layer.h 2.6 KB
Newer Older
X
Xin Pan 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
// Copyright (c) 2018 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

X
Xin Pan 已提交
17
#include <string>
X
Xin Pan 已提交
18
#include <vector>
X
Xin Pan 已提交
19
#include "paddle/fluid/framework/op_desc.h"
X
Xin Pan 已提交
20
#include "paddle/fluid/framework/operator.h"
X
Xin Pan 已提交
21 22
#include "paddle/fluid/framework/scope.h"
#include "paddle/fluid/framework/var_desc.h"
X
Xin Pan 已提交
23 24 25 26 27
#include "paddle/fluid/platform/enforce.h"

namespace paddle {
namespace imperative {

X
Xin Pan 已提交
28 29
class OpBase;

X
Xin Pan 已提交
30
class VarBase {
X
Xin Pan 已提交
31
 public:
X
Xin Pan 已提交
32 33 34 35 36 37 38
  VarBase()
      : pre_op_(nullptr),
        pre_op_out_idx_(-1),
        var_desc_(nullptr),
        var_(nullptr),
        grads_(nullptr) {}

X
polish  
Xin Pan 已提交
39
  virtual ~VarBase() {}
X
Xin Pan 已提交
40 41 42 43 44 45

  void ApplyGrad(framework::Scope* scope, framework::Variable* grad);

  void RunBackward(framework::Scope* scope);

  framework::LoDTensor& Grad();
X
Xin Pan 已提交
46

X
Xin Pan 已提交
47
  OpBase* pre_op_;
X
Xin Pan 已提交
48 49
  int pre_op_out_idx_;

X
Xin Pan 已提交
50
  framework::VarDesc* var_desc_;
X
Xin Pan 已提交
51 52
  framework::Variable* var_;
  framework::Variable* grads_;
X
Xin Pan 已提交
53 54
};

X
Xin Pan 已提交
55 56
class OpBase {
 public:
X
Xin Pan 已提交
57 58
  OpBase()
      : input_vars_(new std::vector<VarBase*>()),
X
Xin Pan 已提交
59 60 61 62 63 64
        output_vars_(new std::vector<VarBase*>()),
        pre_ops_(new std::vector<OpBase*>()),
        pre_ops_out_idx_(new std::vector<int>()),
        op_desc_(nullptr),
        grad_op_desc_(nullptr) {}

X
Xin Pan 已提交
65 66 67
  virtual ~OpBase() {
    delete input_vars_;
    delete output_vars_;
X
Xin Pan 已提交
68 69 70 71 72 73

    delete pre_ops_;
    delete pre_ops_out_idx_;

    if (grad_op_desc_) delete grad_op_desc_;
    if (grad_to_var_) delete grad_to_var_;
X
Xin Pan 已提交
74
  }
X
Xin Pan 已提交
75

X
Xin Pan 已提交
76 77
  std::vector<framework::Variable*> ApplyGrad(framework::Scope* scope);

X
Xin Pan 已提交
78 79
  std::vector<VarBase*>* input_vars_;
  std::vector<VarBase*>* output_vars_;
X
Xin Pan 已提交
80 81
  std::vector<OpBase*>* pre_ops_;
  std::vector<int>* pre_ops_out_idx_;
X
Xin Pan 已提交
82
  framework::OpDesc* op_desc_;
X
Xin Pan 已提交
83 84 85 86

  framework::OpDesc* grad_op_desc_;
  std::unordered_map<std::string, std::string>* grad_to_var_;
  framework::BlockDesc* block_;
X
Xin Pan 已提交
87 88
};

X
Xin Pan 已提交
89 90 91 92
class Layer {
 public:
  virtual ~Layer() {}

X
Xin Pan 已提交
93 94
  virtual std::vector<VarBase> Forward(const std::vector<VarBase>& inputs) {
    std::vector<VarBase> vars;
X
Xin Pan 已提交
95 96
    return vars;
  }
X
Xin Pan 已提交
97

X
Xin Pan 已提交
98
  virtual void Backward() { LOG(ERROR) << "To support customize"; }
X
Xin Pan 已提交
99 100 101 102
};

}  // namespace imperative
}  // namespace paddle