Gradient Clipping is a method where the error derivative is changed or clipped to a threshold during back propagation through network and the clipped gradients are used to update the weights