#hbDNNQuantiShift

Quantization/Dequantization shift data, this quantization/dequantization method has been deprecated in J6 and is only retained for compatibility with J5.

Input: If the floating-point data data is collected, and the corresponding shift data is shift, then the inference data sent to the model is $data[i] * (1 < < shift[i])$ rounded.

Output:: If the corresponding shift data of the inference result data is shift, then the final inference result is $data / (1 < < shift)$.

The shiftLen is determined by data according to per-axis or per-tensor quantization/dequantization. When the data is quantized/dequantized by per-tensor, the shiftLen is equal to 1, ignoring quantizeAxis. Otherwise the quantizeAxis represents the dimension index of the data quantization axis, and the shiftLen is equal to the size of the quantizeAxis dimension of the data.

• Member