🧠 Tiny MLP Gradient-Check Lab: Live Backprop on 2D Toy Datasets
Train a 2-hidden-layer tanh network live with Adam on moons, circles, XOR and spiral data. Watch the decision boundary form, track loss and accuracy, and verify backpropagation against a numerical gradient.
Machine Learning & Neural Networks2DModerate60 FPS📱 Mobile-adapted
⚙ Under the hood
Train a 2-hidden-layer tanh network live with Adam on moons, circles, XOR and spiral data. Watch the decision boundary form, track loss and accuracy, and verify backpropagation against a numerical gradient.
neural networkbackpropagationgradient checkdecision boundarymachine learningAdam
2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install