Generative AI Leader Google Cloud's Generative AI Offerings Practice Test 5 — 15 Questions

Question 1

A healthcare company is building a chatbot to answer patient queries using Vertex AI Agent Builder. They want to ensure the chatbot only uses approved medical references and does not generate unverified advice. How should they configure the agent?

Accepted Answer

Set up grounding with a private data store containing verified medical documents. Grounding with a curated data store ensures the chatbot only retrieves information from approved sources. Option B is wrong because safety filters block categories but not unverified content. Option C is wrong because high temperature increases creativity, risking unverified answers. Option D is wrong because model monitoring detects drift but does not restrict sources.

Answer

Enable strict safety filters to block any medical advice

Answer

Increase the temperature parameter to get more diverse responses

Answer

Use Vertex AI Model Monitoring to track answer accuracy

Question 2

Which TWO actions can help reduce latency for an online prediction endpoint served by a large language model on Vertex AI? (Select TWO.)

Accepted Answer

Enable response caching for frequent similar queries. Using a smaller model (A) and enabling response caching (D) can reduce latency. B increases latency. C reduces latency but is not a direct action (the platform handles it). E does not exist as a standard optimization.

Answer

Increase the max input token limit to capture more context

Answer

Use automatic scaling to add more replicas

Answer

Set the model to preemptible instances

Question 3

Which THREE capabilities does Vertex AI Agent Builder provide out of the box? (Select THREE.)

Accepted Answer

Automatic escalation to human agents when confidence is low. Agent Builder includes pre-built agents (A), grounding (D), and automatic fallback (E). B is done via Model Garden, not Agent Builder directly. C is a feature of Vertex AI Search.

Answer

Fine-tuning of foundation models using custom datasets

Answer

Enterprise search over internal documents

Question 4

Which TWO safety features are available in Vertex AI Gemini API? (Select TWO.)

Accepted Answer

Safety filters for categories like hate speech and harassment. Vertex AI Gemini API provides safety filters (A) and content restrictions (C). B is not a standard Gemini API feature. D is a separate service. E is not a specific safety feature.

Answer

Model-level encryption at rest

Answer

Automatic redaction of personally identifiable information (PII)

Answer

Integration with Cloud Data Loss Prevention (DLP)

Question 5

A data scientist runs the above command to upload a model to Vertex AI Model Registry. The model is a TensorFlow 2.6 model trained on tabular data. After deployment to an endpoint, the prediction latency is higher than expected. What is the most likely cause?

Accepted Answer

The container image used is CPU-only, but a GPU-accelerated image would improve latency. The command uses a tf2-cpu image; GPU-optimized images offer faster inference for many models. Option A is wrong because the command is correct. Option B is wrong because the artifact URI is a directory, and the command is correct. Option C is wrong because the region is specified.

Answer

The artifact URI points to a single file instead of a directory

Answer

The model should be uploaded with a different display name

Answer

The model is uploaded to the wrong region

Question 6

A machine learning engineer submits the above batch prediction job for a large language model. The job is expected to process 100,000 instances. The job takes much longer than expected. Which change would most likely reduce the execution time?

Accepted Answer

Increase maxReplicaCount to 10. Increasing maxReplicaCount allows the job to use more workers for parallel processing, reducing time for large jobs. Option A is wrong because n1-standard-4 might be underpowered; but increasing replicas is more impactful. Option C is wrong because larger batch size can improve throughput but may cause memory issues. Option D is wrong because increasing startingReplicaCount alone without maxReplicaCount doesn't help scalability.

Answer

Increase startingReplicaCount to 10 without changing maxReplicaCount

Answer

Increase the machine type to n1-standard-16

Answer

Decrease the batch size to 1

Question 7

A developer is configuring a Vertex AI Agent Builder agent to use grounding. They receive the above error when calling the API. What is the most likely cause?

Accepted Answer

The data store has not been created in the specified project. The error says the data store does not exist. The developer likely created a data store in a different project or did not create one. Option A is wrong because authentication errors give different messages. Option C is wrong because permissions errors are 403. Option D is wrong because a race condition would not produce 'does not exist'.

Answer

The data store was just created and is not yet propagated

Answer

The agent is not authenticated to access the data store

Answer

The grounding configuration is missing required permissions

Question 8

A company wants to build a chatbot that can answer questions about its internal knowledge base using natural language. Which Google Cloud Generative AI offering should they use to quickly prototype and deploy this chatbot with minimal coding?

Accepted Answer

Generative AI Studio. Generative AI Studio provides a no-code/low-code environment to prototype and deploy chatbots with foundation models.

Answer

Vertex AI Endpoints

Answer

Cloud Natural Language API

Answer

Vertex AI Model Garden

Question 9

A data scientist is using the Vertex AI PaLM API for text generation. They notice that the model occasionally generates toxic content. Which parameter should they adjust to reduce the likelihood of toxic outputs?

Accepted Answer

safety_settings. The safety_settings parameter allows specifying thresholds for categories like toxic content to block or filter responses.

Answer

max_output_tokens

Answer

temperature

Answer

top_k

Question 10

An organization is deploying a summarization model on Vertex AI and needs to ensure that the model's responses are consistent and avoid hallucinations. They have a labeled dataset of source documents and human-written summaries. Which approach would best align the model with their quality requirements?

Accepted Answer

Perform supervised fine-tuning using their labeled dataset. Supervised fine-tuning on a high-quality dataset specific to the task reduces hallucinations and improves consistency.

Answer

Deploy the model with a larger max_output_tokens

Answer

Use prompt engineering with few-shot examples

Answer

Increase the temperature to 0.9

Question 11

A developer is using the Vertex AI PaLM API and receives a 429 Resource Exhausted error. What is the most likely cause?

Accepted Answer

The user has exceeded the allowed number of requests per minute. 429 errors indicate rate limiting or quota exhaustion for the API.

Answer

The request payload is too large

Answer

The model is not available in the current region

Answer

The API key is invalid

Question 12

A team wants to fine-tune a PaLM 2 model with their own data on Vertex AI. What is the recommended way to prepare the training data?

Accepted Answer

JSON Lines file with 'input_text' and 'output_text' keys. Fine-tuning for PaLM expects data in JSON Lines format with 'input_text' and 'output_text' fields.

Answer

TFRecord files

Answer

CSV file with prompt and completion columns

Answer

Pickle serialized objects

Question 13

A company is using Vertex AI Model Registry to manage multiple versions of its custom generative model. They want to automatically route a percentage of traffic to a new model version for testing. What should they do?

Accepted Answer

Deploy both versions to the same endpoint and adjust traffic split settings. Vertex AI Endpoints support traffic splitting between model versions.

Answer

Set up a Cloud Tasks queue to distribute requests

Answer

Create a new endpoint for each version

Answer

Use a load balancer in front of the endpoints

Question 14

A data scientist is comparing two fine-tuned models on Vertex AI Model Evaluation. They want to choose the model with better factual accuracy for a medical Q&A task. Which evaluation metric should they prioritize?

Accepted Answer

exact_match. The 'exact_match' metric measures whether the generated answer matches the ground truth exactly, which is suitable for factual accuracy.

Answer

pairwise_rouge

Answer

ROUGE-L

Answer

BLEU

Question 15

A developer needs to use the Vertex AI PaLM API to generate text embeddings for a large corpus of documents. Which model should they use?

Accepted Answer

textembedding-gecko@001. textembedding-gecko is the dedicated model for text embeddings.

Answer

codey-bison@001

Answer

text-bison@001

Answer

chat-bison@001