Skip to content
The loss curve

Neuron

Weighted sum of inputs followed by a non-linearity: σ(Σ w_i·x_i + b). The smallest learnable unit in a neural network.

Companion explanation in Step by Token, chapter 6.

Where this term gets built