How do i get word2vec similarity from the mean vector?

Question

For example, there are words for 'apple', 'banana', and 'orange'.

We will execute the code below to save the distance between apple and banana.

model.similarity('apple', 'banana')

But what I want to know is the similarity between 'apple' and 'whole fruits'. How do i get the similarity of apples and whole fruits?

I already got vectors for the whole fruit. e.g. whole fruits=[0, 0.4, 0.2, 0.2, 0.5, .....]

Answer 1

model.similarity calculates cosine similarity behind the scenes between the embedding vectors for the words. If you have already have the vectors for "apple" and "whole fruits" then you can get the cosine similarity using sklearn's pairwise cosine similarity function.