GloVe Nearest neighbors (NLP)

Question

I´m trying to obtain the nearest neighbours(NLP) of 'frog' based on the nlp.stanford.edu/projects/glove & the gensim library, but I can't obtain their(standford's web) results¡¡. Why they are diffrent from theirs?? Can anyone advice/explain me??

This is my python code:

import gensim
## nlp.stanford.edu/projects/glove >> glove.6B.zip >> glove.6B.XXd.txt

new_model = gensim.models.keyedvectors.load_word2vec_format('./glove.6B/glove.6B.300d.txt',binary=False,no_header=True)

print(new_model)

print(new_model.most_similar('frog'))

and these are my results for 300d, 200d, 100d & 50d word vectors, how ever anyone is the same that the ones in stanford website. Anyone knows the reason??

nearest neighbors

Thanks so much.