1Z0-1127-25 Fundamentals of Large Language Models Practice Test 1 — 15 Questions

Question 1

Which TWO factors most significantly influence the computational cost of fine-tuning a large language model?

Accepted Answer

Number of model parameters. Option B (Number of model parameters) is correct because the parameter count determines the size of the weight matrices and gradients, so both the forward pass and backpropagation FLOPs scale roughly linearly with the number of parameters, making it a primary driver of fine-tuning compute. Option C (Maximum sequence length) is correct because self-attention cost scales quadratically with sequence length (O(n²) in attention scores) and activations grow with sequence length, so longer maximum sequences substantially increase memory and compute per step. Option A (Batch size) affects throughput and memory but is a tunable hyperparameter rather than a fundamental cost driver, since total work is roughly proportional to examples processed regardless of how they are batched. Option D (Quantization bits) reduces memory and can speed up inference, but it is an optimization choice rather than a primary determinant of the underlying computational cost. Option E (Dataset size) influences total training time by determining the number of steps, but it does not change the per-step computational cost that the question targets.

Answer

Batch size

Answer

Quantization bits

Answer

Dataset size

Question 2

A data scientist is using a large language model to summarize customer support tickets. The model occasionally generates summaries that include hallucinated details not present in the original ticket. Which technique would best reduce hallucinations while maintaining summary quality?

Accepted Answer

Implement retrieval-augmented generation (RAG) to ground the model in relevant documents.. Retrieval-Augmented Generation (RAG) reduces hallucinations by grounding the model's output in external, verifiable documents retrieved from a knowledge base. Instead of relying solely on the model's parametric memory, RAG fetches relevant context (e.g., the original ticket) at inference time, ensuring the summary is factually aligned with the source. This maintains summary quality because the model can still generate fluent text while being constrained to the retrieved evidence.

Answer

Use a longer system prompt instructing the model to be factual.

Answer

Fine-tune the model on a large corpus of general text to improve its knowledge.

Answer

Increase the temperature parameter to 0.9 to encourage more deterministic outputs.

Question 3

Refer to the exhibit. A developer ran the OCI CLI command shown and received the JSON output. What does the output indicate about the model's confidence and why?

Accepted Answer

The model is highly confident the text is positive, as indicated by the 0.98 score.. The JSON output shows a sentiment score of 0.98 for 'positive', which is very close to 1.0, indicating the model is highly confident that the text is positive. In sentiment analysis models, scores represent probabilities for each class, and a value near 1.0 for one class with much lower scores for others reflects strong confidence.

Answer

The model is uncertain because all scores are roughly equal.

Answer

The model is neutral because the neutral score is lowest.

Answer

The model is unsure because the scores are probabilities that sum to 1.

Question 4

A healthcare startup is building an AI assistant to help doctors draft clinical notes from patient-physician conversations. They have a large language model that is fine-tuned on medical data. During testing, they notice the model occasionally generates plausible-sounding but incorrect medical recommendations. The startup wants to deploy the assistant to assist doctors, not replace them. They have the following options: (A) Deploy the model as-is and rely on doctors to catch errors, (B) Add a disclaimer that the model may make mistakes, (C) Implement a fact-checking pipeline that cross-references outputs with a trusted medical knowledge base before presenting to doctors, (D) Reduce the model's temperature to 0 to ensure deterministic outputs. Which option best balances safety and utility?

Accepted Answer

Implement a fact-checking pipeline that cross-references outputs with a trusted medical knowledge base.. Implementing a fact-checking pipeline that cross-references outputs with a trusted medical knowledge base directly mitigates the risk of hallucinated medical recommendations while preserving the assistant's utility. This approach leverages retrieval-augmented generation (RAG) principles to ground the model's outputs in verified facts, ensuring safety without sacrificing the flexibility needed for drafting clinical notes.

Answer

Add a disclaimer that the model may make mistakes.

Answer

Deploy the model as-is and rely on doctors to catch errors.

Answer

Reduce the model's temperature to 0 to ensure deterministic outputs.

Question 5

A company wants to build a customer support chatbot using OCI Generative AI. They have a large number of historical support tickets. Which approach is most effective for leveraging this data to improve the chatbot's responses?

Accepted Answer

Fine-tune the Cohere Command model on the historical tickets using OCI Data Science.. Fine-tuning the Cohere Command model on the historical support tickets using OCI Data Science is the most effective approach because it adapts the model's weights to the specific domain language, terminology, and resolution patterns found in the company's data. This supervised learning process creates a specialized model that can generate accurate, context-aware responses for customer support queries, unlike generic prompting methods that lack deep domain adaptation.

Answer

Use a pre-loaded prompt template from the OCI console.

Answer

Increase the temperature parameter to 1.0 to encourage diverse responses.

Answer

Use zero-shot prompting with the base model and include few-shot examples in the prompt.

Question 6

A data scientist is evaluating different models for a summarization task. Which two metrics are commonly used to evaluate the quality of generated summaries?

Accepted Answer

ROUGE. ROUGE (C) is correct because it is the standard recall-oriented metric family (ROUGE-N, ROUGE-L, ROUGE-S) that measures n-gram and longest-common-subsequence overlap between a generated summary and reference summaries, which is exactly what summarization evaluation requires. BLEU (E) is correct because it computes precision of n-grams between the generated text and references with a brevity penalty, and it is widely used alongside ROUGE to assess the fluency and fidelity of generated summaries. F1 score (A) is not specific to summarization; it is a general classification metric combining precision and recall over predicted labels, not text overlap. Mean Average Precision (B) is a ranking/retrieval metric that averages precision at relevant ranks, so it does not measure summary quality. Perplexity (D) measures how well a language model predicts a token sequence and is used for model training/intrinsic language modeling, not for comparing generated summaries against references.

Answer

F1 score

Answer

Mean Average Precision

Answer

Perplexity

Question 7

Refer to the exhibit. A developer runs this command and sees that the 'cohere.embed-english-v3.0' model is INACTIVE. What is the most likely cause?

Accepted Answer

The model has been deprecated and is no longer available.. The 'cohere.embed-english-v3.0' model is listed as INACTIVE because Oracle Cloud Infrastructure (OCI) has deprecated it, meaning it is no longer available for inference. When a model is deprecated, its status changes to INACTIVE, and any attempt to invoke it will fail, even if the region, policies, and compartment permissions are correctly configured.

Answer

The model is not supported in the current region.

Answer

The API call lacks the required OCI policy for the model.

Answer

The compartment does not have access to the model.

Question 8

Refer to the exhibit. A developer encounters this error. Which action should they take to resolve the issue?

Accepted Answer

Wait and retry after some time.. The error indicates a rate limit or throttling issue, typically returned by the OCI Generative AI service when the API request quota is exceeded. Waiting and retrying after the cooldown period allows the rate limit to reset, which is the correct resolution for transient throttling errors.

Answer

Change the model to cohere.command-light.

Answer

Increase the max-tokens value.

Answer

Decrease the temperature to 0.0.

Question 9

Refer to the exhibit. A user in group GenAIUsers reports that they cannot call the OCI Generative AI API. What is the most likely issue?

Accepted Answer

The compartment ID in the policy does not match the user's compartment.. The policy is scoped to a specific compartment ID, but the user's compartment does not match that ID. For OCI IAM policies to grant access to resources like the Generative AI API, the policy must be written for the compartment where the resource resides or where the user operates. Since the user is in a different compartment, the policy does not apply, causing the API call to fail.

Answer

The policy statement is missing the 'inspect' verb.

Answer

The policy is in INACTIVE state.

Answer

The user is not in the group GenAIUsers.

Question 10

A startup needs to deploy a large language model for a customer support chatbot that requires low latency and cost efficiency. They are evaluating OCI Generative AI models. Which model type is most appropriate?

Accepted Answer

Instruct model (e.g., cohere.command). The startup requires low latency and cost efficiency for a customer support chatbot. Instruct models like cohere.command are specifically fine-tuned to follow conversational instructions and generate concise, task-oriented responses, making them ideal for interactive chatbot applications. They balance performance and cost better than base models, which lack instruction-following capability, and embedding models, which are designed for semantic search rather than text generation.

Answer

Embedding model (e.g., cohere.embed)

Answer

Image generation model

Answer

Base model (e.g., cohere.base)

Question 11

An AI engineer is testing a large language model on OCI Generative AI and receives this error: 'Token limit exceeded. Maximum context length is 4096 tokens.' The prompt is 4000 tokens long. What is the most effective way to resolve the issue without losing important context?

Accepted Answer

Reduce the prompt length by summarizing or trimming less relevant information.. The error indicates that the combined prompt and generated output exceed the model's maximum context length of 4096 tokens. Since the prompt alone is 4000 tokens, there is very little room for the model to generate a response. Trimming or summarizing less relevant parts of the prompt directly reduces the token count, allowing the model to produce a complete output without exceeding the limit. This approach preserves the most critical context while staying within the model's constraints.

Answer

Switch to a model with a larger context window, if available.

Answer

Increase the max_tokens parameter in the API call.

Answer

Split the prompt into multiple requests and combine outputs.

Question 12

Which TWO factors most directly impact the consistency of text generated by an LLM when the same prompt is used multiple times?

Accepted Answer

Top_p. Top_p (A) is correct because nucleus sampling restricts the token pool to the smallest set whose cumulative probability reaches the top_p threshold, so changing top_p directly changes how much randomness enters token selection and thus how consistent repeated outputs are. Temperature (D) is correct because it scales the logits before the softmax, directly flattening or sharpening the probability distribution and therefore controlling the degree of variation between runs of the same prompt. Batch size (B) affects throughput and memory, not the sampling distribution, so it does not directly change output consistency. Max_tokens (C) only caps the generated length and does not alter the per-token sampling probabilities. Seed (E) can improve reproducibility, but it is not one of the two factors that most directly govern the stochastic sampling behavior itself.

Answer

Batch size

Answer

Max_tokens

Answer

Seed

Question 13

An OCI administrator wants to limit which users can invoke a specific LLM endpoint. Which resource type should be used?

Accepted Answer

IAM policies. IAM policies are the correct resource type because they define authorization rules that control which users, groups, or service principals can invoke specific OCI APIs, including the LLM inference endpoint. By writing a policy with a verb like 'inspect' or 'use' on the resource type 'ai-service-language-family' or a specific model family, the administrator can grant or deny access to the LLM endpoint. This is the standard OCI mechanism for fine-grained access control to AI services.

Answer

OCI Audit

Answer

OCI Vault

Answer

Network security groups

Question 14

Which THREE factors should be considered when choosing between a fine-tuning and a prompt engineering approach?

Accepted Answer

Need for model personalization. Option B is correct because fine-tuning is the stronger choice when the use case demands deep model personalization—adapting weights to a specific domain, tone, or proprietary task—whereas prompt engineering can only steer a frozen model through instructions and examples. Option D is correct because fine-tuning typically requires a substantial volume of high-quality labeled examples for supervised training, while prompt engineering can often succeed with few or zero labeled examples, so the amount of labeled data directly drives the decision. Option E is correct because fine-tuning consumes significant GPU compute for training runs and often for hosting the tuned model, making the available budget a key cost factor, whereas prompt engineering mainly incurs inference token costs. Option A is not among the marked answers because latency is generally an inference-time concern that can be addressed through model selection, caching, or serving configuration rather than being a defining factor between fine-tuning and prompting. Option C is also not marked because foundation model availability in OCI is a platform prerequisite for either approach, not a differentiator that determines which approach to choose.

Answer

Latency requirements

Answer

Availability of foundation model in OCI

Question 15

A machine learning engineer is fine-tuning a model on OCI Data Science and notices that the training loss decreases but then suddenly increases. What is the most likely cause?

Accepted Answer

Reduce model size. The sudden increase in training loss after a period of decrease is often due to the model being too large for the dataset. An oversized model can create a complex loss landscape with erratic gradients, leading to divergence. Reducing the model size simplifies the architecture, resulting in more stable convergence and preventing the loss from spiking.

Answer

Add dropout

Answer

Increase batch size

Answer

Increase learning rate