How to get the optimal L for l-diversity

Question

I am a student trying to learn anonymisation of data by using a program called ARX. We use the dataset Adult and have to apply 5, 10, 15, 20, 25, 30, 40,50 and 75 for k-anonymit