1Z0-1127-25 Practice Test 23 — 15 Questions

Question 1

A developer in the GenAIDevelopers group tries to call the OCI Generative AI inference API but receives an unauthorized error. Which statement best explains the issue?

Accepted Answer

The developer lacks the required permission to 'use generative-ai-inference' to invoke the model.. The OCI Generative AI inference API requires the specific 'use generative-ai-inference' permission in an IAM policy. Without this permission, any API call to invoke a model will result in an unauthorized error, regardless of other permissions the developer might have. The error indicates the developer's group policy does not grant the necessary verb for the inference resource.

Answer

The developer does not have permission to use the generative-ai-model-family.

Answer

The policy does not allow any operations in the compartment.

Answer

The compartment name in the policy is incorrect.

Question 2

A developer is tuning parameters for a text-generation model and wants to reduce the likelihood of the same phrase appearing repeatedly. Which parameter should be adjusted?

Accepted Answer

Frequency penalty. Frequency penalty reduces the model's tendency to repeat tokens or phrases by penalizing tokens that have already appeared in the generated text.

Answer

Top-k

Answer

Stop sequences

Answer

Presence penalty

Question 3

An architect needs to ensure that an LLM deployed in OCI does not reveal sensitive information in its outputs. Which technique should be used?

Accepted Answer

Output filtering via custom inference wrapper. Output filtering via a custom inference wrapper allows the architect to inspect and sanitize the model's generated text before it reaches the user, preventing the leakage of sensitive information such as PII, credentials, or internal data. This technique operates at the application layer, intercepting the LLM's response and applying rules or regex patterns to redact or block prohibited content, which is essential for compliance and data security in production deployments.

Answer

Limiting max tokens

Answer

OCI Data Safe masking

Answer

Input sanitization

Question 4

Which THREE components are required to deploy a custom generative AI model on OCI Data Science model deployment?

Accepted Answer

An inference script (e.g., score.py) to handle prediction requests. Option B is correct because OCI Data Science model deployments require an inference script such as score.py that defines the predict function and handles incoming prediction requests through the model artifact's runtime. Option C is correct because the model artifact (a ZIP or folder containing the serialized model files, dependencies, and the inference script) is the packaged unit that OCI Data Science deploys and serves. Option E is correct because a deployment configuration is required to specify the shape/resources (for example, VM.Standard.E4.Flex with a given OCPU and memory), the environment (conda environment or container image), and other runtime settings for the deployment. Option A is not required because OCI Data Science model deployment provisions its own managed endpoint and load balancing internally, so the user does not supply a load balancer. Option D is not required because authentication to the deployment endpoint is handled through OCI IAM policies, resource principals, or auth tokens, not through an API signing key configured as a deployment component.

Answer

A load balancer to distribute traffic

Answer

An API signing key for authentication

Question 5

A data scientist wants to generate a response that lists top-3 product features in a bulleted format with no introductory text. Which prompt engineering technique should be used?

Accepted Answer

Include a clear task instruction specifying the format. Including a clear task instruction specifying the format is the most direct and reliable way to get a bulleted list with no introductory text. The instruction explicitly tells the model to output only the bullet points in the desired format. This leverages the model's instruction-following ability.

Answer

Set temperature to 0 to ensure deterministic output

Answer

Provide few-shot examples of desired output

Answer

Use chain-of-thought prompting

Question 6

An OCI user wants to reduce the cost of running a generative AI model while maintaining output quality. Which THREE strategies can help achieve this?

Accepted Answer

Use a smaller model from the same family. Using a smaller model from the same family (e.g., switching from Llama 3 70B to Llama 3 8B) reduces the number of parameters and computational resources required per inference, directly lowering cost. Smaller models often retain strong performance on many tasks, especially when the task complexity does not demand the full capacity of the larger model, thus maintaining output quality while reducing token processing costs.

Answer

Use greedy decoding instead of sampling

Answer

Increase max tokens to ensure complete answers

Question 7

Which sampling strategy selects the token with the highest probability at each step, resulting in deterministic and often repetitive outputs?

Accepted Answer

Greedy decoding. Greedy decoding selects the token with the highest probability at each step, making it deterministic and often leading to repetitive outputs because it always chooses the most likely next token without considering future alternatives. This contrasts with stochastic methods that introduce randomness or explore multiple paths.

Answer

Beam search

Answer

Temperature sampling

Answer

Top-k sampling

Question 8

A developer is using the OCI Generative AI SDK to call a custom fine-tuned model. They get an HTTP 404 error. What is the most likely issue?

Accepted Answer

The model endpoint ID is incorrect. An HTTP 404 error indicates that the requested resource was not found on the server. When using the OCI Generative AI SDK to call a custom fine-tuned model, the model endpoint ID is a unique identifier that maps to the deployed model's endpoint. If this ID is incorrect, the API request cannot locate the model, resulting in a 404 error. The other options would typically cause different HTTP status codes, such as 401 for invalid API keys or 400 for missing required parameters.

Answer

The API key is invalid

Answer

The compartment OCID is missing in the request

Answer

The region is not specified

Question 9

A data scientist is using the OCI Generative AI Playground to test a model for a text generation task. They want to control the output to be more focused and avoid repeating the same phrases. Which THREE parameter adjustments should they consider?

Accepted Answer

Increase the presence penalty. Option E (Decrease the temperature) is correct because lowering temperature makes the model's probability distribution sharper, so sampling favors high-probability tokens and produces more focused, deterministic output rather than wandering into tangential text. Option D (Increase the frequency penalty) is correct because it penalizes tokens in proportion to how many times they have already appeared in the generated text, directly discouraging the repetition of the same phrases. Option A (Increase the presence penalty) is correct because it applies a flat penalty to any token that has already appeared at least once, pushing the model toward new topics and wording and thereby reducing repeated phrases. Option B (Increase the temperature) is wrong because raising temperature flattens the distribution and makes output more random and less focused, the opposite of the goal. Option C (Increase the max tokens) is wrong because it only extends the allowed output length and does not influence focus or repetition.

Answer

Increase the temperature

Answer

Increase the max tokens

Question 10

Which THREE steps are required to deploy a custom generative AI model using OCI Data Science Model Deployment?

Accepted Answer

Create a model artifact (e.g., pickle, ONNX) with inference code. Option B is correct because deploying a custom generative AI model in OCI Data Science Model Deployment requires a model artifact containing the serialized model (e.g., pickle, ONNX) along with the inference code (such as a score.py defining the predict function) that the deployment's model server will invoke. Option D is correct because the model artifact must be stored in an OCI Object Storage bucket, and the deployment references it via the artifact URI (for example, oci://bucket@namespace/path/to/artifact) so the service can download and load it. Option E is correct because the actual deployment is created with the OCI Data Science Model Deployment service, which provisions the managed HTTP endpoint and model server that hosts the artifact. Options A and C are not required: fine-tuning via the OCI Generative AI service and registering the model in the OCI Generative AI service are separate workflows for the managed Generative AI offering, not steps for deploying a custom model through OCI Data Science Model Deployment.

Answer

Fine-tune the model using OCI Generative AI service

Answer

Register the model in OCI Generative AI service

Question 11

A team is planning to use OCI Generative AI to summarize large documents. They need to choose between on-demand (pay-as-you-go) and dedicated cluster pricing. Which THREE factors should they consider when deciding? (Choose three.)

Accepted Answer

Throughput requirements and consistency. Option A (Throughput requirements and consistency) is correct because dedicated AI clusters provide reserved, guaranteed capacity (measured in dedicated AI cluster units) that delivers consistent throughput, whereas on-demand inference is subject to shared-capacity limits and throttling, so high or steady throughput needs favor a dedicated cluster. Option B (Latency sensitivity of the application) is correct because latency-sensitive summarization workloads benefit from the dedicated cluster's provisioned, non-shared inference endpoints, which avoid the queuing variability of on-demand endpoints. Option E (Cost predictability and budget constraints) is correct because dedicated clusters are billed on a committed hourly/period basis (predictable, fixed cost) while on-demand is pay-per-token (variable), so budget forecasting drives the choice. Option C (Model accuracy on summarization tasks) does not belong because accuracy is a function of the chosen foundation model and prompting, not the pricing/consumption model. Option D (Data residency requirements) does not belong because residency is determined by the region and endpoint you select, not by whether you use on-demand or dedicated cluster pricing.

Answer

Model accuracy on summarization tasks

Answer

Data residency requirements

Question 12

Which THREE factors should be considered when designing a chunking strategy for a RAG application?

Accepted Answer

Desired granularity of retrieval. Option A (Desired granularity of retrieval) is correct because chunk size directly determines whether retrieval returns broad context or precise, focused passages, so the target granularity must drive the chunking design. Option D (Document structure) is correct because chunk boundaries should respect natural divisions such as headings, sections, paragraphs, or code blocks, so the splitter must adapt to the source format rather than cutting blindly at fixed lengths. Option E (Embedding model's maximum input tokens) is correct because every chunk must fit within the embedding model's token limit; exceeding it causes truncation or errors, so chunk size must be bounded by that maximum. Option B (Number of GPUs available) does not belong because GPU count affects training or inference throughput, not the logical chunking strategy. Option C (Database indexing method) does not belong because the vector index type (for example HNSW or IVF) is chosen after chunking and concerns search performance, not how documents are split.

Answer

Number of GPUs available

Answer

Database indexing method

Question 13

A security administrator wrote the above IAM policy for a compartment named MyCompartment. Users in the GenerativeAIUsers group can successfully list dedicated AI clusters and models in MyCompartment, but when they try to create an inference endpoint using a model from a different compartment (SharedModels), they get an authorization error. What is the most likely missing policy statement?

Accepted Answer

ALLOW GROUP GenerativeAIUsers TO USE generative-ai-models IN COMPARTMENT SharedModels. The error occurs because the user has permission to list models in MyCompartment but not to use a model from SharedModels when creating an inference endpoint. The missing policy must grant the USE permission on generative-ai-models in the SharedModels compartment, as creating an endpoint requires the ability to reference and use the model resource from that compartment. Option D correctly provides this permission.

Answer

ALLOW GROUP GenerativeAIUsers TO MANAGE generative-ai-models IN COMPARTMENT SharedModels

Answer

ALLOW GROUP GenerativeAIUsers TO USE generative-ai-family IN TENANCY

Answer

ALLOW GROUP GenerativeAIUsers TO USE generative-ai-dedicated-ai-clusters IN COMPARTMENT SharedModels

Question 14

A company is deploying a chatbot powered by OCI Generative AI. They want to inject the conversation history into the model prompt to maintain context. However, they notice that after a long conversation, the model starts to ignore earlier messages. What is the most likely cause?

Accepted Answer

The model has a limited context window size.. The model's context window size limits the total number of tokens (input + output) it can process at once. When the conversation history grows beyond this limit, older messages are truncated or dropped, causing the model to lose context from earlier parts of the conversation. This is a fundamental constraint of transformer-based models like those used in OCI Generative AI.

Answer

The model's max_tokens limit is too low, truncating the prompt.

Answer

The top_p parameter is set to 1, causing deterministic output.

Answer

The temperature setting is too high, causing randomness.

Question 15

Which three statements about transformer architecture are correct? (Choose three.)

Accepted Answer

The softmax function is used in the attention mechanism to normalize attention scores.. Option A is correct because in scaled dot-product attention the raw compatibility scores (QK^T / sqrt(d_k)) are passed through a softmax along the key dimension, converting them into a probability distribution that sums to 1 so the weighted sum of values is properly normalized. Option C is correct because transformers contain no recurrence or convolution, so without positional encodings (sinusoidal or learned) added to the input embeddings, the model would be permutation-invariant and could not distinguish token order. Option D is correct because the self-attention layer computes pairwise query-key similarities and uses them as weights over the value vectors, letting each token aggregate information from all other tokens according to learned relevance. Option B is not correct because the position-wise feed-forward network shares the same weight matrices across all token positions, applying an identical MLP to each position independently. Option E is not correct because GPT models are decoder-only architectures that use masked self-attention, whereas the encoder-decoder structure is characteristic of the original transformer and models like T5 or BART.

Answer

The feed-forward network applies a different set of weights for each token position.

Answer

The encoder-decoder structure is used in GPT models.