Hyperparameter

A configuration value set before the learning process begins, such as learning rate, batch size, or number of layers. Hyperparameters are tuned to optimise model performance.