I am a self educated enthusiast in machine learning and currently trying to
set up my first LSTM NN in WEKA. It is not clear to me yet, whether the LSTM
layer in WEKA is a complete LSTM layer (containing forget, input, and output
gates) or an LSTM node (or cell) that needs to be stacked with others in
order to construct a layer. What partly confuses me, is that in the LSTM
layer configuration, there is a "gate activation function" parameter (what
does it do?).