Hi,
I am following along the cuDNN Documentation provided online and the cudnnActivationBackward() function has a different order of input parameters, is this correct? or should it follow the same format as the other backward propagating functions (such as cudnnPoolingBackward)?
http://docs.nvidia.com/deeplearning/sdk/cudnn-developer-guide/index.html#cudnnActivationBackward
Please Let me know,
Thank you
-Liam