Try it out here: https://train-your-own-neural-network.vercel.app/
You should get something like this once you've trained your model:
You can see the model is able to correctly classify Xs and Os with over 90% confidence each time.
The app features a helpful visualiser which displays the activations and weights in your neural network.
You'll see patterns emerging in the visualiser as you train your model and it picks out patterns in the input data.
Before:
After:
Another thing you'll notice in the visualiser, is the neurons change colour as you draw in the input square.
These colours display activations in the neural network.
A fully green neuron, means it is fully activated with a value of 1. Whereas red indicates a value of 0.
As you plan around with the neural network, you'll notice the activations in the input layer correspond to the input data shown in the top right square. And the hidden (middle) layer then extracts patterns e.g. lines/edges in the drawings.