This repo contains the python scripts used to train the weights and save them in a file used in my master thesis.
The weights for the Airline or the C. elegans problems are trained by running airline.py or celegans.py with --train --save as arguments.
Training weights for other problems is possible. You can use the barbaLib.py module to use the custom activation functions or save the weights and adapt the code according to your specific problem.
Those scripts feature the possibility to choose the type of RNN, the input size, the number of time steps, the number of hidden state, the size of the output and more from a command argument.
This project is licensed under the General Public License, version 3.0 or later - see the COPYING file for details.