Open davidhorak opened 6 years ago
Is there a way how to mutate the current weights and biases stored in the network? I.e. when a network is cloned, is there a way how to slightly modify the values (mutate)?
dirty fix would be to keep one network trained one more epoch.
Is there a way how to mutate the current weights and biases stored in the network? I.e. when a network is cloned, is there a way how to slightly modify the values (mutate)?