HACKING.md

Modules:

• activation maximisation (activemax)
• layer activation
• CONV/FC filter visualisation (filtervis)

Ideas:

Extend models by adding a custom bound method, rather than passing it to an object of the custom class?

from types import MethodType

model = models.alexnet(pretrained=True)

def custom_func(self):
    print('bound custom func!')

model.custom_func = MethodType(custom_func, model)

model.custom_func()  # prints out 'bound custon func!'

TODO:

• add test script to travis config