Google Cloud Professional Machine Learning Engineer — Question 151

You recently developed a deep learning model. To test your new model, you trained it for a few epochs on a large dataset. You observe that the training and validation losses barely changed during the training run. You want to quickly debug your model. What should you do first?

Answer options

• A. Verify that your model can obtain a low loss on a small subset of the dataset
• B. Add handcrafted features to inject your domain knowledge into the model
• C. Use the Vertex AI hyperparameter tuning service to identify a better learning rate
• D. Use hardware accelerators and train your model for more epochs

Correct answer: A

Explanation

Verifying that your model can achieve a low loss on a small subset of the dataset is crucial as it helps confirm that the model is capable of learning. If the model cannot perform well on a simpler task, it suggests fundamental issues with the model architecture or data. The other options, while potentially useful, do not address the immediate need to assess the model's basic functionality.