I got AssertionError when checking the similarity.

[kohei0219]

Hello!!
Thank you for providing the great library.

I tried to check the nearest_neighbors after I trained a model.
Then, AssertionError was shown in here.
The reason is that all elements of in_vs are 0, so the result of norm is also 0.

The model that I produced has 5 meanings for each word.
I guessed that a word that was checked only had less than 5 meanings. Is this interpretation correct?

In order to run the function, I used below code instead of @Assert.

if isnan(sim[s, v]) sim[s, v] = -Inf end

The result was what I expected.

nearest_neighbors(vm, dict, "apple", 1, 10)
("peach", 1, 0.957842), ("plum", 1, 0.952987), ("cherry", 1, 0.94981), ("lemon", 5, 0.947042), ("pear", 1, 0.945165), ("sweet", 2, 0.943617), ("quince", 1, 0.942606), ("blackberry", 1, 0.94111), ("melon", 1, 0.940196), ("pomegranate", 1, 0.940196)

I'm using Julia version 1.1.1. I think your project runs correctly if you modified this part.