How to implement multilabel classification for OCR (Hindi Characters)

Question

I have all Hindi Vowels and Consonants (क, का, कि, कु and so on..). How can I use Multilabel Classification in this using tensorflow or pytorch?