That's right, you basically want to add the contributions of each gradient of the output volume to get the full picture of how a weight affected the entire outp

That's right, you basically want to add the contributions of each gradient of the output volume to get the full picture of how a weight affected the entire output. In practice, this can implemented as a convolution of the gradients for one feature map on an input feature map: https://github.com/augustt198/ml/blob/master/ml/nn.py#L201 https://github.com/augustt198/ml/blob/master/ml/nn.py#L201