Hyperparameter
A setting you choose before training that shapes how the model learns — like the learning rate, batch size, or number of layers. Unlike weights, the model never learns these; you pick them.
Think of It Like This
The oven settings you dial in before baking, not the cake itself.
Weights are learned from the data; hyperparameters are the knobs you set around that learning, from the learning rate to the batch size to how many layers deep the network runs. Good choices can be the difference between a model that trains smoothly and one that never converges, so teams search over them with grids, random sampling, or smarter tuners. The catch is that the settings interact, so tuning is part science, part patience.