Hyperparameters

These are configuration settings that are external to the model and whose values cannot be learned from the data during training. They must be set manually by the user before training begins and significantly impact model performance.