optimum

Drop mask for training in all cases for BetterTransformer & precise documentation

#1250

Merged