Neural Network Architecture Simulator

Define layer sizes, pick an activation function, and visualize a feed-forward network. Edge thickness reflects illustrative weight magnitudes.

Total Parameters
—

Understanding the Visualization

Each column is a layer. Lines connect neurons between adjacent layers — this is how information flows in a feed-forward network. Parameter count includes weights and biases:

For layers \([n_0, n_1, \ldots, n_L]\), parameters = \(\sum_{l=0}^{L-1} (n_l \times n_{l+1} + n_{l+1})\).

Learn more in ML Advanced — Neural Networks.