#2 Tensors

I quite enjoyed building a convolutional neural network on the MNIST

database, but I really would like to see those images which creates difficulties with

the model which is currently 98.6% okay. To do that, I have moved my searches

to  German-language YouTube and found another series of tutorials on pytorch MNIST.

I do speak some German, but one can be follow with captions in English. It's a 

very different approach, but it aims to give a theoretical understanding of what is going on.

Not for the faint-hearted, but very interesting. (It has alreay confirmed to me that

MNIST images come with a label saying what they represent, something I had

conjectured about.)

Below, #2  on Tensors:

                                                                          *     *     *

Below, a 'toy'  one can use to familiarize oneself with neural networks. There are 

blue dots, and there are orange ones. One wants to draw a separation between the

two; the problem can be made more or less complicated, but one can add hidden layers at

will.

https://playground.tensorflow.org

Still getting a test loss; shoud I have more training ‽

Helpful:

                                                                    *     *     *

   The Given:

                                              

                                                                             

 Test Data:

                                                                         

Our output after training: