I converted a tensorflow model to a caffemodel, however all the weights and activation were floating point values. I want to somehow quantize my caffemodel to a fix point 16 bit
