Arrange the following steps in addressing the vanishing/exploding gradients problem in deep neural networks from first to last: A) Implement normalization techniques, B) Train the network, C) Initialize weights appropriately, D) Monitor gradient behavior during training.

First, you should initialize the weights properly so that gradients start with reasonable magnitudes. Next, you apply normalization techniques such as batch or layer normalization to keep the signals stable across layers. After that, you monitor the gradients during training to detect any vanishing or exploding patterns early. Finally, you proceed to train the network, adjusting learning rates or other hyperparameters based on what you observed. For example, initializing with Xavier and using batch normalization helps maintain gradient flow before you start the main training loop.