performance is really not that good

[wasifmasood]

Hello team,

thanks a lot for making your contribution open-source. I did some quick tests and got really bad results. E.g., look at the following example where I am measuring cosine distance between "cancer" and "leukemia", got a score of 0.27. Similarly the cosine distance between dog and dobermann is only 0.367. I am doing it the following way:

model = sent2vec.Sent2vecModel() model.load_model(model_path) emb1 = model.embed_sentence("dog") emb2 = model.embed_sentence("dobermann") cosine_similarities = cosine_similarity(emb1, emb2).flatten() print(cosine_similarities)

Can you pls comment whether I am doing something wrong here.