1Z0-1127-25 Deploying and Managing Generative AI on OCI — Timed 15 Questions

Question 1

A data scientist is fine-tuning a generative AI model on OCI Data Science using a custom container with GPU resources. The training job fails with an out-of-memory error despite the GPU instance having sufficient memory. The job works fine on a smaller dataset. What is the most likely cause?

15 minute time limit — choose an answer to begin.

Accepted Answer

The batch size is too large for the GPU memory. The most likely cause is that the batch size is too large for the GPU memory. Even though the GPU instance has sufficient total memory, a batch size that exceeds the available GPU memory (after accounting for model parameters, gradients, and optimizer states) will trigger an out-of-memory (OOM) error. Reducing the batch size allows the model to fit within the GPU's memory limits, which explains why the job works on a smaller dataset but fails on a larger one.

Answer

The training script has a memory leak

Answer

The GPU instance is not supported by OCI Data Science

Answer

The model is not compatible with the PyTorch version

Question 2

An organization wants to deploy a generative AI chatbot using OCI Generative AI service. The chatbot must comply with data residency requirements by ensuring that all data processing occurs within a specific geographic region. What is the best practice to achieve this?

Accepted Answer

Use a dedicated AI cluster in the required region. OCI Generative AI service allows you to provision a dedicated AI cluster within a specific region, ensuring all model inference and data processing remain within that geographic boundary. This dedicated cluster is isolated from other regions and complies with data residency requirements by design, as no data leaves the chosen region during processing.

Answer

Enable cross-region replication for disaster recovery

Answer

Configure a tenancy-wide policy to restrict region usage

Answer

Use IAM policies to block access from other regions

Question 3

A team has deployed a generative AI model using OCI Data Science model deployment. The endpoint is behind a load balancer. Users report that after 5 minutes of inactivity, the first request takes over 30 seconds to respond, while subsequent requests are fast. What is the most likely cause and solution?

Accepted Answer

The model deployment has an idle timeout that scales down to zero; configure a minimum number of instances or use a warm-up request. The described behavior—first request after 5 minutes of inactivity taking over 30 seconds, with subsequent requests fast—is a classic symptom of an idle timeout that scales the model deployment to zero instances. OCI Data Science model deployments support auto-scaling with an idle timeout (default 5 minutes) that can reduce the number of instances to zero when no requests are received. When a new request arrives, it must wait for a new instance to spin up, causing the delay. The solution is to configure a minimum number of instances (e.g., 1) to keep the model warm, or use a warm-up request to prevent the idle timeout from triggering.

Answer

The load balancer is scaling based on CPU utilization; increase the CPU threshold

Answer

The VCN has a network latency issue; use a different availability domain

Answer

The inference code has a lazy initialization; pre-load the model in the deployment script

Question 4

A machine learning engineer is deploying a fine-tuned Llama 2 model on OCI Data Science model deployment. The deployment fails with an error: 'Model artifact exceeds the maximum allowed size of 10 GB.' The model files total 12 GB. What is the best approach to resolve this?

Accepted Answer

Store the model in Object Storage and reference it in the deployment configuration. OCI Data Science model deployment has a hard limit of 10 GB for the model artifact uploaded directly. By storing the model in Object Storage and referencing it in the deployment configuration, you bypass this limit entirely, as the deployment service can load the model from Object Storage at runtime without requiring the artifact to be part of the deployment package.

Answer

Use a different model that is smaller than 10 GB

Answer

Increase the model deployment artifact size limit via a service request

Answer

Compress the model artifact to under 10 GB using gzip

Question 5

Which TWO actions are recommended best practices for managing costs when using OCI Generative AI dedicated AI clusters?

Accepted Answer

Use preemptible instances for non-critical inference workloads. Option B is correct because preemptible instances cost significantly less than on-demand capacity and are well suited to non-critical inference workloads that can tolerate interruption, helping reduce overall spend on dedicated AI clusters. Option C is correct because autoscaling dynamically adjusts the number of nodes in the dedicated AI cluster to match actual demand, so you avoid paying for idle capacity during low-traffic periods while still scaling up when needed. Option A is not recommended because provisioning a fixed number of nodes sized for peak load means paying for maximum capacity even when utilization is low. Option D is not a valid cost practice for dedicated AI clusters, since clusters are provisioned capacity rather than a resource you simply stop and start like a compute instance. Option E is incorrect because preemptible instances are a lower-cost option, and pay-as-you-go billing is not a substitute that reduces dedicated AI cluster costs in the way described.

Answer

Provision a fixed number of nodes to handle peak load

Answer

Stop the dedicated AI cluster when not in use

Answer

Use pay-as-you-go billing instead of preemptible instances

Question 6

An organization wants to fine-tune a large language model on OCI using their proprietary data. They are concerned about data privacy and want to ensure that fine-tuning data does not leave the OCI region. Which OCI service should they use to securely store and manage their training data?

Accepted Answer

OCI Object Storage. C is correct because OCI Object Storage is a regional service that stores data within a specific OCI region, ensuring that fine-tuning data does not leave that region. It provides secure, durable, and scalable storage for large datasets, such as training data for LLMs, with encryption at rest and in transit, and supports direct integration with OCI Data Science and Generative AI services for fine-tuning workflows.

Answer

OCI Block Volume

Answer

OCI File Storage

Answer

Oracle Autonomous Database

Question 7

A data scientist is deploying a custom generative AI model using OCI Data Science. After deploying the model to an endpoint, they notice that inference requests are failing with a timeout error when the payload size exceeds 1 MB. What is the most likely cause and solution?

Accepted Answer

The model deployment has a default payload size limit of ~1 MB; increase the payload limit in the deployment configuration.. OCI Data Science model deployments have a default payload size limit of approximately 1 MB. When inference requests exceed this limit, the load balancer or gateway times out the request. The solution is to increase the payload limit in the deployment configuration, which can be adjusted via the OCI console or API by modifying the `maximumRequestPayloadSize` setting.

Answer

The load balancer is misconfigured; reconfigure the load balancer timeout settings.

Answer

The model server lacks sufficient memory; scale out to more instances.

Answer

The model is not optimized for large payloads; use AutoML to optimize the model.

Question 8

A team is deploying a generative AI model using OCI Functions for serverless inference. They are experiencing cold start latency of over 10 seconds for the first invocation after idle periods. What is the best strategy to reduce cold start latency?

Accepted Answer

Use provisioned concurrency to keep a set number of function instances warm.. Provisioned concurrency in OCI Functions pre-initializes a specified number of function instances and keeps them warm, so the first invocation after an idle period does not trigger a cold start. This directly eliminates the container initialization and runtime bootstrap overhead that causes the 10+ second latency. It is the only option that addresses the root cause of cold starts rather than attempting to mask or shift the problem.

Answer

Migrate the inference to OCI Data Flow for better performance.

Answer

Reduce the function timeout to force faster execution.

Answer

Increase the memory allocation for the function.

Question 9

A company is deploying a large generative AI model on OCI using GPU compute instances. They want to optimize inference cost while maintaining acceptable latency. Which TWO strategies should they implement?

Accepted Answer

Select the smallest GPU instance type that meets latency requirements.. Option B is correct because right-sizing the GPU instance to the smallest type that still satisfies the latency SLA avoids paying for excess GPU capacity, directly reducing inference cost while keeping acceptable latency. Option E is correct because an inference endpoint with auto-scaling dynamically adjusts the number of replicas to match actual demand, so you avoid over-provisioning during low traffic and scale up only when needed, optimizing cost without violating latency targets. Option A is not appropriate because provisioned concurrency keeps capacity always warm and billed even when idle, which raises cost rather than optimizing it. Option C is wrong because increasing max-tokens produces longer outputs, consuming more compute and memory bandwidth per request and increasing latency and cost. Option D is wrong because deploying on multiple large GPU instances over-provisions fixed capacity for peak load, wasting money during non-peak periods instead of scaling elastically.

Answer

Enable provisioned concurrency on all models.

Answer

Increase the max-tokens parameter to generate longer responses.

Answer

Deploy the model on multiple large GPU instances to handle peak load.

Question 10

A global enterprise is deploying a generative AI application that requires high availability across multiple OCI regions. The application must automatically fail over to a secondary region if the primary region becomes unavailable. What is the recommended architecture to achieve this?

Accepted Answer

Deploy endpoints in two regions behind an OCI Load Balancer with cross-region failover. OCI Load Balancer supports cross-region failover by distributing traffic across backend sets in multiple regions, enabling automatic failover to a secondary region when the primary region becomes unavailable. This architecture ensures high availability for generative AI applications by leveraging health checks and failover policies at the load balancer level, which is the recommended approach for multi-region active-passive setups.

Answer

Deploy OCI Generative AI endpoints in two regions and use a global DNS round-robin

Answer

Use OCI Streaming to replicate requests between regions

Answer

Use DNS failover with a single endpoint in the primary region

Question 11

A financial services company is concerned about data privacy when using OCI Generative AI service for processing sensitive customer data. They want to ensure that their data is not used to improve the model and is encrypted at rest and in transit. Which combination of OCI features should they implement?

Accepted Answer

Provision a dedicated endpoint, configure data privacy opt-out, and use OCI Vault for encryption keys. It addresses all three requirements: a dedicated endpoint ensures network isolation and encryption in transit, data privacy opt-out prevents OCI from using customer data for model improvement, and OCI Vault integration allows customers to manage their own encryption keys for data at rest, meeting the financial services company's strict data privacy and encryption needs.

Answer

Use OCI Object Storage buckets with encryption to store prompts and responses

Answer

Deploy the model in an OCI Data Science project with a private endpoint

Answer

Use the on-demand API with default encryption

Question 12

Which TWO factors should be considered when selecting a base model for fine-tuning on OCI Generative AI service?

Accepted Answer

The model's size and number of parameters. Option B is correct because the model's size and parameter count directly affect fine-tuning feasibility on OCI Generative AI — larger models demand more compute, memory, and time, and OCI imposes limits on which base models can be fine-tuned and with what dedicated AI cluster capacity. Option C is correct because the model's license and terms of use govern whether you may legally fine-tune, modify, and commercially deploy the resulting custom model, which is a mandatory consideration before selecting a base model. Option A is not a selection factor because the base model's original training dataset size is not exposed or adjustable during fine-tuning and does not determine fine-tuning suitability. Option D is not relevant because OCI Generative AI manages the training framework internally; customers do not choose PyTorch versus TensorFlow when fine-tuning a hosted base model. Option E is not a primary selection factor because built-in features like content filtering are platform-level guardrails applied regardless of which base model is chosen, not a differentiator for fine-tuning.

Answer

The model's training dataset size

Answer

The model's training framework (PyTorch vs TensorFlow)

Answer

The model's built-in features like content filtering

Question 13

Which TWO methods can be used to invoke a generative AI model deployed on OCI?

Accepted Answer

Sending HTTP requests to the model endpoint URL. Option C is correct because a generative AI model deployed on OCI exposes an inference endpoint URL, and you invoke it by sending HTTP requests (typically REST POST with a JSON payload containing the prompt) to that endpoint. Option E is correct because the OCI SDKs (such as the Python or Java SDK) provide programmatic clients, e.g., the GenerativeAiInferenceClient, that authenticate with OCI credentials and call the same model endpoint, making SDK-based invocation a supported method. Options A and D are incorrect because OCI Notifications and OCI Events are messaging/event-routing services used to publish and react to state changes, not to send inference prompts to a model. Option B is incorrect because the OCI Console is an administrative web UI for managing resources and configurations, not a supported channel for invoking a deployed generative AI model's inference API.

Answer

Using OCI Notifications service

Answer

Using the OCI Console web interface

Answer

Using OCI Events service

Question 14

Given the CLI output from `oci generative-ai model list`, what can be determined about the model 'my-fine-tuned-model'?

Accepted Answer

It was created by fine-tuning an existing base model. The CLI output from `oci generative-ai model list` includes a model named 'my-fine-tuned-model'. In OCI Generative AI, models listed with custom names that are not part of the base model catalog (e.g., cohere.command, meta.llama) indicate they were created by fine-tuning a base model using your own dataset. The presence of a custom name without a base model prefix confirms it is a fine-tuned model, not a pre-built one.

Answer

It is a pre-built model provided by OCI

Answer

It has been deployed to an endpoint

Answer

It is currently being trained

Question 15

A developer receives the above error when trying to send a request to a model endpoint. What is the most likely reason?

Accepted Answer

The endpoint was deleted by an administrator. The error message indicates that the model endpoint is not found. In OCI Generative AI, when an administrator deletes an endpoint, subsequent requests to that endpoint's URL return a 404 Not Found error. This is the most likely reason because the endpoint resource no longer exists in the tenancy, and the request cannot be routed to any model.

Answer

The network connection to OCI is down

Answer

The API key is invalid

Answer

The model is still being deployed