lstm_unit_op.cc 4.0 KB
Newer Older
Z
zchen0211 已提交
1 2
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.

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
Z
zchen0211 已提交
6

L
Luo Tao 已提交
7
    http://www.apache.org/licenses/LICENSE-2.0
Z
zchen0211 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. */
Z
zchen0211 已提交
14

Y
Yi Wang 已提交
15
#include "paddle/fluid/operators/lstm_unit_op.h"
Z
zchen0211 已提交
16 17 18 19 20 21 22 23

namespace paddle {
namespace operators {

class LstmUnitOp : public framework::OperatorWithKernel {
 public:
  using framework::OperatorWithKernel::OperatorWithKernel;

24
  void InferShape(framework::InferShapeContext* ctx) const override {
Q
Qiao Longfei 已提交
25 26 27 28 29 30 31 32 33 34 35 36
    PADDLE_ENFORCE(ctx->HasInput("X"), "Input(X) of LSTM should not be null.");
    PADDLE_ENFORCE(ctx->HasInput("C_prev"),
                   "Input(C_prev) of LSTM should not be null.");
    PADDLE_ENFORCE(ctx->HasOutput("C"),
                   "Output(C) of LSTM should not be null.");
    PADDLE_ENFORCE(ctx->HasOutput("H"),
                   "Output(H) of LSTM should not be null.");

    auto x_dims = ctx->GetInputDim("X");
    auto c_prev_dims = ctx->GetInputDim("C_prev");

    PADDLE_ENFORCE_EQ(x_dims.size(), 2, "Input(X)'s rank must be 2.");
Y
Yang Yang(Tony) 已提交
37 38 39 40
    PADDLE_ENFORCE_EQ(x_dims[0], c_prev_dims[0],
                      "Batch size of inputs and states must be equal");
    PADDLE_ENFORCE_EQ(x_dims[1], c_prev_dims[1] * 4,
                      "Dimension of FC should equal to prev state * 4");
Z
zchen0211 已提交
41

Q
Qiao Longfei 已提交
42 43 44 45
    int b_size = c_prev_dims[0];  // batch size
    int s_dim = c_prev_dims[1];   // state dim
    ctx->SetOutputDim("C", {b_size, s_dim});
    ctx->SetOutputDim("H", {b_size, s_dim});
Z
zchen0211 已提交
46 47 48 49 50
  }
};

class LstmUnitOpMaker : public framework::OpProtoAndCheckerMaker {
 public:
51
  LstmUnitOpMaker(OpProto* proto, OpAttrChecker* op_checker)
Z
zchen0211 已提交
52
      : OpProtoAndCheckerMaker(proto, op_checker) {
Y
yangyaming 已提交
53 54 55 56
    AddInput("X",
             "Lstm unit only applies non-linear activations, please make sure"
             "that linear tranformation has already been applied to `X`. "
             "Linear tranformation can be applied by adding a `fc` layer");
Z
zchen0211 已提交
57 58 59 60 61
    AddInput(
        "C_prev",
        "The cell state tensor of last time-step in the Lstm Unit operator.");
    AddOutput("C", "The cell tensor of Lstm Unit operator.");
    AddOutput("H", "The hidden state tensor of Lstm Unit operator.");
K
kexinzhao 已提交
62 63 64 65 66 67
    AddAttr<float>("forget_bias",
                   "(float, default 0.0) "
                   "The forget bias of Lstm Unit.")
        .SetDefault(0.0);
    AddComment(R"DOC(
Lstm Unit Operator
Z
zchen0211 已提交
68

Q
Qiao Longfei 已提交
69
Equation:
K
kexinzhao 已提交
70 71 72 73 74 75

$$
i, f, o, j = split(X) \\
C = C_{prev} * sigm(f + forget\_bias) + sigm(i) * tanh(j) \\
H = C * sigm(o)
$$
Q
Qiao Longfei 已提交
76

Z
zchen0211 已提交
77 78 79 80 81 82 83 84
)DOC");
  }
};

class LstmUnitGradOp : public framework::OperatorWithKernel {
 public:
  using framework::OperatorWithKernel::OperatorWithKernel;

85
  void InferShape(framework::InferShapeContext* ctx) const override {
Q
Qiao Longfei 已提交
86 87 88 89 90 91 92
    PADDLE_ENFORCE(ctx->HasInput(framework::GradVarName("C")),
                   "Input(C@GRAD) should not be null");
    PADDLE_ENFORCE(ctx->HasInput(framework::GradVarName("H")),
                   "Input(H@GRAD) should not be null");
    ctx->SetOutputDim(framework::GradVarName("X"), ctx->GetInputDim("X"));
    ctx->SetOutputDim(framework::GradVarName("C_prev"),
                      ctx->GetInputDim("C_prev"));
Z
zchen0211 已提交
93 94 95 96 97 98 99
  }
};

}  // namespace operators
}  // namespace paddle

namespace ops = paddle::operators;
100 101
REGISTER_OP(lstm_unit, ops::LstmUnitOp, ops::LstmUnitOpMaker, lstm_unit_grad,
            ops::LstmUnitGradOp);
Z
zchen0211 已提交
102
REGISTER_OP_CPU_KERNEL(lstm_unit,
103 104
                       ops::LstmUnitKernel<paddle::platform::CPUPlace, float>,
                       ops::LstmUnitKernel<paddle::platform::CPUPlace, double>);
Z
lstm  
zchen0211 已提交
105
REGISTER_OP_CPU_KERNEL(
106 107
    lstm_unit_grad, ops::LstmUnitGradKernel<paddle::platform::CPUPlace, float>,
    ops::LstmUnitGradKernel<paddle::platform::CPUPlace, double>);