1Z0-1127-25 Prompt Engineering Practice Test 9 — 15 Questions

Question 1

Which TWO are best practices for prompt management in production environments?

Accepted Answer

Maintain a prompt library with reusable templates. Option B is correct because maintaining a prompt library with reusable templates lets teams standardize, share, and iterate on prompts across applications, reducing duplication and drift in production. Option C is correct because storing prompts in a version-controlled repository provides change history, peer review, rollback capability, and traceability, which are essential for auditing and safely evolving prompts in production. Option A is not a best practice because system prompts are a key mechanism for defining model behavior, safety constraints, and role, so avoiding them weakens control rather than simplifying management. Option D is wrong because hard-coded strings make prompts difficult to update, test, and version independently of code deployments. Option E is wrong because different use cases require tailored prompts; reusing one prompt everywhere degrades quality and flexibility.

Answer

Avoid using system prompts to keep prompts simple

Answer

Keep all prompts as hard-coded strings in the application code

Answer

Use the same prompt for all use cases to reduce complexity

Question 2

A financial services firm uses an OCI Generative AI Cohere Command model to summarize quarterly earnings transcripts. Analysts report that summaries are accurate but frequently 900+ words when the downstream report template only allows 250 words. The team's prompt currently reads: 'Summarize the following transcript.' Which prompt change MOST reliably constrains output length while preserving the key financial figures?

Accepted Answer

Add 'Limit the summary to approximately 250 words, and prioritize revenue, EPS, and guidance figures.' and set max_tokens to a value that caps generation.. Combining an explicit word-count instruction with the max_tokens generation parameter is the reliable way to bound summary length while directing the model to keep high-value content. Instructions alone can be ignored; the parameter alone can truncate mid-thought. Together they enforce the limit and preserve the financial figures the analysts need.

Answer

Add a stop sequence containing the word 'summary' so generation halts after the first paragraph.

Answer

Increase temperature to 1.0 so the model varies its output and naturally produces shorter summaries.

Answer

Switch to a larger context window model so the transcript fits entirely and the model self-compresses the content.

Question 3

A developer is prototyping an OCI Generative AI chatbot that must always respond in valid JSON with keys 'intent' and 'confidence'. The model sometimes wraps the JSON in markdown code fences or adds a conversational preamble. Which prompt engineering technique MOST directly prevents this formatting drift?

Accepted Answer

Provide a strict output schema instruction, include a one-shot example of the exact JSON with no preamble or code fences, and set temperature low.. A strict schema instruction paired with a one-shot example of the exact desired output and a low temperature setting is the most direct way to eliminate markdown fences and preamble text. The example demonstrates the format, the instruction forbids deviations, and low temperature reduces the randomness that produces formatting variance.

Answer

Set the stop sequence to '```' so the model stops before emitting a code fence.

Answer

Increase the model's context window so the full schema description fits in the prompt.

Answer

Provide a single clear instruction such as 'Please return JSON.'

Question 4

A developer is using the OCI Generative AI Playground with a Meta Llama 3.1 70B model to draft marketing copy. They need the same prompt to yield noticeably different creative variations each time it is submitted, while still remaining coherent. Which parameter change should they make first?

Accepted Answer

Increase the temperature value in the model parameters.. Temperature directly controls sampling randomness by rescaling the model's probability distribution before a token is chosen. A higher temperature flattens that distribution so less probable words become more likely, which yields diverse creative drafts from an identical prompt. Length limits, repetition penalties, and candidate-pool truncation affect other behaviours and do not drive variation across runs.

Answer

Increase the frequency penalty value in the model parameters.

Answer

Increase the max tokens value in the model parameters.

Answer

Increase the top k value in the model parameters.

Question 5

A support automation team feeds 40 pages of product documentation into the prompt of an OCI Generative AI Cohere Command R model and asks a single question about a clause buried on page 32. The model answers confidently but incorrectly. The team wants to keep using the same model and improve accuracy without changing the underlying documents. Which change is MOST appropriate?

Accepted Answer

Retrieve only the most relevant passages and place them near the end of the prompt before the question.. Accuracy with long inputs improves when the relevant evidence is retrieved and positioned close to the query, countering the lost-in-the-middle attention pattern. Retrieval plus strategic placement gives the model a short reasoning path to the grounded clause. Decoding parameters and polite instructions about reading carefully do not change how attention is distributed across a long context.

Answer

Raise the max tokens value so the model has room to reason through all 40 pages before answering.

Answer

Increase the temperature so the model explores a wider set of interpretations of the documentation.

Answer

Add a system instruction telling the model to read the entire documentation carefully before answering.

Question 6

A team is building a financial summarisation assistant on OCI Generative AI. They want the model to output a fixed set of fields every time, and they are choosing between few-shot examples and a schema description in the prompt. Which TWO practices reliably improve adherence to the required output structure? (Choose two.)

Accepted Answer

Describe each required field, its data type, and whether it is mandatory in the instruction text.. Structured output adherence improves when the model receives both a specification and a template. Describing each field, its type, and its mandatory status defines the contract, while two or three complete examples show the model exactly how that contract looks in practice. Randomness and larger output budgets undermine consistency rather than strengthen it.

Answer

Set the temperature to its maximum value so the model can explore many possible field layouts.

Answer

Remove all examples and rely solely on the model's pre-trained knowledge of financial summaries.

Answer

Increase the maximum output tokens dramatically to give the model more room for extra fields.

Question 7

A support team using OCI Generative AI with a Meta Llama chat model finds that short user questions like 'How do I reset my password?' produce vague answers that skip the actual reset steps. They want to improve answer completeness without changing the model or the retrieval documents. Which prompt engineering change is BEST?

Accepted Answer

Instruct the model in the system prompt to always answer with numbered step-by-step procedures when the user asks a procedural question.. The most reliable fix is to change the prompt so the model is explicitly told how to answer procedural questions. System-prompt instructions shape output structure and content selection far more than sampling parameters or input trimming. Because the team cannot change the model or documents, prompt-level guidance is the only lever that directly targets the missing step-by-step detail.

Answer

Shorten the user prompt to a single keyword like 'password reset' before sending it to the model.

Answer

Increase the max_tokens parameter so the model has more room to write a longer answer.

Answer

Lower the temperature to 0 so the model produces deterministic answers.

Question 8

A data engineer is building a classification pipeline on OCI Generative AI using a Meta Llama chat model. They need the model to always respond with exactly one of four labels: BILLING, TECHNICAL, ACCOUNT, or OTHER. During testing, the model sometimes returns sentences like 'This seems like a billing issue.' Which prompt engineering technique is MOST reliable for forcing a single-token label output?

Accepted Answer

Add a system instruction telling the model to answer only with the label, and end the prompt with 'Label:' so the model continues with the label token.. Constraining output to a fixed label set is best achieved by combining a system-level instruction with a primed assistant prefix such as 'Label:'. The prefix acts as a completion cue, so the model continues with the label token instead of a full sentence. Sampling tweaks, token limits, and streaming do not enforce vocabulary constraints.

Answer

Enable streaming responses so the model emits the label first and the rest of the sentence can be ignored.

Answer

Increase the top_p value to 0.99 so the model samples more broadly and eventually picks the correct label.

Answer

Set the max_tokens parameter to 1 so only the first token of the response is returned.

Question 9

A developer is iterating on a prompt for an OCI Generative AI model used to classify customer emails into five categories. The prompt currently includes a long narrative describing the company's history. The accuracy is inconsistent. Which change is the MOST direct improvement to prompt clarity?

Accepted Answer

Replace the narrative with a concise task instruction, the five category definitions, and one labelled example per category.. Classification quality hinges on unambiguous task instructions and discriminative category definitions backed by representative examples. Irrelevant narrative competes for attention and blurs the decision boundaries. Removing filler and replacing it with a crisp instruction, five definitions, and one labelled example per class gives the model the signal it needs to assign labels consistently.

Answer

Append the company history again at the end of the prompt so the model sees it twice.

Answer

Increase the top p value so the model samples from a broader set of category labels.

Answer

Add a persona statement that the model is a helpful assistant before the narrative.

Question 10

A developer is building a document summarization feature in OCI Generative AI using a Cohere Command model. They need the summary to always be exactly three bullet points, each under 20 words. Which prompt engineering technique is MOST appropriate?

Accepted Answer

Add explicit formatting constraints to the prompt, such as 'Return exactly three bullet points, each no longer than 20 words.'. Explicitly stating the required number of bullets and the per-bullet word limit in the prompt is the most reliable way to control summary structure with a Cohere Command model on OCI. Formatting constraints work best when they are concrete, numeric, and stated alongside the task instruction, because the model can verify its own output against clear rules during generation.

Answer

Set the stop sequence to a newline character so the model stops after each bullet.

Answer

Split the document into chunks and summarize each chunk separately, then concatenate the results.

Answer

Raise the top_p value to 1.0 to encourage the model to explore more phrasing options.

Question 11

A solutions architect is configuring a generation request against an OCI Generative AI Cohere Command model. The team wants short, deterministic answers for a FAQ bot where the same question should always produce the same response. Which parameter change BEST supports this goal?

Accepted Answer

Set temperature to 0 and lower top_p to a small value such as 0.1.. Deterministic, repeatable generation is achieved by suppressing sampling randomness. Temperature 0 forces greedy decoding, and a small top_p narrows the candidate set further. Penalties, length caps, and higher temperatures all either preserve or increase variability, so they work against the FAQ bot's requirement.

Answer

Increase max_tokens so the model has more room to settle on a stable answer.

Answer

Set the frequency penalty to a high value so repeated tokens are discouraged.

Answer

Increase temperature to 1.0 and keep top_p at the default.

Question 12

A financial services team uses an OCI Generative AI Cohere Command model to summarize quarterly earnings call transcripts. They want summaries that stay strictly grounded in the transcript and refuse to speculate on future revenue. Which prompt engineering technique should they apply to BEST achieve this?

Accepted Answer

Use a system prompt that instructs the model to answer only from the provided transcript and to respond "not stated" if information is absent.. Grounding a model in a specific source document is achieved through explicit prompt instructions that define the allowed evidence and a fallback behavior when the answer is missing. A system prompt scoping the model to the transcript and requiring a "not stated" response prevents speculation. Sampling parameters like temperature, top_k, and max_tokens affect randomness and length, not factual grounding.

Answer

Lower the top_k value to 1 so the model always picks the single most probable token.

Answer

Set the temperature parameter to 1.0 and add "be creative" to the prompt.

Answer

Increase the max_tokens parameter to the model's maximum so the summary can include every detail.

Question 13

A team is deploying a retrieval-augmented generation (RAG) pipeline on OCI Generative AI. They notice the model sometimes answers using its own prior knowledge instead of the retrieved passages, producing plausible but uncited claims. They want the model to ground every statement in the provided context. Which prompt engineering change is BEST?

Accepted Answer

Add an instruction that the model must answer only from the provided context and must say 'not found in context' if the answer is absent.. Grounding failures in RAG are best addressed by instructing the model to answer strictly from the retrieved context and to indicate when the answer is not present. This constrains the model's generation to the supplied evidence and produces a detectable fallback phrase. Parameter tuning and context trimming do not change the model's tendency to blend prior knowledge with retrieved facts.

Answer

Increase the temperature so the model produces more diverse answers.

Answer

Shorten the retrieved passages by truncating each to 100 tokens before inserting them into the prompt.

Answer

Reduce the number of retrieved passages to one so the model has less to read.

Question 14

A team is building a retrieval-augmented generation (RAG) assistant on OCI Generative AI. Users report that answers sometimes cite facts not present in the retrieved context, a behavior known as hallucination. Which TWO prompt engineering changes are MOST likely to reduce unsupported claims? (Choose two.)

Accepted Answer

Add a system instruction stating that the model must answer only from the provided context and must reply 'I do not know' when the context is insufficient.. Reducing hallucination in RAG requires two things: an explicit grounding instruction with a safe fallback, and context presented with delimiters plus a citation requirement. Together they constrain the model to the retrieved evidence and make gaps visible. Sampling changes, longer outputs, and removing context all increase the risk of unsupported claims.

Answer

Raise the temperature to 0.9 so the model explores more of the context before answering.

Answer

Remove the retrieved context and rely on the model's pretrained knowledge to answer the question.

Answer

Increase the model's max_tokens so it has more space to explain its reasoning before answering.

Question 15

A developer is building a customer support assistant on OCI Generative AI using a Meta Llama model. They want the assistant to always answer in the persona of a calm, empathetic agent and to never discuss competitor products. They notice that adding these rules to the user message works inconsistently across sessions. What should they do to make the behavior reliable?

Accepted Answer

Move the persona and competitor-exclusion rules into the system prompt so they apply to every turn of the conversation.. Persistent behavioral rules such as persona, tone, and topic exclusions belong in the system prompt, which the model treats as standing instructions across all turns. User messages are task-level input and can be overridden or diluted. Sampling parameters and stop sequences affect randomness and truncation, not policy adherence.

Answer

Repeat the rules at the end of every user message to reinforce them.

Answer

Add the rules to a stop sequence list so the model halts before mentioning competitors.

Answer

Increase the temperature so the model has more freedom to express empathy.