Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
s920243400
PaddleDetection
提交
67308822
P
PaddleDetection
项目概览
s920243400
/
PaddleDetection
与 Fork 源项目一致
Fork自
PaddlePaddle / PaddleDetection
通知
2
Star
0
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
PaddleDetection
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
提交
67308822
编写于
9月 30, 2018
作者:
M
minqiyang
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Add selected_rows merge for clip_by_norm op
test=develop
上级
2f5a7cc4
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
25 addition
and
2 deletion
+25
-2
paddle/fluid/operators/CMakeLists.txt
paddle/fluid/operators/CMakeLists.txt
+2
-1
paddle/fluid/operators/clip_by_norm_op.h
paddle/fluid/operators/clip_by_norm_op.h
+23
-1
未找到文件。
paddle/fluid/operators/CMakeLists.txt
浏览文件 @
67308822
...
...
@@ -229,7 +229,7 @@ if(WITH_DISTRIBUTE)
op_library
(
${
dist_op
}
DEPS
${
DISTRIBUTE_DEPS
}
)
set_source_files_properties
(
${
dist_op
}
.cc PROPERTIES COMPILE_FLAGS
${
DISTRIBUTE_COMPILE_FLAGS
}
)
endforeach
()
#set_source_files_properties(send_recv_op_test.cc PROPERTIES COMPILE_FLAGS ${DISTRIBUTE_COMPILE_FLAGS})
#cc_test(test_send_recv SRCS send_recv_op_test.cc DEPS prefetch_op send_op
# listen_and_serv_op sum_op executor SERIAL)
...
...
@@ -267,6 +267,7 @@ if (WITH_GPU AND TENSORRT_FOUND)
else
()
set
(
DEPS_OPS
${
DEPS_OPS
}
tensorrt_engine_op
)
endif
()
op_library
(
clip_by_norm_op DEPS selected_rows_functor
)
op_library
(
sum_op DEPS selected_rows_functor
)
op_library
(
sgd_op DEPS selected_rows_functor
)
op_library
(
print_op DEPS lod_tensor
)
...
...
paddle/fluid/operators/clip_by_norm_op.h
浏览文件 @
67308822
...
...
@@ -16,6 +16,7 @@ limitations under the License. */
#include "paddle/fluid/framework/eigen.h"
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/operators/math/selected_rows_functor.h"
#include "paddle/fluid/platform/transform.h"
namespace
paddle
{
...
...
@@ -31,10 +32,31 @@ class ClipByNormKernel : public framework::OpKernel<T> {
public:
void
Compute
(
const
framework
::
ExecutionContext
&
context
)
const
override
{
auto
max_norm
=
context
.
Attr
<
T
>
(
"max_norm"
);
auto
*
input
=
context
.
Input
<
Tensor
>
(
"X"
);
auto
in_var
=
context
.
InputVar
(
"X"
);
auto
*
output
=
context
.
Output
<
Tensor
>
(
"Out"
);
output
->
mutable_data
<
T
>
(
context
.
GetPlace
());
const
Tensor
*
input
=
nullptr
;
if
(
in_var
->
IsType
<
framework
::
LoDTensor
>
())
{
input
=
context
.
Input
<
Tensor
>
(
"X"
);
}
else
if
(
in_var
->
IsType
<
framework
::
SelectedRows
>
())
{
auto
*
x
=
context
.
Input
<
framework
::
SelectedRows
>
(
"X"
);
// merge ids in selected rows first
math
::
scatter
::
MergeAdd
<
DeviceContext
,
T
>
merge_func
;
auto
*
merged_input
=
const_cast
<
framework
::
Scope
&>
(
context
.
scope
())
.
Var
()
->
GetMutable
<
framework
::
SelectedRows
>
();
merge_func
(
context
.
template
device_context
<
DeviceContext
>(),
*
x
,
merged_input
);
input
=
&
(
merged_input
->
value
());
}
else
{
PADDLE_THROW
(
"Unexpected branch, input variable type is %s"
,
in_var
->
Type
().
name
());
}
PADDLE_ENFORCE_NOT_NULL
(
input
);
auto
x
=
EigenVector
<
T
>::
Flatten
(
*
input
);
auto
out
=
EigenVector
<
T
>::
Flatten
(
*
output
);
auto
x_norm
=
x
.
square
().
sum
().
sqrt
();
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录