mirror of https://github.com/alibaba/MNN.git
51 lines
1.7 KiB
C++
51 lines
1.7 KiB
C++
|
//
|
||
|
|
// ShapeDet.cpp
|
||
|
|
// MNN
|
||
|
|
//
|
||
|
|
// Created by MNN on 2019/01/10.
|
||
|
|
// Copyright © 2018, Alibaba Group Holding Limited
|
||
|
|
//
|
||
|
|
|
||
|
|
#include "shape/SizeComputer.hpp"
|
||
|
|
#include "core/Macro.h"
|
||
|
|
|
||
|
|
namespace MNN {
|
||
|
|
class DynamicQuantComputer : public SizeComputer {
|
||
|
|
virtual bool onComputeSize(const MNN::Op* op, const std::vector<Tensor*>& inputs,
|
||
|
|
const std::vector<Tensor*>& outputs) const override {
|
||
|
|
MNN_ASSERT(outputs.size() == 3);
|
||
|
|
if (inputs.size() != 1) {
|
||
|
|
MNN_ERROR("DynamicQuant only accept 1 input\n");
|
||
|
|
return false;
|
||
|
|
}
|
||
|
|
auto input = inputs[0];
|
||
|
|
auto output = outputs[0];
|
||
|
|
int dimSize = input->dimensions();
|
||
|
|
output->buffer().dimensions = dimSize;
|
||
|
|
for (int i = 0; i < dimSize; ++i) {
|
||
|
|
output->buffer().dim[i].extent = input->buffer().dim[i].extent;
|
||
|
|
}
|
||
|
|
auto scale = outputs[1];
|
||
|
|
auto zeroPoint = outputs[2];
|
||
|
|
scale->buffer().dimensions = 1;
|
||
|
|
zeroPoint->buffer().dimensions = 1;
|
||
|
|
scale->buffer().dim[0].extent = 1;
|
||
|
|
zeroPoint->buffer().dim[0].extent = 1;
|
||
|
|
|
||
|
|
TensorUtils::getDescribe(output)->dimensionFormat = TensorUtils::getDescribe(inputs[0])->dimensionFormat;
|
||
|
|
output->buffer().type = halide_type_of<int8_t>();
|
||
|
|
|
||
|
|
TensorUtils::getDescribe(scale)->dimensionFormat = TensorUtils::getDescribe(inputs[0])->dimensionFormat;
|
||
|
|
scale->buffer().type = halide_type_of<float>();
|
||
|
|
|
||
|
|
TensorUtils::getDescribe(output)->dimensionFormat = TensorUtils::getDescribe(inputs[0])->dimensionFormat;
|
||
|
|
zeroPoint->buffer().type = halide_type_of<float>();
|
||
|
|
|
||
|
|
return true;
|
||
|
|
}
|
||
|
|
};
|
||
|
|
|
||
|
|
REGISTER_SHAPE(DynamicQuantComputer, OpType_DynamicQuant);
|
||
|
|
|
||
|
|
} // namespace MNN
|