Closed njss closed 7 years ago
Hello @njss ,
The prediction can run on CPU (will be very slow), but is strongly reommended to train the network with GPU, otherwise it will take too long to train. And also for adversarial training, there is part of the code that used lasagne.layers.dnn.Conv2DDNNLayer
which needs cuDNN to work, if you wish you can replace it with lasagne.layers.Conv2DLayer
in order to train with CPU.
Best, Junting
Is mandatory to run with a NVIDIA GPU? Or can we run it using only the cpu?