Is there a way I can convert and use Google's universal-sentence-encoder (available through TF hub) in pytorch?
