NeuralODE for Dampled Pendulum and MNIST classification
Repository structure
-
Classification problem on MNIST dataset with flags:
--network
: "resnet" or "odenet"--use_adjoint
: whether to use adjoint method or not for NeuralODE--tol
: error tolerance for the ODE solver--max_epochs
: max epochs for training--data_aug
: whether to use data augmentation or not--lr
: learning rate--batch_size
: batch size
Links