mirror of https://github.com/alibaba/MNN.git
53 lines
1.6 KiB
C++
53 lines
1.6 KiB
C++
|
//
|
||
|
|
// ShapeBatchMatMul.cpp
|
||
|
|
// MNN
|
||
|
|
//
|
||
|
|
// Created by MNN on 2019/03/25.
|
||
|
|
// Copyright © 2018, Alibaba Group Holding Limited
|
||
|
|
//
|
||
|
|
|
||
|
|
#include "Macro.h"
|
||
|
|
#include "SizeComputer.hpp"
|
||
|
|
#include "TensorUtils.hpp"
|
||
|
|
|
||
|
|
namespace MNN {
|
||
|
|
|
||
|
|
class BatchMatMulComputer : public SizeComputer {
|
||
|
|
public:
|
||
|
|
virtual bool onComputeSize(const MNN::Op* op, const std::vector<Tensor*>& inputs,
|
||
|
|
const std::vector<Tensor*>& outputs) const override {
|
||
|
|
MNN_ASSERT(2 == inputs.size());
|
||
|
|
MNN_ASSERT(1 == outputs.size());
|
||
|
|
|
||
|
|
auto param = op->main_as_BatchMatMulParam();
|
||
|
|
// [TODO] now only support: adj_x and adj_y are false!
|
||
|
|
MNN_ASSERT(!param->adjX());
|
||
|
|
MNN_ASSERT(!param->adjY());
|
||
|
|
|
||
|
|
auto input0 = inputs[0];
|
||
|
|
auto input1 = inputs[1];
|
||
|
|
MNN_ASSERT(input0->dimensions() == input1->dimensions());
|
||
|
|
|
||
|
|
const int dimensions = input0->dimensions();
|
||
|
|
MNN_ASSERT(dimensions >= 2);
|
||
|
|
for (int i = 0; i < dimensions - 2; ++i) {
|
||
|
|
MNN_ASSERT(input0->length(i) == input1->length(i));
|
||
|
|
}
|
||
|
|
|
||
|
|
const int input0LastDimSize = input0->length(dimensions - 1);
|
||
|
|
const int input1LastDimSize = input1->length(dimensions - 1);
|
||
|
|
const int input1LastSecondDimSize = input1->length(dimensions - 2);
|
||
|
|
MNN_ASSERT(input0LastDimSize == input1LastSecondDimSize);
|
||
|
|
|
||
|
|
auto output = outputs[0];
|
||
|
|
TensorUtils::copyShape(input0, output, true);
|
||
|
|
output->setLength(dimensions - 1, input1LastDimSize);
|
||
|
|
|
||
|
|
return true;
|
||
|
|
}
|
||
|
|
};
|
||
|
|
|
||
|
|
REGISTER_SHAPE(BatchMatMulComputer, OpType_BatchMatMul);
|
||
|
|
|
||
|
|
} // namespace MNN
|