This release is debugged (as far as I can tell), and can play chess at a poor level.
It needs to train more, but is a fully-functioning model.
It can almost beat the Chess.com AI level 1, but not quite.
Such is why I am publishing it as a separate project.
I will create a new project for its subsequent iterations, as I consider this one to be 'complete,' at least within the code.
The only quirks in this model are:
No check[mate] detection
It causes a memory leak if left for too long (it uses about .5GB per hour when learning)