, WGU NUTRITION D440; COMPLETE STUDY EXAM
WITH ALL CHAPTERS COVERED FOR BETTER
REVISION WITH NEW UPDATES OF 2026 | GRADED
A+ | GUARANTEED PASS | COMBINED WITH
REVISION MATERIALS
Question
What is the primary purpose of using "temperature" in conjunction with "top_p" in a
language model's generation settings?
A) They serve the same function and using both is redundant
B) Temperature controls the shape of the probability distribution, while top_p controls the
size of the sampling pool, together providing fine-grained control over output diversity
C) Temperature controls the size of the sampling pool, while top_p controls the shape of
the distribution
D) They are both used to control output length
Answer: B) Temperature controls the shape of the probability distribution, while top_p
controls the size of the sampling pool, together providing fine-grained control over output
diversity
Expert Rationale: Temperature and topp are complementary parameters that work
together to shape the generation process. Temperature scales the logits before the softmax
function, effectively "flattening" or "sharpening" the probability distribution—higher
temperatures make less probable tokens more likely, while lower temperatures make the
distribution more peaked around the most probable tokens. Topp (nucleus sampling) then
selects the smallest set of tokens whose cumulative probability exceeds the threshold p,
dynamically determining the sampling pool size. Together, they provide nuanced control:
temperature affects the relative probabilities, and top_p filters out the long tail of
improbable tokens. In WGU D685, students learn that these parameters are not redundant
but synergistic. Option A is incorrect because they serve different functions. Option C
reverses their roles. Option D is incorrect because neither parameter directly controls
output length.
Question
A prompt engineer is optimizing a prompt for a customer service chatbot. The model
,sometimes responds with overly technical jargon that confuses users. What is the most
effective intervention?
A) Increase the temperature to encourage more varied and simpler language
B) Add a system instruction specifying the target audience's reading level, such as "Use a
6th-grade reading level and avoid technical jargon"
C) Decrease the max tokens to force shorter, simpler responses
D) Increase the frequency penalty to reduce repetition of technical terms
Answer: B) Add a system instruction specifying the target audience's reading level, such as
"Use a 6th-grade reading level and avoid technical jargon"
Expert Rationale: The most direct and effective way to control the complexity of language
is through explicit instructions in the system prompt. By specifying the desired reading level
and explicitly prohibiting technical jargon, the engineer provides clear behavioral guidance.
This is more reliable than indirect methods. In WGU D685, audience adaptation is a core
principle—prompts should be tailored to the end-user's needs and capabilities. Increasing
temperature (A) would increase randomness, not simplify language. Decreasing max
tokens (C) might truncate responses but does not address vocabulary complexity.
Frequency penalty (D) addresses repetition, not terminology choice. Explicit reading level
instructions are the gold standard.
Question
What is the role of "semantic search" in a Retrieval-Augmented Generation (RAG) pipeline?
A) To rank and retrieve the most semantically relevant documents from a knowledge base
based on the user's query
B) To generate the final response to the user
C) To encrypt the retrieved documents for security
D) To translate the user's query into multiple languages
Answer: A) To rank and retrieve the most semantically relevant documents from a
knowledge base based on the user's query
Expert Rationale: Semantic search is the retrieval engine of a RAG pipeline. It uses
embeddings—numerical vector representations of text—to compute the semantic
similarity between the user's query and documents in a knowledge base. The documents
with the highest similarity scores are retrieved and provided to the language model as
context for grounding the response. This ensures that the model has access to the most
relevant information. In WGU D685, students learn that the quality of semantic retrieval
, directly impacts the quality of the model's output. Option B describes the generation step,
not retrieval. Option C is incorrect. Option D is incorrect. Semantic search is about
relevance ranking, not translation or encryption.
Question
What is the significance of "prompt engineering" in the context of AI bias mitigation?
A) Prompt engineering can introduce biases but cannot mitigate them
B) Prompt engineering is a primary tool for mitigating bias by instructing the model to
consider multiple perspectives, avoid stereotypes, and produce balanced outputs
C) Bias mitigation is solely the responsibility of the training data, not prompt engineering
D) Prompt engineering has no effect on bias
Answer: B) Prompt engineering is a primary tool for mitigating bias by instructing the model
to consider multiple perspectives, avoid stereotypes, and produce balanced outputs
Expert Rationale: While training data and model architecture are significant sources of
bias, prompt engineering is a powerful and practical tool for mitigation. By including explicit
instructions—such as "Avoid stereotypes," "Consider multiple perspectives," "Present
balanced viewpoints," or "Do not make assumptions based on gender, race, or ethnicity"—
engineers can guide the model toward more fair and equitable outputs. In WGU D685, bias
mitigation is a core ethical responsibility. Option A is incorrect; prompt engineering can
both introduce and mitigate bias. Option C is incorrect; prompt engineering is a key
mitigation layer. Option D is false. Ethical prompt design actively reduces bias.
Question
What is the purpose of "response truncation" in a quality control pipeline?
A) To cut off the model's response at a specified token limit to prevent overly long outputs
B) To increase the model's creativity by truncating responses
C) To encrypt the response for security
D) To translate the response into another language
Answer: A) To cut off the model's response at a specified token limit to prevent overly long
outputs
Expert Rationale: Response truncation is a quality control mechanism where the model's
output is programmatically cut off at a predefined token limit. This prevents excessively
long outputs that may be costly, irrelevant, or difficult to parse. In WGU D685, students
WITH ALL CHAPTERS COVERED FOR BETTER
REVISION WITH NEW UPDATES OF 2026 | GRADED
A+ | GUARANTEED PASS | COMBINED WITH
REVISION MATERIALS
Question
What is the primary purpose of using "temperature" in conjunction with "top_p" in a
language model's generation settings?
A) They serve the same function and using both is redundant
B) Temperature controls the shape of the probability distribution, while top_p controls the
size of the sampling pool, together providing fine-grained control over output diversity
C) Temperature controls the size of the sampling pool, while top_p controls the shape of
the distribution
D) They are both used to control output length
Answer: B) Temperature controls the shape of the probability distribution, while top_p
controls the size of the sampling pool, together providing fine-grained control over output
diversity
Expert Rationale: Temperature and topp are complementary parameters that work
together to shape the generation process. Temperature scales the logits before the softmax
function, effectively "flattening" or "sharpening" the probability distribution—higher
temperatures make less probable tokens more likely, while lower temperatures make the
distribution more peaked around the most probable tokens. Topp (nucleus sampling) then
selects the smallest set of tokens whose cumulative probability exceeds the threshold p,
dynamically determining the sampling pool size. Together, they provide nuanced control:
temperature affects the relative probabilities, and top_p filters out the long tail of
improbable tokens. In WGU D685, students learn that these parameters are not redundant
but synergistic. Option A is incorrect because they serve different functions. Option C
reverses their roles. Option D is incorrect because neither parameter directly controls
output length.
Question
A prompt engineer is optimizing a prompt for a customer service chatbot. The model
,sometimes responds with overly technical jargon that confuses users. What is the most
effective intervention?
A) Increase the temperature to encourage more varied and simpler language
B) Add a system instruction specifying the target audience's reading level, such as "Use a
6th-grade reading level and avoid technical jargon"
C) Decrease the max tokens to force shorter, simpler responses
D) Increase the frequency penalty to reduce repetition of technical terms
Answer: B) Add a system instruction specifying the target audience's reading level, such as
"Use a 6th-grade reading level and avoid technical jargon"
Expert Rationale: The most direct and effective way to control the complexity of language
is through explicit instructions in the system prompt. By specifying the desired reading level
and explicitly prohibiting technical jargon, the engineer provides clear behavioral guidance.
This is more reliable than indirect methods. In WGU D685, audience adaptation is a core
principle—prompts should be tailored to the end-user's needs and capabilities. Increasing
temperature (A) would increase randomness, not simplify language. Decreasing max
tokens (C) might truncate responses but does not address vocabulary complexity.
Frequency penalty (D) addresses repetition, not terminology choice. Explicit reading level
instructions are the gold standard.
Question
What is the role of "semantic search" in a Retrieval-Augmented Generation (RAG) pipeline?
A) To rank and retrieve the most semantically relevant documents from a knowledge base
based on the user's query
B) To generate the final response to the user
C) To encrypt the retrieved documents for security
D) To translate the user's query into multiple languages
Answer: A) To rank and retrieve the most semantically relevant documents from a
knowledge base based on the user's query
Expert Rationale: Semantic search is the retrieval engine of a RAG pipeline. It uses
embeddings—numerical vector representations of text—to compute the semantic
similarity between the user's query and documents in a knowledge base. The documents
with the highest similarity scores are retrieved and provided to the language model as
context for grounding the response. This ensures that the model has access to the most
relevant information. In WGU D685, students learn that the quality of semantic retrieval
, directly impacts the quality of the model's output. Option B describes the generation step,
not retrieval. Option C is incorrect. Option D is incorrect. Semantic search is about
relevance ranking, not translation or encryption.
Question
What is the significance of "prompt engineering" in the context of AI bias mitigation?
A) Prompt engineering can introduce biases but cannot mitigate them
B) Prompt engineering is a primary tool for mitigating bias by instructing the model to
consider multiple perspectives, avoid stereotypes, and produce balanced outputs
C) Bias mitigation is solely the responsibility of the training data, not prompt engineering
D) Prompt engineering has no effect on bias
Answer: B) Prompt engineering is a primary tool for mitigating bias by instructing the model
to consider multiple perspectives, avoid stereotypes, and produce balanced outputs
Expert Rationale: While training data and model architecture are significant sources of
bias, prompt engineering is a powerful and practical tool for mitigation. By including explicit
instructions—such as "Avoid stereotypes," "Consider multiple perspectives," "Present
balanced viewpoints," or "Do not make assumptions based on gender, race, or ethnicity"—
engineers can guide the model toward more fair and equitable outputs. In WGU D685, bias
mitigation is a core ethical responsibility. Option A is incorrect; prompt engineering can
both introduce and mitigate bias. Option C is incorrect; prompt engineering is a key
mitigation layer. Option D is false. Ethical prompt design actively reduces bias.
Question
What is the purpose of "response truncation" in a quality control pipeline?
A) To cut off the model's response at a specified token limit to prevent overly long outputs
B) To increase the model's creativity by truncating responses
C) To encrypt the response for security
D) To translate the response into another language
Answer: A) To cut off the model's response at a specified token limit to prevent overly long
outputs
Expert Rationale: Response truncation is a quality control mechanism where the model's
output is programmatically cut off at a predefined token limit. This prevents excessively
long outputs that may be costly, irrelevant, or difficult to parse. In WGU D685, students
