Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

delete final state pre name #5191

Closed
wants to merge 3 commits into from
Closed

delete final state pre name #5191

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
5605a42
delete final state
wanghuancoder Aug 25, 2022
7731106
refine
wanghuancoder Aug 25, 2022
ef53aa3
Merge branch 'develop' into fix_new_python_api_cn
wanghuancoder Aug 26, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions docs/dev_guides/api_contributing_guides/new_cpp_op_cn.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -771,7 +771,7 @@ def trace(x, offset=0, axis1=0, axis2=1, name=None):
__check_input(input, offset, axis1, axis2)

if in_dygraph_mode():
return _C_ops.final_state_trace( x, offset, axis1, axis2 )
return _C_ops.trace( x, offset, axis1, axis2 )

helper = LayerHelper('trace', **locals())
out = helper.create_variable_for_type_inference(dtype=x.dtype)
Expand Down Expand Up @@ -1068,7 +1068,7 @@ Paddle 支持动态图和静态图两种模式，在 YAML 配置文件中完成
- 动态图前向函数与反向节点（Autograd API）：在 C++ API 的基础上进行了封装，组成一个提供自动微分功能的 C++函数接口。
- 注：生成的相关代码在`paddle/fluid/eager/api/generated/eager_generated`目录下
- Python-C 接口：将支持自动微分功能的 C++的函数接口（Autograd API）暴露到 Python 层供 Python API 调用。
- 注：生成的 Python-C 接口代码在`paddle/fluid/pybind/eager_final_state_op_function_impl.h`中
- 注：生成的 Python-C 接口代码在`paddle/fluid/pybind/eager_op_function.cc`中
- 静态图的执行流程与动态图不同，所以生成的代码也与动态图有较大差异。静态图由于是先组网后计算，Python API 主要负责组网，算子的调度和 kernel 计算由静态图执行器来完成，因此自动生成的代码是将配置文件中的算子信息注册到框架内供执行器调度，主要包括[OpMaker](https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/fluid/framework/op_proto_maker.h)（静态图中定义算子的输入、输出以及属性等信息）和`REGISTER_OPERATOR`(将算子名称以及 OpMaker 等信息进行注册)等静态图算子注册组件，具体的代码逻辑可参考`paddle/fluid/operators/generated_op.cc`

**注意：由于代码自动生成在编译时进行，所以查看上述生成代码需要先完成[框架的编译](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/install/compile/fromsource.html)。**
6 changes: 3 additions & 3 deletions docs/dev_guides/api_contributing_guides/new_python_api_cn.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -77,7 +77,7 @@ def trace(x, offset=0, axis1=0, axis2=1, name=None):
# 为了突出重点，省略部分代码
# 动态图分支，直接调用算子对应的 Python C 函数
if in_dygraph_mode():
return _C_ops.final_state_trace( x, offset, axis1, axis2 )
return _C_ops.trace( x, offset, axis1, axis2 )

# 静态图分支
## 输入参数检查
Expand All @@ -103,11 +103,11 @@ def trace(x, offset=0, axis1=0, axis2=1, name=None):
```Python
# 动态图分支，直接调用算子对应的 Python C 函数
if in_dygraph_mode():
return _C_ops.final_state_trace( x, offset, axis1, axis2 )
return _C_ops.trace( x, offset, axis1, axis2 )
```
_C_ops 是 Python/paddle/_C_ops.py，其中从 paddle 编译得到的二进制文件中 import 了 C++ 算子对应的 Python C 函数。

在动态图模式下，Python C 的调用函数名为 final_state_ + 算子名，然后将参数按照 YAML 配置文件中定义的输入参数顺序传入即可。
在动态图模式下，Python C 的调用函数名为 算子名，然后将参数按照 YAML 配置文件中定义的输入参数顺序传入即可。

> 注意：由于目前动态图正处在重构升级阶段，所以现有算子的代码会分别有新旧动态图两个代码分支，其中 `in_dygraph_mode()` 表示新动态图分支（默认），`_in_legacy_dygraph()`为旧动态图分支，**在新增算子时无需添加旧动态图分支代码**。

Expand Down