Embeddings were created through applying word2vec skipgram to a corpus of wikipedia non-stub articles from a December 2015 English dump with the following parameters: -cbow 0 -size 200 -window 10 -negative 25 -hs 0 -sample 1e-4 -threads 20 -binary 0 -iter 10 The format is standard for word embeddings textual representations and can be read simply, e.g. in a line-by-line basis. First line is dimensionality of the embedding matrix (number of words, number of dimensions): 2847710 200 The following lines: word1 v_1_1 ... v_1_200 . . . word2847710 v_2847710_1 ... v_2847710_200 where v_i_j is the value of the i-th word embedding in the j-th dimension.