Workshop Tutorial example not working

[knoxnox]

Hello!

Recently I've installed DragoNN with the help of anaconda (as explained on the installation page).
Generally it looks working fine without any kind of errors during execution.

However, when I wanted worked through the workshop_tutorial.ipynb file from examples folder I've observed different results from the ones suggested in the notebook.
What I mean by this is that in notebook there are presented three different models for class predictions of simulated data.
First 2 models (one layer one filter; one layer multi filter) were mentioned as not so good, 3rd one (multi layer multi filter CNN) according to the notebook should give very good results.
However, when I trained that 3rd model the result was not different from the previous models. Particularly the train accuracy was growing, train loss was decreasing, but at same time validation parameters were not changing a lot they were randomly jumping around some fixed values.
After this I've redone training of the model several times, the result was not any better.

I haven't done any changes to the code.

Attaching screen of the train results:

Any ideas what can be wrong?
Thanks in advance!