【CUDA Kernel No.74、75】lookup_table、lookup_table_grad算子Kernel修复 -part (#75645)

* Add lookup_table_kernel.h

* Add lookup_table_grad_kernel.h

* Bug fix

* Delete CPUKernel

* 修改路径、去除cpu声明

* Update lookup_table_grad_kernel.cc

* CI

* Fix

* Fix codestyle

* Fix ci

* Fix ci

* save

* Revert "Fix ci"

This reverts commit 644eacc.

* Revert "Fix ci"

This reverts commit f7bf793.

* Revert "Fix codestyle"

This reverts commit cbba545.

* fix

* 忘记改了一个报错

Author: Le-soleile

paddle/phi/kernels/cpu/lookup_table_grad_kernel.cc

@@ -14,15 +14,13 @@
 
 #include <string>
 #include <vector>
-
+#include "paddle/phi/backends/cpu/cpu_context.h"
 #include "paddle/phi/core/dense_tensor.h"
 #include "paddle/phi/core/enforce.h"
+#include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/funcs/blas/blas.h"
 #include "paddle/phi/kernels/funcs/eigen/common.h"
 
-#include "paddle/phi/backends/cpu/cpu_context.h"
-#include "paddle/phi/core/kernel_registry.h"
-
 namespace phi {
 
 constexpr int64_t kNoPadding = -1;

paddle/phi/kernels/cpu/lookup_table_kernel.cc

@@ -14,15 +14,13 @@
 
 #include <string>
 #include <vector>
-
+#include "paddle/phi/backends/cpu/cpu_context.h"
 #include "paddle/phi/core/dense_tensor.h"
 #include "paddle/phi/core/enforce.h"
+#include "paddle/phi/core/kernel_registry.h"
 #include "paddle/phi/kernels/funcs/blas/blas.h"
 #include "paddle/phi/kernels/funcs/eigen/common.h"
 
-#include "paddle/phi/backends/cpu/cpu_context.h"
-#include "paddle/phi/core/kernel_registry.h"
-
 namespace phi {
 
 constexpr int64_t kNoPadding = -1;

paddle/phi/kernels/gpu/lookup_table_grad_kernel.cu

@@ -12,6 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#include "paddle/phi/kernels/gpu/lookup_table_grad_kernel.h"
 #include "paddle/phi/backends/gpu/gpu_primitives.h"
 #include "paddle/phi/common/memory_utils.h"
 #include "paddle/phi/core/kernel_registry.h"

paddle/phi/kernels/gpu/lookup_table_grad_kernel.h

@@ -0,0 +1,64 @@
+// Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
+//
+// 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.
+
+#pragma once
+
+#include "paddle/phi/core/dense_tensor.h"
+#include "paddle/phi/core/device_context.h"
+#include "paddle/phi/core/selected_rows.h"
+
+namespace phi {
+
+template <typename T, typename Context>
+void LookupTableGradCUDAKernel(const Context& dev_ctx,
+                               const DenseTensor& w,
+                               const DenseTensor& ids_in,
+                               const DenseTensor& out_grad,
+                               bool is_sparse,
+                               bool is_distributed,
+                               int64_t padding_idx,
+                               bool remote_prefetch,
+                               const std::string& entry_config,
+                               bool is_test,
+                               const std::string& entry,
+                               const std::string& table_class,
+                               const std::vector<std::string>& table_names,
+                               int trainer_id,
+                               bool grad_inplace,
+                               const std::vector<std::string>& epmap,
+                               const std::vector<int64_t>& height_sections,
+                               DenseTensor* w_grad);
+
+template <typename T, typename Context>
+void LookupTableSparseGradCUDAKernel(
+    const Context& dev_ctx,
+    const DenseTensor& w,
+    const DenseTensor& ids_in,
+    const DenseTensor& out_grad,
+    bool is_sparse,
+    bool is_distributed,
+    int64_t padding_idx,
+    bool remote_prefetch,
+    const std::string& entry_config,
+    bool is_test,
+    const std::string& entry,
+    const std::string& table_class,
+    const std::vector<std::string>& table_names,
+    int trainer_id,
+    bool grad_inplace,
+    const std::vector<std::string>& epmap,
+    const std::vector<int64_t>& height_sections,
+    SelectedRows* w_grad);
+
+}  // namespace phi

paddle/phi/kernels/gpu/lookup_table_kernel.cu

@@ -12,6 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#include "paddle/phi/kernels/gpu/lookup_table_kernel.h"
 #include "paddle/phi/backends/gpu/gpu_primitives.h"
 #include "paddle/phi/common/memory_utils.h"
 #include "paddle/phi/core/kernel_registry.h"

paddle/phi/kernels/gpu/lookup_table_kernel.h

@@ -0,0 +1,44 @@
+// Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
+//
+// 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.
+
+#pragma once
+
+#include <string>
+#include <vector>
+#include "paddle/phi/core/dense_tensor.h"
+#include "paddle/phi/core/enforce.h"
+
+namespace phi {
+
+template <typename T, typename Context>
+void LookupTableCUDAKernel(const Context &dev_ctx,
+                           const DenseTensor &w,
+                           const DenseTensor &ids_in,
+                           bool is_sparse,
+                           bool is_distributed,
+                           int64_t padding_idx,
+                           bool remote_prefetch,
+                           const std::string &entry_config,
+                           bool is_test,
+                           const std::string &entry,
+                           const std::string &table_class,
+                           const std::vector<std::string> &table_names,
+                           int trainer_id,
+                           bool grad_inplace,
+                           const std::vector<std::string> &epmap,
+                           const std::vector<int64_t> &height_sections,
+                           DenseTensor *out);
+
+}  // namespace phi