
Episode 479: Luis Ceze on the Apache TVM Machine Learning Compiler
Software Engineering Radio - the podcast for professional software developers
00:00
How to Optimize Your Models With TVM?
So where t v m shines is reallys doing optimisations like ok? Once you have an operator, or you fuse two operators in optimising your moto, theyall get represented as looks over data and performing operations. The second thing is picking the right instructions. You want to make sure that you keep the data in registers as much as possible. And when you do that, now i can put my comparer architecture had to acton my core back nosing compular architecture. That's a good oeu interesting to know about, like exibus and how it works with decisionries. So one aspect of t v m, i think, is the portability of your models,
Transcript
Play full episode