Stefan Grosser

Will be done on branch 'leaky' @martinjm97 will need your help to put this in Julia

With the new UI for layers, make the parameter entries over the canvas, on select.

I think all we need to do is save the model and then provide skeleton code for loading it

True, we can just export the saved model files

Switch to a lighter weight plot library?

As in a < select > instead of an < input >

@martinjm97 would we want an "advanced" parameters checkbox somewhere which would reveal stuff like this? Or should we stick with no toggles in our UI? I just don't want there...

Take the global state of these files and refactor to something similar to params_object.ts This would be a good approach to handle future versions, where data set sizes might be...

Should probably also come first before adding super nodes.

This might mean abandoning the tensorflow plotting, as we will want to plot loss and accuracy from multiple models on one graph for comparison.