XOR Visual Lab

Epoch 0 Error: – Correct: 0/4

Input A

Input B

What just happened?

Toggle the inputs or train the network to see each step of the forward pass explained here.

All inputs

A	B	Expected	Network

Train the network

Learning rate 0.50

Auto-train (10 epochs per step) Freeze visuals

Training status

Sparkline shows the last 120 error values.

About this lab

This interactive demo uses a 2‑2‑1 neural network with sigmoid activations to learn XOR. Watch weights change (edge thickness), activations glow (node fill), and follow the maths in plain language.

Built for hands-on learning. View the source, remix it, and share what you discover.