Neural Network Explorer
Explore how a 2-layer neural network computes its output. Adjust weights, change activation functions, and see the forward pass propagate through the network in real time.
Network Visualization
Network Controls
Activation
Hidden Size
Input x
Input y
Forward Pass
Hidden 1
—
Hidden 2
—
Hidden 3
—
Output
—
σ(W · x + b)
How it works: Inputs
x₁, x₂ are multiplied
by weights, summed with biases, and passed through an
activation function. Adjust sliders to see how each
weight affects the output.
What's happening here? A neural network
is like a series of pipes and faucets. The
inputs (x₁, x₂) are water flowing in.
The weights (sliders) are faucets that
control how much water passes through each pipe. The
hidden layer neurons are mixing
chambers that combine the flows. The
activation function (ReLU, Sigmoid,
etc.) is like a filter that decides "only let enough
water through if it's strong enough." The
output is what comes out the other end.
Green connections mean a weight is pushing the signal
forward; red means it's pushing against it. Try sliding
weights to zero — that connection shuts off completely.
Change the activation and watch how the numbers change.
Premium · From the makers of ReLU.chat
On-Device Chatbot Builder Toolkit
Interactive demos teach one concept at a time. The
On-Device Chatbot Builder Toolkit gives you the full
system: frameworks, templates, and evaluation guides for
building your own privacy-first chatbot that runs
entirely on-device.
Get the Toolkit · $29
