capsule-net-pytorch

capsule-net-pytorch copied to clipboard

Test error results reported are actually loss figures and are not comparable with paper reported accuracies.

3

Thanks for a very nicely presented implementation of Capsule Networks. I especially appreciate the tensorboard plots. Unfortunately I believe you have mixed up **"test error"** with **"test loss"** when reporting...

geefer

\# [1 x primary_unit_size x num_classes x output_unit_size x num_primary_unit] \# == [1 x 8 x 10 x 16 x 1152]

zhangjcqq

Deviations of this implementation from the original paper

2

I have come across the following main deviation between the paper(https://arxiv.org/abs/1710.09829) and the implementation of this repo: In the paper: there are 32 primary capsule layer with 6x6 capsules each...

InnovArul

Migrate existing code to work in PyTorch version 0.4.0. PyTorch 0.4.0 introduced a major core changes that changed the previous API version. Status: I have started work on this and...

cedrickchee

Continue training?

2

Train CapsulesNet is known as time-consuming and not yet optimized. So, I want an option of loading pre-trained weights (e.g. provided file path) and continue training.

hiepph

Memory required explodes when the CapsLayer instance is called

1

@cedrickchee The memory required by the network increases from 3 GB to about 25 GB after the finishing of the Primary caps layer and after Capslayer is called. I wonder...

arshiyaaggarwal

RGB 256*256 image

1

Can you tell us how can we use your code to classify RGB images. Our dataset is like that: class1: 0001.jpg 0002.jpg Class2: 001.jpg 002.jpg

deep0learning