Function variables unrecognized and training gives incorrect predicitons #1
Description
Hello,
First of all, I find the repository very interesting. Thanks for sharing it.
When I load your latest commit, the function defined in DL_model.py, specifically the Class DL defines the variables s_all_inputs, m_all_inputs, s_out_targets, m_out_targets etc. but when you call it in the Main_ML.py script the varibles are called differently. This on its own is not a problem to change but when I run your latest commit with these changes I cannot reproduce the results that you get in your article.
When I go back to your first commit, the training of the model works and the results are much better but I still do not get close to the accuracy that you get in the paper.
It could be the case that I am using the CPU for training since I do not have an NVIDIA GPU but an AMD one.
I am trying different combinations of hyperparameters in order to reproduce the accuracy in your paper but if you could run the latest commit on your side and confirm that it works as expected, that would be of great help.
I am currently looking into the possibilites of using your model and adapting it to a problem for my Master's thesis so any help would be appreciated.
Kind regards,
Faruk