I am looking to compute the gradient of a neural network with one output node with respect to the input to the neural network. Is there some way to hack cudnnConvolutionBackwardData, cudnnActivationBackward etc to do this?
Many thanks
I am looking to compute the gradient of a neural network with one output node with respect to the input to the neural network. Is there some way to hack cudnnConvolutionBackwardData, cudnnActivationBackward etc to do this?
Many thanks