DeepCaps-PyTorch
DeepCaps-PyTorch copied to clipboard
detached dynamic routing and model mode change
I finally have been able to check the performance of the model with FashionMNIST with and without the detached version. The detached version proved to have a steadier learning curve. Sorry it took so long. Also, I noticed that the model's mode was not changed during the testing. It should be set to eval mode in order to notify the batchNorm layers that the model is now being used for testing.