Remove classifier.rnn module and DeepGRU classifier from the package.
There are many reasons for this, but mainly:
Due to the fact that DeepGRU is based around a torch workflow, this doesn't really fit in well with other parts of Sequentia such as the preprocessing and datasets module. It also has a very noticeably different interface to the other models in the classifiers module.
Sequentia was initially designed to provide easy-to-use ML algorithms which do not require much configuration. Although DeepGRU takes away most of the work in terms of designing the network architecture, there is still much setup that needs to be done to get a neural network running properly.
If you are interested in still using DeepGRU, the implementation that was used in sequentia.rnn will be made available here.
Remove all torch dependencies and torch-related code.
Major changes
Remove
classifier.rnn
module andDeepGRU
classifier from the package.If you are interested in still using DeepGRU, the implementation that was used in
sequentia.rnn
will be made available here.Remove all
torch
dependencies andtorch
-related code.Minor changes
/notebooks
.