details about binarization #11
Description
As far as I am understanding the paper and the implementation, the input image is not binarized, the weight is binarized by bnn binary operation, while the outputs are still real number, am I correct?
{{ message }}