Prelosses and Postlosses

[Naman9639]

Hey!
I am new in this field and I was going through the code but I am not able to understand what exactly does the preloss and post loss mean here. Can someone help?

Thanks

[ryujaehun]

I understand it that preloss is adaptaion loss about label_a(Number of classes per task) and postloss is meta-learning loss about lable_b(Number of test example per class).

[asker-github]

I understand it that preloss is adaptaion loss about label_a(Number of classes per task) and postloss is meta-learning loss about lable_b(Number of test example per class).
Hello, I have a question. Why does postoss get higher and higher in training? I think that should mean the accuracy.