Neural Networks package in Wolfram Mathematica not parallel?

Question

I just created a VERY big neural network, albeit on very powerful hardware, and imagine my shock and disappointment when I realized that NeuralFit [] from the NeuralNetworks package seems to use only one core and not even the full one least I was heartbroken. Should I really write an entire NN implementation from scratch? Or am I missing something simple?

My network took 200 inputs for 2 hidden layers of 300 neurons to produce 100 outputs. I understand that we are talking about trillions of calculations, but as long as I know that my equipment is a weak point, you can update it. He should train well with such a network if he is left alone for a while (a 4-gigabyte 8-stream machine with 24 GB of 2000 MHz CL7 memory with SSD-RAID-0 drives on SATA-III - I'm sure).

Ideas? Suggestions? Thank you in advance for your entry.

Answer 1

You can directly contact the author of the package, he is a very affordable guy and can make some suggestions.

Answer 2

. . ( , ). , , . , , , . , . Wolfram. .

9 10, ( ). , HelpBrowser , .

Answer 3

I'm not sure how you wrote the code or how it is written inside the package you are using; try using vectorization, it really speeds up linear algebra calculations. In the ml-class.org course , you can see how this is done.