CoLT5-attention
CoLT5-attention copied to clipboard
Implementation of the conditionally routed attention in the CoLT5 architecture, in Pytorch
Hello, I'm trying to apply this to cv, using the code from Simple Vit give me this error.  There seems to be an error in this code 
what hardware are used? how long does the training take? thx
Update autocast function from deprecated argument to the new one.