Put it all together You've made it through all six lessons. You now understand the core ideas behind every AI language model, from simple word prediction all the way to the full transformer architecture.
This final exam tests what you've learned. There are no tricks. If you understood the lessons, you'll do great.
Take your time. And remember, you started from zero.
What is the fundamental operation of a language model? A Understanding the meaning of sentences B Predicting the next token given previous tokens C Translating between languages D Memorizing all text from the internet
Why do models use tokens instead of whole words? A Tokens are easier for humans to read B Tokens make the model run faster on GPUs C Subword tokens handle unknown words and keep vocabulary size manageable D Words are too long for computers to process
What does the loss function measure during training? A How fast the model runs B How many parameters the model has C How wrong the model's predictions are D How much data the model has seen
What property makes embeddings useful? A They compress text to save storage space B Similar words end up near each other in embedding space C They make text invisible to hackers D They translate words into different languages automatically
What does the attention mechanism do? A Makes the model pay attention to the user's request B Lets each token figure out which other tokens are most relevant to it C Removes unimportant words from the input D Speeds up the model by skipping unnecessary computation
What does temperature control? A How hot the computer gets B How creative vs. predictable the AI's word choices are C How fast the model generates text D How many words the model knows
In a transformer, what is the correct order of operations? A Tokenize → Embed → (Attention + Feed-Forward) × N → Predict B Embed → Tokenize → Predict → Attention C Attention → Embed → Tokenize → Feed-Forward D Predict → Attention → Tokenize → Embed
How does a model learn? A A human teaches it every answer B It memorizes a list of rules C It makes predictions, measures its mistakes, and adjusts its numbers to be less wrong D It copies answers from the internet in real time
Your review Answer 8 remaining questions above to finish this review.
Finish the review ← The full picture Back to the course →