Regarding the backward propagation in the custom loss function

nxsEdson / MLCR

Multi-label Co-regularization for Semi-supervised Facial Action Unit Recognition (NeurIPS 2019)

63 stars 18 forks source link

Regarding the backward propagation in the custom loss function #9

Open rohankrgupta opened 3 years ago

rohankrgupta commented 3 years ago

Hi, I am rather new and trying to learn new things from other people's code. I am not able to find examples of code snippets for training with custom loss function in pytorch. Could you suggest me some code snippets/projects from where I may get the idea.