Hi,
I cannot find the right result for “approximate_kn” function. All the previous methods work fine (all tests passed). I keep having [51, 701, 3028] instead of [51, 701, 2140]. I calculated by hand the cosine_similarity for 2140 and 3028 and it happens that it is bigger for 3028 than for 2140. So my result seems to be better than the “grader result”. I don’t understand this !
Many thanks in advance,
Matthieu