For example, if I choose two window size, 5 and 50, and train the word2vec model, will the 50 one takes more time to train? Will the embeddings of the 50 one concentrates mo