1Z0-1127-25 Using OCI Generative AI Service Practice Test 3 — 15 Questions

Question 1

Which TWO configurations are required to use a custom fine-tuned model on OCI Gen AI?

Accepted Answer

A dedicated AI cluster. Option B (a dedicated AI cluster) is correct because OCI Generative AI fine-tuning and hosting of custom models requires a dedicated AI cluster, which provides the isolated compute infrastructure (GPU capacity) needed to train and serve the custom fine-tuned model. Option C (training data in Object Storage) is correct because the fine-tuning process reads the training dataset from an OCI Object Storage bucket, which must be provided as the data source when creating the fine-tuning job. Option A (a security list) is not required specifically for this scenario, since access is governed by IAM policies and the cluster's networking configuration rather than a security list defined for the fine-tuning use case. Option D (an API key) is not required, as authentication to OCI services uses IAM credentials, instance principals, or resource principals rather than an API key for fine-tuning. Option E (a serverless endpoint) is not required, because custom fine-tuned models are hosted on the dedicated AI cluster rather than a serverless endpoint.

Answer

A security list

Answer

An API key

Answer

A serverless endpoint

Question 2

Which THREE factors should be considered when choosing a base model for OCI Gen AI?

Accepted Answer

Availability in the desired region. When choosing a base model for OCI Generative AI, availability in the desired region (A) is essential because OCI Gen AI models are hosted per-region, and a model not offered in your target region cannot be used regardless of its capabilities. Cost per token (B) directly affects the operational expense of inference, so it must be weighed against budget and expected request volume. Model size (D) matters because it influences latency, throughput, resource requirements, and often the quality/capability trade-off, making it a key selection factor. The remaining options are not among the three correct factors: supported languages (C) and training data format (E) are not the specified selection criteria in this context, even though language support can be relevant in general model evaluation.

Answer

Supported languages

Answer

Training data format

Question 3

Which TWO steps are necessary to deploy a fine-tuned model on a dedicated AI cluster?

Accepted Answer

Create the dedicated AI cluster. To deploy a fine-tuned model on a dedicated AI cluster, you must first create the dedicated AI cluster (Option A), since this provisions the GPU/accelerator infrastructure that hosts the model and is the foundational resource required before any deployment can occur. You must then create a model deployment endpoint (Option E), which registers the model artifacts and exposes an inference endpoint on that cluster so the fine-tuned model can actually serve predictions. Options B, C, and D are not required steps: a VCN with a public subnet is not mandatory because the cluster can be provisioned in a private subnet or use an existing network configuration, uploading artifacts to Object Storage is not a distinct required deployment step since model artifacts are referenced during deployment creation, and obtaining a third-party license is irrelevant unless the specific base model's licensing demands it, which is not a general deployment requirement.

Answer

Set up a VCN with public subnet

Answer

Upload the model artifacts to OCI Object Storage

Answer

Obtain a third-party license

Question 4

Refer to the exhibit. A user receives this error when calling the OCI Gen AI inference endpoint. What is the most likely cause?

Accepted Answer

The model is not deployed in the region. The error indicates that the model is not available in the specified region. OCI Gen AI models are deployed regionally, and each region supports only a specific subset of models. If the user calls an endpoint in a region where the requested model has not been deployed, the service returns an error because the model's inference endpoint does not exist in that region's routing table.

Answer

The region name is misspelled

Answer

The API key is expired

Answer

The model name is incorrect

Question 5

Refer to the exhibit. A user runs 'oci generative-ai model list' and sees this output. They then try to use 'cohere.command-light' but get an error. What is the most likely reason?

Accepted Answer

The model is in INACTIVE state. The error occurs because the model 'cohere.command-light' is in an INACTIVE state. In OCI Generative AI, models must be in ACTIVE state to be used. The 'oci generative-ai model list' output would show the lifecycle state, and if it's INACTIVE, the model cannot be invoked. Therefore, the most likely reason is that the model is inactive.

Answer

The API key does not have access

Answer

The model is not listed

Answer

The region is wrong

Question 6

A team is building a Retrieval-Augmented Generation (RAG) pipeline using OCI Generative AI. They need to store and retrieve document embeddings for semantic search. Which OCI service is most appropriate as the vector store?

Accepted Answer

OCI Search with OpenSearch. OCI Search with OpenSearch is the most appropriate vector store for a RAG pipeline because it natively supports storing and querying high-dimensional vector embeddings using the k-nearest neighbor (k-NN) algorithm. It integrates directly with OCI Generative AI to enable semantic search over ingested documents, providing the required similarity search capabilities for retrieval-augmented generation.

Answer

OCI Streaming

Answer

OCI Object Storage

Answer

OCI Autonomous Database with AI Vector Search

Question 7

Which authentication method should be used to securely call the OCI Generative AI API from a microservice running on OCI Compute?

Accepted Answer

OCI API signing key. The OCI API signing key method uses a pre-generated RSA key pair to sign each HTTP request cryptographically, which is the standard and most secure way to authenticate calls to OCI services, including the Generative AI API, from a microservice running on OCI Compute. This method does not rely on external identity providers or token exchanges, making it ideal for server-to-server communication within OCI.

Answer

OAuth 2.0 client credentials

Answer

SAML 2.0 assertion

Answer

Instance principal

Question 8

A company runs batch inference jobs daily using the OCI Generative AI service. The current cost is higher than expected. Which change would most effectively reduce cost while maintaining throughput?

Accepted Answer

Switch from on-demand to dedicated AI cluster with batch endpoint.. Switching from on-demand to a dedicated AI cluster with a batch endpoint reduces cost because dedicated clusters provide reserved capacity at a lower per-token rate compared to on-demand pay-per-token pricing, and batch endpoints allow you to process multiple inference requests in a single job, amortizing overhead and reducing idle time. This combination directly addresses the high cost of per-request on-demand pricing while maintaining the same throughput for daily batch jobs.

Answer

Reduce the max token limit for all requests.

Answer

Use a larger model to reduce retries.

Answer

Increase the number of parallel requests to improve efficiency.

Question 9

Which TWO factors are most important when deciding between on-demand and dedicated AI clusters for OCI GenAI?

Accepted Answer

Number of concurrent requests. Option D (Number of concurrent requests) is correct because on-demand AI clusters in OCI GenAI are designed for variable, bursty workloads with shared capacity, while dedicated AI clusters provide guaranteed, isolated compute that is necessary when sustained high concurrency would otherwise cause throttling or queueing. Option E (Latency requirements) is correct because dedicated clusters eliminate the noisy-neighbor effects and resource contention inherent in multi-tenant on-demand infrastructure, delivering the predictable, low-latency inference that strict SLAs demand. Option A (Fine-tuning capability) is not the deciding factor, since fine-tuning is available through the OCI Generative AI service independent of whether the underlying cluster is on-demand or dedicated. Option B (Model size) does not determine the choice, as both deployment modes can host the same supported foundation models. Option C (Data residency) is also not the differentiator, because data residency is governed by the OCI region and compartment configuration rather than by the on-demand versus dedicated cluster selection.

Answer

Fine-tuning capability

Answer

Model size

Answer

Data residency

Question 10

Refer to the exhibit. The group DataScientists can run inference but cannot fine-tune a model on a dedicated AI cluster. Which additional policy statement is required to allow fine-tuning?

Accepted Answer

Allow group DataScientists to manage dedicated-ai-clusters in compartment ABC. To fine-tune a model on a dedicated AI cluster, the user must have the `manage` permission on the `dedicated-ai-clusters` resource type in the compartment. The `inspect` permission (Option A) only allows viewing metadata, not performing write operations like fine-tuning. The `use` permission (Option D) is insufficient for management tasks. Option C is invalid because `generative-ai-fine-tune` is not a valid resource type in OCI IAM policies.

Answer

Allow group DataScientists to inspect dedicated-ai-clusters in compartment ABC

Answer

Allow group DataScientists to use generative-ai-fine-tune in compartment ABC

Answer

Allow group DataScientists to use dedicated-ai-clusters in compartment ABC

Question 11

A healthcare company wants to use OCI Generative AI to summarize patient medical records while ensuring PHI compliance. Which OCI service feature should they enable?

Accepted Answer

Use the data masking capability within OCI Generative AI. OCI Generative AI includes a built-in data masking capability specifically designed to automatically detect and redact sensitive information such as Protected Health Information (PHI) from input prompts and generated outputs. This feature ensures compliance with healthcare regulations like HIPAA by preventing PHI from being stored, logged, or exposed during inference, without requiring external preprocessing or post-processing.

Answer

Configure a Virtual Cloud Network (VCN) with private subnets

Answer

Deploy a Web Application Firewall (WAF) in front of the API

Answer

Set up Identity and Access Management (IAM) policies to restrict access

Question 12

A developer wants to generate text embeddings using OCI Generative AI. Which model endpoint should they call?

Accepted Answer

POST /v1/embed. The correct endpoint for generating text embeddings in OCI Generative AI is POST /v1/embed. Embeddings convert text into dense vector representations that capture semantic meaning, enabling tasks like similarity search and clustering. The other endpoints serve different purposes: /v1/generate for text generation, /v1/summarize for summarization, and /v1/chat for conversational interactions.

Answer

POST /v1/generate

Answer

POST /v1/summarize

Answer

POST /v1/chat

Question 13

A company wants to use OCI Generative AI to analyze legal documents and extract key clauses. Which model type is best suited for this task?

Accepted Answer

Cohere Summarize. Cohere Summarize is specifically designed to condense long documents into concise summaries, making it ideal for extracting key clauses from legal documents. Unlike other Cohere models, Summarize focuses on distilling the most important information from text, which aligns with the task of identifying and extracting critical clauses.

Answer

Cohere Command (generate)

Answer

Cohere Chat

Answer

Cohere Embed

Question 14

Which TWO actions are required to use a custom fine-tuned model via OCI Generative AI? (Choose two.)

Accepted Answer

Deploy the model to an endpoint. Option E is correct because a custom fine-tuned model in OCI Generative AI only exists after the fine-tuning job has run to completion successfully; the resulting fine-tuned model (with its own OCID) is the artifact you then serve. Option A is correct because a fine-tuned model must be deployed to a dedicated AI cluster endpoint before it can be invoked for inference; the deployment creates the endpoint that applications call. Option B is not required, since a private endpoint is an optional networking choice for restricting access, not a prerequisite for using a fine-tuned model. Option C is not required, as cross-region replication is not a mandatory step for fine-tuned model usage. Option D is not required, because granting access to other tenancies is an optional cross-tenancy sharing configuration, not part of the standard fine-tune-and-deploy workflow.

Answer

Provision a private endpoint for the model

Answer

Enable cross-region replication

Answer

Grant access to other tenancies

Question 15

Which THREE parameters can be adjusted to reduce repetition in generated text? (Choose three.)

Accepted Answer

presence_penalty. presence_penalty (A) is correct because it applies a flat penalty to any token that has already appeared at least once, which discourages the model from reusing the same tokens and thereby reduces repetition. frequency_penalty (D) is correct because it scales the penalty by how many times a token has already occurred, so frequently repeated tokens are progressively suppressed, directly countering loops and verbatim repetition. top_k (B) is correct because restricting sampling to the k highest-probability tokens truncates the long tail of low-probability continuations that often produce degenerate, repetitive output, tightening the distribution the model samples from. max_tokens (C) only caps the length of the generated output and does not change the probability distribution, so it cannot reduce repetition within the text. temperature (E) controls the randomness/flatness of the sampling distribution but does not specifically penalize or track repeated tokens, so it is not a targeted repetition-reduction parameter.

Answer

max_tokens

Answer

temperature