Quiz – Tokens and Windows

Question 1 of 5

1. Question

When an AI system receives text, what does it do first with the words?

It breaks everything down into smaller units called tokens

Correct Incorrect Correct answer

It reads the words exactly the way humans do

Correct Incorrect Correct answer

It stores the raw text unchanged in a database

Correct Incorrect Correct answer

It translates the text into a spoken audio signal

Correct Incorrect Correct answer

Question 2 of 5

2. Question

What is a context window?

The maximum amount of text a model can consider at once, measured in tokens

Correct Incorrect Correct answer

The visual interface where users type their prompts

Correct Incorrect Correct answer

The number of languages a model was trained on

Correct Incorrect Correct answer

The speed at which a model returns a response

Correct Incorrect Correct answer

Question 3 of 5

3. Question

Which landmark 2017 paper is credited with revolutionizing tokenization and language processing?

Attention Is All You Need

Correct Incorrect Correct answer

Learning From Data

Correct Incorrect Correct answer

Deep Vision Networks

Correct Incorrect Correct answer

The Turing Test Revisited

Correct Incorrect Correct answer

Question 4 of 5

4. Question

What range of context window sizes does the capsule say current models typically support?

From two thousand to one hundred thousand tokens

Correct Incorrect Correct answer

From ten to fifty tokens

Correct Incorrect Correct answer

From one million to ten million tokens

Correct Incorrect Correct answer

Exactly five hundred tokens for every model

Correct Incorrect Correct answer

Question 5 of 5

5. Question

Why can multilingual tokenization introduce bias?

Different languages tokenize differently, so non-English speakers may see different performance

Correct Incorrect Correct answer

Tokens are only ever created for the English language

Correct Incorrect Correct answer

Larger context windows automatically remove all language differences

Correct Incorrect Correct answer

Tokenization deletes any text that is not in English

Correct Incorrect Correct answer