Does text preprocessing or cleaning required for Bert model or others one?

Can someone suggest me, is there any text preprocessing required for pre train model like Bert, T5 etc.