Pytorch Implementation of the paper StackedCapsuleAutoEncoders. link
- The code currently supports MNIST dataset only
- All the operations are vectorised for supporting batch operations except the template transformations part (Pytorch doesn't have batch support for image transformation functions)
- The code for the set transformers part (setmodules.py) is used from the official repository [https://github.com/juho-lee/set_transformer].
- Feel free to cite this repository if you want to use the code.