Watch It Learn — XOR

A single neuron can only draw one straight line. Watch it try, fail, and then watch a 2-layer network bend the line into something that actually works.

speed
Epoch
0
Loss
0.0000
Status
Paused

Decision Boundary

Background color = what the network predicts across the whole input space

Loss Curve

Should drop toward zero — or plateau, if the model has hit its ceiling

Live Predictions

Rounded predictions vs. actual XOR truth table
InputPredictedRoundedActualMatch