diff --git a/mindspore/ccsrc/kernel/common_utils.cc b/mindspore/ccsrc/kernel/common_utils.cc
index 497bb397f72051260e82f4edccff1f93a735bf54..60ddf7b2fe1065decf2d2bd3e8d98dae249b025d 100644
--- a/mindspore/ccsrc/kernel/common_utils.cc
+++ b/mindspore/ccsrc/kernel/common_utils.cc
@@ -559,21 +559,24 @@ void DeduplicateIndexedSlices(const SparseGradient &origin_sparse_grad, SparseGr
   size_t unique_indices_size = 0;
   for (size_t i = 0; i < origin_sparse_grad.indices_size_; ++i) {
     int index = origin_sparse_grad.indices_[i];
-    if (index < 0 || (size_t)index >= first_dim) {
+    if (index < 0 || IntToSize(index) >= first_dim) {
       continue;
     }
     auto iter = index_map.find(index);
     if (iter == index_map.end()) {
       index_map[index] = unique_indices_size;
       unique_grad->indices_[unique_indices_size] = index;
-      for (size_t j = unique_indices_size * outer_dim, k = i * outer_dim; j < (unique_indices_size + 1) * outer_dim;
-           ++j, ++k) {
+      size_t start_index = unique_indices_size * outer_dim;
+      size_t end_index = start_index + outer_dim;
+      for (size_t j = start_index, k = i * outer_dim; j < end_index; ++j, ++k) {
         unique_grad->value_[j] = origin_sparse_grad.value_[k];
       }
       unique_indices_size++;
     } else {
       size_t first_index = iter->second;
-      for (size_t j = first_index * outer_dim, k = i * outer_dim; j < (first_index + 1) * outer_dim; ++j, ++k) {
+      size_t start_index = first_index * outer_dim;
+      size_t end_index = start_index + outer_dim;
+      for (size_t j = start_index, k = i * outer_dim; j < end_index; ++j, ++k) {
         unique_grad->value_[j] += origin_sparse_grad.value_[k];
       }
     }
diff --git a/mindspore/ccsrc/kernel/cpu/cpu_kernel.h b/mindspore/ccsrc/kernel/cpu/cpu_kernel.h
index a391f7586357a35447808c814828651422afe4a4..083652984014df72f50ab070cdd6f9c58eb69908 100644
--- a/mindspore/ccsrc/kernel/cpu/cpu_kernel.h
+++ b/mindspore/ccsrc/kernel/cpu/cpu_kernel.h
@@ -49,6 +49,7 @@ const char AXIS[] = "axis";
 const char BEGIN[] = "begin";
 const char END[] = "end";
 const char SIZE[] = "size";
+const char USE_NESTEROV[] = "use_nesterov";
 
 class CPUKernel : public kernel::KernelMod {
  public:
diff --git a/mindspore/ccsrc/kernel/cpu/sparse_apply_adam_cpu_kernel.cc b/mindspore/ccsrc/kernel/cpu/sparse_apply_adam_cpu_kernel.cc
new file mode 100644
index 0000000000000000000000000000000000000000..4d03645578d7c9118e9327b5c3d528e935dff8e2
--- /dev/null
+++ b/mindspore/ccsrc/kernel/cpu/sparse_apply_adam_cpu_kernel.cc
@@ -0,0 +1,131 @@
+/**
+ * Copyright 2020 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 "kernel/cpu/sparse_apply_adam_cpu_kernel.h"
+#include "device/cpu/cpu_device_address.h"
+
+namespace mindspore {
+namespace kernel {
+namespace {
+constexpr size_t kSparseApplyAdamInputSize = 11;
+}  // namespace
+
+void SparseApplyAdamCPUKernel::InitKernel(const CNodePtr &kernel_node) {
+  MS_EXCEPTION_IF_NULL(kernel_node);
+  std::vector<size_t> var_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 0);
+  std::vector<size_t> m_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 1);
+  std::vector<size_t> v_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 2);
+  std::vector<size_t> grad_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 9);
+  std::vector<size_t> indices_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 10);
+  if (!IsSameShape(var_shape, m_shape)) {
+    MS_LOG(EXCEPTION) << "var and m should have the same shape";
+  }
+  if (!IsSameShape(var_shape, v_shape)) {
+    MS_LOG(EXCEPTION) << "var and v should have the same shape";
+  }
+  if (var_shape.empty()) {
+    MS_LOG(EXCEPTION) << "var must be at least 1D";
+  }
+  var_first_dim_size_ = var_shape[0];
+  for (size_t i = 1; i < var_shape.size(); ++i) {
+    if (var_shape[i] != grad_shape[i]) {
+      MS_LOG(EXCEPTION) << "The shape of var and grad must equal in dimension " << i;
+    }
+    var_outer_dim_size_ *= var_shape[i];
+  }
+  if (indices_shape.size() != 1) {
+    MS_LOG(EXCEPTION) << "indices must be 1D";
+  }
+  indices_size_ = indices_shape[0];
+  if (grad_shape[0] != indices_size_) {
+    MS_LOG(ERROR) << "The first dimension of grad shape must be equal to indices";
+  }
+  if (AnfAlgo::HasNodeAttr(USE_NESTEROV, kernel_node)) {
+    use_nesterov_ = AnfAlgo::GetNodeAttr<bool>(kernel_node, "use_nesterov");
+  }
+}
+
+void SparseApplyAdamCPUKernel::UpdateSparseMomentum(const SparseGradient &unique_sparse_grad, float *m, float *m_t,
+                                                    float *v, float beta1, float beta2) {
+  MS_EXCEPTION_IF_NULL(m);
+  MS_EXCEPTION_IF_NULL(m_t);
+  MS_EXCEPTION_IF_NULL(v);
+  for (size_t i = 0; i < unique_sparse_grad.indices_size_; ++i) {
+    int index = unique_sparse_grad.indices_[i];
+    if (index < 0 || IntToSize(index) >= var_first_dim_size_) {
+      MS_LOG(EXCEPTION) << "Index " << index << " in indices is out of range after unique process";
+    }
+    size_t start_index = var_outer_dim_size_ * index;
+    size_t end_index = start_index + var_outer_dim_size_;
+    for (size_t j = start_index, k = var_outer_dim_size_ * i; j < end_index; ++j, ++k) {
+      auto summed_grad = unique_sparse_grad.value_[k];
+      m[j] += (1 - beta1) * summed_grad;
+      v[j] += (1 - beta2) * summed_grad * summed_grad;
+      if (use_nesterov_) {
+        m_t[j] = m[j] * beta1 + (1 - beta1) * summed_grad;
+      }
+    }
+  }
+}
+
+bool SparseApplyAdamCPUKernel::Launch(const std::vector<kernel::AddressPtr> &inputs,
+                                      const std::vector<kernel::AddressPtr> & /*workspace*/,
+                                      const std::vector<kernel::AddressPtr> & /*outputs*/) {
+  if (inputs.size() < kSparseApplyAdamInputSize) {
+    MS_LOG(EXCEPTION) << "Error input size!";
+  }
+
+  auto var = reinterpret_cast<float *>(inputs[0]->addr);
+  auto m = reinterpret_cast<float *>(inputs[1]->addr);
+  auto v = reinterpret_cast<float *>(inputs[2]->addr);
+  auto beta1_power = reinterpret_cast<float *>(inputs[3]->addr)[0];
+  if (beta1_power == 1) {
+    MS_LOG(EXCEPTION) << "The beta1_power should not be 1";
+  }
+  auto beta2_power = reinterpret_cast<float *>(inputs[4]->addr)[0];
+  auto lr = reinterpret_cast<float *>(inputs[5]->addr)[0];
+  auto beta1 = reinterpret_cast<float *>(inputs[6]->addr)[0];
+  auto beta2 = reinterpret_cast<float *>(inputs[7]->addr)[0];
+  auto epsilon = reinterpret_cast<float *>(inputs[8]->addr)[0];
+  auto grad = reinterpret_cast<float *>(inputs[9]->addr);
+  auto indices = reinterpret_cast<int *>(inputs[10]->addr);
+
+  std::vector<float> new_grad;
+  new_grad.reserve(indices_size_ * var_outer_dim_size_);
+  std::vector<int> new_indices;
+  new_indices.reserve(indices_size_);
+  SparseGradient unique_sparse_grad({new_grad.data(), new_indices.data(), indices_size_});
+  DeduplicateIndexedSlices(SparseGradient({grad, indices, indices_size_}), &unique_sparse_grad, var_first_dim_size_,
+                           var_outer_dim_size_);
+  size_t total_dim_size = var_first_dim_size_ * var_outer_dim_size_;
+  // Update momentum
+  lr = lr * std::sqrt(1 - beta2_power) / (1 - beta1_power);
+  for (size_t i = 0; i < total_dim_size; ++i) {
+    m[i] *= beta1;
+    v[i] *= beta2;
+  }
+  std::vector<float> m_t(m, m + total_dim_size);
+  UpdateSparseMomentum(unique_sparse_grad, m, m_t.data(), v, beta1, beta2);
+  // Update weight
+  if (use_nesterov_) {
+    m = m_t.data();
+  }
+  for (size_t i = 0; i < total_dim_size; ++i) {
+    var[i] -= lr * m[i] / (std::sqrt(v[i]) + epsilon);
+  }
+  return true;
+}
+}  // namespace kernel
+}  // namespace mindspore
diff --git a/mindspore/ccsrc/kernel/cpu/sparse_apply_adam_cpu_kernel.h b/mindspore/ccsrc/kernel/cpu/sparse_apply_adam_cpu_kernel.h
new file mode 100644
index 0000000000000000000000000000000000000000..ea1ce54995aa237adc6fda4c839e3dcca2bfdbbc
--- /dev/null
+++ b/mindspore/ccsrc/kernel/cpu/sparse_apply_adam_cpu_kernel.h
@@ -0,0 +1,66 @@
+/**
+ * Copyright 2020 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_KERNEL_CPU_SPARSE_APPLY_ADAM_CPU_KERNEL_H_
+#define MINDSPORE_CCSRC_KERNEL_CPU_SPARSE_APPLY_ADAM_CPU_KERNEL_H_
+
+#include <vector>
+#include <memory>
+#include "kernel/cpu/cpu_kernel.h"
+#include "kernel/cpu/cpu_kernel_factory.h"
+#include "kernel/common_utils.h"
+
+namespace mindspore {
+namespace kernel {
+class SparseApplyAdamCPUKernel : public CPUKernel {
+ public:
+  SparseApplyAdamCPUKernel() = default;
+  ~SparseApplyAdamCPUKernel() 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;
+
+ private:
+  void UpdateSparseMomentum(const SparseGradient &unique_sparse_grad, float *m, float *m_t, float *v, float beta1,
+                            float beta2);
+  size_t indices_size_{0};
+  size_t var_first_dim_size_{0};
+  size_t var_outer_dim_size_{1};
+  bool use_nesterov_{false};
+};
+
+MS_REG_CPU_KERNEL(SparseApplyAdam,
+                  KernelAttr()
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeInt32)
+                    .AddOutputAttr(kNumberTypeFloat32)
+                    .AddOutputAttr(kNumberTypeFloat32)
+                    .AddOutputAttr(kNumberTypeFloat32),
+                  SparseApplyAdamCPUKernel);
+}  // namespace kernel
+}  // namespace mindspore
+
+#endif  // MINDSPORE_CCSRC_KERNEL_CPU_SPARSE_APPLY_ADAM_CPU_KERNEL_H_
diff --git a/mindspore/ccsrc/kernel/cpu/sparse_apply_ftrl_cpu_kernel.cc b/mindspore/ccsrc/kernel/cpu/sparse_apply_ftrl_cpu_kernel.cc
index af1f301080f2835ac725674d7fe899637284b6c0..4d96d31f426d8020fd0730c24757da192e5bd985 100644
--- a/mindspore/ccsrc/kernel/cpu/sparse_apply_ftrl_cpu_kernel.cc
+++ b/mindspore/ccsrc/kernel/cpu/sparse_apply_ftrl_cpu_kernel.cc
@@ -84,19 +84,22 @@ bool SparseApplyFtrlCPUKernel::Launch(const std::vector<kernel::AddressPtr> &inp
   auto grad = reinterpret_cast<float *>(inputs[3]->addr);
   auto indices = reinterpret_cast<int *>(inputs[4]->addr);
 
-  std::vector<float> new_grad(indices_size_ * var_outer_dim_size_);
-  std::vector<int> new_indices(indices_size_);
+  std::vector<float> new_grad;
+  new_grad.reserve(indices_size_ * var_outer_dim_size_);
+  std::vector<int> new_indices;
+  new_indices.reserve(indices_size_);
   SparseGradient unique_sparse_grad({new_grad.data(), new_indices.data(), indices_size_});
   DeduplicateIndexedSlices(SparseGradient({grad, indices, indices_size_}), &unique_sparse_grad, var_first_dim_size_,
                            var_outer_dim_size_);
 
   for (size_t i = 0; i < unique_sparse_grad.indices_size_; ++i) {
     int index = unique_sparse_grad.indices_[i];
-    if (index < 0 || (size_t)index >= var_first_dim_size_) {
+    if (index < 0 || IntToSize(index) >= var_first_dim_size_) {
       MS_LOG(EXCEPTION) << "Index " << index << " in indices is out of range after unique process";
     }
-    for (size_t j = var_outer_dim_size_ * index, k = var_outer_dim_size_ * i; j < var_outer_dim_size_ * (index + 1);
-         ++j, ++k) {
+    size_t start_index = var_outer_dim_size_ * index;
+    size_t end_index = start_index + var_outer_dim_size_;
+    for (size_t j = start_index, k = var_outer_dim_size_ * i; j < end_index; ++j, ++k) {
       auto summed_grad = unique_sparse_grad.value_[k];
       auto accum_new = accum[j] + summed_grad * summed_grad;
       if (lr_power_ == -0.5) {
diff --git a/mindspore/ccsrc/kernel/cpu/sparse_apply_lazy_adam_cpu_kernel.cc b/mindspore/ccsrc/kernel/cpu/sparse_apply_lazy_adam_cpu_kernel.cc
new file mode 100644
index 0000000000000000000000000000000000000000..c0e091f02bab2047fbc12da26ed66c3013b9cf42
--- /dev/null
+++ b/mindspore/ccsrc/kernel/cpu/sparse_apply_lazy_adam_cpu_kernel.cc
@@ -0,0 +1,113 @@
+/**
+ * Copyright 2020 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 "kernel/cpu/sparse_apply_lazy_adam_cpu_kernel.h"
+#include "kernel/common_utils.h"
+#include "device/cpu/cpu_device_address.h"
+
+namespace mindspore {
+namespace kernel {
+namespace {
+constexpr size_t kSparseApplyLazyAdamInputSize = 11;
+}  // namespace
+
+void SparseApplyLazyAdamCPUKernel::InitKernel(const CNodePtr &kernel_node) {
+  MS_EXCEPTION_IF_NULL(kernel_node);
+  std::vector<size_t> var_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 0);
+  std::vector<size_t> m_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 1);
+  std::vector<size_t> v_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 2);
+  std::vector<size_t> grad_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 9);
+  std::vector<size_t> indices_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 10);
+  if (!IsSameShape(var_shape, m_shape)) {
+    MS_LOG(EXCEPTION) << "var and m should have the same shape";
+  }
+  if (!IsSameShape(var_shape, v_shape)) {
+    MS_LOG(EXCEPTION) << "var and v should have the same shape";
+  }
+  if (var_shape.empty()) {
+    MS_LOG(EXCEPTION) << "var must be at least 1D";
+  }
+  var_first_dim_size_ = var_shape[0];
+  for (size_t i = 1; i < var_shape.size(); ++i) {
+    if (var_shape[i] != grad_shape[i]) {
+      MS_LOG(EXCEPTION) << "The shape of var and grad must equal in dimension " << i;
+    }
+    var_outer_dim_size_ *= var_shape[i];
+  }
+  if (indices_shape.size() != 1) {
+    MS_LOG(EXCEPTION) << "indices must be 1D";
+  }
+  indices_size_ = indices_shape[0];
+  if (grad_shape[0] != indices_size_) {
+    MS_LOG(ERROR) << "The first dimension of grad shape must be equal to indices";
+  }
+  if (AnfAlgo::HasNodeAttr(USE_NESTEROV, kernel_node)) {
+    use_nesterov_ = AnfAlgo::GetNodeAttr<bool>(kernel_node, "use_nesterov");
+  }
+}
+
+bool SparseApplyLazyAdamCPUKernel::Launch(const std::vector<kernel::AddressPtr> &inputs,
+                                          const std::vector<kernel::AddressPtr> & /*workspace*/,
+                                          const std::vector<kernel::AddressPtr> & /*outputs*/) {
+  if (inputs.size() < kSparseApplyLazyAdamInputSize) {
+    MS_LOG(EXCEPTION) << "Error input size!";
+  }
+
+  auto var = reinterpret_cast<float *>(inputs[0]->addr);
+  auto m = reinterpret_cast<float *>(inputs[1]->addr);
+  auto v = reinterpret_cast<float *>(inputs[2]->addr);
+  auto beta1_power = reinterpret_cast<float *>(inputs[3]->addr)[0];
+  if (beta1_power == 1) {
+    MS_LOG(EXCEPTION) << "The beta1_power should not be 1";
+  }
+  auto beta2_power = reinterpret_cast<float *>(inputs[4]->addr)[0];
+  auto lr = reinterpret_cast<float *>(inputs[5]->addr)[0];
+  auto beta1 = reinterpret_cast<float *>(inputs[6]->addr)[0];
+  auto beta2 = reinterpret_cast<float *>(inputs[7]->addr)[0];
+  auto epsilon = reinterpret_cast<float *>(inputs[8]->addr)[0];
+  auto grad = reinterpret_cast<float *>(inputs[9]->addr);
+  auto indices = reinterpret_cast<int *>(inputs[10]->addr);
+
+  std::vector<float> new_grad;
+  new_grad.reserve(indices_size_ * var_outer_dim_size_);
+  std::vector<int> new_indices;
+  new_indices.reserve(indices_size_);
+  SparseGradient unique_sparse_grad({new_grad.data(), new_indices.data(), indices_size_});
+  DeduplicateIndexedSlices(SparseGradient({grad, indices, indices_size_}), &unique_sparse_grad, var_first_dim_size_,
+                           var_outer_dim_size_);
+
+  lr = lr * std::sqrt(1 - beta2_power) / (1 - beta1_power);
+  for (size_t i = 0; i < unique_sparse_grad.indices_size_; ++i) {
+    int index = unique_sparse_grad.indices_[i];
+    if (index < 0 || IntToSize(index) >= var_first_dim_size_) {
+      MS_LOG(EXCEPTION) << "Index " << index << " in indices is out of range";
+    }
+    size_t start_index = var_outer_dim_size_ * index;
+    size_t end_index = start_index + var_outer_dim_size_;
+    for (size_t j = start_index, k = var_outer_dim_size_ * i; j < end_index; ++j, ++k) {
+      auto summed_grad = unique_sparse_grad.value_[k];
+      m[j] = beta1 * m[j] + (1 - beta1) * summed_grad;
+      v[j] = beta2 * v[j] + (1 - beta2) * summed_grad * summed_grad;
+      if (use_nesterov_) {
+        var[j] -= lr * (m[j] * beta1 + (1 - beta1) * summed_grad) / (std::sqrt(v[j]) + epsilon);
+      } else {
+        var[j] -= lr * m[j] / (std::sqrt(v[j]) + epsilon);
+      }
+    }
+  }
+  return true;
+}
+}  // namespace kernel
+}  // namespace mindspore
diff --git a/mindspore/ccsrc/kernel/cpu/sparse_apply_lazy_adam_cpu_kernel.h b/mindspore/ccsrc/kernel/cpu/sparse_apply_lazy_adam_cpu_kernel.h
new file mode 100644
index 0000000000000000000000000000000000000000..0a521815616fe6a165f0cbe57892d40e13d92240
--- /dev/null
+++ b/mindspore/ccsrc/kernel/cpu/sparse_apply_lazy_adam_cpu_kernel.h
@@ -0,0 +1,63 @@
+/**
+ * Copyright 2020 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_KERNEL_CPU_SPARSE_APPLY_LAZY_ADAM_CPU_KERNEL_H_
+#define MINDSPORE_CCSRC_KERNEL_CPU_SPARSE_APPLY_LAZY_ADAM_CPU_KERNEL_H_
+
+#include <vector>
+#include <memory>
+#include "kernel/cpu/cpu_kernel.h"
+#include "kernel/cpu/cpu_kernel_factory.h"
+
+namespace mindspore {
+namespace kernel {
+class SparseApplyLazyAdamCPUKernel : public CPUKernel {
+ public:
+  SparseApplyLazyAdamCPUKernel() = default;
+  ~SparseApplyLazyAdamCPUKernel() 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;
+
+ private:
+  size_t indices_size_{0};
+  size_t var_first_dim_size_{0};
+  size_t var_outer_dim_size_{1};
+  bool use_nesterov_{false};
+};
+
+MS_REG_CPU_KERNEL(SparseApplyLazyAdam,
+                  KernelAttr()
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeFloat32)
+                    .AddInputAttr(kNumberTypeInt32)
+                    .AddOutputAttr(kNumberTypeFloat32)
+                    .AddOutputAttr(kNumberTypeFloat32)
+                    .AddOutputAttr(kNumberTypeFloat32),
+                  SparseApplyLazyAdamCPUKernel);
+}  // namespace kernel
+}  // namespace mindspore
+
+#endif  // MINDSPORE_CCSRC_KERNEL_CPU_SPARSE_APPLY_LAZY_ADAM_CPU_KERNEL_H_
diff --git a/tests/st/ops/cpu/test_sparse_apply_adam_op.py b/tests/st/ops/cpu/test_sparse_apply_adam_op.py
new file mode 100644
index 0000000000000000000000000000000000000000..e81ac470b76e07aeae57bffeea7ddafb8dbfedab
--- /dev/null
+++ b/tests/st/ops/cpu/test_sparse_apply_adam_op.py
@@ -0,0 +1,46 @@
+# Copyright 2020 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 mindspore.context as context
+import mindspore.nn as nn
+from mindspore import Tensor
+from mindspore.common.parameter import Parameter
+from mindspore.ops import operations as P
+import mindspore.common.dtype as mstype
+
+
+class Net(nn.Cell):
+    def __init__(self):
+        super(Net, self).__init__()
+        self.sparse_apply_adam = P.SparseApplyAdam()
+        self.var = Parameter(Tensor(np.ones([3, 3, 3]).astype(np.float32)), name="var")
+        self.m = Parameter(Tensor(np.ones([3, 3, 3]).astype(np.float32)), name="m")
+        self.v = Parameter(Tensor(np.ones([3, 3, 3]).astype(np.float32)), name="v")
+
+    def construct(self, beta1_power, beta2_power, lr, beta1, beta2, epsilon, grad, indices):
+        out = self.sparse_apply_adam(self.var, self.m, self.v, beta1_power, beta2_power, lr, beta1, beta2, epsilon,
+                                     grad, indices)
+        return out
+
+
+def test_net():
+    gradient = Tensor(np.random.rand(3, 3, 3).astype(np.float32))
+    indices = Tensor([0, 1, 2], mstype.int32)
+
+    context.set_context(mode=context.GRAPH_MODE, device_target="CPU")
+    sparse_apply_adam = Net()
+    output = sparse_apply_adam(0.9, 0.999, 0.001, 0.9, 0.999, 1e-8, gradient, indices)
+    print(output[0].asnumpy())