I've split the dataset generation and training to separate functions.
It's a bit messy at the moment, each dataset only works with certain loss functions. Currently it will just crash for one reason or another if the wrong ones are mixed.
I have also added 3 more loss functions. They each have slightly different graphing requirements that we will need to add.
I've split the dataset generation and training to separate functions.
It's a bit messy at the moment, each dataset only works with certain loss functions. Currently it will just crash for one reason or another if the wrong ones are mixed.
I have also added 3 more loss functions. They each have slightly different graphing requirements that we will need to add.
main.py is currently set up to use Scandal.