Why do you normalize the vectors different compared to the original word2vec code? med = sqrt(med / vector_size); len = sqrt(len);
Why do you normalize the vectors different compared to the original word2vec code?
med = sqrt(med / vector_size);
len = sqrt(len);