!14732 Add L2NormalizeGrad for CPU

From: @he-botao
Reviewed-by: @wuxuejian,@liangchenghui
Signed-off-by:

mindspore/ccsrc/backend/kernel_compiler/cpu/l2normalize_grad_cpu_kernel.cc

@@ -0,0 +1,169 @@
+/**
+ * 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.
+ * 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 "backend/kernel_compiler/cpu/l2normalize_grad_cpu_kernel.h"
+#include "runtime/device/cpu/cpu_device_address.h"
+
+namespace mindspore {
+namespace kernel {
+template <typename T>
+void L2NormalizeGradCPUKernel<T>::InitKernel(const CNodePtr &kernel_node) {
+  MS_EXCEPTION_IF_NULL(kernel_node);
+  CheckIONumber(kernel_node);
+  for (size_t i = 0; i < INPUT_SIZE; i++) {
+    input_shape_list_.emplace_back(AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, i));
+  }
+  auto output_shape = AnfAlgo::GetOutputInferShape(kernel_node, 0);
+  CheckInputShape(output_shape);
+
+  int output_dim_length = output_shape.size();
+  dim_elem_num_list_.resize(output_dim_length, 1);
+  for (int i = output_dim_length - 2; i >= 0; i--) {
+    dim_elem_num_list_[i] = output_shape[i + 1] * dim_elem_num_list_[i + 1];
+  }
+
+  int axis = LongToInt(AnfAlgo::GetNodeAttr<int64_t>(kernel_node, "axis"));
+  int input_dim_length = SizeToInt(input_shape_list_[0].size());
+  axis_ = axis < 0 ? (axis + input_dim_length) : axis;
+  epsilon_ = static_cast<T>(AnfAlgo::GetNodeAttr<float>(kernel_node, "epsilon"));
+}
+
+template <typename T>
+bool L2NormalizeGradCPUKernel<T>::Launch(const std::vector<AddressPtr> &inputs,
+                                         const std::vector<AddressPtr> &workspace,
+                                         const std::vector<AddressPtr> &outputs) {
+  auto input_x = reinterpret_cast<T *>(inputs[0]->addr);
+  auto y = reinterpret_cast<T *>(inputs[1]->addr);
+  auto dout = reinterpret_cast<T *>(inputs[2]->addr);
+  auto output = reinterpret_cast<T *>(outputs[0]->addr);
+  auto output_size = outputs[0]->size / sizeof(T);
+  auto task = [&](size_t start, size_t end) {
+    for (size_t i = start; i < end; i++) {
+      std::vector<size_t> high_dim_index;
+      OneDimIndexToHighDimIndex(i, &high_dim_index);
+      std::vector<T> input_x_vector;
+      GetVector(&input_x_vector, high_dim_index, input_x);
+      std::vector<T> dout_vector;
+      GetVector(&dout_vector, high_dim_index, dout);
+      std::vector<T> y_vector;
+      GetVector(&y_vector, high_dim_index, y);
+      GetOutput(input_x_vector, y_vector, dout_vector, high_dim_index, &output[i]);
+    }
+  };
+  CPUKernelUtils::ParallelFor(task, output_size);
+  return true;
+}
+
+template <typename T>
+void L2NormalizeGradCPUKernel<T>::CheckInputShape(const std::vector<size_t> &output_shape) {
+  for (const auto &shape : input_shape_list_) {
+    if (output_shape != shape) {
+      MS_LOG(EXCEPTION) << "Input shape and output shape should be same.";
+    }
+  }
+  auto input_x_shape = input_shape_list_[0];
+  if (input_x_shape.size() != 0) {
+    if (std::any_of(input_x_shape.begin(), input_x_shape.end(), [](size_t i) { return i == 0; })) {
+      MS_LOG(EXCEPTION) << "L2NormalizeCPUKernel input is null.";
+    }
+  }
+}
+
+template <typename T>
+void L2NormalizeGradCPUKernel<T>::CheckIONumber(const CNodePtr &kernel_node) {
+  size_t input_num = AnfAlgo::GetInputTensorNum(kernel_node);
+  if (input_num != INPUT_SIZE) {
+    MS_LOG(EXCEPTION) << "Input number is " << input_num << ", but L2NormalizeGradCPUKernel needs 3 input.";
+  }
+  size_t output_num = AnfAlgo::GetOutputTensorNum(kernel_node);
+  if (output_num != OUTPUT_SIZE) {
+    MS_LOG(EXCEPTION) << "Output number is " << output_num << ", but L2NormalizeGradCPUKernel needs 1 output.";
+  }
+}
+
+template <typename T>
+void L2NormalizeGradCPUKernel<T>::OneDimIndexToHighDimIndex(size_t one_dim_index, std::vector<size_t> *high_dim_index) {
+  for (const auto &item : dim_elem_num_list_) {
+    high_dim_index->push_back(one_dim_index / item);
+    one_dim_index %= item;
+  }
+}
+
+template <typename T>
+void L2NormalizeGradCPUKernel<T>::HighDimIndexToOneDimIndex(size_t *one_dim_index,
+                                                            const std::vector<size_t> &high_dim_index) {
+  *one_dim_index = 0;
+  int len = high_dim_index.size();
+  for (int i = 0; i < len; i++) {
+    *one_dim_index += high_dim_index[i] * dim_elem_num_list_[i];
+  }
+}
+
+template <typename T>
+void L2NormalizeGradCPUKernel<T>::GetVector(std::vector<T> *x_vector, const std::vector<size_t> &high_dim_index,
+                                            const T *x) {
+  auto x_shape = input_shape_list_[0];
+  for (size_t i = 0; i < x_shape[axis_]; i++) {
+    size_t oneDimIndex = 0;
+    std::vector<size_t> tmp_high_dim_index = high_dim_index;
+    tmp_high_dim_index[axis_] = i;
+    HighDimIndexToOneDimIndex(&oneDimIndex, tmp_high_dim_index);
+    x_vector->push_back(x[oneDimIndex]);
+  }
+}
+
+template <typename T>
+void L2NormalizeGradCPUKernel<T>::GetSumOfProduct(const std::vector<T> &x_vector, const std::vector<T> &y_vector,
+                                                  T *ss) {
+  size_t len = x_vector.size();
+  std::vector<T> tmp_vector(len);
+  for (size_t i = 0; i < len; i++) {
+    tmp_vector[i] = x_vector[i] * y_vector[i];
+  }
+  if (len % 2 == 1) {
+    tmp_vector[0] += tmp_vector[len - 1];
+  }
+  for (size_t stride = len / 2; stride > 0; stride >>= 1) {
+    for (size_t i = 0; i < stride; i++) {
+      tmp_vector[i] += tmp_vector[i + stride];
+    }
+    if (stride > 2 && stride % 2 == 1) {
+      tmp_vector[0] += tmp_vector[stride - 1];
+    }
+  }
+  *ss = tmp_vector[0];
+}
+
+template <typename T>
+void L2NormalizeGradCPUKernel<T>::GetOutput(const std::vector<T> &input_x_vector, const std::vector<T> &y_vector,
+                                            const std::vector<T> &dout_vector,
+                                            const std::vector<size_t> &high_dim_index, T *output) {
+  size_t axis_index = high_dim_index[axis_];
+  T dout = dout_vector[axis_index];
+  T y = y_vector[axis_index];
+  T tmp_sum1;
+  GetSumOfProduct(y_vector, dout_vector, &tmp_sum1);
+  T tmp_sum2;
+  GetSumOfProduct(input_x_vector, input_x_vector, &tmp_sum2);
+  tmp_sum2 = sqrt(tmp_sum2);
+  if (tmp_sum2 >= epsilon_) {
+    *output = (dout - y * tmp_sum1) / tmp_sum2;
+  } else {
+    *output = (dout - y * tmp_sum1) / epsilon_;
+  }
+}
+}  // namespace kernel
+}  // namespace mindspore

mindspore/ccsrc/backend/kernel_compiler/cpu/l2normalize_grad_cpu_kernel.h

@@ -0,0 +1,70 @@
+/**
+ * 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.
+ * 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_L2NORMALIZE_GRAD_CPU_KERNEL_H_
+#define MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_L2NORMALIZE_GRAD_CPU_KERNEL_H_
+
+#include <vector>
+#include "backend/kernel_compiler/cpu/cpu_kernel.h"
+#include "backend/kernel_compiler/cpu/cpu_kernel_factory.h"
+
+namespace mindspore {
+namespace kernel {
+constexpr size_t INPUT_SIZE = 3;
+constexpr size_t OUTPUT_SIZE = 1;
+template <typename T>
+class L2NormalizeGradCPUKernel : public CPUKernel {
+ public:
+  L2NormalizeGradCPUKernel() = default;
+  ~L2NormalizeGradCPUKernel() override = default;
+
+  bool Launch(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &workspace,
+              const std::vector<AddressPtr> &outputs) override;
+
+  void InitKernel(const CNodePtr &kernel_node) override;
+
+ private:
+  void CheckInputShape(const std::vector<size_t> &output_shape);
+  void CheckIONumber(const CNodePtr &kernel_node);
+  void OneDimIndexToHighDimIndex(size_t one_dim_index, std::vector<size_t> *high_dim_index);
+  void HighDimIndexToOneDimIndex(size_t *one_dim_index, const std::vector<size_t> &high_dim_index);
+  void GetVector(std::vector<T> *x_vector, const std::vector<size_t> &high_dim_index, const T *x);
+  void GetSumOfProduct(const std::vector<T> &x_vector, const std::vector<T> &y_vector, T *ss);
+  void GetOutput(const std::vector<T> &input_x_vector, const std::vector<T> &y_vector,
+                 const std::vector<T> &dout_vector, const std::vector<size_t> &high_dim_index, T *output);
+  std::vector<std::vector<size_t>> input_shape_list_;
+  std::vector<size_t> dim_elem_num_list_;
+  int axis_{0};
+  T epsilon_{0};
+};
+
+MS_REG_CPU_KERNEL_T(L2NormalizeGrad,
+                    KernelAttr()
+                      .AddInputAttr(kNumberTypeFloat32)
+                      .AddInputAttr(kNumberTypeFloat32)
+                      .AddInputAttr(kNumberTypeFloat32)
+                      .AddOutputAttr(kNumberTypeFloat32),
+                    L2NormalizeGradCPUKernel, float);
+
+MS_REG_CPU_KERNEL_T(L2NormalizeGrad,
+                    KernelAttr()
+                      .AddInputAttr(kNumberTypeFloat16)
+                      .AddInputAttr(kNumberTypeFloat16)
+                      .AddInputAttr(kNumberTypeFloat16)
+                      .AddOutputAttr(kNumberTypeFloat16),
+                    L2NormalizeGradCPUKernel, float16);
+}  // namespace kernel
+}  // namespace mindspore
+#endif  // MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_CPU_L2NORMALIZE_GRAD_CPU_KERNEL_H_

tests/st/ops/cpu/test_l2normalize_grad_op.py

@@ -0,0 +1,53 @@
+# 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.
+# 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.
+# ============================================================================
+
+import numpy as np
+import pytest
+
+import mindspore.context as context
+import mindspore.common.dtype as mstype
+import mindspore.nn as nn
+from mindspore import Tensor
+from mindspore.ops.operations import _grad_ops as G
+
+
+class Net(nn.Cell):
+    def __init__(self, axis=0, epsilon=1e-4):
+        super(Net, self).__init__()
+        self.ops = G.L2NormalizeGrad(axis, epsilon)
+
+    def construct(self, input_x, output, dout):
+        return self.ops(input_x, output, dout)
+
+
+@pytest.mark.level0
+@pytest.mark.platform_x86_cpu_training
+@pytest.mark.env_onecard
+def test_net01():
+    context.set_context(mode=context.GRAPH_MODE, device_target='CPU')
+    axis = 1
+    net = Net(axis)
+    input_x = np.arange(24).astype(np.float32).reshape((2, 3, 4))
+    dout = np.arange(24, 48).astype(np.float32).reshape((2, 3, 4))
+    output = input_x / np.sqrt(np.sum(input_x**2, axis=axis, keepdims=True))
+    except_asn = (dout - output * np.sum(output * dout, axis=axis, keepdims=True)
+                  ) / np.sqrt(np.sum(input_x**2, axis=axis, keepdims=True))
+    input_x = Tensor(input_x, mstype.float32)
+    output = Tensor(output, mstype.float32)
+    dout = Tensor(dout, mstype.float32)
+    net_output = net(input_x, output, dout).asnumpy()
+    precision = np.ones(shape=(2, 3, 4), dtype=np.float32) * 1.0e-5
+    absolute_error = np.abs(except_asn-net_output)
+    assert np.all(absolute_error < precision)