forked from mindspore-Ecosystem/mindspore
回退 'Pull Request !36155 : [assistant][ResizeNearestNeighborV2][ResizeNearestNeighborV2Grad] cast shape type size_t to int64_t for ResizeNearestNeighborV2 & ResizeNearestNeighborV2Grad'
This commit is contained in:
Gitee
parent
f98be788a3
commit
8c444c00fc
|
|
@ -85,6 +85,8 @@ ConstInputToAttrInfoRegistry::ConstInputToAttrInfoRegistry() {
|
|||
Register(kErfOpName, {1});
|
||||
Register(kSparseApplyAdagradOpName, {2});
|
||||
Register(kResizeNearestNeighborGradOpName, {1});
|
||||
Register(kResizeNearestNeighborV2OpName, {1});
|
||||
Register(kResizeNearestNeighborV2GradOpName, {1});
|
||||
Register(kApplyRMSPropOpname, {5, 6, 7});
|
||||
Register(kResizeBilinearV2OpName, {1});
|
||||
Register(kReduceProdOpName, {1});
|
||||
|
|
|
|||
|
|
@ -390,8 +390,6 @@ constexpr auto kHcomOpTypeReceive = "HcomReceive";
|
|||
constexpr auto kHcomOpTypeReduceScatter = "HcomReduceScatter";
|
||||
|
||||
// attr key name
|
||||
constexpr auto kAttrAlignCorners = "align_corners";
|
||||
constexpr auto kAttrHalfPixelCenters = "half_pixel_centers";
|
||||
constexpr auto kAttrInputNames = "input_names";
|
||||
constexpr auto kAttrAttrNames = "attr_names";
|
||||
constexpr auto kAttrIsAiCpuKernel = "is_AICPU_kernel";
|
||||
|
|
|
|||
|
|
@ -17,7 +17,6 @@
|
|||
#define MINDSPORE_CCSRC_KERNEL_COMMON_UTILS_H_
|
||||
|
||||
#include <dirent.h>
|
||||
#include <limits>
|
||||
#include <memory>
|
||||
#include <unordered_map>
|
||||
#include <unordered_set>
|
||||
|
|
@ -165,23 +164,6 @@ std::string GetProcessorStr(const AnfNodePtr &anf_node);
|
|||
Processor GetProcessorFromContext();
|
||||
std::string GetStrProcessorFromContext();
|
||||
float Scaling(size_t in_size, size_t out_size, bool align_corners);
|
||||
inline float Scaler(const size_t x, const float scale, bool half_pixel_centers) {
|
||||
if (half_pixel_centers) {
|
||||
/**
|
||||
* function with a std::floor(), so instead of subtracting the 0.5 as we
|
||||
* do in HalfPixelScale, we leave it as is, as the std::floor does the
|
||||
* correct thing.
|
||||
* */
|
||||
return (static_cast<float>(x) + 0.5f) * scale;
|
||||
} else {
|
||||
/**
|
||||
* Older incorrect scaling method that causes all resizes to have a slight
|
||||
* translation leading to inconsistent results. For example, a flip then a
|
||||
* resize gives different results then a resize then a flip.
|
||||
* */
|
||||
return static_cast<float>(x) * scale;
|
||||
}
|
||||
}
|
||||
float ScaleGrid(const int x, const float scale);
|
||||
FusionType GetFusionTypeByName(const std::string &name);
|
||||
std::string GetFusionNameByType(const kernel::FusionType &type);
|
||||
|
|
|
|||
|
|
@ -62,8 +62,6 @@ constexpr auto kGather = "Gather";
|
|||
constexpr auto kIdentity = "Identity";
|
||||
constexpr auto kIdentityN = "IdentityN";
|
||||
constexpr auto kRandomChoiceWithMask = "RandomChoiceWithMask";
|
||||
constexpr auto kResizeNearestNeighborV2 = "ResizeNearestNeighborV2";
|
||||
constexpr auto kResizeNearestNeighborV2Grad = "ResizeNearestNeighborV2Grad";
|
||||
constexpr auto kUpdateCache = "UpdateCache";
|
||||
constexpr auto kCacheSwapTable = "CacheSwapTable";
|
||||
constexpr auto kSubAndFilter = "SubAndFilter";
|
||||
|
|
|
|||
|
|
@ -1,5 +1,5 @@
|
|||
/**
|
||||
* Copyright 2021-2022 Huawei Technologies Co., Ltd
|
||||
* Copyright 2021 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
|
|
@ -16,17 +16,8 @@
|
|||
|
||||
#ifndef MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_EIGEN_EIGEN_COMMON_UTILS_H_
|
||||
#define MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_EIGEN_EIGEN_COMMON_UTILS_H_
|
||||
#include <algorithm>
|
||||
#include <functional>
|
||||
#include <vector>
|
||||
#include "Eigen/Core"
|
||||
#include "Eigen/Dense"
|
||||
#include "unsupported/Eigen/CXX11/Tensor"
|
||||
|
||||
#ifdef _WIN32
|
||||
#undef ERROR
|
||||
#endif
|
||||
|
||||
#include "Eigen/Core"
|
||||
namespace mindspore {
|
||||
namespace kernel {
|
||||
using Eigen::ColMajor;
|
||||
|
|
@ -44,115 +35,6 @@ template <typename T>
|
|||
using MatrixSquare = Eigen::Matrix<T, Dynamic, Dynamic, RowMajor>;
|
||||
template <typename T>
|
||||
using ComplexMatrixSquare = Eigen::Matrix<std::complex<T>, Dynamic, Dynamic, RowMajor>;
|
||||
|
||||
template <typename T, int NDIMS = kDim1, typename IndexType = Eigen::DenseIndex>
|
||||
struct TTypes {
|
||||
// Rank-<NDIMS> tensor of scalar type T.
|
||||
typedef Eigen::TensorMap<Eigen::Tensor<T, NDIMS, Eigen::RowMajor, IndexType>, Eigen::Aligned> Tensor;
|
||||
typedef Eigen::TensorMap<Eigen::Tensor<const T, NDIMS, Eigen::RowMajor, IndexType>, Eigen::Aligned> ConstTensor;
|
||||
|
||||
// Rank-1 tensor (vector) of scalar type T.
|
||||
typedef Eigen::TensorMap<Eigen::Tensor<T, kDim1, Eigen::RowMajor, IndexType>, Eigen::Aligned> Flat;
|
||||
typedef Eigen::TensorMap<Eigen::Tensor<const T, kDim1, Eigen::RowMajor, IndexType>, Eigen::Aligned> ConstFlat;
|
||||
typedef Eigen::TensorMap<Eigen::Tensor<T, kDim1, Eigen::RowMajor, IndexType>, Eigen::Aligned> Vec;
|
||||
typedef Eigen::TensorMap<Eigen::Tensor<const T, kDim1, Eigen::RowMajor, IndexType>, Eigen::Aligned> ConstVec;
|
||||
|
||||
// Rank-2 tensor (matrix) of scalar type T.
|
||||
typedef Eigen::TensorMap<Eigen::Tensor<T, kDim2, Eigen::RowMajor, IndexType>, Eigen::Aligned> Matrix;
|
||||
typedef Eigen::TensorMap<Eigen::Tensor<const T, kDim2, Eigen::RowMajor, IndexType>, Eigen::Aligned> ConstMatrix;
|
||||
};
|
||||
|
||||
class EigenTensor {
|
||||
public:
|
||||
EigenTensor() = delete;
|
||||
EigenTensor(ShapeVector &shape, void *data_ptr) : tensor_shape(shape), tensor_data_ptr(data_ptr) {}
|
||||
~EigenTensor() = default;
|
||||
|
||||
/*
|
||||
* Eigen vec
|
||||
* @return Eigen vec
|
||||
*/
|
||||
template <typename T>
|
||||
typename TTypes<T>::Vec vec() {
|
||||
return tensor<T, 1>();
|
||||
}
|
||||
|
||||
/*
|
||||
* Eigen matrix
|
||||
* @return Eigen matrix
|
||||
*/
|
||||
template <typename T>
|
||||
typename TTypes<T>::Matrix matrix() {
|
||||
return tensor<T, kDim2>();
|
||||
}
|
||||
|
||||
/*
|
||||
* Eigen ConstMatrix
|
||||
* @return Eigen ConstMatrix
|
||||
*/
|
||||
template <typename T>
|
||||
typename TTypes<T>::ConstMatrix matrix() const {
|
||||
return tensor<T, kDim2>();
|
||||
}
|
||||
|
||||
/*
|
||||
* Eigen tensor
|
||||
* @return Eigen tensor
|
||||
*/
|
||||
template <typename T, size_t NDIMS>
|
||||
typename TTypes<T, NDIMS>::Tensor tensor() {
|
||||
return typename TTypes<T, NDIMS>::Tensor(reinterpret_cast<T *>(tensor_data_ptr), AsEigenDSizes<NDIMS>());
|
||||
}
|
||||
|
||||
/*
|
||||
* Eigen ConstTensor
|
||||
* @return Eigen ConstTensor
|
||||
*/
|
||||
template <typename T, size_t NDIMS>
|
||||
typename TTypes<T, NDIMS>::ConstTensor tensor() const {
|
||||
return typename TTypes<T, NDIMS>::ConstTensor(reinterpret_cast<const T *>(tensor_data_ptr), AsEigenDSizes<NDIMS>());
|
||||
}
|
||||
|
||||
/*
|
||||
* Eigen Flat
|
||||
* @return Eigen Flat
|
||||
*/
|
||||
template <typename T>
|
||||
typename TTypes<T>::Flat flat() {
|
||||
return typename TTypes<T>::Flat(
|
||||
reinterpret_cast<T *>(tensor_data_ptr),
|
||||
{std::accumulate(tensor_shape.begin(), tensor_shape.end(), 1, std::multiplies<int64_t>())});
|
||||
}
|
||||
|
||||
/*
|
||||
* which case we pad the rest of the sizes with 1.
|
||||
* @return Eigen::DSizes: pad the rest of the sizes with 1
|
||||
*/
|
||||
template <int NDIMS, typename IndexType>
|
||||
Eigen::DSizes<IndexType, NDIMS> AsEigenDSizesWithPadding() const {
|
||||
Eigen::DSizes<IndexType, NDIMS> dsizes;
|
||||
for (size_t d = 0; d < tensor_shape.size(); d++) {
|
||||
dsizes[d] = static_cast<IndexType>(tensor_shape[d]);
|
||||
}
|
||||
for (size_t d = tensor_shape.size(); d < NDIMS; d++) {
|
||||
dsizes[d] = 1;
|
||||
}
|
||||
return dsizes;
|
||||
}
|
||||
|
||||
/*
|
||||
* Fill `*dsizes` from `*this`
|
||||
* @return Eigen::DSizes: pad the rest of the sizes with 1
|
||||
*/
|
||||
template <int NDIMS, typename IndexType = Eigen::DenseIndex>
|
||||
Eigen::DSizes<IndexType, NDIMS> AsEigenDSizes() const {
|
||||
return AsEigenDSizesWithPadding<NDIMS, IndexType>();
|
||||
}
|
||||
|
||||
private:
|
||||
ShapeVector tensor_shape;
|
||||
void *tensor_data_ptr;
|
||||
};
|
||||
} // namespace kernel
|
||||
} // namespace mindspore
|
||||
#endif // MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_EIGEN_EIGEN_COMMON_UTILS_H_
|
||||
|
|
|
|||
|
|
@ -1,166 +0,0 @@
|
|||
/**
|
||||
* Copyright 2022 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#include "plugin/device/cpu/kernel/resize_nearest_neighbor_v2_cpu_kernel.h"
|
||||
#include <string>
|
||||
#include "kernel/common_utils.h"
|
||||
#include "plugin/device/cpu/hal/device/cpu_device_address.h"
|
||||
#include "plugin/device/cpu/kernel/eigen/eigen_common_utils.h"
|
||||
|
||||
namespace mindspore {
|
||||
namespace kernel {
|
||||
namespace {
|
||||
constexpr size_t kResizeNearestNeighborV2InputsNum = 2;
|
||||
constexpr size_t kResizeNearestNeighborV2OutputNum = 1;
|
||||
} // namespace
|
||||
|
||||
void ResizeNearestNeighborV2CpuKernelMod::InitKernel(const CNodePtr &kernel_node) {
|
||||
MS_EXCEPTION_IF_NULL(kernel_node);
|
||||
cnode_ptr_ = kernel_node;
|
||||
kernel_name_ = common::AnfAlgo::GetCNodeName(kernel_node);
|
||||
y_type_ = AnfAlgo::GetOutputDeviceDataType(kernel_node, kIndex0);
|
||||
x_shape_ = AnfAlgo::GetInputDeviceShape(kernel_node, kIndex0);
|
||||
y_shape_ = AnfAlgo::GetOutputDeviceShape(kernel_node, kIndex0);
|
||||
auto size_shape = AnfAlgo::GetInputDeviceShape(kernel_node, kIndex1);
|
||||
if (x_shape_.size() != kShape4dDims) {
|
||||
MS_EXCEPTION(ValueError) << "For '" << kernel_name_ << "', the dimension of 'x' should be " << kShape4dDims
|
||||
<< ", but got " << x_shape_.size();
|
||||
}
|
||||
if (size_shape.size() != kShape1dDims) {
|
||||
MS_EXCEPTION(ValueError) << "For '" << kernel_name_ << "', the dimension of 'size' should be " << kShape1dDims
|
||||
<< ", but got " << size_shape.size();
|
||||
}
|
||||
align_corners_ = common::AnfAlgo::GetNodeAttr<bool>(kernel_node, kAttrAlignCorners);
|
||||
half_pixel_centers_ = common::AnfAlgo::GetNodeAttr<bool>(kernel_node, kAttrHalfPixelCenters);
|
||||
std::string data_format = common::AnfAlgo::GetNodeAttr<std::string>(kernel_node, kAttrFormat);
|
||||
|
||||
if (data_format.compare(kOpFormat_NCHW) == 0) {
|
||||
dim_idx_map_ = {{'N', kIndex0}, {'C', kIndex1}, {'H', kIndex2}, {'W', kIndex3}};
|
||||
} else if (data_format.compare(kOpFormat_NHWC) == 0) {
|
||||
dim_idx_map_ = {{'N', kIndex0}, {'H', kIndex1}, {'W', kIndex2}, {'C', kIndex3}};
|
||||
} else {
|
||||
MS_EXCEPTION(ValueError) << "For '" << kernel_name_ << "', the attr of 'data_format' only support ["
|
||||
<< kOpFormat_NCHW << ", " << kOpFormat_NHWC << "].";
|
||||
}
|
||||
}
|
||||
|
||||
bool ResizeNearestNeighborV2CpuKernelMod::Launch(const std::vector<kernel::AddressPtr> &inputs,
|
||||
const std::vector<kernel::AddressPtr> &,
|
||||
const std::vector<kernel::AddressPtr> &outputs) {
|
||||
CHECK_KERNEL_INPUTS_NUM(inputs.size(), kResizeNearestNeighborV2InputsNum, kernel_name_);
|
||||
CHECK_KERNEL_OUTPUTS_NUM(outputs.size(), kResizeNearestNeighborV2OutputNum, kernel_name_);
|
||||
bool res = false;
|
||||
switch (y_type_) {
|
||||
case kNumberTypeUInt8:
|
||||
res = LaunchKernel<uint8_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeUInt16:
|
||||
res = LaunchKernel<uint16_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeInt8:
|
||||
res = LaunchKernel<int8_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeInt16:
|
||||
res = LaunchKernel<int16_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeInt32:
|
||||
res = LaunchKernel<int32_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeInt64:
|
||||
res = LaunchKernel<int64_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeFloat16:
|
||||
res = LaunchKernel<float16>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeFloat32:
|
||||
res = LaunchKernel<float>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeFloat64:
|
||||
res = LaunchKernel<double>(inputs, outputs);
|
||||
break;
|
||||
default:
|
||||
MS_EXCEPTION(TypeError)
|
||||
<< "For '" << kernel_name_
|
||||
<< "', the dtype of 'x' should be float16, float32, float64, int32, int64, int16, int8, uint16 or uin8 but got "
|
||||
<< TypeIdLabel(y_type_);
|
||||
}
|
||||
return res;
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
bool ResizeNearestNeighborV2CpuKernelMod::LaunchKernel(const std::vector<kernel::AddressPtr> &inputs,
|
||||
const std::vector<kernel::AddressPtr> &outputs) {
|
||||
const int64_t batch_size = x_shape_[dim_idx_map_['N']];
|
||||
const int64_t in_height = x_shape_[dim_idx_map_['H']];
|
||||
const int64_t in_width = x_shape_[dim_idx_map_['W']];
|
||||
const int64_t channels = x_shape_[dim_idx_map_['C']];
|
||||
const int64_t out_height = y_shape_[dim_idx_map_['H']];
|
||||
const int64_t out_width = y_shape_[dim_idx_map_['W']];
|
||||
|
||||
const float height_scale = Scaling(in_height, out_height, align_corners_);
|
||||
const float width_scale = Scaling(in_width, out_width, align_corners_);
|
||||
|
||||
auto x_4d = EigenTensor(x_shape_, inputs[kIndex0]->addr).tensor<T, kDim4>();
|
||||
auto y_4d = EigenTensor(y_shape_, outputs[kIndex0]->addr).tensor<T, kDim4>();
|
||||
for (int64_t b = 0; b < batch_size; ++b) {
|
||||
for (int64_t y = 0; y < out_height; ++y) {
|
||||
int64_t in_y =
|
||||
std::min((align_corners_) ? static_cast<int64_t>(roundf(Scaler(y, height_scale, half_pixel_centers_)))
|
||||
: static_cast<int64_t>(floorf(Scaler(y, height_scale, half_pixel_centers_))),
|
||||
in_height - 1);
|
||||
if (half_pixel_centers_) {
|
||||
in_y = std::max(static_cast<int64_t>(0), in_y);
|
||||
}
|
||||
for (int64_t x = 0; x < out_width; ++x) {
|
||||
int64_t in_x =
|
||||
std::min((align_corners_) ? static_cast<int64_t>(roundf(Scaler(x, width_scale, half_pixel_centers_)))
|
||||
: static_cast<int64_t>(floorf(Scaler(x, width_scale, half_pixel_centers_))),
|
||||
in_width - 1);
|
||||
if (half_pixel_centers_) {
|
||||
in_x = std::max(static_cast<int64_t>(0), in_x);
|
||||
}
|
||||
// data_format = NHWC
|
||||
if (dim_idx_map_['C'] == kIndex3) {
|
||||
std::copy_n(&x_4d(b, in_y, in_x, 0), channels, &y_4d(b, y, x, 0));
|
||||
} else {
|
||||
// data_format = NCHW
|
||||
for (int64_t c = 0; c < channels; ++c) {
|
||||
y_4d(b, c, y, x) = x_4d(b, c, in_y, in_x);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
return true;
|
||||
}
|
||||
|
||||
std::vector<KernelAttr> ResizeNearestNeighborV2CpuKernelMod::GetOpSupport() {
|
||||
static std::vector<KernelAttr> support_list = {
|
||||
KernelAttr().AddInputAttr(kNumberTypeUInt8).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt8),
|
||||
KernelAttr().AddInputAttr(kNumberTypeInt8).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt8),
|
||||
KernelAttr().AddInputAttr(kNumberTypeUInt16).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt16),
|
||||
KernelAttr().AddInputAttr(kNumberTypeInt16).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt16),
|
||||
KernelAttr().AddInputAttr(kNumberTypeInt32).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt32),
|
||||
KernelAttr().AddInputAttr(kNumberTypeInt64).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt64),
|
||||
KernelAttr().AddInputAttr(kNumberTypeFloat16).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat16),
|
||||
KernelAttr().AddInputAttr(kNumberTypeFloat32).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat32),
|
||||
KernelAttr().AddInputAttr(kNumberTypeFloat64).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat64)};
|
||||
|
||||
return support_list;
|
||||
}
|
||||
MS_KERNEL_FACTORY_REG(NativeCpuKernelMod, ResizeNearestNeighborV2, ResizeNearestNeighborV2CpuKernelMod);
|
||||
} // namespace kernel
|
||||
} // namespace mindspore
|
||||
|
|
@ -1,57 +0,0 @@
|
|||
/**
|
||||
* Copyright 2022 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_RESIZE_NEAREST_NEIGHBOR_V2_CPU_KERNEL_H_
|
||||
#define MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_RESIZE_NEAREST_NEIGHBOR_V2_CPU_KERNEL_H_
|
||||
|
||||
#include <algorithm>
|
||||
#include <unordered_map>
|
||||
#include <memory>
|
||||
#include <vector>
|
||||
#include "plugin/device/cpu/kernel/cpu_kernel.h"
|
||||
#include "kernel/common_utils.h"
|
||||
#include "plugin/factory/ms_factory.h"
|
||||
|
||||
namespace mindspore {
|
||||
namespace kernel {
|
||||
class ResizeNearestNeighborV2CpuKernelMod : public DeprecatedNativeCpuKernelMod {
|
||||
public:
|
||||
ResizeNearestNeighborV2CpuKernelMod() = default;
|
||||
~ResizeNearestNeighborV2CpuKernelMod() override = default;
|
||||
|
||||
void InitKernel(const CNodePtr &kernel_node) override;
|
||||
|
||||
bool Launch(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &workspace,
|
||||
const std::vector<AddressPtr> &outputs) override;
|
||||
|
||||
protected:
|
||||
std::vector<KernelAttr> GetOpSupport() override;
|
||||
|
||||
private:
|
||||
template <typename T>
|
||||
bool LaunchKernel(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &outputs);
|
||||
|
||||
TypeId y_type_{kTypeUnknown};
|
||||
bool align_corners_{false};
|
||||
bool half_pixel_centers_{false};
|
||||
std::vector<int64_t> x_shape_;
|
||||
std::vector<int64_t> y_shape_;
|
||||
std::unordered_map<char, size_t> dim_idx_map_;
|
||||
};
|
||||
} // namespace kernel
|
||||
} // namespace mindspore
|
||||
|
||||
#endif // MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_RESIZE_NEAREST_NEIGHBOR_V2_CPU_KERNEL_H_
|
||||
|
|
@ -1,162 +0,0 @@
|
|||
/**
|
||||
* Copyright 2022 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#include <string>
|
||||
#include "plugin/device/cpu/kernel/resize_nearest_neighbor_v2_grad_cpu_kernel.h"
|
||||
#include "kernel/common_utils.h"
|
||||
#include "plugin/device/cpu/hal/device/cpu_device_address.h"
|
||||
#include "plugin/device/cpu/kernel/eigen/eigen_common_utils.h"
|
||||
|
||||
namespace mindspore {
|
||||
namespace kernel {
|
||||
namespace {
|
||||
constexpr size_t kResizeNearestNeighborV2GradInputsNum = 2;
|
||||
constexpr size_t kResizeNearestNeighborV2GradOutputNum = 1;
|
||||
} // namespace
|
||||
|
||||
void ResizeNearestNeighborV2GradCpuKernelMod::InitKernel(const CNodePtr &kernel_node) {
|
||||
MS_EXCEPTION_IF_NULL(kernel_node);
|
||||
cnode_ptr_ = kernel_node;
|
||||
kernel_name_ = common::AnfAlgo::GetCNodeName(kernel_node);
|
||||
y_type_ = AnfAlgo::GetOutputDeviceDataType(kernel_node, kIndex0);
|
||||
grads_shape_ = AnfAlgo::GetInputDeviceShape(kernel_node, kIndex0);
|
||||
y_shape_ = AnfAlgo::GetOutputDeviceShape(kernel_node, kIndex0);
|
||||
auto size_shape = common::AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, kIndex1);
|
||||
if (grads_shape_.size() != kShape4dDims) {
|
||||
MS_EXCEPTION(ValueError) << "For '" << kernel_name_ << "', the dimension of 'x' should be " << kShape4dDims
|
||||
<< ", but got " << grads_shape_.size();
|
||||
}
|
||||
if (size_shape.size() != kShape1dDims) {
|
||||
MS_EXCEPTION(ValueError) << "For '" << kernel_name_ << "', the dimension of 'size' should be " << kShape1dDims
|
||||
<< ", but got " << size_shape.size();
|
||||
}
|
||||
align_corners_ = common::AnfAlgo::GetNodeAttr<bool>(kernel_node, kAttrAlignCorners);
|
||||
half_pixel_centers_ = common::AnfAlgo::GetNodeAttr<bool>(kernel_node, kAttrHalfPixelCenters);
|
||||
std::string data_format = common::AnfAlgo::GetNodeAttr<std::string>(kernel_node, kAttrFormat);
|
||||
|
||||
if (data_format.compare(kOpFormat_NCHW) == 0) {
|
||||
dim_idx_map_ = {{'N', kIndex0}, {'C', kIndex1}, {'H', kIndex2}, {'W', kIndex3}};
|
||||
} else if (data_format.compare(kOpFormat_NHWC) == 0) {
|
||||
dim_idx_map_ = {{'N', kIndex0}, {'H', kIndex1}, {'W', kIndex2}, {'C', kIndex3}};
|
||||
} else {
|
||||
MS_EXCEPTION(ValueError) << "For '" << kernel_name_ << "', the attr of 'data_format' only support ["
|
||||
<< kOpFormat_NCHW << ", " << kOpFormat_NHWC << "].";
|
||||
}
|
||||
}
|
||||
|
||||
bool ResizeNearestNeighborV2GradCpuKernelMod::Launch(const std::vector<AddressPtr> &inputs,
|
||||
const std::vector<AddressPtr> &workspace,
|
||||
const std::vector<AddressPtr> &outputs) {
|
||||
CHECK_KERNEL_INPUTS_NUM(inputs.size(), kResizeNearestNeighborV2GradInputsNum, kernel_name_);
|
||||
CHECK_KERNEL_OUTPUTS_NUM(outputs.size(), kResizeNearestNeighborV2GradOutputNum, kernel_name_);
|
||||
bool res = false;
|
||||
switch (y_type_) {
|
||||
case kNumberTypeUInt8:
|
||||
res = LaunchKernel<uint8_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeUInt16:
|
||||
res = LaunchKernel<uint16_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeInt8:
|
||||
res = LaunchKernel<int8_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeInt16:
|
||||
res = LaunchKernel<int16_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeInt32:
|
||||
res = LaunchKernel<int32_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeInt64:
|
||||
res = LaunchKernel<int64_t>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeFloat16:
|
||||
res = LaunchKernel<float16>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeFloat32:
|
||||
res = LaunchKernel<float>(inputs, outputs);
|
||||
break;
|
||||
case kNumberTypeFloat64:
|
||||
res = LaunchKernel<double>(inputs, outputs);
|
||||
break;
|
||||
default:
|
||||
MS_EXCEPTION(TypeError)
|
||||
<< "For '" << kernel_name_
|
||||
<< "', the dtype of 'x' should be float16, float32, float64, int32, int64, int16, int8, uint16 or uin8 but got "
|
||||
<< TypeIdLabel(y_type_);
|
||||
break;
|
||||
}
|
||||
return res;
|
||||
}
|
||||
template <typename T>
|
||||
bool ResizeNearestNeighborV2GradCpuKernelMod::LaunchKernel(const std::vector<kernel::AddressPtr> &inputs,
|
||||
const std::vector<kernel::AddressPtr> &outputs) {
|
||||
const int64_t batch_size = grads_shape_[dim_idx_map_['N']];
|
||||
const int64_t in_height = grads_shape_[dim_idx_map_['H']];
|
||||
const int64_t in_width = grads_shape_[dim_idx_map_['W']];
|
||||
const int64_t channels = grads_shape_[dim_idx_map_['C']];
|
||||
const int64_t out_height = y_shape_[dim_idx_map_['H']];
|
||||
const int64_t out_width = y_shape_[dim_idx_map_['W']];
|
||||
|
||||
const float height_scale = Scaling(out_height, in_height, align_corners_);
|
||||
const float width_scale = Scaling(out_width, in_width, align_corners_);
|
||||
|
||||
auto grads_4d = EigenTensor(grads_shape_, inputs[kIndex0]->addr).tensor<T, kDim4>();
|
||||
auto y_4d = EigenTensor(y_shape_, outputs[kIndex0]->addr).tensor<T, kDim4>();
|
||||
y_4d.setZero();
|
||||
|
||||
for (int64_t y = 0; y < in_height; ++y) {
|
||||
int64_t out_y =
|
||||
std::min((align_corners_) ? static_cast<int64_t>(roundf(Scaler(y, height_scale, half_pixel_centers_)))
|
||||
: static_cast<int64_t>(floorf(Scaler(y, height_scale, half_pixel_centers_))),
|
||||
out_height - 1);
|
||||
for (int64_t x = 0; x < in_width; ++x) {
|
||||
int64_t out_x =
|
||||
std::min((align_corners_) ? static_cast<int64_t>(roundf(Scaler(x, width_scale, half_pixel_centers_)))
|
||||
: static_cast<int64_t>(floorf(Scaler(x, width_scale, half_pixel_centers_))),
|
||||
out_width - 1);
|
||||
for (int64_t b = 0; b < batch_size; ++b) {
|
||||
for (int64_t c = 0; c < channels; ++c) {
|
||||
// data_format = NHWC
|
||||
if (dim_idx_map_['C'] == kIndex3) {
|
||||
y_4d(b, out_y, out_x, c) += grads_4d(b, y, x, c);
|
||||
} else {
|
||||
// data_format = NCHW
|
||||
y_4d(b, c, out_y, out_x) += grads_4d(b, c, y, x);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
return true;
|
||||
}
|
||||
|
||||
std::vector<KernelAttr> ResizeNearestNeighborV2GradCpuKernelMod::GetOpSupport() {
|
||||
static std::vector<KernelAttr> support_list = {
|
||||
KernelAttr().AddInputAttr(kNumberTypeUInt8).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt8),
|
||||
KernelAttr().AddInputAttr(kNumberTypeInt8).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt8),
|
||||
KernelAttr().AddInputAttr(kNumberTypeUInt16).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeUInt16),
|
||||
KernelAttr().AddInputAttr(kNumberTypeInt16).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt16),
|
||||
KernelAttr().AddInputAttr(kNumberTypeInt32).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt32),
|
||||
KernelAttr().AddInputAttr(kNumberTypeInt64).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt64),
|
||||
KernelAttr().AddInputAttr(kNumberTypeFloat16).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat16),
|
||||
KernelAttr().AddInputAttr(kNumberTypeFloat32).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat32),
|
||||
KernelAttr().AddInputAttr(kNumberTypeFloat64).AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeFloat64)};
|
||||
|
||||
return support_list;
|
||||
}
|
||||
MS_KERNEL_FACTORY_REG(NativeCpuKernelMod, ResizeNearestNeighborV2Grad, ResizeNearestNeighborV2GradCpuKernelMod);
|
||||
} // namespace kernel
|
||||
} // namespace mindspore
|
||||
|
|
@ -1,55 +0,0 @@
|
|||
/**
|
||||
* Copyright 2022 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_RESIZE_NEAREST_NEIGHBOR_V2_GRAD_CPU_KERNEL_H_
|
||||
#define MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_RESIZE_NEAREST_NEIGHBOR_V2_GRAD_CPU_KERNEL_H_
|
||||
|
||||
#include <algorithm>
|
||||
#include <memory>
|
||||
#include <unordered_map>
|
||||
#include <vector>
|
||||
#include "kernel/common_utils.h"
|
||||
#include "plugin/device/cpu/kernel/cpu_kernel.h"
|
||||
#include "plugin/factory/ms_factory.h"
|
||||
|
||||
namespace mindspore {
|
||||
namespace kernel {
|
||||
class ResizeNearestNeighborV2GradCpuKernelMod : public DeprecatedNativeCpuKernelMod {
|
||||
public:
|
||||
ResizeNearestNeighborV2GradCpuKernelMod() = default;
|
||||
~ResizeNearestNeighborV2GradCpuKernelMod() override = default;
|
||||
|
||||
void InitKernel(const CNodePtr &kernel_node) override;
|
||||
|
||||
bool Launch(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &workspace,
|
||||
const std::vector<AddressPtr> &outputs) override;
|
||||
|
||||
protected:
|
||||
std::vector<KernelAttr> GetOpSupport() override;
|
||||
|
||||
private:
|
||||
template <typename T>
|
||||
bool LaunchKernel(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &outputs);
|
||||
TypeId y_type_{kTypeUnknown};
|
||||
bool align_corners_{false};
|
||||
bool half_pixel_centers_{false};
|
||||
std::vector<int64_t> grads_shape_;
|
||||
std::vector<int64_t> y_shape_;
|
||||
std::unordered_map<char, size_t> dim_idx_map_;
|
||||
};
|
||||
} // namespace kernel
|
||||
} // namespace mindspore
|
||||
#endif // MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_RESIZE_NEAREST_NEIGHBOR_V2_GRAD_CPU_KERNEL_H_
|
||||
|
|
@ -80,8 +80,6 @@ PrimShapeDependMap &GetHostDependsMap() {
|
|||
static const auto &kNonDeterministicInts = prim::kPrimNonDeterministicInts->name();
|
||||
static const auto &kSliceGrad = prim::kPrimSliceGrad->name();
|
||||
static const auto &kReshape = prim::kPrimReshape->name();
|
||||
static const auto &kResizeNearestNeighborV2 = prim::kPrimResizeNearestNeighborV2->name();
|
||||
static const auto &kResizeNearestNeighborV2Grad = prim::kPrimResizeNearestNeighborV2Grad->name();
|
||||
static const auto &kScatterNd = prim::kPrimScatterNd->name();
|
||||
static const auto &kTruncatedNormal = prim::kPrimTruncatedNormal->name();
|
||||
static const auto &kRandomGamma = prim::kPrimRandomGamma->name();
|
||||
|
|
@ -124,8 +122,6 @@ PrimShapeDependMap &GetHostDependsMap() {
|
|||
{kTile, ShapeSet{1}},
|
||||
{kTopK, ShapeSet{1}},
|
||||
{kReshape, ShapeSet{1}},
|
||||
{kResizeNearestNeighborV2, ShapeSet{1}},
|
||||
{kResizeNearestNeighborV2Grad, ShapeSet{1}},
|
||||
{kScatterNd, ShapeSet{2}},
|
||||
{kSliceGrad, ShapeSet{2, 3}},
|
||||
{kFillV2, ShapeSet{0}},
|
||||
|
|
|
|||
|
|
@ -449,8 +449,6 @@ GVAR_DEF(PrimitivePtr, kPrimParallelResizeBilinearGrad, std::make_shared<Primiti
|
|||
GVAR_DEF(PrimitivePtr, kPrimResizeGrad, std::make_shared<Primitive>("ResizeGrad"));
|
||||
GVAR_DEF(PrimitivePtr, kPrimResizeNearestNeighbor, std::make_shared<Primitive>("ResizeNearestNeighbor"));
|
||||
GVAR_DEF(PrimitivePtr, kPrimResizeNearestNeighborGrad, std::make_shared<Primitive>("ResizeNearestNeighborGrad"));
|
||||
GVAR_DEF(PrimitivePtr, kPrimResizeNearestNeighborV2, std::make_shared<Primitive>("ResizeNearestNeighborV2"));
|
||||
GVAR_DEF(PrimitivePtr, kPrimResizeNearestNeighborV2Grad, std::make_shared<Primitive>("ResizeNearestNeighborV2Grad"));
|
||||
GVAR_DEF(PrimitivePtr, kPrimDynamicResizeNearestNeighbor, std::make_shared<Primitive>("DynamicResizeNearestNeighbor"));
|
||||
GVAR_DEF(PrimitivePtr, kPrimResizeLinear1D, std::make_shared<Primitive>("ResizeLinear1D"));
|
||||
GVAR_DEF(PrimitivePtr, kPrimResizeLinear1DGrad, std::make_shared<Primitive>("ResizeLinear1DGrad"));
|
||||
|
|
|
|||
|
|
@ -1,121 +0,0 @@
|
|||
/**
|
||||
* Copyright 2022 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#include "ops/grad/resize_nearest_neighbor_v2_grad.h"
|
||||
#include <string>
|
||||
#include <algorithm>
|
||||
#include <memory>
|
||||
#include <map>
|
||||
#include <set>
|
||||
#include <vector>
|
||||
#include "ops/op_utils.h"
|
||||
#include "utils/check_convert_utils.h"
|
||||
#include "abstract/ops/primitive_infer_map.h"
|
||||
#include "mindapi/src/helper.h"
|
||||
|
||||
namespace mindspore {
|
||||
namespace ops {
|
||||
namespace {
|
||||
#define IsNoneOrAnyValue(value_ptr) ((value_ptr->isa<None>()) || (value_ptr->isa<AnyValue>()))
|
||||
abstract::ShapePtr ResizeNearestNeighborV2GradInferShape(const PrimitivePtr &primitive,
|
||||
const std::vector<AbstractBasePtr> &input_args) {
|
||||
auto prim_name = primitive->name();
|
||||
auto grads_shape = CheckAndConvertUtils::ConvertShapePtrToShapeMap(input_args[kInputIndex0]->BuildShape())[kShape];
|
||||
auto size_shape = CheckAndConvertUtils::ConvertShapePtrToShapeMap(input_args[kInputIndex1]->BuildShape())[kShape];
|
||||
auto size_ptr = input_args[kInputIndex1]->BuildValue();
|
||||
|
||||
(void)CheckAndConvertUtils::CheckInteger("dimension of grads", SizeToLong(grads_shape.size()), kEqual,
|
||||
SizeToLong(kDim4), prim_name);
|
||||
(void)CheckAndConvertUtils::CheckInteger("dimension of size", SizeToLong(size_shape.size()), kEqual,
|
||||
SizeToLong(kDim1), prim_name);
|
||||
|
||||
auto data_format = CheckAndConvertUtils::GetAndCheckFormat(primitive->GetAttr(kFormat));
|
||||
std::map<char, size_t> dim_idx_map;
|
||||
auto align_corners_ptr = primitive->GetAttr(kAlignCorners);
|
||||
MS_EXCEPTION_IF_NULL(align_corners_ptr);
|
||||
auto align_corners = GetValue<bool>(align_corners_ptr);
|
||||
auto half_pixel_centers_ptr = primitive->GetAttr(kHalfPixelCenters);
|
||||
MS_EXCEPTION_IF_NULL(half_pixel_centers_ptr);
|
||||
auto half_pixel_centers = GetValue<bool>(half_pixel_centers_ptr);
|
||||
if (align_corners && half_pixel_centers) {
|
||||
MS_EXCEPTION(ValueError) << "For '" << prim_name << ". If half_pixel_centers is True, align_corners must be False.";
|
||||
}
|
||||
|
||||
if (data_format == Format::NCHW) {
|
||||
dim_idx_map = {{'N', kInputIndex0}, {'C', kInputIndex1}, {'H', kInputIndex2}, {'W', kInputIndex3}};
|
||||
} else if (data_format == Format::NHWC) {
|
||||
dim_idx_map = {{'N', kInputIndex0}, {'H', kInputIndex1}, {'W', kInputIndex2}, {'C', kInputIndex3}};
|
||||
} else {
|
||||
MS_EXCEPTION(ValueError) << "For '" << prim_name << "', the attr of 'data_format' only support [" << kFormatNCHW
|
||||
<< ", " << kFormatNHWC << "]. But get '" << data_format << "'.";
|
||||
}
|
||||
|
||||
bool is_compile = IsNoneOrAnyValue(size_ptr);
|
||||
ShapeVector y_shape(kDim4);
|
||||
if (is_compile) {
|
||||
y_shape[dim_idx_map['N']] = grads_shape[dim_idx_map['N']];
|
||||
y_shape[dim_idx_map['C']] = grads_shape[dim_idx_map['C']];
|
||||
y_shape[dim_idx_map['H']] = abstract::Shape::SHP_ANY;
|
||||
y_shape[dim_idx_map['W']] = abstract::Shape::SHP_ANY;
|
||||
ShapeVector y_shape_min(y_shape);
|
||||
y_shape_min[dim_idx_map['H']] = 0;
|
||||
y_shape_min[dim_idx_map['W']] = 0;
|
||||
ShapeVector y_shape_max(grads_shape);
|
||||
return std::make_shared<abstract::Shape>(y_shape, y_shape_min, y_shape_max);
|
||||
} else {
|
||||
MS_EXCEPTION_IF_NULL(size_ptr);
|
||||
auto size_value = CheckAndConvertUtils::CheckTensorIntValue("input size", size_ptr, prim_name);
|
||||
if (size_value.size() != kDim2) {
|
||||
MS_EXCEPTION(ValueError) << "For '" << prim_name << "', the elements number of 'size' should be 2, but get "
|
||||
<< size_value.size() << " number.";
|
||||
}
|
||||
y_shape[dim_idx_map['N']] = grads_shape[dim_idx_map['N']];
|
||||
y_shape[dim_idx_map['C']] = grads_shape[dim_idx_map['C']];
|
||||
y_shape[dim_idx_map['H']] = size_value.front();
|
||||
y_shape[dim_idx_map['W']] = size_value.back();
|
||||
}
|
||||
return std::make_shared<abstract::Shape>(y_shape);
|
||||
}
|
||||
|
||||
TypePtr ResizeNearestNeighborV2GradInferType(const PrimitivePtr &primitive,
|
||||
const std::vector<AbstractBasePtr> &input_args) {
|
||||
std::set<TypePtr> support_types = {kInt8, kInt16, kInt32, kInt64, kUInt8, kUInt16, kFloat16, kFloat32, kFloat64};
|
||||
auto grads_type = CheckAndConvertUtils::CheckTensorTypeValid("grads", input_args[kInputIndex0]->BuildType(),
|
||||
support_types, primitive->name());
|
||||
(void)CheckAndConvertUtils::CheckTensorTypeValid("size", input_args[kInputIndex1]->BuildType(), {kInt32, kInt64},
|
||||
primitive->name());
|
||||
return grads_type;
|
||||
}
|
||||
} // namespace
|
||||
|
||||
AbstractBasePtr ResizeNearestNeighborV2GradInfer(const abstract::AnalysisEnginePtr &, const PrimitivePtr &primitive,
|
||||
const std::vector<AbstractBasePtr> &input_args) {
|
||||
MS_EXCEPTION_IF_NULL(primitive);
|
||||
auto prim_name = primitive->name();
|
||||
constexpr int64_t input_num = 2;
|
||||
CheckAndConvertUtils::CheckInputArgs(input_args, kEqual, input_num, prim_name);
|
||||
(void)CheckAndConvertUtils::CheckArgs<abstract::AbstractTensor>(prim_name, input_args, kInputIndex0);
|
||||
(void)CheckAndConvertUtils::CheckArgs<abstract::AbstractTensor>(prim_name, input_args, kInputIndex1);
|
||||
auto infer_shape = ResizeNearestNeighborV2GradInferShape(primitive, input_args);
|
||||
auto infer_type = ResizeNearestNeighborV2GradInferType(primitive, input_args);
|
||||
return abstract::MakeAbstract(infer_shape, infer_type);
|
||||
}
|
||||
|
||||
MIND_API_OPERATOR_IMPL(ResizeNearestNeighborV2Grad, BaseOperator);
|
||||
REGISTER_PRIMITIVE_EVAL_IMPL(ResizeNearestNeighborV2Grad, prim::kPrimResizeNearestNeighborV2Grad,
|
||||
ResizeNearestNeighborV2GradInfer, nullptr, true);
|
||||
} // namespace ops
|
||||
} // namespace mindspore
|
||||
|
|
@ -1,46 +0,0 @@
|
|||
/**
|
||||
* Copyright 2022 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
#ifndef MINDSPORE_CORE_OPS_RESIZE_NEAREST_NEIGHBOR_V2_GRAD_H_
|
||||
#define MINDSPORE_CORE_OPS_RESIZE_NEAREST_NEIGHBOR_V2_GRAD_H_
|
||||
|
||||
#include <memory>
|
||||
#include <vector>
|
||||
|
||||
#include "abstract/abstract_value.h"
|
||||
#include "ops/base_operator.h"
|
||||
#include "utils/check_convert_utils.h"
|
||||
#include "mindapi/base/types.h"
|
||||
|
||||
namespace mindspore {
|
||||
namespace ops {
|
||||
constexpr auto kNameResizeNearestNeighborV2Grad = "ResizeNearestNeighborV2Grad";
|
||||
/// \brief the grad operation of @ref mindspore.ops.ResizeNearestNeighborV2
|
||||
/// Refer to Python API @ref mindspore._grad_ops.ResizeNearestNeighborV2Grad for more details.
|
||||
class MIND_API ResizeNearestNeighborV2Grad : public BaseOperator {
|
||||
public:
|
||||
MIND_API_BASE_MEMBER(ResizeNearestNeighborV2Grad);
|
||||
|
||||
/// \brief Constructor.
|
||||
ResizeNearestNeighborV2Grad() : BaseOperator(kNameResizeNearestNeighborV2Grad) {}
|
||||
};
|
||||
|
||||
AbstractBasePtr ResizeNearestNeighborV2GradInfer(const abstract::AnalysisEnginePtr &, const PrimitivePtr &primitive,
|
||||
const std::vector<AbstractBasePtr> &input_args);
|
||||
using PrimResizeNearestNeighborV2GradPtr = std::shared_ptr<ResizeNearestNeighborV2Grad>;
|
||||
} // namespace ops
|
||||
} // namespace mindspore
|
||||
|
||||
#endif // MINDSPORE_CORE_OPS_RESIZE_NEAREST_NEIGHBOR_V2_GRAD_H_
|
||||
|
|
@ -275,8 +275,6 @@ constexpr auto kSymmetric = "symmetric";
|
|||
constexpr auto kDstType = "dst_type";
|
||||
constexpr auto kNone = "none";
|
||||
constexpr auto kMean = "mean";
|
||||
constexpr auto kFormatNCHW = "NCHW";
|
||||
constexpr auto kFormatNHWC = "NHWC";
|
||||
constexpr auto kBatchMean = "batchmean";
|
||||
constexpr auto kSum = "sum";
|
||||
constexpr auto kIndices = "indices";
|
||||
|
|
|
|||
|
|
@ -1,121 +0,0 @@
|
|||
/**
|
||||
* Copyright 2022 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#include "ops/resize_nearest_neighbor_v2.h"
|
||||
#include <string>
|
||||
#include <algorithm>
|
||||
#include <memory>
|
||||
#include <map>
|
||||
#include <set>
|
||||
#include <vector>
|
||||
#include "ops/op_utils.h"
|
||||
#include "utils/check_convert_utils.h"
|
||||
#include "abstract/ops/primitive_infer_map.h"
|
||||
#include "mindapi/src/helper.h"
|
||||
#include "mindapi/base/types.h"
|
||||
|
||||
namespace mindspore {
|
||||
namespace ops {
|
||||
namespace {
|
||||
#define IsSameType(source_type, cmp_type) (cmp_type->equal(source_type))
|
||||
#define IsNoneOrAnyValue(value_ptr) ((value_ptr->isa<None>()) || (value_ptr->isa<AnyValue>()))
|
||||
|
||||
abstract::ShapePtr ResizeNearestNeighborV2InferShape(const PrimitivePtr &primitive,
|
||||
const std::vector<AbstractBasePtr> &input_args) {
|
||||
auto prim_name = primitive->name();
|
||||
auto x_shape = CheckAndConvertUtils::ConvertShapePtrToShapeMap(input_args[kInputIndex0]->BuildShape())[kShape];
|
||||
auto size_shape = CheckAndConvertUtils::ConvertShapePtrToShapeMap(input_args[kInputIndex1]->BuildShape())[kShape];
|
||||
auto size_ptr = input_args[kInputIndex1]->BuildValue();
|
||||
|
||||
(void)CheckAndConvertUtils::CheckInteger("dimension of x", SizeToLong(x_shape.size()), kEqual, SizeToLong(kDim4),
|
||||
prim_name);
|
||||
(void)CheckAndConvertUtils::CheckInteger("dimension of size", SizeToLong(size_shape.size()), kEqual,
|
||||
SizeToLong(kDim1), prim_name);
|
||||
|
||||
auto align_corners_ptr = primitive->GetAttr(kAlignCorners);
|
||||
MS_EXCEPTION_IF_NULL(align_corners_ptr);
|
||||
auto align_corners = GetValue<bool>(align_corners_ptr);
|
||||
auto half_pixel_centers_ptr = primitive->GetAttr(kHalfPixelCenters);
|
||||
MS_EXCEPTION_IF_NULL(half_pixel_centers_ptr);
|
||||
auto half_pixel_centers = GetValue<bool>(half_pixel_centers_ptr);
|
||||
if (align_corners && half_pixel_centers) {
|
||||
MS_EXCEPTION(ValueError) << "For '" << prim_name << ". If half_pixel_centers is True, align_corners must be False.";
|
||||
}
|
||||
|
||||
auto data_format = CheckAndConvertUtils::GetAndCheckFormat(primitive->GetAttr(kFormat));
|
||||
std::map<char, size_t> dim_idx_map;
|
||||
|
||||
if (data_format == Format::NCHW) {
|
||||
dim_idx_map = {{'N', kInputIndex0}, {'C', kInputIndex1}, {'H', kInputIndex2}, {'W', kInputIndex3}};
|
||||
} else if (data_format == Format::NHWC) {
|
||||
dim_idx_map = {{'N', kInputIndex0}, {'H', kInputIndex1}, {'W', kInputIndex2}, {'C', kInputIndex3}};
|
||||
} else {
|
||||
MS_EXCEPTION(ValueError) << "For '" << prim_name << "', the attr of 'data_format' only support [" << kFormatNCHW
|
||||
<< ", " << kFormatNHWC << "]. But get '" << data_format << "'.";
|
||||
}
|
||||
|
||||
bool is_compile = IsNoneOrAnyValue(size_ptr);
|
||||
ShapeVector y_shape(kDim4);
|
||||
y_shape[dim_idx_map['N']] = x_shape[dim_idx_map['N']];
|
||||
y_shape[dim_idx_map['C']] = x_shape[dim_idx_map['C']];
|
||||
if (is_compile) {
|
||||
y_shape[dim_idx_map['H']] = abstract::Shape::SHP_ANY;
|
||||
y_shape[dim_idx_map['W']] = abstract::Shape::SHP_ANY;
|
||||
ShapeVector y_shape_min(y_shape);
|
||||
y_shape_min[dim_idx_map['H']] = 0;
|
||||
y_shape_min[dim_idx_map['W']] = 0;
|
||||
ShapeVector y_shape_max(x_shape);
|
||||
return std::make_shared<abstract::Shape>(y_shape, y_shape_min, y_shape_max);
|
||||
} else {
|
||||
MS_EXCEPTION_IF_NULL(size_ptr);
|
||||
auto size_value = CheckAndConvertUtils::CheckTensorIntValue("input size", size_ptr, prim_name);
|
||||
if (size_value.size() != kDim2) {
|
||||
MS_EXCEPTION(ValueError) << "For '" << prim_name << "', the elements number of 'size' should be 2, but get "
|
||||
<< size_value.size() << " number.";
|
||||
}
|
||||
y_shape[dim_idx_map['H']] = size_value.front();
|
||||
y_shape[dim_idx_map['W']] = size_value.back();
|
||||
}
|
||||
return std::make_shared<abstract::Shape>(y_shape);
|
||||
}
|
||||
|
||||
TypePtr ResizeNearestNeighborV2InferType(const PrimitivePtr &primitive,
|
||||
const std::vector<AbstractBasePtr> &input_args) {
|
||||
std::set<TypePtr> support_types = {kInt8, kInt16, kInt32, kInt64, kUInt8, kUInt16, kFloat16, kFloat32, kFloat64};
|
||||
auto start_type = CheckAndConvertUtils::CheckTensorTypeValid("x", input_args[kInputIndex0]->BuildType(),
|
||||
support_types, primitive->name());
|
||||
(void)CheckAndConvertUtils::CheckTensorTypeValid("size", input_args[kInputIndex1]->BuildType(), {kInt32, kInt64},
|
||||
primitive->name());
|
||||
return start_type;
|
||||
}
|
||||
} // namespace
|
||||
|
||||
AbstractBasePtr ResizeNearestNeighborV2Infer(const abstract::AnalysisEnginePtr &, const PrimitivePtr &primitive,
|
||||
const std::vector<AbstractBasePtr> &input_args) {
|
||||
MS_EXCEPTION_IF_NULL(primitive);
|
||||
auto prim_name = primitive->name();
|
||||
constexpr int64_t input_num = 2;
|
||||
CheckAndConvertUtils::CheckInputArgs(input_args, kEqual, input_num, prim_name);
|
||||
auto infer_type = ResizeNearestNeighborV2InferType(primitive, input_args);
|
||||
auto infer_shape = ResizeNearestNeighborV2InferShape(primitive, input_args);
|
||||
return abstract::MakeAbstract(infer_shape, infer_type);
|
||||
}
|
||||
|
||||
MIND_API_OPERATOR_IMPL(ResizeNearestNeighborV2, BaseOperator);
|
||||
REGISTER_PRIMITIVE_EVAL_IMPL(ResizeNearestNeighborV2, prim::kPrimResizeNearestNeighborV2, ResizeNearestNeighborV2Infer,
|
||||
nullptr, true);
|
||||
} // namespace ops
|
||||
} // namespace mindspore
|
||||
|
|
@ -1,46 +0,0 @@
|
|||
/**
|
||||
* Copyright 2022 Huawei Technologies Co., Ltd
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
#ifndef MINDSPORE_CORE_OPS_RESIZE_NEAREST_NEIGHBOR_V2_H_
|
||||
#define MINDSPORE_CORE_OPS_RESIZE_NEAREST_NEIGHBOR_V2_H_
|
||||
|
||||
#include <memory>
|
||||
#include <vector>
|
||||
|
||||
#include "abstract/abstract_value.h"
|
||||
#include "ops/base_operator.h"
|
||||
#include "utils/check_convert_utils.h"
|
||||
#include "mindapi/base/types.h"
|
||||
|
||||
namespace mindspore {
|
||||
namespace ops {
|
||||
constexpr auto kNameResizeNearestNeighborV2 = "ResizeNearestNeighborV2";
|
||||
/// \brief Resizes the input tensor by using the nearest neighbor algorithm.
|
||||
/// Refer to Python API @ref mindspore.ops.ResizeNearestNeighborV2 for more details.
|
||||
class MIND_API ResizeNearestNeighborV2 : public BaseOperator {
|
||||
public:
|
||||
MIND_API_BASE_MEMBER(ResizeNearestNeighborV2);
|
||||
|
||||
/// \brief Constructor.
|
||||
ResizeNearestNeighborV2() : BaseOperator(kNameResizeNearestNeighborV2) {}
|
||||
};
|
||||
|
||||
AbstractBasePtr ResizeNearestNeighborV2Infer(const abstract::AnalysisEnginePtr &, const PrimitivePtr &primitive,
|
||||
const std::vector<AbstractBasePtr> &input_args);
|
||||
using PrimResizeNearestNeighborV2Ptr = std::shared_ptr<ResizeNearestNeighborV2>;
|
||||
} // namespace ops
|
||||
} // namespace mindspore
|
||||
|
||||
#endif // MINDSPORE_CORE_OPS_RESIZE_NEAREST_NEIGHBOR_V2_H_
|
||||
|
|
@ -16,7 +16,6 @@
|
|||
"""array_ops"""
|
||||
|
||||
from mindspore import Tensor
|
||||
from mindspore.ops.primitive import constexpr
|
||||
from ...common import dtype as mstype
|
||||
from ...numpy.array_ops import where
|
||||
from .._grad.grad_math_ops import binop_grad_common
|
||||
|
|
@ -25,7 +24,6 @@ from ..composite.multitype_ops.zeros_like_impl import zeros_like
|
|||
from ..operations.array_ops import Tril
|
||||
from ..operations.array_ops import MatrixDiagV3
|
||||
from ..operations.array_ops import MatrixDiagPartV3
|
||||
from ..operations.array_ops import ResizeNearestNeighborV2
|
||||
from ..operations.array_ops import MatrixSetDiagV3
|
||||
from ..operations.array_ops import Triu
|
||||
from ..operations.array_ops import IdentityN
|
||||
|
|
@ -37,12 +35,6 @@ from ..operations.array_ops import Expand
|
|||
from .. import functional as F
|
||||
from .. import operations as P
|
||||
from .._utils.utils import is_shape_unknown
|
||||
from ..operations import _grad_ops as G
|
||||
|
||||
|
||||
@constexpr
|
||||
def _create_tensor(data, dtype):
|
||||
return Tensor(data, dtype=dtype)
|
||||
|
||||
|
||||
def _segment_min_or_max_grad(segment_sum_op, input_x, segment_ids, output, dout):
|
||||
|
|
@ -272,25 +264,6 @@ def get_bprop_identity_n(self):
|
|||
return bprop
|
||||
|
||||
|
||||
@bprop_getters.register(ResizeNearestNeighborV2)
|
||||
def get_bprop_resize_nearest_neighbor_v2(self):
|
||||
"""Generate bprop for ResizeNearestNeighborV2"""
|
||||
align_corners = self.align_corners
|
||||
half_pixel_centers = self.half_pixel_centers
|
||||
data_format = self.data_format
|
||||
grad_op = G.ResizeNearestNeighborV2Grad(align_corners, half_pixel_centers, data_format)
|
||||
|
||||
def bprop(x, size, output, dout):
|
||||
x_shape = P.Shape()(x)
|
||||
grad_in_size = x_shape[1:3]
|
||||
if data_format == 'NCHW':
|
||||
grad_in_size = x_shape[2:4]
|
||||
dx = grad_op(dout, _create_tensor(grad_in_size, mstype.int32))
|
||||
return dx, zeros_like(grad_in_size)
|
||||
|
||||
return bprop
|
||||
|
||||
|
||||
@bprop_getters.register(P.ExtractVolumePatches)
|
||||
def get_bprop_extract_volume_patches(self):
|
||||
"""Generate bprop for ExtractVolumePatches"""
|
||||
|
|
|
|||
|
|
@ -153,8 +153,6 @@ from .reduce_prod import _reduce_prod_aicpu
|
|||
from .reduce_mean import _reduce_mean_aicpu
|
||||
from .resize_bilinear import _resize_bilinear_aicpu
|
||||
from .resize_bilinear_grad import _resize_bilinear_grad_aicpu
|
||||
from .resize_nearest_neighbor_v2 import _resize_nearest_neighbor_v2_aicpu
|
||||
from .resize_nearest_neighbor_v2_grad import _resize_nearest_neighbor_v2_grad_aicpu
|
||||
from .scatter_elements import _scatter_elements_aicpu
|
||||
from .non_max_suppression import _non_max_suppression_aicpu
|
||||
from .square import _square_aicpu
|
||||
|
|
|
|||
|
|
@ -1,42 +0,0 @@
|
|||
# Copyright 2022 Huawei Technologies Co., Ltd
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
# ============================================================================
|
||||
|
||||
"""ResizeNearestNeighborV2 op"""
|
||||
from mindspore.ops.op_info_register import op_info_register, AiCPURegOp, DataType
|
||||
|
||||
resize_nearest_neighbor_v2_op_info = AiCPURegOp("ResizeNearestNeighborV2") \
|
||||
.fusion_type("OPAQUE") \
|
||||
.attr("align_corners", "bool") \
|
||||
.attr("half_pixel_centers", "bool") \
|
||||
.attr("format", "str") \
|
||||
.input(0, "x", "required") \
|
||||
.input(1, "size", "required") \
|
||||
.output(0, "y", "dynamic") \
|
||||
.dtype_format(DataType.I8_Default, DataType.I32_Default, DataType.I8_Default) \
|
||||
.dtype_format(DataType.U8_Default, DataType.I32_Default, DataType.U8_Default) \
|
||||
.dtype_format(DataType.I16_Default, DataType.I32_Default, DataType.I16_Default) \
|
||||
.dtype_format(DataType.U16_Default, DataType.I32_Default, DataType.U16_Default) \
|
||||
.dtype_format(DataType.I32_Default, DataType.I32_Default, DataType.I32_Default) \
|
||||
.dtype_format(DataType.I64_Default, DataType.I32_Default, DataType.I64_Default) \
|
||||
.dtype_format(DataType.F16_Default, DataType.I32_Default, DataType.F16_Default) \
|
||||
.dtype_format(DataType.F32_Default, DataType.I32_Default, DataType.F32_Default) \
|
||||
.dtype_format(DataType.F64_Default, DataType.I32_Default, DataType.F64_Default) \
|
||||
.get_op_info()
|
||||
|
||||
|
||||
@op_info_register(resize_nearest_neighbor_v2_op_info)
|
||||
def _resize_nearest_neighbor_v2_aicpu():
|
||||
"""ResizeNearestNeighborV2 AiCPU register"""
|
||||
return
|
||||
|
|
@ -1,42 +0,0 @@
|
|||
# Copyright 2022 Huawei Technologies Co., Ltd
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
# ============================================================================
|
||||
|
||||
"""ResizeNearestNeighborV2Grad op"""
|
||||
from mindspore.ops.op_info_register import op_info_register, AiCPURegOp, DataType
|
||||
|
||||
resize_nearest_neighbor_v2_grad_op_info = AiCPURegOp("ResizeNearestNeighborV2Grad") \
|
||||
.fusion_type("OPAQUE") \
|
||||
.attr("align_corners", "bool") \
|
||||
.attr("half_pixel_centers", "bool") \
|
||||
.attr("format", "str") \
|
||||
.input(0, "grads", "required") \
|
||||
.input(1, "size", "required") \
|
||||
.output(0, "y", "required") \
|
||||
.dtype_format(DataType.I8_Default, DataType.I32_Default, DataType.I8_Default) \
|
||||
.dtype_format(DataType.U8_Default, DataType.I32_Default, DataType.U8_Default) \
|
||||
.dtype_format(DataType.I16_Default, DataType.I32_Default, DataType.I16_Default) \
|
||||
.dtype_format(DataType.U16_Default, DataType.I32_Default, DataType.U16_Default) \
|
||||
.dtype_format(DataType.I32_Default, DataType.I32_Default, DataType.I32_Default) \
|
||||
.dtype_format(DataType.I64_Default, DataType.I32_Default, DataType.I64_Default) \
|
||||
.dtype_format(DataType.F16_Default, DataType.I32_Default, DataType.F16_Default) \
|
||||
.dtype_format(DataType.F32_Default, DataType.I32_Default, DataType.F32_Default) \
|
||||
.dtype_format(DataType.F64_Default, DataType.I32_Default, DataType.F64_Default) \
|
||||
.get_op_info()
|
||||
|
||||
|
||||
@op_info_register(resize_nearest_neighbor_v2_grad_op_info)
|
||||
def _resize_nearest_neighbor_v2_grad_aicpu():
|
||||
"""ResizeNearestNeighborV2Grad AiCPU register"""
|
||||
return
|
||||
|
|
@ -1879,29 +1879,6 @@ class ResizeLinear1DGrad(Primitive):
|
|||
"coordinate_transformation_mode", self.name)
|
||||
|
||||
|
||||
class ResizeNearestNeighborV2Grad(Primitive):
|
||||
"""
|
||||
Compute gradient of `ResizeNearestNeighborV2` operator.
|
||||
|
||||
Args:
|
||||
align_corners (bool): Whether the centers of the 4 corner pixels of the input
|
||||
and output tensors are aligned. Default: False.
|
||||
half_pixel_centers (bool): Default :False.
|
||||
data_format: An optional `string` that describes the format of the input `x` Defaults to `NHWC`.
|
||||
"""
|
||||
|
||||
@prim_attr_register
|
||||
def __init__(self, align_corners=False, half_pixel_centers=False, data_format='NHWC'):
|
||||
"""Initialize ResizeNearestNeighborV2Grad"""
|
||||
self.init_prim_io_names(inputs=['grads', 'size'], outputs=['y'])
|
||||
|
||||
validator.check_value_type('align_corners', align_corners, [bool], self.name)
|
||||
validator.check_value_type('half_pixel_centers', half_pixel_centers, [bool], self.name)
|
||||
validator.check_value_type('data_format', data_format, [str], self.name)
|
||||
self.format = validator.check_string(data_format, ['NHWC', 'NCHW'], 'data_format', self.name)
|
||||
self.add_prim_attr('data_format', self.format)
|
||||
|
||||
|
||||
class ROIAlignGrad(PrimitiveWithInfer):
|
||||
"""
|
||||
ROIAlignGrad operator.
|
||||
|
|
|
|||
|
|
@ -4181,73 +4181,6 @@ class ResizeNearestNeighbor(Primitive):
|
|||
self.init_prim_io_names(inputs=['image_in'], outputs=['image_out'])
|
||||
|
||||
|
||||
class ResizeNearestNeighborV2(Primitive):
|
||||
r"""
|
||||
Resizes the input tensor to specific size by using the nearest neighbor algorithm.
|
||||
|
||||
Resizes the input tensor to a given size by using the nearest neighbor algorithm. The nearest
|
||||
neighbor algorithm selects the value of the nearest point and does not consider the
|
||||
values of neighboring points at all, yielding a piecewise-constant interpolant.
|
||||
|
||||
Args:
|
||||
align_corners: An optional `bool`. Defaults to `False`.
|
||||
If true, the centers of the 4 corner pixels of the input and output tensors are
|
||||
aligned, preserving the values at the corner pixels.
|
||||
half_pixel_centers: An optional `bool`. Defaults to `False`.
|
||||
data_format: An optional `string` that describes the format of the input `x`. Defaults to `NHWC`.
|
||||
|
||||
Inputs:
|
||||
- **x** (Tensor) - 4-D with shape `[batch, height, width, channels]` or `[batch, channels, height, width]`
|
||||
depending on the attr 'data_format'. Support type [`int8`, `uint8`, `int16`, `uint16`, `int32`, `int64`,
|
||||
`float16`, `float32`, `float64`].
|
||||
- **size** (Tensor) - A 1-D int32 Tensor of 2 elements: [`new_height, new_width`]. The new size for the images.
|
||||
|
||||
Outputs:
|
||||
Tensor `y`, has the same type as input `x` with the shape of `[batch, channels, new_height, new_width]` or
|
||||
`[batch, new_height, new_width, channels]` depending on attr 'data_format'.
|
||||
|
||||
Raises:
|
||||
TypeError: If `x` or `size` is not a Tensor.
|
||||
TypeError: If `x` data type not in support list.
|
||||
TypeError: If `size` data type is not int32.
|
||||
TypeError: If `align_corners` or `half_pixel_centers` is not `bool` value.
|
||||
TypeError: If `data_format` is not `str`.
|
||||
ValueError: If `data_format` not in [`NHWC`, `NCHW`].
|
||||
ValueError: If any value of `size` is non positive.
|
||||
ValueError: If the dimension of `x` is not 4.
|
||||
ValueError: If the dimension of `size` is not 1.
|
||||
ValueError: If the elements number of `size` is not 2.
|
||||
ValueError: If attr `half_pixel_centers` and `align_corners` are True at the same time.
|
||||
|
||||
Supported Platforms:
|
||||
``Ascend`` ``CPU``
|
||||
|
||||
Examples:
|
||||
>>> input_tensor = Tensor(np.ones((1, 4, 4, 1)), mstype.float32)
|
||||
>>> size = Tensor([2, 2], mstype.int32)
|
||||
>>> resize = ops.ResizeNearestNeighborV2()
|
||||
>>> output = resize(input_tensor, size)
|
||||
>>> print(output)
|
||||
[[[[1.]
|
||||
[1.]]
|
||||
[[1.]
|
||||
[1.]]]]
|
||||
>>> print(output.shape)
|
||||
(1, 2, 2, 1)
|
||||
"""
|
||||
|
||||
@prim_attr_register
|
||||
def __init__(self, align_corners=False, half_pixel_centers=False, data_format='NHWC'):
|
||||
"""Initialize ResizeNearestNeighborV2"""
|
||||
self.init_prim_io_names(inputs=['x', 'size'], outputs=['y'])
|
||||
|
||||
validator.check_bool(align_corners, 'align_corners', self.name)
|
||||
validator.check_bool(half_pixel_centers, 'half_pixel_centers', self.name)
|
||||
validator.check_value_type('data_format', data_format, [str], self.name)
|
||||
self.format = validator.check_string(data_format, ['NHWC', 'NCHW'], 'data_format', self.name)
|
||||
self.add_prim_attr('data_format', self.format)
|
||||
|
||||
|
||||
class GatherNd(Primitive):
|
||||
r"""
|
||||
Gathers slices from a tensor by indices.
|
||||
|
|
|
|||
|
|
@ -51,8 +51,6 @@ from mindspore.ops.operations.random_ops import NonDeterministicInts
|
|||
from mindspore.ops.operations.random_ops import TruncatedNormal
|
||||
from mindspore.ops.operations.other_ops import SampleDistortedBoundingBoxV2
|
||||
from mindspore.ops.operations.array_ops import Triu
|
||||
from mindspore.ops.operations.array_ops import ResizeNearestNeighborV2
|
||||
from mindspore.ops.operations._grad_ops import ResizeNearestNeighborV2Grad
|
||||
from mindspore.ops.operations.array_ops import MatrixDiagV3
|
||||
from mindspore.ops.operations.array_ops import MatrixDiagPartV3
|
||||
from mindspore.ops.operations.array_ops import MatrixSetDiagV3
|
||||
|
|
@ -2688,16 +2686,6 @@ test_case_nn_ops = [
|
|||
'desc_inputs': [Tensor([[[[1, 2, 3, 4, 5]]]], mstype.float32), Tensor([[[[1, 2, 3, 4, 5]]]], mstype.float32)],
|
||||
'desc_bprop': [Tensor([[[[1, 2, 3, 4, 5]]]], mstype.float32)],
|
||||
'skip': ['backward']}),
|
||||
('ResizeNearestNeighborV2', {
|
||||
'block': ResizeNearestNeighborV2(),
|
||||
'desc_inputs': [Tensor(np.random.rand(16, 16, 32, 32).astype(np.float32)),
|
||||
Tensor(np.array([8, 8]).astype(np.int32))],
|
||||
'desc_bprop': [Tensor(np.random.rand(16, 16, 8, 8).astype(np.float32))]}),
|
||||
('ResizeNearestNeighborV2Grad', {
|
||||
'block': ResizeNearestNeighborV2Grad(),
|
||||
'desc_inputs': [Tensor(np.random.rand(16, 16, 8, 8).astype(np.float32)),
|
||||
Tensor(np.array([32, 32]).astype(np.int32))],
|
||||
'skip': ['backward']}),
|
||||
('ROIAlign', {
|
||||
'block': P.ROIAlign(7, 7, 0.03125, 2),
|
||||
'desc_inputs': [[2, 256, 192, 320], [1024, 5]],
|
||||
|
|
|
|||
Loading…
Reference in New Issue