Consider the following function

def average$\_$vectorizer$\_$Word$2$Vec$($doc$)$:

feature$\_$vector $=$ np$.$zeros$(($vec$\_$size,$),$ dtype$=$"float$64")$

words$\_$in$\_$vocab $=[$word for word in doc.split$()$ if word in words$]$

for word in words$\_$in$\_$vocab:

feature$\_$vector $+=$ np$.$array$($word$\_$vector$\_$dict$[$word$])$

if len$($words$\_$in$\_$vocab$)!=0$:

feature$\_$vector $/=$ len$($words$\_$in$\_$vocab$)$

return feature$\_$vector

What does the below code output?

if len$($words$\_$in$\_$vocab$)!=0$:

feature$\_$vector $/=$ len$($words$\_$in$\_$vocab$)$

The count of the words in the document

The sum of the feature vectors of words in the document

The average feature vector of the document

The individual word vectors concatenated into a single vector