I used the bitserial_conv2d_nchw and bitserial_dense in x86, but I found such op didn’t support GPU platform. Why not implement it in GPU? Is it hard to support such thing?
I used the bitserial_conv2d_nchw and bitserial_dense in x86, but I found such op didn’t support GPU platform. Why not implement it in GPU? Is it hard to support such thing?