Skip to content
ml-recall

#phrases

2 questions from Text Representation & Embeddings, Tokenization

Timed test

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

Take a test
Q 1 / 2Score 0 / 0

Question 1: Single choice

mediumAI-checkedText Representation & Embeddings

To learn vectors for phrases such as "new york", word2vec first merges frequent bigrams into single tokens, scoring each bigram by

score(wi,wj)=count(wiwj)−δcount(wi)×count(wj),δ=5.\mathrm{score}(w_i, w_j) = \frac{\mathrm{count}(w_i w_j) - \delta}{\mathrm{count}(w_i) \times \mathrm{count}(w_j)}, \quad \delta = 5.

bigram count of the bigram count of wiw_i count of wjw_j
new york 500 1,000 600
of the 10,000 50,000 100,000
kaya toast 3 3 3

Which bigram gets the highest score?

Choose one option