Because the externally injected loss function will empty the brain of the model. Models need to decide for themselves what they should learn. Eventually, afte

Because the externally injected loss function will empty the brain of the model. Models need to decide for themselves what they should learn. Eventually, after entering the open world, reinforcement learning/genetic algorithms are still the only perpetual training solution. https://github.com/dmf-archive/PILF https://github.com/dmf-archive/PILF