[ Sandbox ]
The Playground
No lessons, no rules — every visualization in the library, free to explore.
Neural Network Playground
Live
Dataset
needs ~3 hidden neurons
Hidden neurons
4 neurons in 1 hidden layer
Epoch
0
Loss
0.000
Acc
0%
A real network training in your browser — 2 inputs → 4 tanh neurons → sigmoid output, trained by backpropagation. The background color is its decision boundary forming live.
Train a real network live
[ Same components appear inside lessons — sandbox is for exploration ]