I use mixmatch to train my own data set, which is simpler than cifar10 image. After running several epoch, the loss function of training and trainstats will become larger, and the accuracy of test set will be reduced

[ghost]

I face the same question : the loss of train becomes smaller , but the loss of val becomes larger. My own dataset has 200 classes

[Lotk1103]

I face the same question : the loss of train becomes smaller , but the loss of val becomes larger. My own dataset has 200 classes

Hey, do you solve this problem? I meet the same question, is it realted to dataset imbalance?

[SuperJunier666]

我也遇见了这个问题，我是用这个方法做分割