Is it possible to fine tune zero shot text classification model for our data set?

I would like to know that is it possible to fine tune zero shot text classification model for our data set? if yes what should i do