EquiConvPytorch icon indicating copy to clipboard operation
EquiConvPytorch copied to clipboard

training time

Open rshivansh opened this issue 5 years ago • 1 comments

Hi, thanks for the code. I am using equi_conv in my own code to train resnet50 on custom data. however I find it is really slow, so I wanted to check with you. Is that the case?

rshivansh avatar Jan 24 '21 12:01 rshivansh

@rshivansh Yes, the code is very slow. If you want to optimize it, you have to create a lookup table for the offsets at each layer of your model.

palver7 avatar Jan 24 '21 14:01 palver7