Add metadata for dataset used to train model (#1)

improving usage example

fixing typo in readme

adding link to paraphrases dataset inkoziev/paraphrases

first release of the tiny model for paraphrase detection

initial commit