conditional_block_op.cc 9.9 KB
Newer Older
1
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved.
Y
Yu Yang 已提交
2

L
Luo Tao 已提交
3 4 5
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
Y
Yu Yang 已提交
6

L
Luo Tao 已提交
7
    http://www.apache.org/licenses/LICENSE-2.0
Y
Yu Yang 已提交
8

L
Luo Tao 已提交
9 10 11 12 13
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. */
Y
Yu Yang 已提交
14
#include <algorithm>
Y
Yi Wang 已提交
15 16
#include "paddle/fluid/framework/executor.h"
#include "paddle/fluid/framework/op_registry.h"
S
sneaxiy 已提交
17
#include "paddle/fluid/framework/var_type.h"
Y
Yu Yang 已提交
18 19 20 21 22 23 24 25 26 27 28 29 30 31

namespace paddle {
namespace operators {

class ConditionalOp : public framework::OperatorBase {
 public:
  ConditionalOp(const std::string &type,
                const framework::VariableNameMap &inputs,
                const framework::VariableNameMap &outputs,
                const framework::AttributeMap &attrs)
      : OperatorBase(type, inputs, outputs, attrs) {}

 protected:
  std::vector<const framework::LoDTensor *> InputTensors(
32
      const framework::Scope &scope, const std::string &in_name) const {
Y
Yu Yang 已提交
33
    std::vector<const framework::LoDTensor *> retv;
34
    auto xs = Inputs(in_name);
Y
Yu Yang 已提交
35 36 37 38 39 40 41 42 43 44
    retv.resize(xs.size(), nullptr);
    std::transform(
        xs.begin(), xs.end(), retv.begin(),
        [&scope](const std::string &var_name) -> const framework::LoDTensor * {
          auto *var = scope.FindVar(var_name);
          PADDLE_ENFORCE(var != nullptr, "Cannot find variable %s", var_name);
          return &var->Get<framework::LoDTensor>();
        });
    return retv;
  }
45 46 47 48 49 50

  bool ScalarCondition(
      const std::vector<const framework::LoDTensor *> &ips) const {
    if (!(ips.size() == 1UL && ips[0]->IsInitialized())) {
      PADDLE_THROW("should have one initialized input as condition");
    }
Y
Yu Yang 已提交
51 52 53 54 55 56

    PADDLE_ENFORCE(ips[0]->type() == framework::proto::VarType::BOOL &&
                       ips[0]->numel() == 1,
                   "condition input's data type should be bool, "
                   "numel should be 1, actual numel is %d",
                   ips[0]->numel());
F
fengjiayi 已提交
57
    bool res = false;
58 59 60 61 62 63 64 65 66 67 68
    if (platform::is_gpu_place(ips[0]->place())) {
#ifdef PADDLE_WITH_CUDA
      framework::LoDTensor cpu_tensor;
      framework::TensorCopy(*ips[0], platform::CPUPlace(), &cpu_tensor);
      platform::DeviceContextPool::Instance().Get(ips[0]->place())->Wait();
      res = cpu_tensor.data<bool>()[0];
#endif
    } else {
      res = ips[0]->data<bool>()[0];
    }
    return res;
69
  }
Y
Yu Yang 已提交
70 71 72 73 74 75 76 77 78
};

class ConditionalBlockOp : public ConditionalOp {
 public:
  ConditionalBlockOp(const std::string &type,
                     const framework::VariableNameMap &inputs,
                     const framework::VariableNameMap &outputs,
                     const framework::AttributeMap &attrs)
      : ConditionalOp(type, inputs, outputs, attrs) {}
79 80 81 82

 private:
  void RunImpl(const framework::Scope &scope,
               const platform::Place &dev_place) const override {
83 84
    bool need_run;
    if (Attr<bool>("is_scalar_condition")) {
85 86 87 88
      // When is_scalar_condition is True, the conditional variable is a scalar,
      // whether need to execute the operators in sub-block depends on the
      // conditional variable (Cond).
      auto xs = InputTensors(scope, "Cond");
89 90
      need_run = ScalarCondition(xs);
    } else {
91 92 93 94
      // When is_scalar_condition is False, the conditional variable maybe a
      // vector or tensor, whether need to execute the operators in sub-block
      // depends on the input variables (Input).
      auto xs = InputTensors(scope, "Input");
95 96 97 98
      need_run = std::all_of(
          xs.begin(), xs.end(),
          [](const framework::LoDTensor *t) { return t->numel() != 0; });
    }
Y
Yu Yang 已提交
99 100 101 102 103 104 105 106 107

    if (need_run) {
      auto *scope_var = scope.FindVar(Output("Scope"));
      PADDLE_ENFORCE(scope_var != nullptr, "Must set scope");
      auto *scopes = scope_var->GetMutable<std::vector<framework::Scope *>>();
      scopes->resize(1);
      scopes->front() = &scope.NewScope();
      auto &cur_scope = *scopes->front();

D
dzhwinter 已提交
108
      framework::Executor exec(dev_place);
Y
Yu Yang 已提交
109
      auto *block = Attr<framework::BlockDesc *>("sub_block");
Y
Yu Yang 已提交
110 111 112 113 114 115 116
      exec.Run(*block->Program(), &cur_scope, block->ID(), false);
    }
  }
};

class ConditionalBlockOpProtoMaker : public framework::OpProtoAndCheckerMaker {
 public:
Y
Yu Yang 已提交
117
  void Make() override {
118 119
    AddInput("Cond",
             "The conditional variable of this operator. If Cond is empty, the "
Y
Yu Yang 已提交
120 121
             "whole sub-block will not be executed.")
        .AsDuplicable();
122
    AddInput("Input", "The input variables of the sub-block.").AsDuplicable();
Y
Yu Yang 已提交
123 124 125 126 127
    AddOutput("Out", "The output variables of the sub-block.").AsDuplicable();
    AddOutput("Scope",
              "(std::vector<Scope*>) The step scope of conditional block. To "
              "unify the conditional block, rnn and while op, the type of "
              "scope is std::vector<Scope*>");
Y
Yu Yang 已提交
128
    AddAttr<framework::BlockDesc *>(
129
        "sub_block", "The step block of conditional block operator");
130
    AddAttr<bool>("is_scalar_condition",
131 132
                  "The conditional variable (Cond) is used as scalar "
                  "condition.")
133
        .SetDefault(false);
Y
Yu Yang 已提交
134 135
    AddComment(R"DOC(Conditional block operator

136 137 138 139 140 141 142
If `is_scalar_condition` is True, the conditional variable (Cond) is a scalar,
run the operators in sub-block if Cond is True.

If `is_scalar_condition` is False, the conditional variable (Cond) is a vector or
tensor, run the operators in sub-block if all of input variables are not empty.


Y
Yu Yang 已提交
143 144 145 146 147 148 149 150 151 152 153
)DOC");
  }
};

class ConditionalBlockGradOp : public ConditionalOp {
 public:
  ConditionalBlockGradOp(const std::string &type,
                         const framework::VariableNameMap &inputs,
                         const framework::VariableNameMap &outputs,
                         const framework::AttributeMap &attrs)
      : ConditionalOp(type, inputs, outputs, attrs) {}
154 155 156 157

 private:
  void RunImpl(const framework::Scope &scope,
               const platform::Place &dev_place) const override {
158 159
    bool need_run;
    if (Attr<bool>("is_scalar_condition")) {
160
      auto xs = this->InputTensors(scope, "Cond");
161 162
      need_run = ScalarCondition(xs);
    } else {
163
      auto xs = this->InputTensors(scope, "Input");
164 165 166 167
      need_run = std::all_of(
          xs.begin(), xs.end(),
          [](const framework::LoDTensor *t) { return t->numel() != 0; });
    }
Y
Yu Yang 已提交
168 169 170 171 172 173 174

    if (need_run) {
      auto *scope_var = scope.FindVar(Input("Scope"));
      PADDLE_ENFORCE(scope_var != nullptr, "Must set scope");
      auto &scopes = scope_var->Get<std::vector<framework::Scope *>>();
      framework::Scope &cur_scope = *scopes[0];

D
dzhwinter 已提交
175
      framework::Executor exec(dev_place);
Y
Yu Yang 已提交
176
      auto *block = Attr<framework::BlockDesc *>("sub_block");
Y
Yu Yang 已提交
177 178
      exec.Run(*block->Program(), &cur_scope, block->ID(), false);

179 180
      AssignLocalGradientToGlobal(dev_place, cur_scope, Inputs("Input"),
                                  Outputs(framework::GradVarName("Input")));
Y
Yu Yang 已提交
181

182 183
      AssignLocalGradientToGlobal(dev_place, cur_scope, Inputs("Cond"),
                                  Outputs(framework::GradVarName("Cond")));
Y
Yu Yang 已提交
184 185 186 187 188
    }
  }

 private:
  void AssignLocalGradientToGlobal(
D
dzhwinter 已提交
189
      const platform::Place &place, const framework::Scope &cur_scope,
Y
Yu Yang 已提交
190 191 192 193 194 195 196 197 198 199
      const std::vector<std::string> &p_names,
      const std::vector<std::string> &pg_names) const {
    for (size_t i = 0; i < p_names.size(); ++i) {
      auto out_grad_name = pg_names[i];
      auto in_grad_name = framework::GradVarName(p_names[i]);
      auto *in_var = cur_scope.FindVar(in_grad_name);
      if (in_var == nullptr) {
        continue;
      }
      auto new_in_grad_name = cur_scope.Rename(in_grad_name);
Y
Yiqun Liu 已提交
200 201 202
      auto assign = framework::OpRegistry::CreateOp(
          "assign", {{"X", {new_in_grad_name}}}, {{"Out", {out_grad_name}}},
          framework::AttributeMap{});
D
dzhwinter 已提交
203
      assign->Run(cur_scope, place);
Y
Yu Yang 已提交
204 205 206 207 208 209 210 211
      cur_scope.Rename(new_in_grad_name, in_grad_name);
    }
  }
};

class ConditionalBlockGradInferShape : public framework::InferShapeBase {
 public:
  void operator()(framework::InferShapeContext *context) const override {
212 213 214 215 216
    PADDLE_ENFORCE(context->HasInputs("Cond"));
    if (context->HasInputs("Input")) {
      PADDLE_ENFORCE(context->HasOutputs(framework::GradVarName("Input")));
      context->SetOutputsDim(framework::GradVarName("Input"),
                             context->GetInputsDim("Input"));
Y
Yu Yang 已提交
217
    }
218 219 220
    if (context->HasOutputs(framework::GradVarName("Cond"))) {
      context->SetOutputsDim(framework::GradVarName("Cond"),
                             context->GetInputsDim("Cond"));
221
    }
Y
Yu Yang 已提交
222 223 224 225 226 227 228 229
  }
};

class ConditionalBlockGradMaker : public framework::SingleGradOpDescMaker {
 public:
  using framework::SingleGradOpDescMaker::SingleGradOpDescMaker;

 protected:
Y
Yu Yang 已提交
230 231
  std::unique_ptr<framework::OpDesc> Apply() const override {
    auto grad_op = new framework::OpDesc();
Y
Yu Yang 已提交
232
    grad_op->SetType("conditional_block_grad");
233 234
    grad_op->SetInput("Cond", Input("Cond"));
    grad_op->SetInput("Input", Input("Input"));
Y
Yu Yang 已提交
235 236 237
    grad_op->SetInput("Out", Output("Out"));
    grad_op->SetInput(framework::GradVarName("Out"), OutputGrad("Out"));
    grad_op->SetInput("Scope", Output("Scope"));
238 239 240 241
    grad_op->SetOutput(framework::GradVarName("Cond"),
                       InputGrad("Cond", false));
    grad_op->SetOutput(framework::GradVarName("Input"),
                       InputGrad("Input", false));
A
Abhinav Arora 已提交
242
    grad_op->SetBlockAttr("sub_block", this->grad_block_[0]);
243
    grad_op->SetAttr("is_scalar_condition", GetAttr("is_scalar_condition"));
Y
Yu Yang 已提交
244
    return std::unique_ptr<framework::OpDesc>(grad_op);
Y
Yu Yang 已提交
245 246 247 248 249 250 251 252 253 254 255 256
  }
};

}  // namespace operators
}  // namespace paddle

namespace ops = paddle::operators;
REGISTER_OPERATOR(conditional_block, ops::ConditionalBlockOp,
                  ops::ConditionalBlockOpProtoMaker,
                  ops::ConditionalBlockGradMaker);
REGISTER_OPERATOR(conditional_block_grad, ops::ConditionalBlockGradOp,
                  ops::ConditionalBlockGradInferShape);