Have you finished the quantization?
qinziqiao opened this issue · comments
I have read your source code of ristretto. You still use the floating-point and do not replace it with 8-bit fixed point number, although you said you would do it on introduction. Right ? Is it the reason that Caffe not support this data-type?