What distinguishes hyperparameters from parameters in machine learning models?

A$.$ Parameters are tuned by the model during training, while hyperparameters are set manually before training

B$.$ Hyperparameters dictate how the algorithm should improve the model, while parameters are adjusted during training to minimise the loss function

C$.$ Hyperparameters are initialised randomly or to some default value, while parameters are updated iteratively to minimise the cost function

D$.$ Hyperparameters are specific to the optimization technique, while parameters are specific to the model architecture