wehub-resource-sync
69 lines
2.5 KiB
C++
69 lines
2.5 KiB
C++
//
|
|
// SaveHalfFloat.cpp
|
|
// MNNConverter
|
|
//
|
|
// Created by MNN on 2021/08/11.
|
|
// Copyright © 2018, Alibaba Group Holding Limited
|
|
//
|
|
|
|
#include "CommonUtils.hpp"
|
|
#include "half.hpp"
|
|
#include <math.h>
|
|
|
|
void CastParamsToHalf(std::unique_ptr<MNN::OpT>& op) {
|
|
const auto opType = op->type;
|
|
switch (opType) {
|
|
case MNN::OpType_Convolution:
|
|
case MNN::OpType_ConvolutionDepthwise:
|
|
case MNN::OpType_Deconvolution:
|
|
case MNN::OpType_DeconvolutionDepthwise:
|
|
{
|
|
auto param = op->main.AsConvolution2D();
|
|
if (param->quanParameter != nullptr) {
|
|
break;
|
|
}
|
|
const int weightSize = param->weight.size();
|
|
// const int biasSize = param->bias.size();
|
|
std::vector<half_float::half> quantizedFp16Weight;
|
|
quantizedFp16Weight.resize(weightSize);
|
|
std::transform(param->weight.begin(), param->weight.end(), quantizedFp16Weight.begin(),
|
|
[](float w) {
|
|
w = fmaxf(w, -65504.0f);
|
|
w = fminf(w, 65504.0f);
|
|
return half_float::half(w);
|
|
});
|
|
// std::vector<half_float::half> quantizedFp16Bias;
|
|
// quantizedFp16Bias.resize(biasSize);
|
|
// std::transform(param->bias.begin(), param->bias.end(), quantizedFp16Bias.begin(), [](float
|
|
// b){return half_float::half(b); });
|
|
param->weight.clear();
|
|
// param->bias.clear();
|
|
|
|
param->quanParameter.reset(new MNN::IDSTQuanT);
|
|
param->quanParameter->type = 3;
|
|
int8_t* halfWeight = reinterpret_cast<int8_t*>(quantizedFp16Weight.data());
|
|
param->quanParameter->buffer.assign(halfWeight, halfWeight + sizeof(half_float::half) * weightSize);
|
|
break;
|
|
}
|
|
case MNN::OpType_Const: {
|
|
auto blob = op->main.AsBlob();
|
|
if (blob->dataType == MNN::DataType_DT_FLOAT) {
|
|
blob->dataType = MNN::DataType_DT_HALF;
|
|
blob->uint8s.resize(sizeof(half_float::half) * blob->float32s.size());
|
|
auto size = blob->float32s.size();
|
|
auto dst = (half_float::half*)blob->uint8s.data();
|
|
for (int i=0; i<size; ++i) {
|
|
float v = blob->float32s[i];
|
|
v = fmaxf(v, -65504.0f);
|
|
v = fminf(v, 65504.0f);
|
|
dst[i] = v;
|
|
}
|
|
blob->float32s.clear();
|
|
}
|
|
break;
|
|
}
|
|
default:
|
|
break;
|
|
}
|
|
};
|