diff --git a/doc/fluid/dev/new_op_cn.md b/doc/fluid/dev/new_op_cn.md index 0c3f88d9c31e05bec399c64bf6ade56e62e01f68..0fa3252d220a49c88a5ea01a15073e74d6489c99 100644 --- a/doc/fluid/dev/new_op_cn.md +++ b/doc/fluid/dev/new_op_cn.md @@ -54,10 +54,10 @@ -实现新的op都添加至目录[paddle/operators](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/operators)下，文件命名以`*_op.h`（如有） 、 `*_op.cc` 、`*_op.cu`（如有）结尾。**系统会根据文件名自动构建op和其对应的Python扩展。** +实现新的op都添加至目录[paddle/fluid/operators](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/fluid/operators)下，文件命名以`*_op.h`（如有） 、 `*_op.cc` 、`*_op.cu`（如有）结尾。**系统会根据文件名自动构建op和其对应的Python扩展。** -下面以矩阵乘操作，即[MulOp](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/mul_op.cc)为例来介绍如何写带Kernel的Operator。 +下面以矩阵乘操作，即[MulOp](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/operators/mul_op.cc)为例来介绍如何写带Kernel的Operator。 ## 实现C++类 @@ -85,17 +85,17 @@ The equation is: Out = X * Y }; ``` -[`MulOpMaker`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/mul_op.cc#L43)继承自`framework::OpProtoAndCheckerMaker`，构造函数含有2个参数： +[`MulOpMaker`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/operators/mul_op.cc#L76-L127)继承自`framework::OpProtoAndCheckerMaker`，构造函数含有2个参数： - `framework::OpProto` ： 前者存储Op的输入输出和参数属性，将用于Python API接口的生成。 - `framework::OpAttrChecker` ：后者用于检查参数属性的合法性。 构造函数里通过`AddInput`添加输入参数，通过`AddOutput`添加输出参数，通过`AddComment`添加Op的注释。这些函数会将对应内容添加到`OpProto`中。 -上面的代码在`MulOp`中添加两个输入`X`和`Y`，添加了一个输出`Out`，并解释了各自含义，命名请遵守[命名规范](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/name_convention.md)。 +上面的代码在`MulOp`中添加两个输入`X`和`Y`，添加了一个输出`Out`，并解释了各自含义，命名请遵守[命名规范](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/fluid/dev/name_convention.md)。 -再以[`ScaleOp`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/scale_op.cc#L37)为例： +再以[`ScaleOp`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/operators/scale_op.cc#L38-L55)为例： ```cpp template @@ -103,21 +103,21 @@ class ScaleOpMaker : public framework::OpProtoAndCheckerMaker { public: ScaleOpMaker(OpProto *proto, OpAttrChecker *op_checker) : OpProtoAndCheckerMaker(proto, op_checker) { - AddInput("X", "The input tensor of scale operator.").NotInGradient(); - AddOutput("Out", "The output tensor of scale operator.").NotInGradient(); - AddComment(R"DOC(Scale operator -The equation is: Out = scale*X + AddInput("X", "(Tensor) Input tensor of scale operator."); + AddOutput("Out", "(Tensor) Output tensor of scale operator."); + AddComment(R"DOC( +Scale operator +$$Out = scale*X$$ )DOC"); - AddAttr("scale", "scale of scale operator.").SetDefault(1.0); + AddAttr("scale", + "(float, default 1.0)" + "The scaling factor of the scale operator.") + .SetDefault(1.0); } }; ``` -这个例子有两处不同： - -- `AddInput("X","...").NotInGradient()` : 表示`X`这个输入不参与`ScaleOp`对应的梯度Op计算之中，如果Op的某个输入不参与反向梯度的计算，请显示地调用`.NotInGradient()`进行设置。 - -- `AddAttr("scale", "...").SetDefault(1.0);` : 增加`scale`系数，作为参数属性，并且设置默认值为1.0。 +这个例子有`AddAttr("scale", "...").SetDefault(1.0);` : 增加`scale`系数，作为参数属性，并且设置默认值为1.0。 ### 定义Operator类 @@ -205,7 +205,6 @@ MulOp(const std::string &type, const framework::VariableNameMap &inputs, 为了使`OpKernel`的计算过程书写更加简单，并且CPU、CUDA的代码可以复用，我们通常借助 Eigen unsupported Tensor模块来实现`Compute`接口。关于在PaddlePaddle中如何使用Eigen库，请参考[使用文档](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/howto/dev/use_eigen_cn.md)。 - 到此，前向Op实现完成。接下来，需要在`.cc`文件中注册该op和kernel。 反向Op类的定义，反向OpKernel的定义与前向Op类似，这里不再赘述。**但需注意反向Op没有`ProtoMaker`**。 @@ -215,7 +214,9 @@ MulOp(const std::string &type, const framework::VariableNameMap &inputs, ```cpp namespace ops = paddle::operators; - REGISTER_OP(mul, ops::MulOp, ops::MulOpMaker, mul_grad, ops::MulOpGrad); + REGISTER_OPERATOR(mul, ops::MulOp, ops::MulOpMaker, + paddle::framework::DefaultGradOpDescMaker) + REGISTER_OPERATOR(mul_grad, ops::MulGradOp) REGISTER_OP_CPU_KERNEL(mul, ops::MulKernel); REGISTER_OP_CPU_KERNEL(mul_grad, ops::MulGradKernel); @@ -223,8 +224,7 @@ MulOp(const std::string &type, const framework::VariableNameMap &inputs, 在上面的代码中： - - `REGISTER_OP` ： 注册`ops::MulOp`类，类型名为`mul`，该类的`ProtoMaker`为`ops::MulOpMaker`，注册`ops::MulOpGrad`，类型名为`mul_grad`。 - - `REGISTER_OP_WITHOUT_GRADIENT` ： 用于注册没有反向的Op。 + - `REGISTER_OPERATOR` ： 注册`ops::MulOp`类，类型名为`mul`，该类的`ProtoMaker`为`ops::MulOpMaker`，注册`ops::MulOpGrad`，类型名为`mul_grad`。 - `REGISTER_OP_CPU_KERNEL` ：注册`ops::MulKernel`类，并特化模板参数为`paddle::platform::CPUPlace`和`float`类型，同理，注册`ops::MulGradKernel`类。 @@ -255,7 +255,7 @@ make mul_op ## 实现单元测试 -单测包括对比前向Op不同设备(CPU、CUDA)的实现、对比反向OP不同设备(CPU、CUDA)的实现、反向Op的梯度测试。下面介绍介绍[`MulOp`的单元测试](https://github.com/PaddlePaddle/Paddle/blob/develop/python/paddle/v2/framework/tests/test_mul_op.py)。 +单测包括对比前向Op不同设备(CPU、CUDA)的实现、对比反向OP不同设备(CPU、CUDA)的实现、反向Op的梯度测试。下面介绍介绍[`MulOp`的单元测试](https://github.com/PaddlePaddle/Paddle/blob/develop/python/paddle/fluid/tests/unittests/test_mul_op.py)。 ### 前向Operator单测 @@ -315,7 +315,7 @@ Op单元测试继承自`OpTest`。各项更加具体的单元测试在`TestMulOp ### 编译和执行 -`python/paddle/v2/framework/tests` 目录下新增的 `test_*.py` 单元测试会被自动加入工程进行编译。 +`python/paddle/fluid/tests/unittests/` 目录下新增的 `test_*.py` 单元测试会被自动加入工程进行编译。 请注意，**不同于Op的编译测试，运行单元测试测时需要编译整个工程**，并且编译时需要打开`WITH_TESTING`, 即`cmake paddle_dir -DWITH_TESTING=ON`。编译成功后，执行下面的命令来运行单元测试： @@ -331,7 +331,6 @@ ctest -R test_mul_op ## 注意事项 -- 为每个Op创建单独的`*_op.h`（如有）、`*_op.cc`和`*_op.cu`（如有）。不允许一个文件中包含多个Op，这将会导致编译出错。 -- 注册Op时的类型名，需要和该Op的名字一样。即不允许在`A_op.cc`里面，注册`REGISTER_OP(B, ...)`等，这将会导致单元测试出错。 +- 注册Op时的类型名，需要和该Op的名字一样。即不允许在`A_op.cc`里面，注册`REGISTER_OPERATOR(B, ...)`等，这将会导致单元测试出错。 - 如果Op没有实现CUDA Kernel，请不要创建空的`*_op.cu`，这将会导致单元测试出错。 - 如果多个Op依赖一些共用的函数，可以创建非`*_op.*`格式的文件来存放，如`gather.h`文件。 diff --git a/doc/fluid/dev/new_op_en.md b/doc/fluid/dev/new_op_en.md index a566a09131f86251b70d5435d0a483aa2a705b35..c8fb7e3b3bfe200aeed43f2c580b8409a2e9008a 100644 --- a/doc/fluid/dev/new_op_en.md +++ b/doc/fluid/dev/new_op_en.md @@ -61,10 +61,10 @@ Registering the Op | Ops are registered in `.cc` files; For Kernel reg -New Operator implementations are added to the list [paddle/operators](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/operators), with file names in the format `*_op.h` (if applicable), `*_op.cc`, `*_op.cu` (if applicable).** The system will use the naming scheme to automatically build operators and their corresponding Python extensions.** +New Operator implementations are added to the list [paddle/operators](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/fluid/operators), with file names in the format `*_op.h` (if applicable), `*_op.cc`, `*_op.cu` (if applicable).** The system will use the naming scheme to automatically build operators and their corresponding Python extensions.** -Let's take matrix multiplication operator, [MulOp](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/mul_op.cc), as an example to introduce the writing of an Operator with Kernel. +Let's take matrix multiplication operator, [MulOp](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/operators/mul_op.cc), as an example to introduce the writing of an Operator with Kernel. ## Implementing C++ Types @@ -92,17 +92,17 @@ The equation is: Out = X * Y }; ``` -[`MulOpMaker`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/mul_op.cc#L43)is inherited from`framework::OpProtoAndCheckerMaker`, consisting of 2 variables in the constructor： +[`MulOpMaker`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/operators/mul_op.cc#L76-L127)is inherited from`framework::OpProtoAndCheckerMaker`, consisting of 2 variables in the constructor： - `framework::OpProto` stores Operator input and variable attribute, used for generating Python API interfaces. - `framework::OpAttrChecker` is used to validate variable attributes. The constructor utilizes `AddInput`, `AddOutput`, and `AddComment`, so that the corresponding information will be added to `OpProto`. -The code above adds two inputs `X` and `Y` to `MulOp`, an output `Out`, and their corresponding descriptions, in accordance to Paddle's [naming convention](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/name_convention.md). +The code above adds two inputs `X` and `Y` to `MulOp`, an output `Out`, and their corresponding descriptions, in accordance to Paddle's [naming convention](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/fluid/dev/name_convention.md). -An additional example [`ScaleOp`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/scale_op.cc#L37) is implemented as follows: +An additional example [`ScaleOp`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/operators/scale_op.cc#L38-L55) is implemented as follows: ```cpp template @@ -120,11 +120,7 @@ The equation is: Out = scale*X }; ``` -There are two changes in this example: - -- `AddInput("X","...").NotInGradient()` expresses that input `X` is not involved in `ScaleOp`'s corresponding computation. If an input to an operator is not participating in back-propagation, please explicitly set `.NotInGradient()`. - -- `AddAttr("scale", "...").SetDefault(1.0);` adds `scale`constant as an attribute, and sets the default value to 1.0. +Note `AddAttr("scale", "...").SetDefault(1.0);` adds `scale`constant as an attribute, and sets the default value to 1.0. ### Defining Operator @@ -154,7 +150,7 @@ class MulOp : public framework::OperatorWithKernel { }; ``` -[`MulOp`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/mul_op.cc#L22) is inherited from `OperatorWithKernel`. Its `public` member +[`MulOp`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/operators/mul_op.cc#L24) is inherited from `OperatorWithKernel`. Its `public` member ```cpp using framework::OperatorWithKernel::OperatorWithKernel; @@ -209,7 +205,7 @@ Usually `OpProtoMaker` and `Op`'s type definitions are written in `.cc` files, w Note that **different devices (CPU, CUDA)share one Op definition; whether or not they share the same `OpKernel` depends on whether `Compute` calls functions can support both devices.** -`MulOp`'s CPU and CUDA share the same `Kernel`. A non-sharing `OpKernel` example can be seen in [`OnehotCrossEntropyOpKernel`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/operators/cross_entropy_op.h#L43). +`MulOp`'s CPU and CUDA share the same `Kernel`. A non-sharing `OpKernel` example can be seen in [`OnehotCrossEntropyOpKernel`](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/operators/cross_entropy_op.cc). To ease the writing of `OpKernel` compute, and for reusing code cross-device, [`Eigen-unsupported Tensor`](https://bitbucket.org/eigen/eigen/src/default/unsupported/Eigen/CXX11/src/Tensor/README.md?fileviewer=file-view-default) module is used to implement `Compute` interface. To learn about how the Eigen library is used in PaddlePaddle, please see [usage document](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/howto/dev/use_eigen_cn.md). @@ -224,7 +220,9 @@ The definition of its corresponding backward operator, if applicable, is similar ```cpp namespace ops = paddle::operators; - REGISTER_OP(mul, ops::MulOp, ops::MulOpMaker, mul_grad, ops::MulOpGrad); + REGISTER_OPERATOR(mul, ops::MulOp, ops::MulOpMaker, + paddle::framework::DefaultGradOpDescMaker) + REGISTER_OPERATOR(mul_grad, ops::MulGradOp) REGISTER_OP_CPU_KERNEL(mul, ops::MulKernel); REGISTER_OP_CPU_KERNEL(mul_grad, @@ -233,9 +231,8 @@ The definition of its corresponding backward operator, if applicable, is similar In that code block, - - `REGISTER_OP` registers the `ops::MulOp` class, type named `mul`, its type `ProtoMaker` is `ops::MulOpMaker`, registering `ops::MulOpGrad` as `mul_grad`. + - `REGISTER_OPERATOR` registers the `ops::MulOp` class, type named `mul`, its type `ProtoMaker` is `ops::MulOpMaker`, registering `ops::MulOpGrad` as `mul_grad`. - `REGISTER_OP_WITHOUT_GRADIENT` registers an operator without gradient. - - `REGISTER_OP_CPU_KERNEL` registers `ops::MulKernel` class and specialized template types `paddle::platform::CPUPlace` and `float`, which also registers `ops::MulGradKernel`. @@ -275,7 +272,7 @@ Unit tests for an operator include 3. a scaling test for the backward operator. -Here, we introduce the [unit tests for `MulOp`](https://github.com/PaddlePaddle/Paddle/blob/develop/python/paddle/v2/framework/tests/test_mul_op.py). +Here, we introduce the [unit tests for `MulOp`](https://github.com/PaddlePaddle/Paddle/blob/develop/python/paddle/fluid/tests/unittests/test_mul_op.py). ### Testing Forward Operators @@ -339,7 +336,7 @@ Some key points in checking gradient above include: ### Compiling and Running -Any new unit testing file of the format `test_*.py` added to the director `python/paddle/v2/framework/tests` is automatically added to the project to compile. +Any new unit testing file of the format `test_*.py` added to the director `python/paddle/fluid/tests/unittests/` is automatically added to the project to compile. Note that **unlike the compile test for Ops, running unit tests requires compiling the entire project** and requires compiling with flag `WITH_TESTING` on i.e. `cmake paddle_dir -DWITH_TESTING=ON`. @@ -357,7 +354,6 @@ ctest -R test_mul_op ## Remarks -- Every `*_op.h` (if applicable), `*_op.cc`, and `*_op.cu` (if applicable) must be created for a unique Op. Compiling will fail if multiple operators are included per file. -- The type with which an operator is registered needs to be identical to the Op's name. Registering `REGISTER_OP(B, ...)` in `A_op.cc` will cause unit testing failures. +- The type with which an operator is registered needs to be identical to the Op's name. Registering `REGISTER_OPERATOR(B, ...)` in `A_op.cc` will cause unit testing failures. - If the operator does not implement a CUDA kernel, please refrain from creating an empty `*_op.cu` file, or else unit tests will fail. - If multiple operators rely on some shared methods, a file NOT named `*_op.*` can be created to store them, such as `gather.h`.