🎓 LLM Final Exam

⏱️ 35 min 📚 Final Exam 🎯 Pass: 70%
1 / 21

Question 1 of 20

What does LLM stand for?

A
Logical Learning Machine
B
Large Language Model
C
Linear Logic Module
D
Layered Learning Method

Question 2 of 20

How does an LLM generate text?

A
It searches a database for pre-written answers
B
It predicts the next word based on the words before it
C
It copies text directly from the internet
D
It asks a human expert for the answer

Question 3 of 20

What is a token in the context of LLMs?

A
A single letter
B
An entire paragraph
C
A word or part of a word the model processes
D
A type of neural network layer

Question 4 of 20

What is self-supervised learning?

A
The model is given correct answers by humans for every input
B
The model learns by predicting missing or next words without human labels
C
The model only trains on labeled datasets
D
The model learns through trial and error in a game

Question 5 of 20

What is the context window of an LLM?

A
The total number of parameters in the model
B
The limit on how many tokens the model can consider at once
C
The speed at which text is processed
D
The number of training examples used

Question 6 of 20

What does "temperature" control in an LLM?

A
How fast the model runs
B
The number of layers in the model
C
How many tokens are in the context window
D
How creative or random the model's output is

Question 7 of 20

What is a prompt?

A
The input text you give to the model to get a response
B
A type of training data format
C
The model's internal memory system
D
A reward signal used during training

Question 8 of 20

What is "hallucination" in LLMs?

A
When the model refuses to answer a question
B
When the model generates false information confidently
C
When the model runs out of tokens
D
When the model copies text from training data

Question 9 of 20

What feature of transformers helps them understand words in context?

A
Fixed grammar rules
B
A dictionary lookup table
C
Self-attention mechanism
D
Convolutional filters

Question 10 of 20

Which of these is a real-world use case for LLMs?

A
Controlling physical robots in real time
B
Summarizing documents and answering questions
C
Storing files in a database
D
Running operating system processes

Question 11 of 20

What is fine-tuning an LLM?

A
Training the model from scratch on new data
B
Further training a pre-trained model on a specific task or dataset
C
Reducing the model's size to run faster
D
Adjusting the temperature setting

Question 12 of 20

Which is a known limitation of LLMs?

A
They can only understand one language
B
They have a knowledge cutoff and may not know recent events
C
They cannot generate text longer than one sentence
D
They always refuse sensitive questions

Question 13 of 20

What is bias in an LLM?

A
When the model is too slow to respond
B
A setting that controls output length
C
Unfair or skewed outputs reflecting patterns in the training data
D
When the model uses too many tokens

Question 14 of 20

What is a good best practice when using an LLM?

A
Always trust the output without checking it
B
Verify important facts from the model's response
C
Only use short one-word prompts
D
Use the lowest temperature for all tasks

Question 15 of 20

What does it mean for an LLM to have an "emergent ability"?

A
A skill that suddenly appears once the model is large enough
B
A feature added by developers manually
C
A bug that causes unexpected output
D
The ability to generate images

Question 16 of 20

What is an ethical concern around LLMs?

A
They process text too quickly
B
They can spread misinformation if outputs are not verified
C
They use too little electricity
D
They are too easy to use

Question 17 of 20

Why do transformers use positional encodings?

A
To understand word order since self-attention alone doesn't track position
B
To make the model run faster
C
To reduce the number of parameters
D
To translate between languages

Question 18 of 20

What is one way the future of LLMs is expected to develop?

A
Models will get smaller and less capable over time
B
Models will become more capable and handle more data types like images and audio
C
LLMs will be replaced entirely by rule-based systems
D
LLMs will stop being used in consumer products

Question 19 of 20

Training a large language model requires:

A
A standard laptop and a few hours
B
Only a small labeled dataset
C
Massive computing power and can cost millions of dollars
D
No data — the model creates its own

Question 20 of 20

How do LLMs differ from traditional AI programs?

A
LLMs follow strict keyword rules
B
LLMs can only do one task
C
LLMs require less data than traditional AI
D
LLMs understand meaning and context rather than just matching keywords

Exam Complete!

0%

0/20

Return to Course