Which term describes parameters that are typically set before the training of a machine learning model begins?

The term that describes parameters typically set before the training of a machine learning model begins is hyperparameters. Hyperparameters are critical to the configuration of the learning process and include elements such as learning rate, batch size, number of epochs, and architecture details like the number of layers in a neural network.

Setting hyperparameters correctly is essential because they significantly influence the performance of the model. For instance, choosing an appropriately small learning rate can help the model learn more progressively, while setting the batch size can affect how the model converges.

On the other hand, learned parameters are those that are optimized during the training process, based on the input data. Model parameters often refer to the weights and biases inside the model that are adjusted as the model learns. Antiparameters is not a standard term in machine learning and does not have a defined meaning in this context. Therefore, hyperparameters are the correct answer as they are specifically defined as settings made prior to the actual training phase of a machine learning model.