Neural Network

Training Data & Decision Boundary

Training Progress

Epoch
0
/ 100
Loss
0.0000
Accuracy
0.0%
on training set
Speed
200ms
per iteration

Instructions

Watch a neural network learn in real-time! The visualization shows activations flowing forward (blue) and gradients flowing backward (red) during training.

Adjust the network architecture, learning rate, and dataset to see how different configurations affect learning.

Network Architecture
2 →→→ 1
Input → Hidden → Hidden → Output

How It Works

The network learns by comparing its predictions to the true labels, computing the error, and adjusting weights through backpropagation. Blue nodes show high activation, red show low. Thick edges indicate stronger connections.

Sorry, this demo is best viewed on a larger screen.

You might wanna head back now.

Back to homepage
profile image
home.index