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
PaddlePaddle
Paddle
提交
ae622479
P
Paddle
项目概览
PaddlePaddle
/
Paddle
大约 2 年 前同步成功
通知
2325
Star
20933
Fork
5424
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
1423
列表
看板
标记
里程碑
合并请求
543
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
Paddle
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
1,423
Issue
1,423
列表
看板
标记
里程碑
合并请求
543
合并请求
543
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
未验证
提交
ae622479
编写于
2月 18, 2023
作者:
G
Guo Sheng
提交者:
GitHub
2月 18, 2023
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
[Zero-Dim] Support 0D for kron (#49847)
* Add unittest for zero-dim kron. * Fix zero dim kron.
上级
af23efe0
变更
3
显示空白变更内容
内联
并排
Showing
3 changed file
with
100 addition
and
94 deletion
+100
-94
paddle/phi/kernels/impl/kron_grad_kernel_impl.h
paddle/phi/kernels/impl/kron_grad_kernel_impl.h
+66
-64
paddle/phi/kernels/impl/kron_kernel_impl.h
paddle/phi/kernels/impl/kron_kernel_impl.h
+33
-30
python/paddle/fluid/tests/unittests/test_zero_dim_tensor.py
python/paddle/fluid/tests/unittests/test_zero_dim_tensor.py
+1
-0
未找到文件。
paddle/phi/kernels/impl/kron_grad_kernel_impl.h
浏览文件 @
ae622479
...
@@ -20,15 +20,15 @@ namespace phi {
...
@@ -20,15 +20,15 @@ namespace phi {
template
<
typename
T
>
template
<
typename
T
>
struct
KronGradElemFunctor
{
struct
KronGradElemFunctor
{
KronGradElemFunctor
(
const
T
*
dout
,
KronGradElemFunctor
(
const
T
*
dout
,
const
T
*
A
,
const
T
*
A
,
const
T
*
B
,
const
T
*
B
,
T
*
dout_a
,
T
*
dout_a
,
T
*
dout_b
,
T
*
dout_b
,
const
int64_t
*
stride_dout
,
const
int64_t
*
stride_dout
,
const
int64_t
*
stride_a
,
const
int64_t
*
stride_a
,
const
int64_t
*
stride_b
,
const
int64_t
*
stride_b
,
const
int64_t
*
shape_b
,
const
int64_t
*
shape_b
,
const
int64_t
numel_a
,
const
int64_t
numel_a
,
const
int64_t
numel_b
,
const
int64_t
numel_b
,
const
int
ndims
)
const
int
ndims
)
...
@@ -69,15 +69,15 @@ struct KronGradElemFunctor {
...
@@ -69,15 +69,15 @@ struct KronGradElemFunctor {
}
}
private:
private:
const
T
*
dout_
;
const
T
*
dout_
;
const
T
*
A_
;
const
T
*
A_
;
const
T
*
B_
;
const
T
*
B_
;
T
*
dout_a_
;
T
*
dout_a_
;
T
*
dout_b_
;
T
*
dout_b_
;
const
int64_t
*
stride_dout_
;
const
int64_t
*
stride_dout_
;
const
int64_t
*
stride_a_
;
const
int64_t
*
stride_a_
;
const
int64_t
*
stride_b_
;
const
int64_t
*
stride_b_
;
const
int64_t
*
shape_b_
;
const
int64_t
*
shape_b_
;
const
int64_t
numel_a_
;
const
int64_t
numel_a_
;
const
int64_t
numel_b_
;
const
int64_t
numel_b_
;
const
int
ndims_
;
const
int
ndims_
;
...
@@ -85,15 +85,15 @@ struct KronGradElemFunctor {
...
@@ -85,15 +85,15 @@ struct KronGradElemFunctor {
template
<
typename
T
>
template
<
typename
T
>
struct
KronGradElemFunctor
<
dtype
::
complex
<
T
>>
{
struct
KronGradElemFunctor
<
dtype
::
complex
<
T
>>
{
KronGradElemFunctor
(
const
dtype
::
complex
<
T
>
*
dout
,
KronGradElemFunctor
(
const
dtype
::
complex
<
T
>
*
dout
,
const
dtype
::
complex
<
T
>
*
A
,
const
dtype
::
complex
<
T
>
*
A
,
const
dtype
::
complex
<
T
>
*
B
,
const
dtype
::
complex
<
T
>
*
B
,
dtype
::
complex
<
T
>
*
dout_a
,
dtype
::
complex
<
T
>
*
dout_a
,
dtype
::
complex
<
T
>
*
dout_b
,
dtype
::
complex
<
T
>
*
dout_b
,
const
int64_t
*
stride_dout
,
const
int64_t
*
stride_dout
,
const
int64_t
*
stride_a
,
const
int64_t
*
stride_a
,
const
int64_t
*
stride_b
,
const
int64_t
*
stride_b
,
const
int64_t
*
shape_b
,
const
int64_t
*
shape_b
,
const
int64_t
numel_a
,
const
int64_t
numel_a
,
const
int64_t
numel_b
,
const
int64_t
numel_b
,
const
int
ndims
)
const
int
ndims
)
...
@@ -136,15 +136,15 @@ struct KronGradElemFunctor<dtype::complex<T>> {
...
@@ -136,15 +136,15 @@ struct KronGradElemFunctor<dtype::complex<T>> {
}
}
private:
private:
const
dtype
::
complex
<
T
>
*
dout_
;
const
dtype
::
complex
<
T
>
*
dout_
;
const
dtype
::
complex
<
T
>
*
A_
;
const
dtype
::
complex
<
T
>
*
A_
;
const
dtype
::
complex
<
T
>
*
B_
;
const
dtype
::
complex
<
T
>
*
B_
;
dtype
::
complex
<
T
>
*
dout_a_
;
dtype
::
complex
<
T
>
*
dout_a_
;
dtype
::
complex
<
T
>
*
dout_b_
;
dtype
::
complex
<
T
>
*
dout_b_
;
const
int64_t
*
stride_dout_
;
const
int64_t
*
stride_dout_
;
const
int64_t
*
stride_a_
;
const
int64_t
*
stride_a_
;
const
int64_t
*
stride_b_
;
const
int64_t
*
stride_b_
;
const
int64_t
*
shape_b_
;
const
int64_t
*
shape_b_
;
const
int64_t
numel_a_
;
const
int64_t
numel_a_
;
const
int64_t
numel_b_
;
const
int64_t
numel_b_
;
const
int
ndims_
;
const
int
ndims_
;
...
@@ -152,29 +152,31 @@ struct KronGradElemFunctor<dtype::complex<T>> {
...
@@ -152,29 +152,31 @@ struct KronGradElemFunctor<dtype::complex<T>> {
template
<
typename
Context
,
typename
T
>
template
<
typename
Context
,
typename
T
>
struct
KronGradOpFunctor
{
struct
KronGradOpFunctor
{
void
operator
()(
const
Context
&
dev_ctx
,
void
operator
()(
const
Context
&
dev_ctx
,
const
DenseTensor
&
dout
,
const
DenseTensor
&
dout
,
const
DenseTensor
&
x
,
const
DenseTensor
&
x
,
const
DenseTensor
&
y
,
const
DenseTensor
&
y
,
DenseTensor
*
dx
,
DenseTensor
*
dx
,
DenseTensor
*
dy
)
{
DenseTensor
*
dy
)
{
int
ndims
=
dout
.
dims
().
size
();
int
ndims
=
dout
.
dims
().
size
();
int64_t
numel
=
dout
.
numel
();
int64_t
numel
=
dout
.
numel
();
int64_t
numel_x
=
x
.
numel
();
int64_t
numel_x
=
x
.
numel
();
int64_t
numel_y
=
y
.
numel
();
int64_t
numel_y
=
y
.
numel
();
const
phi
::
DDim
&
dim_x
=
x
.
dims
();
const
phi
::
DDim
&
dim_x
=
x
.
dims
();
const
phi
::
DDim
&
dim_y
=
y
.
dims
();
const
phi
::
DDim
&
dim_y
=
y
.
dims
();
const
phi
::
DDim
&
dim_dout
=
dout
.
dims
();
const
phi
::
DDim
&
dim_dout
=
dout
.
dims
();
const
phi
::
DDim
stride_x
=
dim_x
.
size
()
==
0
?
phi
::
DDim
(
dim_x
)
:
phi
::
stride
(
dim_x
);
const
phi
::
DDim
stride_y
=
dim_y
.
size
()
==
0
?
phi
::
DDim
(
dim_y
)
:
phi
::
stride
(
dim_y
);
const
phi
::
DDim
stride_dout
=
dim_dout
.
size
()
==
0
?
phi
::
DDim
(
dim_dout
)
:
phi
::
stride
(
dim_dout
);
const
phi
::
DDim
stride_x
=
phi
::
stride
(
dim_x
);
const
int64_t
*
p_stride_x
=
nullptr
;
const
phi
::
DDim
stride_y
=
phi
::
stride
(
dim_y
);
const
int64_t
*
p_stride_y
=
nullptr
;
const
phi
::
DDim
stride_dout
=
phi
::
stride
(
dim_dout
);
const
int64_t
*
p_stride_dout
=
nullptr
;
const
int64_t
*
p_shape_y
=
nullptr
;
const
int64_t
*
p_stride_x
=
nullptr
;
const
int64_t
*
p_stride_y
=
nullptr
;
const
int64_t
*
p_stride_dout
=
nullptr
;
const
int64_t
*
p_shape_y
=
nullptr
;
#if defined(__NVCC__) || defined(__HIPCC__)
#if defined(__NVCC__) || defined(__HIPCC__)
thrust
::
device_vector
<
int64_t
>
d_stride_x
(
ndims
);
thrust
::
device_vector
<
int64_t
>
d_stride_x
(
ndims
);
thrust
::
device_vector
<
int64_t
>
d_stride_y
(
ndims
);
thrust
::
device_vector
<
int64_t
>
d_stride_y
(
ndims
);
...
@@ -199,14 +201,14 @@ struct KronGradOpFunctor {
...
@@ -199,14 +201,14 @@ struct KronGradOpFunctor {
// dout_x: dout * kron(ones(X), Y) re-aranged in shape (numel_x, numel_y)
// dout_x: dout * kron(ones(X), Y) re-aranged in shape (numel_x, numel_y)
// dout_y: dout * kron(X, ones(Y)) re-aranged in shaoe (numel_y, numel_x)
// dout_y: dout * kron(X, ones(Y)) re-aranged in shaoe (numel_y, numel_x)
DenseTensor
dout_x
;
DenseTensor
dout_x
;
T
*
p_dout_x
=
nullptr
;
T
*
p_dout_x
=
nullptr
;
if
(
dx
)
{
if
(
dx
)
{
dout_x
.
Resize
({
numel_x
,
numel_y
});
dout_x
.
Resize
({
numel_x
,
numel_y
});
dev_ctx
.
template
Alloc
<
T
>(
&
dout_x
);
dev_ctx
.
template
Alloc
<
T
>(
&
dout_x
);
p_dout_x
=
dout_x
.
data
<
T
>
();
p_dout_x
=
dout_x
.
data
<
T
>
();
}
}
DenseTensor
dout_y
;
DenseTensor
dout_y
;
T
*
p_dout_y
=
nullptr
;
T
*
p_dout_y
=
nullptr
;
if
(
dy
)
{
if
(
dy
)
{
dout_y
.
Resize
({
numel_y
,
numel_x
});
dout_y
.
Resize
({
numel_y
,
numel_x
});
dev_ctx
.
template
Alloc
<
T
>(
&
dout_y
);
dev_ctx
.
template
Alloc
<
T
>(
&
dout_y
);
...
@@ -240,7 +242,7 @@ struct KronGradOpFunctor {
...
@@ -240,7 +242,7 @@ struct KronGradOpFunctor {
dev_ctx
,
dout_y
,
dy
,
kps
::
IdentityFunctor
<
T
>
(),
{
1
});
dev_ctx
,
dout_y
,
dy
,
kps
::
IdentityFunctor
<
T
>
(),
{
1
});
}
}
#else
#else
auto
*
place
=
dev_ctx
.
eigen_device
();
auto
*
place
=
dev_ctx
.
eigen_device
();
Eigen
::
array
<
int
,
1
>
reduce_dim
=
{
1
};
Eigen
::
array
<
int
,
1
>
reduce_dim
=
{
1
};
if
(
dx
)
{
if
(
dx
)
{
auto
eigen_dout_x
=
EigenMatrix
<
T
>::
Reshape
(
dout_x
,
1
);
auto
eigen_dout_x
=
EigenMatrix
<
T
>::
Reshape
(
dout_x
,
1
);
...
@@ -257,12 +259,12 @@ struct KronGradOpFunctor {
...
@@ -257,12 +259,12 @@ struct KronGradOpFunctor {
};
};
template
<
typename
T
,
typename
Context
>
template
<
typename
T
,
typename
Context
>
void
KronGradKernel
(
const
Context
&
ctx
,
void
KronGradKernel
(
const
Context
&
ctx
,
const
DenseTensor
&
x
,
const
DenseTensor
&
x
,
const
DenseTensor
&
y
,
const
DenseTensor
&
y
,
const
DenseTensor
&
out_grad
,
const
DenseTensor
&
out_grad
,
DenseTensor
*
x_grad
,
DenseTensor
*
x_grad
,
DenseTensor
*
y_grad
)
{
DenseTensor
*
y_grad
)
{
if
(
x_grad
)
{
if
(
x_grad
)
{
ctx
.
template
Alloc
<
T
>(
x_grad
);
ctx
.
template
Alloc
<
T
>(
x_grad
);
}
}
...
@@ -274,8 +276,8 @@ void KronGradKernel(const Context& ctx,
...
@@ -274,8 +276,8 @@ void KronGradKernel(const Context& ctx,
DenseTensor
xx
=
UnsqueezeTo
(
x
,
ndims
);
DenseTensor
xx
=
UnsqueezeTo
(
x
,
ndims
);
DenseTensor
yy
=
UnsqueezeTo
(
y
,
ndims
);
DenseTensor
yy
=
UnsqueezeTo
(
y
,
ndims
);
DenseTensor
*
pdxx
=
nullptr
;
DenseTensor
*
pdxx
=
nullptr
;
DenseTensor
*
pdyy
=
nullptr
;
DenseTensor
*
pdyy
=
nullptr
;
DenseTensor
dxx
;
DenseTensor
dxx
;
DenseTensor
dyy
;
DenseTensor
dyy
;
if
(
x_grad
)
{
if
(
x_grad
)
{
...
...
paddle/phi/kernels/impl/kron_kernel_impl.h
浏览文件 @
ae622479
...
@@ -27,8 +27,8 @@
...
@@ -27,8 +27,8 @@
namespace
phi
{
namespace
phi
{
inline
DenseTensor
UnsqueezeTo
(
const
DenseTensor
&
src
,
int
ndims
)
{
inline
DenseTensor
UnsqueezeTo
(
const
DenseTensor
&
src
,
int
ndims
)
{
const
phi
::
DDim
&
shape
=
src
.
dims
();
const
phi
::
DDim
&
shape
=
src
.
dims
();
int
rank
=
shape
.
size
();
int
rank
=
shape
.
size
();
DenseTensor
res
;
DenseTensor
res
;
res
.
ShareDataWith
(
src
);
res
.
ShareDataWith
(
src
);
...
@@ -52,13 +52,13 @@ inline DenseTensor UnsqueezeTo(const DenseTensor& src, int ndims) {
...
@@ -52,13 +52,13 @@ inline DenseTensor UnsqueezeTo(const DenseTensor& src, int ndims) {
template
<
typename
T
>
template
<
typename
T
>
struct
KronElemFunctor
{
struct
KronElemFunctor
{
KronElemFunctor
(
const
T
*
a
,
KronElemFunctor
(
const
T
*
a
,
const
T
*
b
,
const
T
*
b
,
T
*
out
,
T
*
out
,
const
int64_t
*
shape_b
,
const
int64_t
*
shape_b
,
const
int64_t
*
stride_a
,
const
int64_t
*
stride_a
,
const
int64_t
*
stride_b
,
const
int64_t
*
stride_b
,
const
int64_t
*
stride_out
,
const
int64_t
*
stride_out
,
int
ndims
)
int
ndims
)
:
a_
(
a
),
:
a_
(
a
),
b_
(
b
),
b_
(
b
),
...
@@ -86,31 +86,34 @@ struct KronElemFunctor {
...
@@ -86,31 +86,34 @@ struct KronElemFunctor {
}
}
private:
private:
const
T
*
a_
;
const
T
*
a_
;
const
T
*
b_
;
const
T
*
b_
;
T
*
out_
;
T
*
out_
;
const
int64_t
*
shape_b_
;
const
int64_t
*
shape_b_
;
const
int64_t
*
stride_a_
;
const
int64_t
*
stride_a_
;
const
int64_t
*
stride_b_
;
const
int64_t
*
stride_b_
;
const
int64_t
*
stride_out_
;
const
int64_t
*
stride_out_
;
const
int
ndims_
;
const
int
ndims_
;
};
};
template
<
typename
Context
,
typename
T
>
template
<
typename
Context
,
typename
T
>
struct
KronOpFunctor
{
struct
KronOpFunctor
{
void
operator
()(
const
Context
&
dev_ctx
,
void
operator
()(
const
Context
&
dev_ctx
,
const
DenseTensor
&
x
,
const
DenseTensor
&
x
,
const
DenseTensor
&
y
,
const
DenseTensor
&
y
,
DenseTensor
*
out
)
{
DenseTensor
*
out
)
{
int
ndims
=
out
->
dims
().
size
();
int
ndims
=
out
->
dims
().
size
();
int64_t
numel
=
out
->
numel
();
int64_t
numel
=
out
->
numel
();
const
phi
::
DDim
&
dim_x
=
x
.
dims
();
const
phi
::
DDim
&
dim_x
=
x
.
dims
();
const
phi
::
DDim
&
dim_y
=
y
.
dims
();
const
phi
::
DDim
&
dim_y
=
y
.
dims
();
const
phi
::
DDim
&
dim_out
=
out
->
dims
();
const
phi
::
DDim
&
dim_out
=
out
->
dims
();
const
phi
::
DDim
stride_x
=
phi
::
stride
(
dim_x
);
const
phi
::
DDim
stride_x
=
const
phi
::
DDim
stride_y
=
phi
::
stride
(
dim_y
);
dim_x
.
size
()
==
0
?
phi
::
DDim
(
dim_x
)
:
phi
::
stride
(
dim_x
);
const
phi
::
DDim
stride_out
=
phi
::
stride
(
dim_out
);
const
phi
::
DDim
stride_y
=
dim_y
.
size
()
==
0
?
phi
::
DDim
(
dim_y
)
:
phi
::
stride
(
dim_y
);
const
phi
::
DDim
stride_out
=
dim_out
.
size
()
==
0
?
phi
::
DDim
(
dim_out
)
:
phi
::
stride
(
dim_out
);
const
int64_t
*
p_stride_x
=
nullptr
,
*
p_stride_y
=
nullptr
,
const
int64_t
*
p_stride_x
=
nullptr
,
*
p_stride_y
=
nullptr
,
*
p_stride_out
=
nullptr
,
*
p_shape_y
=
nullptr
;
*
p_stride_out
=
nullptr
,
*
p_shape_y
=
nullptr
;
...
@@ -150,10 +153,10 @@ struct KronOpFunctor {
...
@@ -150,10 +153,10 @@ struct KronOpFunctor {
};
};
template
<
typename
T
,
typename
Context
>
template
<
typename
T
,
typename
Context
>
void
KronKernel
(
const
Context
&
ctx
,
void
KronKernel
(
const
Context
&
ctx
,
const
DenseTensor
&
x
,
const
DenseTensor
&
x
,
const
DenseTensor
&
y
,
const
DenseTensor
&
y
,
DenseTensor
*
out
)
{
DenseTensor
*
out
)
{
ctx
.
template
Alloc
<
T
>(
out
);
ctx
.
template
Alloc
<
T
>(
out
);
int
ndims
=
out
->
dims
().
size
();
int
ndims
=
out
->
dims
().
size
();
...
...
python/paddle/fluid/tests/unittests/test_zero_dim_tensor.py
浏览文件 @
ae622479
...
@@ -293,6 +293,7 @@ binary_api_list = [
...
@@ -293,6 +293,7 @@ binary_api_list = [
paddle
.
fmax
,
paddle
.
fmax
,
paddle
.
fmin
,
paddle
.
fmin
,
paddle
.
complex
,
paddle
.
complex
,
paddle
.
kron
,
]
]
binary_int_api_list
=
[
binary_int_api_list
=
[
...
...
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