Backpropagation Visualizer

Step through forward pass, loss computation, backward pass, and weight updates in a 2-2-1 neural network

Inputs & Target

Input x1
0.60
Input x2
0.40
Target
1.00
Learning Rate

Step Controls

Forward Pass
Compute LossStep 2
Backward PassStep 3
Update WeightsStep 4

Loss (MSE)

L = �(y_ - t)�

0.000000

Stats

Epoch 0

Prediction

0.0000

Error

-1.0000

Network Diagram2 ? 2 ? 1 � Sigmoid � MSE

-0.016-0.354-0.502-0.7670.308-0.442x10.600x20.400h10.000b=0.681h20.000b=-0.085y0.000b=0.302InputHidden (s)Output (s)

Legend

Input
Hidden (Sigmoid)
Output (Sigmoid)
?w > 0
?w < 0