Large Language Model (LLM) Knowledge Quiz

1. Which neural network architecture is the foundation for modern Large Language Models like GPT and BERT?

2. Fundamentally, what is the primary mathematical task an LLM performs when generating text?

a) Retrieving whole sentences from a database

b) Predicting the statistical probability of the next token in a sequence

d) Fact-checking statements against the internet in real-time

3. In the context of LLMs, what is a "token"?

a) A cryptocurrency used to pay for API access

b) A security key required to log in to the model

c) The basic unit of text (a character, sub-word, or word) processed by the model

d) A line of Python code used to train the model

4. What does the term "hallucination" refer to in AI?

a) The model generating images instead of text

b) The model confidently generating plausible-sounding but factually incorrect information

c) The model becoming slow due to server overload

d) The model refusing to answer a prompt due to safety filters

5. If you increase the "temperature" parameter when generating text, what is the likely result?

a) The output becomes more deterministic and focused

b) The output becomes more random, diverse, and "creative"

a) The chat interface where you type your prompt

b) The limit on the amount of text (input and output combined) the model can process at one time

c) A time period during which the model is available for free

d) The specific window of dates the training data covers

7. What is "RLHF" (Reinforcement Learning from Human Feedback) primarily used for?

b) Aligning the model's behavior with human values, safety guidelines, and helpfulness

d) Compressing the model to run on smaller devices

8. What are "parameters" in an LLM (e.g., 70B parameters)?

b) The settings you can adjust in the user interface

c) The internal numerical weights and biases learned during training

d) The number of users accessing the model simultaneously

a) Asking the model to perform a task without providing any examples in the prompt

b) Providing zero context or background information

c) When the model gets the answer right on the first try

d) A prompt that contains no text, only images

10. What happens during the "Pre-training" phase of an LLM's lifecycle?

a) The model is tested by human users for bugs

b) The model is fed a massive corpus of text to learn general language patterns and world knowledge

c) The model is specialized for a specific task like medical diagnosis

d) The model is compressed to save storage space