In that case it would make more sense to try and isolate the tasks comprising the last 20% and train on them separately, so that you get more practice.
the good news is that virtually all of the software in the linux ecosystem is missing the last 20%, so there's plenty of opportunities to train.