pytorch

1
#include "caffe2/contrib/fakelowp/int8_quantize_op_nnpi.h"
2
namespace caffe2 {
3

4
REGISTER_CPU_OPERATOR(Int8QuantizeNNPI, int8::Int8QuantizeNNPIOp);
5

6
OPERATOR_SCHEMA(Int8QuantizeNNPI)
7
    .IdenticalTypeAndShape()
8
    .Arg("Y_scale", "Output tensor quantization scale")
9
    .Arg("Y_zero_point", "Output tensor quantization offset")
10
    .NumInputs(1)
11
    .NumOutputs(1)
12
    .Input(0, "X", "FP32 Tensor X.")
13
    .Output(0, "Y", "Int8 Tensor qX representing X with linear quantization.");
14

15
} // namespace caffe2
16