WGU D685 OBJECTIVE ASSESSMENT 2 NEWEST 2025/ 2026 ACTUAL
EXAM| D685 PRACTICAL APPLICATIONS OF PROMPT OA FINAL REAL
EXAM QUESTIONS AND CORRECT VERIFIED ANSWERS/ ALREADY
GRADED A+ (MOST RECENT!!)
Competency 1: Fundamentals of Prompt Engineering Questions 1-15
Question 1
Which of the following best defines prompt engineering?
A. Training a large language model from scratch using custom datasets
B. Designing and optimizing input text to guide a language models output
C. Evaluating the hardware performance of AI inference systems
D. Writing production deployment code for machine learning models
Answer: B
Rationale: Prompt engineering is the practice of designing, refining, and optimizing input
prompts to achieve desired outputs from pre-trained language models. It does not involve
training models from scratch (A), hardware evaluation (C), or deployment coding (D).
Question 2
A language model receives a prompt and generates a response. The model has not been fine-
tuned on new data. This is an example of:
A. Zero-shot learning
B. Supervised fine-tuning
C. Reinforcement learning from human feedback (RLHF)
D. Model pre-training
Answer: A
, Rationale: Zero-shot learning occurs when a model performs a task without any examples or
additional training, relying solely on its pre-trained knowledge. Fine-tuning (B) and RLHF (C)
involve additional training. Pre-training (D) is the initial training phase.
Question 3
Which parameter directly controls the trade-off between randomness and determinism in a
language models output?
A. Max tokens
B. Top_p
C. Temperature
D. Frequency penalty
Answer: C
Rationale: Temperature scales the logits before the softmax function. Lower temperatures
(approaching 0) make the model more deterministic (always choosing the highest probability
token). Higher temperatures (e.g., 0.8 to 1.5) increase randomness and creativity.
Question 4
A prompt engineer sets the temperature parameter to 0.0. What is the most likely effect on the
model's output?
A. The output will be highly random and creative.
B. The output will be nearly deterministic, always choosing the most likely token.
C. The output will be truncated after 0 tokens.
D. The output will include repeated phrases.
Answer: B
Rationale: Temperature 0.0 makes the model greedy, always selecting the token with the highest
probability. This produces deterministic, consistent, and predictable outputs. Randomness (A)
requires higher temperatures. Max tokens (C) controls length. Repetition (D) is addressed by
frequency penalty.
Question 5
Which of the following is the correct description of the top_p (nucleus sampling) parameter?
EXAM| D685 PRACTICAL APPLICATIONS OF PROMPT OA FINAL REAL
EXAM QUESTIONS AND CORRECT VERIFIED ANSWERS/ ALREADY
GRADED A+ (MOST RECENT!!)
Competency 1: Fundamentals of Prompt Engineering Questions 1-15
Question 1
Which of the following best defines prompt engineering?
A. Training a large language model from scratch using custom datasets
B. Designing and optimizing input text to guide a language models output
C. Evaluating the hardware performance of AI inference systems
D. Writing production deployment code for machine learning models
Answer: B
Rationale: Prompt engineering is the practice of designing, refining, and optimizing input
prompts to achieve desired outputs from pre-trained language models. It does not involve
training models from scratch (A), hardware evaluation (C), or deployment coding (D).
Question 2
A language model receives a prompt and generates a response. The model has not been fine-
tuned on new data. This is an example of:
A. Zero-shot learning
B. Supervised fine-tuning
C. Reinforcement learning from human feedback (RLHF)
D. Model pre-training
Answer: A
, Rationale: Zero-shot learning occurs when a model performs a task without any examples or
additional training, relying solely on its pre-trained knowledge. Fine-tuning (B) and RLHF (C)
involve additional training. Pre-training (D) is the initial training phase.
Question 3
Which parameter directly controls the trade-off between randomness and determinism in a
language models output?
A. Max tokens
B. Top_p
C. Temperature
D. Frequency penalty
Answer: C
Rationale: Temperature scales the logits before the softmax function. Lower temperatures
(approaching 0) make the model more deterministic (always choosing the highest probability
token). Higher temperatures (e.g., 0.8 to 1.5) increase randomness and creativity.
Question 4
A prompt engineer sets the temperature parameter to 0.0. What is the most likely effect on the
model's output?
A. The output will be highly random and creative.
B. The output will be nearly deterministic, always choosing the most likely token.
C. The output will be truncated after 0 tokens.
D. The output will include repeated phrases.
Answer: B
Rationale: Temperature 0.0 makes the model greedy, always selecting the token with the highest
probability. This produces deterministic, consistent, and predictable outputs. Randomness (A)
requires higher temperatures. Max tokens (C) controls length. Repetition (D) is addressed by
frequency penalty.
Question 5
Which of the following is the correct description of the top_p (nucleus sampling) parameter?
