Practice computing backpropagation by hand, or explore how gradients flow through a network in real time.
📝
Calculation Mode
Step-by-step quiz: compute forward pass, backward pass, and weight update by hand.
🔍
Interactive Mode
Manipulate weights and inputs in real time. Visualize gradient flow and test your intuition.
Practice computing a forward pass, backward pass, and weight update by hand for a simple neural network.
The activation function is the logistic sigmoid σ(x) = 1/(1+e−x).
The loss function is squared error E = ½(target − output)².
The target is always 0.01 and the learning rate is always η = 0.5.
All values needed for computation are shown on the network diagram. Round answers to 4 decimal places.
Network
Quiz Steps
0 / 12
Quiz Complete!
You computed the full forward pass, backward pass, and weight update for w₆.
Explore backpropagation interactively. Adjust weights, inputs, and hyperparameters to see how gradients flow in real time.
Edge color and thickness show gradient magnitude. Test your intuition with challenges below.