A research lab has fine-tuned a Gemini model to summarize scientific papers. They need to evaluate whether the summaries are faithful to the original papers and do not introduce hallucinated findings. Which evaluation metric is most appropriate?

Measure the ROUGE-L score against a set of human-written reference summaries.

Evaluate the model's perplexity on the source papers to gauge its understanding.

Use a faithfulness metric that checks whether each claim in the summary is supported by the source paper.

Track the model's generation latency to ensure it meets performance requirements.

A media company has built a Gemini-powered article generator. The ML team needs to evaluate whether the generated articles are factually consistent with the source material and free of unsupported claims. Which evaluation approach is most appropriate?

Evaluate the perplexity of the generated text to ensure the model is confident in its outputs.

Measure the BLEU score of the generated articles against a set of human-written reference articles.

Conduct a human evaluation where annotators rate the articles on a 1-5 scale for overall quality.

Use a factual-consistency metric that compares generated claims against the source document.

What is the primary purpose of Model Monitoring on the Gemini Enterprise Agent Platform?

To replace the need for human oversight

To establish continuous evaluation metrics for production models

To manage model deployment rollouts

A model predicting equipment failure was trained on sensor data. The training pipeline applies a scaling transformation that was fit on the training data. In production, the serving pipeline applies a different scaling transformation that was fit on a small sample of production data. The model's performance is degrading, and the feature distributions in the monitoring dashboard appear different from training. What is the root cause and the best fix?

Feature attribution drift; investigate why the model's feature importance changed.

Data drift; retrain the model with more recent production data.

Training-serving skew; ensure the serving pipeline uses the same scaling transformation as training.

Concept drift; retrain the model with new labels.

Google CloudProfessional Machine Learning Engineer

Domain 6Objective 2

Monitoring, Testing, and Troubleshooting AI Solutions PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice Questions (Page 6)

Part of the Monitoring AI solutions domain, which accounts for ~13% of the PROFESSIONAL-MACHINE-LEARNING-ENGINEER exam.

33questions here

7free pages

8concepts

~13%of the exam

Questions 26–30

26application · medium
A research lab has fine-tuned a Gemini model to summarize scientific papers. They need to evaluate whether the summaries are faithful to the original papers and do not introduce hallucinated findings. Which evaluation metric is most appropriate?
Select an answer first
27application · medium
A media company has built a Gemini-powered article generator. The ML team needs to evaluate whether the generated articles are factually consistent with the source material and free of unsupported claims. Which evaluation approach is most appropriate?
Select an answer first
28foundation · easy
What is the primary purpose of Model Monitoring on the Gemini Enterprise Agent Platform?
Select an answer first
29expert · hard
A model predicting equipment failure was trained on sensor data. The training pipeline applies a scaling transformation that was fit on the training data. In production, the serving pipeline applies a different scaling transformation that was fit on a small sample of production data. The model's performance is degrading, and the feature distributions in the monitoring dashboard appear different from training. What is the root cause and the best fix?
Select an answer first
30foundation · easy
A machine learning engineer wants to automatically track the performance of a production model deployed on the Gemini Enterprise Agent Platform. Which feature should they configure to establish continuous evaluation metrics?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to PROFESSIONAL-MACHINE-LEARNING-ENGINEER

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Google Cloud. “PROFESSIONAL-MACHINE-LEARNING-ENGINEER” is a trademark of its owner, used for identification only.

Monitoring, Testing, and Troubleshooting AI Solutions PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice Questions (Page 6)

Questions 26–30

A research lab has fine-tuned a Gemini model to summarize scientific papers. They need to evaluate whether the summaries are faithful to the original papers and do not introduce hallucinated findings. Which evaluation metric is most appropriate?

A media company has built a Gemini-powered article generator. The ML team needs to evaluate whether the generated articles are factually consistent with the source material and free of unsupported claims. Which evaluation approach is most appropriate?

What is the primary purpose of Model Monitoring on the Gemini Enterprise Agent Platform?

A machine learning engineer wants to automatically track the performance of a production model deployed on the Gemini Enterprise Agent Platform. Which feature should they configure to establish continuous evaluation metrics?

Keep studying — Monitoring AI solutions