How to use optimum with encoder-decoder models

Hi @anon32220481, is this related to this issue posted on Github?