Train and test neural networks for handwritten digit recognition using the MNIST dataset.
Configure and train custom neural networks with different architectures.
Train multiple models simultaneously and compare their performance.
Monitor training progress with live loss and accuracy curves.
Draw digits and test the model's prediction capabilities in real-time.