Custom pytorch dataset on a regression problem

Hello !

I have a custom pytorch dataset, which I plugged to fastai with DataBunch.create. And then I use it with ResNet50 on a regression problem where it’s suppose to predict one value. It seems like it apply a sigmoid at the end because I have data.c = 1. Is there anywhere in the code where I could do a monkey patch to change this behavior and only have a linear activation function ?

Thank you