Hyperparameters specify the model's architecture and how it will be trained. Therefore, these hyperparameters are not learned during training phase. In the case of neural networks, some common hyperparameters are: