1Mediumgeneral⏱ 45sWhat problem does the vanishing gradient issue cause in deep networks, and how does ReLU help?📎▼AIt causes training to be too fastBGradients shrink exponentially through many layers during backpropagation with saturating activations (like sigmoid), stalling learning in early layers; ReLU's constant gradient for positive inputs mitigates thisCIt only affects convolutional layersDIt has no effect on training speed or convergence