Last active
August 28, 2019 17:03
-
-
Save mkolod/27e5ca3839bd2f22af8ebce45926262b to your computer and use it in GitHub Desktop.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
We can see that for a matrix that's only 75% sparse, CSR saves hardly any memory, and is more costly computationally. Typically, matrices have to be over 99% sparse (preferably 99.9%) for a sparse representation to be more computationally performant. The requirement for substantial memory savings are a bit less stringent, but 99%+ sparsity is a good general rule of thumb.