oh man someone should make this into a loop where the model can alter it weights after an inference. some kind of descent of the local gradient.
Isn’t that what the Google dreaming thing is?
Isn’t that what the Google dreaming thing is?