Inference performance is the primary reason I’m currently keeping Argos Translate V1 in the master branch. I think in the future I want to move to multilingual models but for now single language pair models are smaller and faster.
1 Like