Open FreezePhoenix opened 6 years ago
Try Tensorflow for large NNs.
@Pummelchen Yeh, um.. that's kinda hard to install on a chromebook or a RaspPi 3.
lol ok, well try to use Binay cost function and/or train only one (1) row in training with Backprop on.
I have 2 layers with 100 neurons each, and currently, it is fairly slow, both to train and to run. Are there any ways to optimize this? The case is that I feed it an array of floating point numbers between 0 and 1.