Skip to content
ml-recall

#sequence-length

2 questions from Tokenization

Timed test

2 questions · 2 min · score and explanations at the end

Take a test
Q 1 / 2Score 0 / 0

Question 1: Short answer

hardAI-checkedTokenization

BPE is trained on a corpus with these word counts:

word hat hats that chat chats at
count 5 2 4 3 2 3

Split into characters, the corpus is 67 tokens long. The first three merges are a t → at, h at → hat and c hat → chat. How many tokens long is the corpus after these three merges (counting every occurrence of every word)?

a whole number