setup (e.g., TensorFlow pre-installed configurations) and use the platform to write, train, and evaluate the model. Exporting the model in a standard format like TensorFlow SavedModel ensures compatibility with deployment pipelines. Training the model locally or manually installing dependencies unnecessarily adds complexity and does not utilize Vertex AI Workbench's benefits.
- A. Correct.
Correct: Vertex AI Workbench allows you to set up notebook instances with TensorFlow pre-installed, which simplifies the environment setup process.
- B. Incorrect.
Incorrect: Training the model locally on your laptop is not recommended for large-scale production-level projects, as it lacks scalability and cloud-specific benefits.
- C. Correct.
Correct: Vertex AI Workbench supports writing and executing TensorFlow code for various stages of the machine learning workflow, including preprocessing, training, and evaluation.
- D. Correct.
Correct: Exporting the trained model as a TensorFlow SavedModel is a standard practice to ensure compatibility with deployment services like Vertex AI or TensorFlow Serving.
- E. Incorrect.
Incorrect: TensorFlow dependencies are pre-installed in Vertex AI Workbench when using the appropriate environment, so manual installation is not necessary unless using a custom setup.