Rank and run-time aware compression of NLP Applications

Urmish Thakker, Jesse Beu, Dibakar Gope +2

Sequence model based NLP applications can be large. Yet, many applications that benefit from them run on small devices with very limited compute and storage capabilities, while sti…