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
0a15b0db
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
0a15b0db
编写于
8月 31, 2023
作者:
Y
yuchen202
提交者:
GitHub
8月 31, 2023
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
[xdoctest] reformat example code with google style in No.36-43 (#56440)
上级
71e28b12
变更
8
隐藏空白更改
内联
并排
Showing
8 changed file
with
843 addition
and
650 deletion
+843
-650
python/paddle/nn/functional/activation.py
python/paddle/nn/functional/activation.py
+290
-226
python/paddle/nn/functional/common.py
python/paddle/nn/functional/common.py
+331
-233
python/paddle/nn/functional/conv.py
python/paddle/nn/functional/conv.py
+59
-59
python/paddle/nn/functional/distance.py
python/paddle/nn/functional/distance.py
+7
-8
python/paddle/nn/functional/extension.py
python/paddle/nn/functional/extension.py
+67
-59
python/paddle/nn/functional/flash_attention.py
python/paddle/nn/functional/flash_attention.py
+9
-11
python/paddle/nn/initializer/uniform.py
python/paddle/nn/initializer/uniform.py
+27
-18
python/paddle/nn/initializer/xavier.py
python/paddle/nn/initializer/xavier.py
+53
-36
未找到文件。
python/paddle/nn/functional/activation.py
浏览文件 @
0a15b0db
...
@@ -49,12 +49,15 @@ def celu(x, alpha=1.0, name=None):
...
@@ -49,12 +49,15 @@ def celu(x, alpha=1.0, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
x = paddle.to_tensor([[-1., 6.], [1., 15.6]])
out = F.celu(x, alpha=0.2)
>>> x = paddle.to_tensor([[-1., 6.], [1., 15.6]])
# [[-0.19865242, 6. ],
>>> out = F.celu(x, alpha=0.2)
# [ 1. , 15.60000038]]
>>> print(out)
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0.19865242, 6. ],
[ 1. , 15.60000038]])
"""
"""
if
alpha
==
0
:
if
alpha
==
0
:
raise
ZeroDivisionError
(
"alpha cannot be 0 for celu"
)
raise
ZeroDivisionError
(
"alpha cannot be 0 for celu"
)
...
@@ -100,13 +103,15 @@ def elu(x, alpha=1.0, name=None):
...
@@ -100,13 +103,15 @@ def elu(x, alpha=1.0, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([[-1., 6.], [1., 15.6]])
>>> x = paddle.to_tensor([[-1., 6.], [1., 15.6]])
out = F.elu(x, alpha=0.2)
>>> out = F.elu(x, alpha=0.2)
# [[-0.12642411 6. ]
>>> print(out)
# [ 1. 15.6 ]]
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0.12642412, 6. ],
[ 1. , 15.60000038]])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -168,16 +173,20 @@ def gelu(x, approximate=False, name=None):
...
@@ -168,16 +173,20 @@ def gelu(x, approximate=False, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
x = paddle.to_tensor([[-1, 0.5], [1, 1.5]])
>>> x = paddle.to_tensor([[-1, 0.5], [1, 1.5]])
out1 = F.gelu(x)
>>> out1 = F.gelu(x)
# [[-0.15865529, 0.34573123],
>>> print(out1)
# [ 0.84134471, 1.39978933]]
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
out2 = F.gelu(x, True)
[[-0.15865529, 0.34573123],
# [[-0.15880799, 0.34571400],
[ 0.84134471, 1.39978933]])
# [ 0.84119201, 1.39957154]]
>>> out2 = F.gelu(x, True)
>>> print(out2)
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0.15880796, 0.34571400],
[ 0.84119201, 1.39957154]])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -223,11 +232,15 @@ def hardshrink(x, threshold=0.5, name=None):
...
@@ -223,11 +232,15 @@ def hardshrink(x, threshold=0.5, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
>>> x = paddle.to_tensor([-1, 0.3, 2.5])
>>> out = F.hardshrink(x)
>>> print(out)
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[-1. , 0. , 2.50000000])
x = paddle.to_tensor([-1, 0.3, 2.5])
out = F.hardshrink(x) # [-1., 0., 2.5]
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -274,11 +287,14 @@ def hardtanh(x, min=-1.0, max=1.0, name=None):
...
@@ -274,11 +287,14 @@ def hardtanh(x, min=-1.0, max=1.0, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-1.5, 0.3, 2.5])
>>> x = paddle.to_tensor([-1.5, 0.3, 2.5])
out = F.hardtanh(x) # [-1., 0.3, 1.]
>>> out = F.hardtanh(x)
>>> print(out)
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[-1. , 0.30000001, 1. ])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -338,11 +354,14 @@ def hardsigmoid(x, slope=0.1666667, offset=0.5, name=None):
...
@@ -338,11 +354,14 @@ def hardsigmoid(x, slope=0.1666667, offset=0.5, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-4., 5., 1.])
>>> x = paddle.to_tensor([-4., 5., 1.])
out = F.hardsigmoid(x) # [0., 1., 0.666667]
>>> out = F.hardsigmoid(x)
>>> print(out)
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[0. , 1. , 0.66666669])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -390,11 +409,14 @@ def hardswish(x, name=None):
...
@@ -390,11 +409,14 @@ def hardswish(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-4., 5., 1.])
>>> x = paddle.to_tensor([-4., 5., 1.])
out = F.hardswish(x) # [0., 5., 0.666667]
>>> out = F.hardswish(x)
>>> print(out)
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[-0. , 5. , 0.66666669])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
return
_C_ops
.
hardswish
(
x
)
return
_C_ops
.
hardswish
(
x
)
...
@@ -442,13 +464,14 @@ def leaky_relu(x, negative_slope=0.01, name=None):
...
@@ -442,13 +464,14 @@ def leaky_relu(x, negative_slope=0.01, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-2., 0., 1.])
>>> x = paddle.to_tensor([-2., 0., 1.])
out = F.leaky_relu(x)
>>> out = F.leaky_relu(x)
print(out)
>>> print(out)
# [-0.02, 0., 1.]
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[-0.02000000, 0. , 1. ])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -502,25 +525,26 @@ def prelu(x, weight, data_format="NCHW", name=None):
...
@@ -502,25 +525,26 @@ def prelu(x, weight, data_format="NCHW", name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
data = paddle.to_tensor([[[[-2.0, 3.0, -4.0, 5.0],
>>> data = paddle.to_tensor([[[[-2.0, 3.0, -4.0, 5.0],
[ 3.0, -4.0, 5.0, -6.0],
... [ 3.0, -4.0, 5.0, -6.0],
[-7.0, -8.0, 8.0, 9.0]],
... [-7.0, -8.0, 8.0, 9.0]],
[[ 1.0, -2.0, -3.0, 4.0],
... [[ 1.0, -2.0, -3.0, 4.0],
[-5.0, 6.0, 7.0, -8.0],
... [-5.0, 6.0, 7.0, -8.0],
[ 6.0, 7.0, 8.0, 9.0]]]], dtype='float32')
... [ 6.0, 7.0, 8.0, 9.0]]]], dtype='float32')
w = paddle.to_tensor([0.25], dtype='float32')
>>> w = paddle.to_tensor([0.25], dtype='float32')
out = F.prelu(data, w)
>>> out = F.prelu(data, w)
print(out)
>>> print(out)
# [[[[-0.5 , 3. , -1. , 5. ],
Tensor(shape=[1, 2, 3, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [ 3. , -1. , 5. , -1.5 ],
[[[[-0.50000000, 3. , -1. , 5. ],
# [-1.75, -2. , 8. , 9. ]],
[ 3. , -1. , 5. , -1.50000000],
# [[ 1. , -0.5 , -0.75, 4. ],
[-1.75000000, -2. , 8. , 9. ]],
# [-1.25, 6. , 7. , -2. ],
[[ 1. , -0.50000000, -0.75000000, 4. ],
# [ 6. , 7. , 8. , 9. ]]]]
[-1.25000000, 6. , 7. , -2. ],
[ 6. , 7. , 8. , 9. ]]]])
"""
"""
assert
(
assert
(
len
(
weight
.
shape
)
==
0
or
len
(
weight
.
shape
)
==
1
len
(
weight
.
shape
)
==
0
or
len
(
weight
.
shape
)
==
1
...
@@ -634,24 +658,24 @@ def rrelu(x, lower=1.0 / 8.0, upper=1.0 / 3.0, training=True, name=None):
...
@@ -634,24 +658,24 @@ def rrelu(x, lower=1.0 / 8.0, upper=1.0 / 3.0, training=True, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
>>> paddle.seed(1)
input_tensor = paddle.to_tensor([[[[-2.0, 3.0, -4.0, 5.0],
>>>
input_tensor = paddle.to_tensor([[[[-2.0, 3.0, -4.0, 5.0],
[ 3.0, -4.0, 5.0, -6.0],
...
[ 3.0, -4.0, 5.0, -6.0],
[-7.0, -8.0, 8.0, 9.0]],
...
[-7.0, -8.0, 8.0, 9.0]],
[[ 1.0, -2.0, -3.0, 4.0],
...
[[ 1.0, -2.0, -3.0, 4.0],
[-5.0, 6.0, 7.0, -8.0],
...
[-5.0, 6.0, 7.0, -8.0],
[ 6.0, 7.0, 8.0, 9.0]]]], dtype='float32')
...
[ 6.0, 7.0, 8.0, 9.0]]]], dtype='float32')
>>> out = F.rrelu(input_tensor, 0.1, 0.3)
out = F.rrelu(input_tensor, 0.1, 0.3
)
>>> print(out
)
print(out)
Tensor(shape=[1, 2, 3, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
#[[[[-0.20000899 3. -0.8810822 5. ]
[[[[-0.20715050, 3. , -1.01193857, 5. ],
# [ 3. -0.55175185 5. -1.0776101 ]
[ 3. , -0.94084597, 5. , -0.65544695],
# [-1.0680687 -1.9896201 8. 9. ]]
[-1.24268556, -2.34339547, 8. , 9. ]],
# [[ 1. -0.5238267 -0.65515125 4. ]
[[ 1. , -0.44942653, -0.68969047, 4. ],
# [-1.3766339 6. 7. -2.3465784 ]
[-1.03736508, 6. , 7. , -0.95799232],
# [ 6. 7. 8. 9. ]]]]
[ 6. , 7. , 8. , 9. ]]]])
"""
"""
if
not
isinstance
(
lower
,
float
)
or
not
isinstance
(
upper
,
float
):
if
not
isinstance
(
lower
,
float
)
or
not
isinstance
(
upper
,
float
):
raise
TypeError
(
raise
TypeError
(
...
@@ -722,13 +746,14 @@ def relu(x, name=None):
...
@@ -722,13 +746,14 @@ def relu(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-2, 0, 1], dtype='float32')
>>> x = paddle.to_tensor([-2, 0, 1], dtype='float32')
out = F.relu(x)
>>> out = F.relu(x)
print(out)
>>> print(out)
# [0., 0., 1.]
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[0., 0., 1.])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -770,11 +795,14 @@ def log_sigmoid(x, name=None):
...
@@ -770,11 +795,14 @@ def log_sigmoid(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([1.0, 2.0, 3.0, 4.0])
>>> x = paddle.to_tensor([1.0, 2.0, 3.0, 4.0])
out = F.log_sigmoid(x) # [-0.313262 -0.126928 -0.0485874 -0.0181499]
>>> out = F.log_sigmoid(x)
>>> print(out)
Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
[-0.31326166, -0.12692805, -0.04858733, -0.01814996])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -830,20 +858,25 @@ def maxout(x, groups, axis=1, name=None):
...
@@ -830,20 +858,25 @@ def maxout(x, groups, axis=1, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
x = paddle.rand([1, 2, 3, 4])
>>> paddle.seed(2023)
# [[[[0.5002636 0.22272532 0.17402348 0.2874594 ]
>>> x = paddle.rand([1, 2, 3, 4])
# [0.95313174 0.6228939 0.7129065 0.7087491 ]
>>> print(x)
# [0.02879342 0.88725346 0.61093384 0.38833922]]
Tensor(shape=[1, 2, 3, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[0.5231306 0.03807496 0.91661984 0.15602879]
[[[[0.86583614, 0.52014720, 0.25960937, 0.90525323],
# [0.666127 0.616567 0.30741522 0.24044901]
[0.42400089, 0.40641287, 0.97020894, 0.74437362],
# [0.7142536 0.7351477 0.31588817 0.23782359]]]]
[0.51785129, 0.73292869, 0.97786582, 0.04315904]],
out = F.maxout(x, groups=2)
[[0.42639419, 0.71958369, 0.20811461, 0.19731510],
# [[[[0.5231306 0.22272532 0.91661984 0.2874594 ]
[0.38424349, 0.14603184, 0.22713774, 0.44607511],
# [0.95313174 0.6228939 0.7129065 0.7087491 ]
[0.21657862, 0.67685395, 0.46460176, 0.92382854]]]])
# [0.7142536 0.88725346 0.61093384 0.38833922]]]]
>>> out = F.maxout(x, groups=2)
>>> print(out)
Tensor(shape=[1, 1, 3, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
[[[[0.86583614, 0.71958369, 0.25960937, 0.90525323],
[0.42400089, 0.40641287, 0.97020894, 0.74437362],
[0.51785129, 0.73292869, 0.97786582, 0.92382854]]]])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
return
_C_ops
.
maxout
(
x
,
groups
,
axis
)
return
_C_ops
.
maxout
(
x
,
groups
,
axis
)
...
@@ -888,13 +921,14 @@ def relu6(x, name=None):
...
@@ -888,13 +921,14 @@ def relu6(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-1, 0.3, 6.5])
>>> x = paddle.to_tensor([-1, 0.3, 6.5])
out = F.relu6(x)
>>> out = F.relu6(x)
print(out)
>>> print(out)
# [0, 0.3, 6]
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[0. , 0.30000001, 6. ])
"""
"""
threshold
=
6.0
threshold
=
6.0
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -945,13 +979,15 @@ def selu(
...
@@ -945,13 +979,15 @@ def selu(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([[0.0, 1.0],[2.0, 3.0]])
>>> x = paddle.to_tensor([[0.0, 1.0],[2.0, 3.0]])
out = F.selu(x)
>>> out = F.selu(x)
print(out)
>>> print(out)
# [[0, 1.050701],[2.101402, 3.152103]]
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
[[0. , 1.05070102],
[2.10140204, 3.15210295]])
"""
"""
if
scale
<=
1.0
:
if
scale
<=
1.0
:
raise
ValueError
(
raise
ValueError
(
...
@@ -1000,11 +1036,14 @@ def silu(x, name=None):
...
@@ -1000,11 +1036,14 @@ def silu(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([1.0, 2.0, 3.0, 4.0])
>>> x = paddle.to_tensor([1.0, 2.0, 3.0, 4.0])
out = F.silu(x) # [ 0.731059, 1.761594, 2.857722, 3.928055 ]
>>> out = F.silu(x)
>>> print(out)
Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
[0.73105860, 1.76159406, 2.85772228, 3.92805505])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -1111,25 +1150,35 @@ def softmax(x, axis=-1, dtype=None, name=None):
...
@@ -1111,25 +1150,35 @@ def softmax(x, axis=-1, dtype=None, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
x = paddle.to_tensor([[[2.0, 3.0, 4.0, 5.0],
>>> x = paddle.to_tensor([[[2.0, 3.0, 4.0, 5.0],
[3.0, 4.0, 5.0, 6.0],
... [3.0, 4.0, 5.0, 6.0],
[7.0, 8.0, 8.0, 9.0]],
... [7.0, 8.0, 8.0, 9.0]],
[[1.0, 2.0, 3.0, 4.0],
... [[1.0, 2.0, 3.0, 4.0],
[5.0, 6.0, 7.0, 8.0],
... [5.0, 6.0, 7.0, 8.0],
[6.0, 7.0, 8.0, 9.0]]],dtype='float32')
... [6.0, 7.0, 8.0, 9.0]]],dtype='float32')
out1 = F.softmax(x)
>>> out1 = F.softmax(x)
out2 = F.softmax(x, dtype='float64')
>>> out2 = F.softmax(x, dtype='float64')
# out1's data type is float32; out2's data type is float64
>>> #out1's data type is float32; out2's data type is float64
# out1 and out2's value is as follows:
>>> #out1 and out2's value is as follows:
# [[[0.0320586 , 0.08714432, 0.23688282, 0.64391426],
>>> print(out1)
# [0.0320586 , 0.08714432, 0.23688282, 0.64391426],
>>> print(out2)
# [0.07232949, 0.19661193, 0.19661193, 0.53444665]],
Tensor(shape=[2, 3, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[0.0320586 , 0.08714432, 0.23688282, 0.64391426],
[[[0.03205860, 0.08714432, 0.23688284, 0.64391428],
# [0.0320586 , 0.08714432, 0.23688282, 0.64391426],
[0.03205860, 0.08714432, 0.23688284, 0.64391428],
# [0.0320586 , 0.08714432, 0.23688282, 0.64391426]]]
[0.07232949, 0.19661194, 0.19661194, 0.53444666]],
[[0.03205860, 0.08714432, 0.23688284, 0.64391428],
[0.03205860, 0.08714432, 0.23688284, 0.64391428],
[0.03205860, 0.08714432, 0.23688284, 0.64391428]]])
Tensor(shape=[2, 3, 4], dtype=float64, place=Place(cpu), stop_gradient=True,
[[[0.03205860, 0.08714432, 0.23688282, 0.64391426],
[0.03205860, 0.08714432, 0.23688282, 0.64391426],
[0.07232949, 0.19661193, 0.19661193, 0.53444665]],
[[0.03205860, 0.08714432, 0.23688282, 0.64391426],
[0.03205860, 0.08714432, 0.23688282, 0.64391426],
[0.03205860, 0.08714432, 0.23688282, 0.64391426]]])
"""
"""
if
(
dtype
is
not
None
)
and
(
not
isinstance
(
dtype
,
core
.
VarDesc
.
VarType
)):
if
(
dtype
is
not
None
)
and
(
not
isinstance
(
dtype
,
core
.
VarDesc
.
VarType
)):
...
@@ -1214,11 +1263,14 @@ def softplus(x, beta=1, threshold=20, name=None):
...
@@ -1214,11 +1263,14 @@ def softplus(x, beta=1, threshold=20, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-0.4, -0.2, 0.1, 0.3], dtype='float32')
>>> x = paddle.to_tensor([-0.4, -0.2, 0.1, 0.3], dtype='float32')
out = F.softplus(x) # [0.513015, 0.598139, 0.744397, 0.854355]
>>> out = F.softplus(x)
>>> print(out)
Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
[0.51301527, 0.59813893, 0.74439669, 0.85435522])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -1264,14 +1316,14 @@ def softshrink(x, threshold=0.5, name=None):
...
@@ -1264,14 +1316,14 @@ def softshrink(x, threshold=0.5, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-0.9, -0.2, 0.1, 0.8])
>>>
x = paddle.to_tensor([-0.9, -0.2, 0.1, 0.8])
out = F.softshrink(x)
>>>
out = F.softshrink(x)
print(out)
>>>
print(out)
# Tensor(shape=[4], dtype=float32, place=Place(gpu:0
), stop_gradient=True,
Tensor(shape=[4], dtype=float32, place=Place(cpu
), stop_gradient=True,
#
[-0.39999998, 0. , 0. , 0.30000001])
[-0.39999998, 0. , 0. , 0.30000001])
"""
"""
if
threshold
<
0
:
if
threshold
<
0
:
raise
ValueError
(
raise
ValueError
(
...
@@ -1315,14 +1367,14 @@ def softsign(x, name=None):
...
@@ -1315,14 +1367,14 @@ def softsign(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-0.4, -0.2, 0.1, 0.3])
>>>
x = paddle.to_tensor([-0.4, -0.2, 0.1, 0.3])
out = F.softsign(x)
>>>
out = F.softsign(x)
print(out)
>>>
print(out)
# Tensor(shape=[4], dtype=float32, place=Place(gpu:0
), stop_gradient=True,
Tensor(shape=[4], dtype=float32, place=Place(cpu
), stop_gradient=True,
#
[-0.28571430, -0.16666666, 0.09090909, 0.23076925])
[-0.28571430, -0.16666666, 0.09090909, 0.23076925])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
return
_C_ops
.
softsign
(
x
)
return
_C_ops
.
softsign
(
x
)
...
@@ -1354,14 +1406,14 @@ def swish(x, name=None):
...
@@ -1354,14 +1406,14 @@ def swish(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-2., 0., 1.])
>>>
x = paddle.to_tensor([-2., 0., 1.])
out = F.swish(x)
>>>
out = F.swish(x)
print(out)
>>>
print(out)
# Tensor(shape=[3], dtype=float32, place=Place(gpu:0
), stop_gradient=True,
Tensor(shape=[3], dtype=float32, place=Place(cpu
), stop_gradient=True,
# [-0.23840584, 0. , 0.73105854
])
[-0.23840584, 0. , 0.73105860
])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
return
_C_ops
.
swish
(
x
)
return
_C_ops
.
swish
(
x
)
...
@@ -1403,11 +1455,14 @@ def mish(x, name=None):
...
@@ -1403,11 +1455,14 @@ def mish(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-5., 0., 5.])
>>> x = paddle.to_tensor([-5., 0., 5.])
out = F.mish(x) # [-0.03357624, 0., 4.99955208]
>>> out = F.mish(x)
>>> print(out)
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[-0.03357624, 0. , 4.99955177])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
return
_C_ops
.
mish
(
x
,
20
)
return
_C_ops
.
mish
(
x
,
20
)
...
@@ -1439,14 +1494,14 @@ def tanhshrink(x, name=None):
...
@@ -1439,14 +1494,14 @@ def tanhshrink(x, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([-0.4, -0.2, 0.1, 0.3])
>>>
x = paddle.to_tensor([-0.4, -0.2, 0.1, 0.3])
out = F.tanhshrink(x)
>>>
out = F.tanhshrink(x)
print(out)
>>>
print(out)
# Tensor(shape=[4], dtype=float32, place=Place(gpu:0
), stop_gradient=True,
Tensor(shape=[4], dtype=float32, place=Place(cpu
), stop_gradient=True,
# [-0.02005106, -0.00262468, 0.00033200
, 0.00868741])
[-0.02005100, -0.00262472, 0.00033201
, 0.00868741])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
return
_C_ops
.
tanh_shrink
(
x
)
return
_C_ops
.
tanh_shrink
(
x
)
...
@@ -1488,14 +1543,14 @@ def thresholded_relu(x, threshold=1.0, name=None):
...
@@ -1488,14 +1543,14 @@ def thresholded_relu(x, threshold=1.0, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([2., 0., 1.])
>>>
x = paddle.to_tensor([2., 0., 1.])
out = F.thresholded_relu(x)
>>>
out = F.thresholded_relu(x)
print(out)
>>>
print(out)
# Tensor(shape=[3], dtype=float32, place=Place(gpu:0
), stop_gradient=True,
Tensor(shape=[3], dtype=float32, place=Place(cpu
), stop_gradient=True,
#
[2., 0., 0.])
[2., 0., 0.])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -1561,26 +1616,35 @@ def log_softmax(x, axis=-1, dtype=None, name=None):
...
@@ -1561,26 +1616,35 @@ def log_softmax(x, axis=-1, dtype=None, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
>>> x = [[[-2.0, 3.0, -4.0, 5.0],
x = [[[-2.0, 3.0, -4.0, 5.0],
... [3.0, -4.0, 5.0, -6.0],
[3.0, -4.0, 5.0, -6.0],
... [-7.0, -8.0, 8.0, 9.0]],
[-7.0, -8.0, 8.0, 9.0]],
... [[1.0, -2.0, -3.0, 4.0],
[[1.0, -2.0, -3.0, 4.0],
... [-5.0, 6.0, 7.0, -8.0],
[-5.0, 6.0, 7.0, -8.0],
... [6.0, 7.0, 8.0, 9.0]]]
[6.0, 7.0, 8.0, 9.0]]]
>>> x = paddle.to_tensor(x)
x = paddle.to_tensor(x)
>>> out1 = F.log_softmax(x)
out1 = F.log_softmax(x)
>>> out2 = F.log_softmax(x, dtype='float64')
out2 = F.log_softmax(x, dtype='float64')
>>> #out1's data type is float32; out2's data type is float64
# out1's data type is float32; out2's data type is float64
>>> #out1 and out2's value is as follows:
# out1 and out2's value is as follows:
>>> print(out1)
# [[[ -7.1278396 -2.1278396 -9.127839 -0.12783948]
Tensor(shape=[2, 3, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [ -2.1270514 -9.127051 -0.12705144 -11.127051 ]
[[[-7.12783957 , -2.12783957 , -9.12783909 , -0.12783945 ],
# [-16.313261 -17.313261 -1.3132617 -0.31326184]]
[-2.12705135 , -9.12705135 , -0.12705141 , -11.12705135],
# [[ -3.0518122 -6.051812 -7.051812 -0.051812 ]
[-16.31326103, -17.31326103, -1.31326187 , -0.31326184 ]],
# [-12.313267 -1.3132664 -0.3132665 -15.313267 ]
[[-3.05181193 , -6.05181217 , -7.05181217 , -0.05181199 ],
# [ -3.4401896 -2.4401896 -1.4401896 -0.44018966]]]
[-12.31326675, -1.31326652 , -0.31326646 , -15.31326675],
[-3.44018984 , -2.44018984 , -1.44018972 , -0.44018975 ]]])
>>> print(out2)
Tensor(shape=[2, 3, 4], dtype=float64, place=Place(cpu), stop_gradient=True,
[[[-7.12783948 , -2.12783948 , -9.12783948 , -0.12783948 ],
[-2.12705141 , -9.12705141 , -0.12705141 , -11.12705141],
[-16.31326180, -17.31326180, -1.31326180 , -0.31326180 ]],
[[-3.05181198 , -6.05181198 , -7.05181198 , -0.05181198 ],
[-12.31326640, -1.31326640 , -0.31326640 , -15.31326640],
[-3.44018970 , -2.44018970 , -1.44018970 , -0.44018970 ]]])
"""
"""
if
(
dtype
is
not
None
)
and
(
not
isinstance
(
dtype
,
core
.
VarDesc
.
VarType
)):
if
(
dtype
is
not
None
)
and
(
not
isinstance
(
dtype
,
core
.
VarDesc
.
VarType
)):
...
@@ -1655,17 +1719,16 @@ def glu(x, axis=-1, name=None):
...
@@ -1655,17 +1719,16 @@ def glu(x, axis=-1, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
from paddle.nn import functional as F
>>> from paddle.nn import functional as F
>>> x = paddle.to_tensor(
x = paddle.to_tensor(
... [[-0.22014759, -1.76358426, 0.80566144, 0.04241343],
[[-0.22014759, -1.76358426, 0.80566144, 0.04241343],
... [-1.94900405, -1.89956081, 0.17134808, -1.11280477]]
[-1.94900405, -1.89956081, 0.17134808, -1.11280477]]
... )
)
>>> print(F.glu(x))
print(F.glu(x))
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
# Tensor(shape=[2, 2], dtype=float32, place=Place(gpu:0), stop_gradient=True,
[[-0.15216254, -0.90048921],
# [[-0.15216254, -0.90048921],
[-1.05778778, -0.46985325]])
# [-1.05778778, -0.46985325]])
"""
"""
check_variable_and_dtype
(
check_variable_and_dtype
(
...
@@ -1727,18 +1790,19 @@ def gumbel_softmax(x, temperature=1.0, hard=False, axis=-1, name=None):
...
@@ -1727,18 +1790,19 @@ def gumbel_softmax(x, temperature=1.0, hard=False, axis=-1, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
logits = paddle.randn([4, 6])
>>> paddle.seed(2023)
temperature = 0.01
>>> logits = paddle.randn([4, 6])
gumbel_softmax = F.gumbel_softmax(logits, temperature)
>>> temperature = 0.01
print(gumbel_softmax)
>>> gumbel_softmax = F.gumbel_softmax(logits, temperature)
# out's value is as follows:
>>> print(gumbel_softmax)
# [[0.00000001, 1. , 0.00000000, 0.00000000, 0.00000006, 0.00000000],
Tensor(shape=[4, 6], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.00000000, 0.00000000, 0.00000000, 0.00000000, 0.00000000, 1. ],
[[0.00000000, 1. , 0.00000000, 0.00000000, 0.00000000, 0.00000000],
# [0.00000062, 0.00000000, 0.00000000, 0.00000000, 0.00000000, 0.99999940],
[0.00000000, 0.00000000, 1. , 0.00000000, 0.00000000, 0.00000000],
# [0.00000000, 0.00000000, 0.00000000, 0.00001258, 0.99998736, 0.00000000]]
[0.00000000, 0.00000004, 0.00000000, 0.00000000, 1. , 0.00000000],
[0.00000000, 1. , 0.00000000, 0.00000000, 0.00000000, 0.00000000]])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
...
python/paddle/nn/functional/common.py
浏览文件 @
0a15b0db
...
@@ -95,11 +95,11 @@ def unfold(x, kernel_sizes, strides=1, paddings=0, dilations=1, name=None):
...
@@ -95,11 +95,11 @@ def unfold(x, kernel_sizes, strides=1, paddings=0, dilations=1, name=None):
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.randn((100,3,224,224))
>>>
x = paddle.randn((100,3,224,224))
y = F.unfold(x, [3, 3], 1, 1, 1)
>>>
y = F.unfold(x, [3, 3], 1, 1, 1)
"""
"""
helper
=
LayerHelper
(
"unfold"
,
**
locals
())
helper
=
LayerHelper
(
"unfold"
,
**
locals
())
...
@@ -348,23 +348,21 @@ def interpolate(
...
@@ -348,23 +348,21 @@ def interpolate(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
input_data = paddle.randn(shape=(2,3,6,10)).astype(paddle.float32)
>>> input_data = paddle.randn(shape=(2,3,6,10)).astype(paddle.float32)
output_1 = F.interpolate(x=input_data, size=[12,12])
>>> output_1 = F.interpolate(x=input_data, size=[12,12])
print(output_1.shape)
>>> print(output_1.shape)
# [2L, 3L, 12L, 12L]
[2, 3, 12, 12]
>>> # given scale
# given scale
>>> output_2 = F.interpolate(x=input_data, scale_factor=[2,1])
output_2 = F.interpolate(x=input_data, scale_factor=[2,1])
>>> print(output_2.shape)
print(output_2.shape)
[2, 3, 12, 10]
# [2L, 3L, 12L, 10L]
>>> # bilinear interp
>>> output_3 = F.interpolate(x=input_data, scale_factor=[2,1], mode="bilinear")
# bilinear interp
>>> print(output_2.shape)
output_3 = F.interpolate(x=input_data, scale_factor=[2,1], mode="bilinear")
[2, 3, 12, 10]
print(output_2.shape)
# [2L, 3L, 12L, 10L]
"""
"""
data_format
=
data_format
.
upper
()
data_format
=
data_format
.
upper
()
resample
=
mode
.
upper
()
resample
=
mode
.
upper
()
...
@@ -877,15 +875,14 @@ def upsample(
...
@@ -877,15 +875,14 @@ def upsample(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn as nn
>>> import paddle.nn as nn
input_data = paddle.randn(shape=(2,3,6,10)).astype(paddle.float32)
upsample_out = paddle.nn.Upsample(size=[12,12])
output = upsample_out(x=input_data)
>>> input_data = paddle.randn(shape=(2,3,6,10)).astype(paddle.float32)
print(output.shape)
>>> upsample_out = paddle.nn.Upsample(size=[12,12])
# [2L, 3L, 12L, 12L]
>>> output = upsample_out(x=input_data)
>>> print(output.shape)
[2, 3, 12, 12]
"""
"""
return
interpolate
(
return
interpolate
(
...
@@ -913,17 +910,16 @@ def bilinear(x1, x2, weight, bias=None, name=None):
...
@@ -913,17 +910,16 @@ def bilinear(x1, x2, weight, bias=None, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x1 = paddle.randn((5, 5)).astype(paddle.float32)
>>> x1 = paddle.randn((5, 5)).astype(paddle.float32)
x2 = paddle.randn((5, 4)).astype(paddle.float32)
>>> x2 = paddle.randn((5, 4)).astype(paddle.float32)
w = paddle.randn((1000, 5, 4)).astype(paddle.float32)
>>> w = paddle.randn((1000, 5, 4)).astype(paddle.float32)
b = paddle.randn((1, 1000)).astype(paddle.float32)
>>> b = paddle.randn((1, 1000)).astype(paddle.float32)
>>> result = F.bilinear(x1, x2, w, b)
result = F.bilinear(x1, x2, w, b)
>>> print(result.shape)
print(result.shape)
[5, 1000]
# [5, 1000]
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
...
@@ -1061,39 +1057,38 @@ def dropout(
...
@@ -1061,39 +1057,38 @@ def dropout(
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
>>> paddle.seed(2023)
x = paddle.to_tensor([[1,2,3], [4,5,6]]).astype(paddle.float32)
>>> x = paddle.to_tensor([[1,2,3], [4,5,6]]).astype(paddle.float32)
y_train = paddle.nn.functional.dropout(x, 0.5)
>>> y_train = paddle.nn.functional.dropout(x, 0.5)
y_test = paddle.nn.functional.dropout(x, 0.5, training=False)
>>> y_test = paddle.nn.functional.dropout(x, 0.5, training=False)
y_0 = paddle.nn.functional.dropout(x, axis=0)
>>> y_0 = paddle.nn.functional.dropout(x, axis=0)
y_1 = paddle.nn.functional.dropout(x, axis=1)
>>> y_1 = paddle.nn.functional.dropout(x, axis=1)
y_01 = paddle.nn.functional.dropout(x, axis=[0,1])
>>> y_01 = paddle.nn.functional.dropout(x, axis=[0,1])
print(x)
>>> print(x)
# Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[1., 2., 3.],
[[1., 2., 3.],
# [4., 5., 6.]])
[4., 5., 6.]])
print(y_train)
>>> print(y_train)
# Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[2. , 0. , 6. ],
[[2., 4., 0.],
# [8. , 0. , 12.]])
[8., 0., 0.]])
print(y_test)
>>> print(y_test)
# Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[1., 2., 3.],
[[1., 2., 3.],
# [4., 5., 6.]])
[4., 5., 6.]])
print(y_0)
>>> print(y_0)
# Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[0. , 0. , 0. ],
[[2., 4., 6.],
# [8. , 10., 12.]])
[8. , 10., 12.]])
print(y_1)
>>> print(y_1)
# Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[2. , 0. , 6. ],
[[2. , 4. , 6. ],
# [8. , 0. , 12.]])
[8. , 10., 12.]])
print(y_01)
>>> print(y_01)
# Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[0. , 0. , 0. ],
[[0., 0., 6.],
# [8. , 0. , 12.]])
[0., 0., 0.]])
"""
"""
if
not
isinstance
(
p
,
(
float
,
int
,
Variable
)):
if
not
isinstance
(
p
,
(
float
,
int
,
Variable
)):
raise
TypeError
(
"p argument should be a number or Variable"
)
raise
TypeError
(
"p argument should be a number or Variable"
)
...
@@ -1258,17 +1253,106 @@ def dropout2d(x, p=0.5, training=True, data_format='NCHW', name=None):
...
@@ -1258,17 +1253,106 @@ def dropout2d(x, p=0.5, training=True, data_format='NCHW', name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
>>> paddle.seed(1)
x = paddle.randn(shape=(2, 3, 4, 5)).astype(paddle.float32)
>>> x = paddle.randn(shape=(2, 3, 4, 5)).astype(paddle.float32)
y_train = paddle.nn.functional.dropout2d(x) #train
>>> y_train = paddle.nn.functional.dropout2d(x) #train
y_test = paddle.nn.functional.dropout2d(x, training=False) #test
>>> y_test = paddle.nn.functional.dropout2d(x, training=False) #test
for i in range(2):
>>> for i in range(2):
for j in range(3):
... for j in range(3):
print(x[i,j,:,:])
... print(x[i,j,:,:])
print(y_train[i,j,:,:]) # may all 0
... print(y_train[i,j,:,:]) # may all 0
print(y_test[i,j,:,:])
... print(y_test[i,j,:,:])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0.30557564, 0.11855337, 0.41220093, -0.09968963, 1.50014710],
[ 1.24004936, -0.92485696, 0.08612321, 1.15149164, -0.09276631],
[ 1.22873247, -1.46587241, -1.30802727, 0.19496460, 1.73776841],
[ 0.40092674, 0.67630458, 0.72265440, 1.31720388, -1.41899264]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0.61115128, 0.23710674, 0.82440186, -0.19937925, 3.00029421],
[ 2.48009872, -1.84971392, 0.17224643, 2.30298328, -0.18553263],
[ 2.45746493, -2.93174481, -2.61605453, 0.38992921, 3.47553682],
[ 0.80185348, 1.35260916, 1.44530880, 2.63440776, -2.83798528]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0.30557564, 0.11855337, 0.41220093, -0.09968963, 1.50014710],
[ 1.24004936, -0.92485696, 0.08612321, 1.15149164, -0.09276631],
[ 1.22873247, -1.46587241, -1.30802727, 0.19496460, 1.73776841],
[ 0.40092674, 0.67630458, 0.72265440, 1.31720388, -1.41899264]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[ 0.88350385, -1.14767575, 0.51043051, -0.10051888, -0.61305630],
[-0.12084112, 0.48506257, -1.13189507, 0.62806708, -0.80003673],
[ 0.51513153, -0.08890446, 0.22753835, 0.11557858, 0.78117645],
[ 1.47505593, 0.84618902, -0.38528305, -1.05887091, 0.16592593]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[ 1.76700771, -2.29535151, 1.02086103, -0.20103776, -1.22611260],
[-0.24168225, 0.97012514, -2.26379013, 1.25613415, -1.60007346],
[ 1.03026307, -0.17780893, 0.45507669, 0.23115715, 1.56235290],
[ 2.95011187, 1.69237804, -0.77056611, -2.11774182, 0.33185187]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[ 0.88350385, -1.14767575, 0.51043051, -0.10051888, -0.61305630],
[-0.12084112, 0.48506257, -1.13189507, 0.62806708, -0.80003673],
[ 0.51513153, -0.08890446, 0.22753835, 0.11557858, 0.78117645],
[ 1.47505593, 0.84618902, -0.38528305, -1.05887091, 0.16592593]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-1.46668839, -0.38117948, 1.18678427, 0.38740095, 0.29117522],
[-0.13538910, -0.14527084, -0.04912176, -0.26063353, 0.23640174],
[ 0.45643106, 0.60587281, -1.03242552, -0.45319262, -1.57911122],
[-0.08732958, -0.75898546, 0.14563090, -1.73751652, -0.89109969]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0., -0., 0. , 0. , 0. ],
[-0., -0., -0., -0., 0. ],
[0. , 0. , -0., -0., -0.],
[-0., -0., 0. , -0., -0.]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-1.46668839, -0.38117948, 1.18678427, 0.38740095, 0.29117522],
[-0.13538910, -0.14527084, -0.04912176, -0.26063353, 0.23640174],
[ 0.45643106, 0.60587281, -1.03242552, -0.45319262, -1.57911122],
[-0.08732958, -0.75898546, 0.14563090, -1.73751652, -0.89109969]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0.32110816, -0.76044011, 0.34456784, -0.39410326, 0.37896338],
[ 0.52747023, 0.72711533, 0.29204839, 0.72493637, 0.31128070],
[ 0.58046782, -1.78499067, -1.67504823, -0.38590902, -0.26243693],
[ 0.96669912, 0.43670532, -0.38109761, 0.78405094, -2.17882323]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0., -0., 0. , -0., 0. ],
[0. , 0. , 0. , 0. , 0. ],
[0. , -0., -0., -0., -0.],
[0. , 0. , -0., 0. , -0.]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[-0.32110816, -0.76044011, 0.34456784, -0.39410326, 0.37896338],
[ 0.52747023, 0.72711533, 0.29204839, 0.72493637, 0.31128070],
[ 0.58046782, -1.78499067, -1.67504823, -0.38590902, -0.26243693],
[ 0.96669912, 0.43670532, -0.38109761, 0.78405094, -2.17882323]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[ 0.17168395, 0.45112833, 0.63307828, 2.38763475, -1.27247131],
[ 0.56171960, -1.09584677, 0.38300961, -0.57512099, 0.31011426],
[-0.95336407, -1.04852903, -0.21312937, -0.53549880, -0.00074209],
[ 2.22819090, 1.12403083, -0.04198794, -1.51167727, -0.42699185]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[0. , 0. , 0. , 0. , -0.],
[0. , -0., 0. , -0., 0. ],
[-0., -0., -0., -0., -0.],
[0. , 0. , -0., -0., -0.]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[ 0.17168395, 0.45112833, 0.63307828, 2.38763475, -1.27247131],
[ 0.56171960, -1.09584677, 0.38300961, -0.57512099, 0.31011426],
[-0.95336407, -1.04852903, -0.21312937, -0.53549880, -0.00074209],
[ 2.22819090, 1.12403083, -0.04198794, -1.51167727, -0.42699185]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[ 0.62503546, -0.20989063, -0.22046235, -0.38679042, -1.02590704],
[ 1.04561794, 1.08428383, -0.52219963, -1.56003857, 0.89213932],
[-0.16578521, 0.14524542, -0.45563069, 0.48180851, 1.35843253],
[ 1.07669640, -0.84535235, -1.18651557, 0.79144061, -0.45565742]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[0. , -0., -0., -0., -0.],
[0. , 0. , -0., -0., 0. ],
[-0., 0. , -0., 0. , 0. ],
[0. , -0., -0., 0. , -0.]])
Tensor(shape=[4, 5], dtype=float32, place=Place(cpu), stop_gradient=True,
[[ 0.62503546, -0.20989063, -0.22046235, -0.38679042, -1.02590704],
[ 1.04561794, 1.08428383, -0.52219963, -1.56003857, 0.89213932],
[-0.16578521, 0.14524542, -0.45563069, 0.48180851, 1.35843253],
[ 1.07669640, -0.84535235, -1.18651557, 0.79144061, -0.45565742]])
"""
"""
input_shape
=
x
.
shape
input_shape
=
x
.
shape
if
len
(
input_shape
)
!=
4
:
if
len
(
input_shape
)
!=
4
:
...
@@ -1317,14 +1401,14 @@ def dropout3d(x, p=0.5, training=True, data_format='NCDHW', name=None):
...
@@ -1317,14 +1401,14 @@ def dropout3d(x, p=0.5, training=True, data_format='NCDHW', name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
x = paddle.randn(shape=(2, 3, 4, 5, 6)).astype(paddle.float32)
>>>
x = paddle.randn(shape=(2, 3, 4, 5, 6)).astype(paddle.float32)
y_train = paddle.nn.functional.dropout3d(x) #train
>>>
y_train = paddle.nn.functional.dropout3d(x) #train
y_test = paddle.nn.functional.dropout3d(x, training=False) #test
>>>
y_test = paddle.nn.functional.dropout3d(x, training=False) #test
print(x[0,0,:,:,:])
>>>
print(x[0,0,:,:,:])
print(y_train[0,0,:,:,:]) # may all 0
>>>
print(y_train[0,0,:,:,:]) # may all 0
print(y_test[0,0,:,:,:])
>>>
print(y_test[0,0,:,:,:])
"""
"""
...
@@ -1371,19 +1455,19 @@ def alpha_dropout(x, p=0.5, training=True, name=None):
...
@@ -1371,19 +1455,19 @@ def alpha_dropout(x, p=0.5, training=True, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
>>> paddle.seed(1)
x = paddle.to_tensor([[-1, 1], [-1, 1]]).astype(paddle.float32)
>>>
x = paddle.to_tensor([[-1, 1], [-1, 1]]).astype(paddle.float32)
y_train = paddle.nn.functional.alpha_dropout(x, 0.5)
>>>
y_train = paddle.nn.functional.alpha_dropout(x, 0.5)
y_test = paddle.nn.functional.alpha_dropout(x, 0.5, training=False)
>>>
y_test = paddle.nn.functional.alpha_dropout(x, 0.5, training=False)
print(y_train)
>>>
print(y_train)
#
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[-0.10721093, -0.77919382
],
[[-0.77919382, 1.66559887
],
# [-0.10721093, 1.66559887]]) (randomly
)
[-0.10721093, -0.77919382]]
)
print(y_test)
>>>
print(y_test)
#
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
#
[[-1., 1.],
[[-1., 1.],
#
[-1., 1.]])
[-1., 1.]])
"""
"""
if
not
isinstance
(
p
,
(
float
,
int
)):
if
not
isinstance
(
p
,
(
float
,
int
)):
raise
TypeError
(
"p argument should be a float or int"
)
raise
TypeError
(
"p argument should be a float or int"
)
...
@@ -1516,32 +1600,35 @@ def pad(x, pad, mode='constant', value=0.0, data_format="NCHW", name=None):
...
@@ -1516,32 +1600,35 @@ def pad(x, pad, mode='constant', value=0.0, data_format="NCHW", name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
# example 1
>>> # example 1
x_shape = (1, 1, 3)
>>> x_shape = (1, 1, 3)
x = paddle.arange(paddle.prod(paddle.to_tensor(x_shape)), dtype="float32").reshape(x_shape) + 1
>>> x = paddle.arange(paddle.prod(paddle.to_tensor(x_shape)), dtype="float32").reshape(x_shape) + 1
y = F.pad(x, [0, 0, 0, 0, 2, 3], value=1, mode='constant', data_format="NCL")
>>> y = F.pad(x, [0, 0, 0, 0, 2, 3], value=1, mode='constant', data_format="NCL")
print(y)
>>> print(y)
# [[[1. 1. 1. 2. 3. 1. 1. 1.]]]
Tensor(shape=[1, 1, 8], dtype=float32, place=Place(cpu), stop_gradient=True,
[[[1., 1., 1., 2., 3., 1., 1., 1.]]])
# example 2
x_shape = (1, 1, 3)
>>> # example 2
x = paddle.arange(paddle.prod(paddle.to_tensor(x_shape)), dtype="float32").reshape(x_shape) + 1
>>> x_shape = (1, 1, 3)
y = F.pad(x, [2, 3], value=1, mode='constant', data_format="NCL")
>>> x = paddle.arange(paddle.prod(paddle.to_tensor(x_shape)), dtype="float32").reshape(x_shape) + 1
print(y)
>>> y = F.pad(x, [2, 3], value=1, mode='constant', data_format="NCL")
# [[[1. 1. 1. 2. 3. 1. 1. 1.]]]
>>> print(y)
Tensor(shape=[1, 1, 8], dtype=float32, place=Place(cpu), stop_gradient=True,
# example 3
[[[1., 1., 1., 2., 3., 1., 1., 1.]]])
x_shape = (1, 1, 2, 3)
x = paddle.arange(paddle.prod(paddle.to_tensor(x_shape)), dtype="float32").reshape(x_shape) + 1
>>> # example 3
y = F.pad(x, [1, 2, 1, 1], value=1, mode='circular')
>>> x_shape = (1, 1, 2, 3)
print(y)
>>> x = paddle.arange(paddle.prod(paddle.to_tensor(x_shape)), dtype="float32").reshape(x_shape) + 1
# [[[[6. 4. 5. 6. 4. 5.]
>>> y = F.pad(x, [1, 2, 1, 1], value=1, mode='circular')
# [3. 1. 2. 3. 1. 2.]
>>> print(y)
# [6. 4. 5. 6. 4. 5.]
Tensor(shape=[1, 1, 4, 6], dtype=float32, place=Place(cpu), stop_gradient=True,
# [3. 1. 2. 3. 1. 2.]]]]
[[[[6., 4., 5., 6., 4., 5.],
[3., 1., 2., 3., 1., 2.],
[6., 4., 5., 6., 4., 5.],
[3., 1., 2., 3., 1., 2.]]]])
"""
"""
assert
mode
in
[
assert
mode
in
[
'reflect'
,
'reflect'
,
...
@@ -1713,16 +1800,18 @@ def zeropad2d(x, padding, data_format="NCHW", name=None):
...
@@ -1713,16 +1800,18 @@ def zeropad2d(x, padding, data_format="NCHW", name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
x_shape = paddle.to_tensor([1, 1, 2, 3])
x = paddle.arange(paddle.prod(x_shape), dtype="float32").reshape(x_shape) + 1
>>> x_shape = paddle.to_tensor([1, 1, 2, 3])
y = F.zeropad2d(x, [1, 2, 1, 1])
>>> x = paddle.arange(paddle.prod(x_shape), dtype="float32").reshape(x_shape) + 1
print(y)
>>> y = F.zeropad2d(x, [1, 2, 1, 1])
# [[[[0. 0. 0. 0. 0. 0.]
>>> print(y)
# [0. 1. 2. 3. 0. 0.]
Tensor(shape=[1, 1, 4, 6], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0. 4. 5. 6. 0. 0.]
[[[[0., 0., 0., 0., 0., 0.],
# [0. 0. 0. 0. 0. 0.]]]]
[0., 1., 2., 3., 0., 0.],
[0., 4., 5., 6., 0., 0.],
[0., 0., 0., 0., 0., 0.]]]])
"""
"""
return
pad
(
return
pad
(
...
@@ -1767,16 +1856,17 @@ def cosine_similarity(x1, x2, axis=1, eps=1e-8):
...
@@ -1767,16 +1856,17 @@ def cosine_similarity(x1, x2, axis=1, eps=1e-8):
Code Examples:
Code Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn as nn
>>>
import paddle.nn as nn
paddle.seed(1)
>>>
paddle.seed(1)
x1 = paddle.randn(shape=[2, 3])
>>>
x1 = paddle.randn(shape=[2, 3])
x2 = paddle.randn(shape=[2, 3])
>>>
x2 = paddle.randn(shape=[2, 3])
result = paddle.nn.functional.cosine_similarity(x1, x2, axis=0)
>>> result = paddle.nn.functional.cosine_similarity(x1, x2, axis=0)
print(result)
>>> print(result)
# [0.97689527, 0.99996042, -0.55138415]
Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
[ 0.97689527, 0.99996042, -0.55138415])
"""
"""
w12
=
sum
(
paddle
.
multiply
(
x1
,
x2
),
axis
=
axis
)
w12
=
sum
(
paddle
.
multiply
(
x1
,
x2
),
axis
=
axis
)
...
@@ -1822,21 +1912,29 @@ def linear(x, weight, bias=None, name=None):
...
@@ -1822,21 +1912,29 @@ def linear(x, weight, bias=None, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
>>> paddle.seed(2023)
x = paddle.randn((3, 2), dtype="float32")
>>> x = paddle.randn((3, 2), dtype="float32")
# x: [[-0.32342386 -1.200079 ]
>>> print(x)
# [ 0.7979031 -0.90978354]
Tensor(shape=[3, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [ 0.40597573 1.8095392 ]]
[[ 0.06132207, 1.11349595],
weight = paddle.full(shape=[2, 4], fill_value="0.5", dtype="float32", name="weight")
[ 0.41906244, -0.24858207],
# weight: [[0.5 0.5 0.5 0.5]
[-1.85169315, -1.50370061]])
# [0.5 0.5 0.5 0.5]]
>>> weight = paddle.full(shape=[2, 4], fill_value="0.5", dtype="float32", name="weight")
bias = paddle.ones(shape=[4], dtype="float32", name="bias")
>>> print(weight)
# bias: [1. 1. 1. 1.]
Tensor(shape=[2, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
y = paddle.nn.functional.linear(x, weight, bias)
[[0.50000000, 0.50000000, 0.50000000, 0.50000000],
# y: [[0.23824859 0.23824859 0.23824859 0.23824859]
[0.50000000, 0.50000000, 0.50000000, 0.50000000]])
# [0.9440598 0.9440598 0.9440598 0.9440598 ]
>>> bias = paddle.ones(shape=[4], dtype="float32", name="bias")
# [2.1077576 2.1077576 2.1077576 2.1077576 ]]
>>> print(bias)
Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
[1., 1., 1., 1.])
>>> y = paddle.nn.functional.linear(x, weight, bias)
>>> print(y)
Tensor(shape=[3, 4], dtype=float32, place=Place(cpu), stop_gradient=True,
[[ 1.58740902, 1.58740902, 1.58740902, 1.58740902],
[ 1.08524013, 1.08524013, 1.08524013, 1.08524013],
[-0.67769694, -0.67769694, -0.67769694, -0.67769694]])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
# TODO(jiabin): using addmm for fast forward route
# TODO(jiabin): using addmm for fast forward route
...
@@ -1921,17 +2019,17 @@ def label_smooth(label, prior_dist=None, epsilon=0.1, name=None):
...
@@ -1921,17 +2019,17 @@ def label_smooth(label, prior_dist=None, epsilon=0.1, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
paddle.disable_static()
>>>
paddle.disable_static()
x = paddle.to_tensor([[[0, 1, 0],
>>>
x = paddle.to_tensor([[[0, 1, 0],
[ 1, 0, 1]]], dtype="float32", stop_gradient=False)
>>>
[ 1, 0, 1]]], dtype="float32", stop_gradient=False)
output = paddle.nn.functional.label_smooth(x)
>>>
output = paddle.nn.functional.label_smooth(x)
print(output)
>>>
print(output)
# Tensor(shape=[1, 2, 3], dtype=float32, place=Place(gpu:0
), stop_gradient=False,
Tensor(shape=[1, 2, 3], dtype=float32, place=Place(cpu
), stop_gradient=False,
#
[[[0.03333334, 0.93333334, 0.03333334],
[[[0.03333334, 0.93333334, 0.03333334],
#
[0.93333334, 0.03333334, 0.93333334]]])
[0.93333334, 0.03333334, 0.93333334]]])
"""
"""
if
epsilon
>
1.0
or
epsilon
<
0.0
:
if
epsilon
>
1.0
or
epsilon
<
0.0
:
raise
ValueError
(
"The value of epsilon must be between 0 and 1."
)
raise
ValueError
(
"The value of epsilon must be between 0 and 1."
)
...
@@ -2002,67 +2100,64 @@ def class_center_sample(label, num_classes, num_samples, group=None):
...
@@ -2002,67 +2100,64 @@ def class_center_sample(label, num_classes, num_samples, group=None):
.. code-block:: python
.. code-block:: python
:name: code-example1
:name: code-example1
# CPU or single GPU
>>> # CPU or single GPU
import paddle
>>> import paddle
num_classes = 20
>>> num_classes = 20
batch_size = 10
>>> batch_size = 10
num_samples = 6
>>> num_samples = 6
label = paddle.randint(low=0, high=num_classes, shape=[batch_size], dtype='int64')
>>> paddle.seed(2023)
remapped_label, sampled_class_index = paddle.nn.functional.class_center_sample(label, num_classes, num_samples)
>>> label = paddle.randint(low=0, high=num_classes, shape=[batch_size], dtype='int64')
>>> remapped_label, sampled_class_index = paddle.nn.functional.class_center_sample(label, num_classes, num_samples)
print(label)
>>> print(label)
print(remapped_label)
Tensor(shape=[10], dtype=int64, place=Place(cpu), stop_gradient=True,
print(sampled_class_index)
[17, 10, 5 , 18, 8 , 8 , 19, 14, 10, 14])
>>> print(remapped_label)
# the output is
Tensor(shape=[10], dtype=int64, place=Place(cpu), stop_gradient=True,
#Tensor(shape=[10], dtype=int64, place=CPUPlace, stop_gradient=True,
[4, 2, 0, 5, 1, 1, 6, 3, 2, 3])
# [11, 5 , 1 , 3 , 12, 2 , 15, 19, 18, 19])
>>> print(sampled_class_index)
#Tensor(shape=[10], dtype=int64, place=CPUPlace, stop_gradient=True,
Tensor(shape=[7], dtype=int64, place=Place(cpu), stop_gradient=True,
# [4, 3, 0, 2, 5, 1, 6, 8, 7, 8])
[5 , 8 , 10, 14, 17, 18, 19])
#Tensor(shape=[9], dtype=int64, place=CPUPlace, stop_gradient=True,
# [1 , 2 , 3 , 5 , 11, 12, 15, 18, 19])
.. code-block:: python
.. code-block:: python
:name: code-example2
:name: code-example2
# required: distributed
>>> # doctest: +REQUIRES(env:DISTRIBUTED)
# Multi GPU, test_class_center_sample.py
>>> # required: distributed
import paddle
>>> # Multi GPU, test_class_center_sample.py
import paddle.distributed as dist
>>> import paddle
strategy = dist.fleet.DistributedStrategy()
>>> import paddle.distributed as dist
dist.fleet.init(is_collective=True, strategy=strategy)
>>> strategy = dist.fleet.DistributedStrategy()
batch_size = 10
>>> dist.fleet.init(is_collective=True, strategy=strategy)
num_samples = 6
>>> batch_size = 10
rank_id = dist.get_rank()
>>> num_samples = 6
# num_classes of each GPU can be different, e.g num_classes_list = [10, 8]
>>> rank_id = dist.get_rank()
num_classes_list = [10, 10]
>>> # num_classes of each GPU can be different, e.g num_classes_list = [10, 8]
num_classes = paddle.sum(paddle.to_tensor(num_classes_list))
>>> num_classes_list = [10, 10]
label = paddle.randint(low=0, high=num_classes.item(), shape=[batch_size], dtype='int64')
>>> num_classes = paddle.sum(paddle.to_tensor(num_classes_list))
label_list = []
>>> label = paddle.randint(low=0, high=num_classes.item(), shape=[batch_size], dtype='int64')
dist.all_gather(label_list, label)
>>> label_list = []
label = paddle.concat(label_list, axis=0)
>>> dist.all_gather(label_list, label)
remapped_label, sampled_class_index = paddle.nn.functional.class_center_sample(label, num_classes_list[rank_id], num_samples)
>>> label = paddle.concat(label_list, axis=0)
>>> remapped_label, sampled_class_index = paddle.nn.functional.class_center_sample(label, num_classes_list[rank_id], num_samples)
print(label)
print(remapped_label)
>>> print(label)
print(sampled_class_index)
>>> print(remapped_label)
>>> print(sampled_class_index)
#python -m paddle.distributed.launch --gpus=0,1 test_class_center_sample.py
>>> #python -m paddle.distributed.launch --gpus=0,1 test_class_center_sample.py
# rank 0 output:
>>> # rank 0 output:
#Tensor(shape=[20], dtype=int64, place=CUDAPlace(0), stop_gradient=True,
Tensor(shape=[20], dtype=int64, place=CUDAPlace(0), stop_gradient=True,
# [10, 17, 15, 11, 9 , 12, 18, 18, 17, 18, 19, 2 , 8 , 13, 11, 13, 9 , 10, 0 , 4 ])
[10, 17, 15, 11, 9 , 12, 18, 18, 17, 18, 19, 2 , 8 , 13, 11, 13, 9 , 10, 0 , 4 ])
#Tensor(shape=[20], dtype=int64, place=CUDAPlace(0), stop_gradient=True,
Tensor(shape=[20], dtype=int64, place=CUDAPlace(0), stop_gradient=True,
# [6 , 11, 10, 7 , 4 , 8 , 12, 12, 11, 12, 13, 1 , 3 , 9 , 7 , 9 , 4 , 6 , 0 , 2 ])
[6 , 11, 10, 7 , 4 , 8 , 12, 12, 11, 12, 13, 1 , 3 , 9 , 7 , 9 , 4 , 6 , 0 , 2 ])
#Tensor(shape=[6], dtype=int64, place=CUDAPlace(0), stop_gradient=True,
Tensor(shape=[6], dtype=int64, place=CUDAPlace(0), stop_gradient=True,
# [0, 2, 4, 8, 9, 3])
[0, 2, 4, 8, 9, 3])
>>> # rank 1 output:
# rank 1 output:
Tensor(shape=[20], dtype=int64, place=CUDAPlace(1), stop_gradient=True,
#Tensor(shape=[20], dtype=int64, place=CUDAPlace(1), stop_gradient=True,
[10, 17, 15, 11, 9 , 12, 18, 18, 17, 18, 19, 2 , 8 , 13, 11, 13, 9 , 10, 0 , 4 ])
# [10, 17, 15, 11, 9 , 12, 18, 18, 17, 18, 19, 2 , 8 , 13, 11, 13, 9 , 10, 0 , 4 ])
Tensor(shape=[20], dtype=int64, place=CUDAPlace(1), stop_gradient=True,
#Tensor(shape=[20], dtype=int64, place=CUDAPlace(1), stop_gradient=True,
[6 , 11, 10, 7 , 4 , 8 , 12, 12, 11, 12, 13, 1 , 3 , 9 , 7 , 9 , 4 , 6 , 0 , 2 ])
# [6 , 11, 10, 7 , 4 , 8 , 12, 12, 11, 12, 13, 1 , 3 , 9 , 7 , 9 , 4 , 6 , 0 , 2 ])
Tensor(shape=[7], dtype=int64, place=CUDAPlace(1), stop_gradient=True,
#Tensor(shape=[7], dtype=int64, place=CUDAPlace(1), stop_gradient=True,
[0, 1, 2, 3, 5, 7, 8])
# [0, 1, 2, 3, 5, 7, 8])
"""
"""
if
not
(
group
is
False
or
group
is
None
or
hasattr
(
group
,
'is_member'
)):
if
not
(
group
is
False
or
group
is
None
or
hasattr
(
group
,
'is_member'
)):
raise
ValueError
(
raise
ValueError
(
...
@@ -2216,12 +2311,15 @@ def fold(
...
@@ -2216,12 +2311,15 @@ def fold(
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.randn([2,3*2*2,12])
>>> x = paddle.randn([2,3*2*2,12])
y = F.fold(x, output_sizes=[4, 5], kernel_sizes=2)
>>> y = F.fold(x, output_sizes=[4, 5], kernel_sizes=2)
# y.shape = [2,3,4,5]
>>> x = paddle.randn([2,3*2*2,12])
>>> y = F.fold(x, output_sizes=[4, 5], kernel_sizes=2)
>>> print(y.shape)
[2, 3, 4, 5]
"""
"""
...
...
python/paddle/nn/functional/conv.py
浏览文件 @
0a15b0db
...
@@ -368,24 +368,24 @@ def conv1d(
...
@@ -368,24 +368,24 @@ def conv1d(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x = paddle.to_tensor([[[4, 8, 1, 9],
>>>
x = paddle.to_tensor([[[4, 8, 1, 9],
[7, 2, 0, 9],
...
[7, 2, 0, 9],
[6, 9, 2, 6]]], dtype="float32")
...
[6, 9, 2, 6]]], dtype="float32")
w = paddle.to_tensor([[[9, 3, 4],
>>>
w = paddle.to_tensor([[[9, 3, 4],
[0, 0, 7],
...
[0, 0, 7],
[2, 5, 6]],
...
[2, 5, 6]],
[[0, 3, 4],
...
[[0, 3, 4],
[2, 9, 7],
...
[2, 9, 7],
[5, 6, 8]]], dtype="float32")
...
[5, 6, 8]]], dtype="float32")
y = F.conv1d(x, w)
>>>
y = F.conv1d(x, w)
print(y)
>>>
print(y)
# Tensor(shape=[1, 2, 2], dtype=float32, place=Place(gpu:0
), stop_gradient=True,
Tensor(shape=[1, 2, 2], dtype=float32, place=Place(cpu
), stop_gradient=True,
#
[[[133., 238.],
[[[133., 238.],
#
[160., 211.]]])
[160., 211.]]])
"""
"""
cudnn_version
=
get_cudnn_version
()
cudnn_version
=
get_cudnn_version
()
if
cudnn_version
is
not
None
:
if
cudnn_version
is
not
None
:
...
@@ -632,16 +632,16 @@ def conv2d(
...
@@ -632,16 +632,16 @@ def conv2d(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x_var = paddle.randn((2, 3, 8, 8), dtype='float32')
>>>
x_var = paddle.randn((2, 3, 8, 8), dtype='float32')
w_var = paddle.randn((6, 3, 3, 3), dtype='float32')
>>>
w_var = paddle.randn((6, 3, 3, 3), dtype='float32')
y_var = F.conv2d(x_var, w_var)
>>>
y_var = F.conv2d(x_var, w_var)
print(y_var.shape)
>>>
print(y_var.shape)
#
[2, 6, 6, 6]
[2, 6, 6, 6]
"""
"""
# entry checks
# entry checks
if
data_format
not
in
[
"NCHW"
,
"NHWC"
]:
if
data_format
not
in
[
"NCHW"
,
"NHWC"
]:
...
@@ -887,20 +887,20 @@ def conv1d_transpose(
...
@@ -887,20 +887,20 @@ def conv1d_transpose(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
# shape: (1, 2, 4)
>>>
# shape: (1, 2, 4)
x = paddle.to_tensor([[[4, 0, 9, 7],
>>>
x = paddle.to_tensor([[[4, 0, 9, 7],
[8, 0, 9, 2,]]], dtype="float32")
>>>
[8, 0, 9, 2,]]], dtype="float32")
# shape: (2, 1, 2)
>>>
# shape: (2, 1, 2)
w = paddle.to_tensor([[[7, 0]],
>>>
w = paddle.to_tensor([[[7, 0]],
[[4, 2]]], dtype="float32")
>>>
[[4, 2]]], dtype="float32")
y = F.conv1d_transpose(x, w)
>>>
y = F.conv1d_transpose(x, w)
print(y)
>>>
print(y)
# Tensor(shape=[1, 1, 5], dtype=float32, place=Place(gpu:0
), stop_gradient=True,
Tensor(shape=[1, 1, 5], dtype=float32, place=Place(cpu
), stop_gradient=True,
#
[[[60., 16., 99., 75., 4. ]]])
[[[60., 16., 99., 75., 4. ]]])
"""
"""
cudnn_version
=
get_cudnn_version
()
cudnn_version
=
get_cudnn_version
()
if
cudnn_version
is
not
None
:
if
cudnn_version
is
not
None
:
...
@@ -1183,16 +1183,16 @@ def conv2d_transpose(
...
@@ -1183,16 +1183,16 @@ def conv2d_transpose(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x_var = paddle.randn((2, 3, 8, 8), dtype='float32')
>>>
x_var = paddle.randn((2, 3, 8, 8), dtype='float32')
w_var = paddle.randn((3, 6, 3, 3), dtype='float32')
>>>
w_var = paddle.randn((3, 6, 3, 3), dtype='float32')
y_var = F.conv2d_transpose(x_var, w_var)
>>>
y_var = F.conv2d_transpose(x_var, w_var)
print(y_var.shape)
>>>
print(y_var.shape)
#
[2, 6, 10, 10]
[2, 6, 10, 10]
"""
"""
if
data_format
not
in
[
'NCHW'
,
'NHWC'
]:
if
data_format
not
in
[
'NCHW'
,
'NHWC'
]:
...
@@ -1476,16 +1476,16 @@ def conv3d(
...
@@ -1476,16 +1476,16 @@ def conv3d(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x_var = paddle.randn((2, 3, 8, 8, 8), dtype='float32')
>>>
x_var = paddle.randn((2, 3, 8, 8, 8), dtype='float32')
w_var = paddle.randn((6, 3, 3, 3, 3), dtype='float32')
>>>
w_var = paddle.randn((6, 3, 3, 3, 3), dtype='float32')
y_var = F.conv3d(x_var, w_var)
>>>
y_var = F.conv3d(x_var, w_var)
print(y_var.shape)
>>>
print(y_var.shape)
#
[2, 6, 6, 6, 6]
[2, 6, 6, 6, 6]
"""
"""
# entry check
# entry check
if
data_format
not
in
[
"NCDHW"
,
"NDHWC"
]:
if
data_format
not
in
[
"NCDHW"
,
"NDHWC"
]:
...
@@ -1688,18 +1688,18 @@ def conv3d_transpose(
...
@@ -1688,18 +1688,18 @@ def conv3d_transpose(
variable storing transposed convolution and non-linearity activation result.
variable storing transposed convolution and non-linearity activation result.
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
x_var = paddle.randn((2, 3, 8, 8, 8), dtype='float32')
>>>
x_var = paddle.randn((2, 3, 8, 8, 8), dtype='float32')
w_var = paddle.randn((3, 6, 3, 3, 3), dtype='float32')
>>>
w_var = paddle.randn((3, 6, 3, 3, 3), dtype='float32')
y_var = F.conv3d_transpose(x_var, w_var)
>>>
y_var = F.conv3d_transpose(x_var, w_var)
print(y_var.shape)
>>>
print(y_var.shape)
#
[2, 6, 10, 10, 10]
[2, 6, 10, 10, 10]
"""
"""
# entry checks
# entry checks
if
data_format
not
in
[
"NCDHW"
,
"NDHWC"
]:
if
data_format
not
in
[
"NCDHW"
,
"NDHWC"
]:
...
...
python/paddle/nn/functional/distance.py
浏览文件 @
0a15b0db
...
@@ -59,14 +59,13 @@ def pairwise_distance(x, y, p=2.0, epsilon=1e-6, keepdim=False, name=None):
...
@@ -59,14 +59,13 @@ def pairwise_distance(x, y, p=2.0, epsilon=1e-6, keepdim=False, name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
x = paddle.to_tensor([[1., 3.], [3., 5.]], dtype=paddle.float64)
>>> x = paddle.to_tensor([[1., 3.], [3., 5.]], dtype=paddle.float64)
y = paddle.to_tensor([[5., 6.], [7., 8.]], dtype=paddle.float64)
>>> y = paddle.to_tensor([[5., 6.], [7., 8.]], dtype=paddle.float64)
distance = paddle.nn.functional.pairwise_distance(x, y)
>>> distance = paddle.nn.functional.pairwise_distance(x, y)
print(distance)
>>> print(distance)
# Tensor(shape=[2], dtype=float64, place=Place(gpu:0), stop_gradient=True,
Tensor(shape=[2], dtype=float64, place=Place(cpu), stop_gradient=True,
# [4.99999860, 4.99999860])
[4.99999860, 4.99999860])
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
sub
=
_C_ops
.
subtract
(
x
,
y
)
sub
=
_C_ops
.
subtract
(
x
,
y
)
...
...
python/paddle/nn/functional/extension.py
浏览文件 @
0a15b0db
...
@@ -55,48 +55,46 @@ def diag_embed(input, offset=0, dim1=-2, dim2=-1):
...
@@ -55,48 +55,46 @@ def diag_embed(input, offset=0, dim1=-2, dim2=-1):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
import paddle.nn.functional as F
>>> import paddle.nn.functional as F
diag_embed_input = paddle.arange(6)
>>> diag_embed_input = paddle.arange(6)
diag_embed_output1 = F.diag_embed(diag_embed_input)
>>> diag_embed_output1 = F.diag_embed(diag_embed_input)
print(diag_embed_output1)
>>> print(diag_embed_output1)
# Tensor(shape=[6, 6], dtype=int64, place=Place(cpu), stop_gradient=True,
Tensor(shape=[6, 6], dtype=int64, place=Place(cpu), stop_gradient=True,
# [[0, 0, 0, 0, 0, 0],
[[0, 0, 0, 0, 0, 0],
# [0, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0],
# [0, 0, 2, 0, 0, 0],
[0, 0, 2, 0, 0, 0],
# [0, 0, 0, 3, 0, 0],
[0, 0, 0, 3, 0, 0],
# [0, 0, 0, 0, 4, 0],
[0, 0, 0, 0, 4, 0],
# [0, 0, 0, 0, 0, 5]])
[0, 0, 0, 0, 0, 5]])
diag_embed_output2 = F.diag_embed(diag_embed_input, offset=-1, dim1=0,dim2=1 )
>>> diag_embed_output2 = F.diag_embed(diag_embed_input, offset=-1, dim1=0,dim2=1 )
print(diag_embed_output2)
>>> print(diag_embed_output2)
# Tensor(shape=[7, 7], dtype=int64, place=Place(cpu), stop_gradient=True,
Tensor(shape=[7, 7], dtype=int64, place=Place(cpu), stop_gradient=True,
# [[0, 0, 0, 0, 0, 0, 0],
[[0, 0, 0, 0, 0, 0, 0],
# [0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0],
# [0, 1, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0],
# [0, 0, 2, 0, 0, 0, 0],
[0, 0, 2, 0, 0, 0, 0],
# [0, 0, 0, 3, 0, 0, 0],
[0, 0, 0, 3, 0, 0, 0],
# [0, 0, 0, 0, 4, 0, 0],
[0, 0, 0, 0, 4, 0, 0],
# [0, 0, 0, 0, 0, 5, 0]])
[0, 0, 0, 0, 0, 5, 0]])
diag_embed_input_2dim = paddle.reshape(diag_embed_input,[2,3])
>>> diag_embed_input_2dim = paddle.reshape(diag_embed_input,[2,3])
print(diag_embed_input_2dim)
>>> print(diag_embed_input_2dim)
# Tensor(shape=[2, 3], dtype=int64, place=Place(cpu), stop_gradient=True,
Tensor(shape=[2, 3], dtype=int64, place=Place(cpu), stop_gradient=True,
# [[0, 1, 2],
[[0, 1, 2],
# [3, 4, 5]])
[3, 4, 5]])
diag_embed_output3 = F.diag_embed(diag_embed_input_2dim,offset= 0, dim1=0, dim2=2 )
>>> diag_embed_output3 = F.diag_embed(diag_embed_input_2dim,offset= 0, dim1=0, dim2=2 )
print(diag_embed_output3)
>>> print(diag_embed_output3)
# Tensor(shape=[3, 2, 3], dtype=int64, place=Place(cpu), stop_gradient=True,
Tensor(shape=[3, 2, 3], dtype=int64, place=Place(cpu), stop_gradient=True,
# [[[0, 0, 0],
[[[0, 0, 0],
# [3, 0, 0]],
[3, 0, 0]],
[[0, 1, 0],
# [[0, 1, 0],
[0, 4, 0]],
# [0, 4, 0]],
[[0, 0, 2],
[0, 0, 5]]])
# [[0, 0, 2],
# [0, 0, 5]]])
"""
"""
if
not
isinstance
(
input
,
Variable
):
if
not
isinstance
(
input
,
Variable
):
input
=
assign
(
input
)
input
=
assign
(
input
)
...
@@ -200,16 +198,16 @@ def sequence_mask(x, maxlen=None, dtype='int64', name=None):
...
@@ -200,16 +198,16 @@ def sequence_mask(x, maxlen=None, dtype='int64', name=None):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
lengths = paddle.to_tensor([10, 9, 8])
>>>
lengths = paddle.to_tensor([10, 9, 8])
mask = paddle.nn.functional.sequence_mask(lengths)
>>>
mask = paddle.nn.functional.sequence_mask(lengths)
print(mask)
>>>
print(mask)
# Tensor(shape=[3, 10], dtype=int64, place=Place(gpu:0
), stop_gradient=True,
Tensor(shape=[3, 10], dtype=int64, place=Place(cpu
), stop_gradient=True,
#
[[1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[[1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
#
[1, 1, 1, 1, 1, 1, 1, 1, 1, 0],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 0],
#
[1, 1, 1, 1, 1, 1, 1, 1, 0, 0]])
[1, 1, 1, 1, 1, 1, 1, 1, 0, 0]])
"""
"""
...
@@ -296,14 +294,24 @@ def gather_tree(ids, parents):
...
@@ -296,14 +294,24 @@ def gather_tree(ids, parents):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
ids = paddle.to_tensor([[[2, 2], [6, 1]], [[3, 9], [6, 1]], [[0, 1], [9, 0]]])
>>>
ids = paddle.to_tensor([[[2, 2], [6, 1]], [[3, 9], [6, 1]], [[0, 1], [9, 0]]])
parents = paddle.to_tensor([[[0, 0], [1, 1]], [[1, 0], [1, 0]], [[0, 0], [0, 1]]])
>>> parents = paddle.to_tensor([[[0, 0], [1, 1]], [[1, 0], [1, 0]], [[0, 0], [0, 1]]])
>>> final_sequences = paddle.nn.functional.gather_tree(ids, parents)
>>> [[[2, 2], [1, 6]], [[3, 3], [6, 1]], [[0, 1], [9, 0]]]
>>> final_sequences = paddle.nn.functional.gather_tree(ids, parents)
>>> print(final_sequences)
Tensor(shape=[3, 2, 2], dtype=int64, place=Place(cpu), stop_gradient=True,
[[[2, 2],
[1, 6]],
[[3, 3],
[6, 1]],
[[0, 1],
[9, 0]]])
final_sequences = paddle.nn.functional.gather_tree(ids, parents)
# [[[2, 2], [1, 6]], [[3, 3], [6, 1]], [[0, 1], [9, 0]]]
"""
"""
if
ids
.
ndim
!=
3
:
if
ids
.
ndim
!=
3
:
...
@@ -388,11 +396,11 @@ def temporal_shift(x, seg_num, shift_ratio=0.25, name=None, data_format="NCHW"):
...
@@ -388,11 +396,11 @@ def temporal_shift(x, seg_num, shift_ratio=0.25, name=None, data_format="NCHW"):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>>
import paddle
import paddle.nn.functional as F
>>>
import paddle.nn.functional as F
input = paddle.randn([6, 4, 2, 2])
>>>
input = paddle.randn([6, 4, 2, 2])
out = F.temporal_shift(x=input, seg_num=2, shift_ratio=0.2)
>>>
out = F.temporal_shift(x=input, seg_num=2, shift_ratio=0.2)
"""
"""
if
data_format
not
in
[
"NCHW"
,
"NHWC"
]:
if
data_format
not
in
[
"NCHW"
,
"NHWC"
]:
raise
ValueError
(
raise
ValueError
(
...
...
python/paddle/nn/functional/flash_attention.py
浏览文件 @
0a15b0db
...
@@ -181,13 +181,12 @@ def flash_attention(
...
@@ -181,13 +181,12 @@ def flash_attention(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
# required: skiptest
>>> import paddle
import paddle
q = paddle.rand((1, 128, 2, 16), dtype=paddle.float16)
>>> paddle.seed(1)
>>> q = paddle.rand((1, 128, 2, 16))
output = paddle.nn.functional.flash_attention(q, q, q, 0.9, False, False)
>>> output = paddle.nn.functional.flash_attention.flash_attention(q, q, q, 0.9, False, False)
print(output)
"""
"""
head_dim
=
query
.
shape
[
3
]
head_dim
=
query
.
shape
[
3
]
sdp_func_name
=
_select_sdp
(
head_dim
)
sdp_func_name
=
_select_sdp
(
head_dim
)
...
@@ -340,13 +339,12 @@ def flash_attn_unpadded(
...
@@ -340,13 +339,12 @@ def flash_attn_unpadded(
Examples:
Examples:
.. code-block:: python
.. code-block:: python
# required: skiptest
>>> import paddle
import paddle
>>> paddle.seed(1)
>>> q = paddle.rand((1, 128, 2, 16))
q = paddle.rand((1, 128, 2, 16), dtype=paddle.float16)
output = paddle.nn.functional
.flash_attn_unpadded(q, q, q, 0.9, False, False)
>>> output = paddle.nn.functional.flash_attention
.flash_attn_unpadded(q, q, q, 0.9, False, False)
print(output)
>>>
print(output)
"""
"""
if
in_dynamic_mode
():
if
in_dynamic_mode
():
(
(
...
...
python/paddle/nn/initializer/uniform.py
浏览文件 @
0a15b0db
...
@@ -158,24 +158,33 @@ class Uniform(UniformInitializer):
...
@@ -158,24 +158,33 @@ class Uniform(UniformInitializer):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
>>> paddle.seed(1)
data = paddle.ones(shape=[3, 1, 2], dtype='float32')
>>> data = paddle.ones(shape=[3, 1, 2], dtype='float32')
weight_attr = paddle.framework.ParamAttr(
>>> weight_attr = paddle.framework.ParamAttr(
name="linear_weight",
... name="linear_weight",
initializer=paddle.nn.initializer.Uniform(low=-0.5, high=0.5))
... initializer=paddle.nn.initializer.Uniform(low=-0.5, high=0.5))
bias_attr = paddle.framework.ParamAttr(
>>> bias_attr = paddle.framework.ParamAttr(
name="linear_bias",
... name="linear_bias",
initializer=paddle.nn.initializer.Uniform(low=-0.5, high=0.5))
... initializer=paddle.nn.initializer.Uniform(low=-0.5, high=0.5))
linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr)
>>> linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr)
# linear.weight: [[-0.46245047 0.05260676]
>>> print(linear.weight)
# [ 0.38054508 0.29169726]]
Parameter containing:
# linear.bias: [-0.2734719 0.23939109]
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=False,
[[-0.48212373, 0.26492310],
res = linear(data)
[ 0.17605734, -0.45379421]])
# res: [[[-0.3553773 0.5836951]]
# [[-0.3553773 0.5836951]]
>>> print(linear.bias)
# [[-0.3553773 0.5836951]]]
Parameter containing:
Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=False,
[-0.11236754, 0.46462214])
>>> res = linear(data)
>>> print(res)
Tensor(shape=[3, 1, 2], dtype=float32, place=Place(cpu), stop_gradient=False,
[[[-0.41843393, 0.27575102]],
[[-0.41843393, 0.27575102]],
[[-0.41843393, 0.27575102]]])
"""
"""
def
__init__
(
self
,
low
=-
1.0
,
high
=
1.0
,
name
=
None
):
def
__init__
(
self
,
low
=-
1.0
,
high
=
1.0
,
name
=
None
):
...
...
python/paddle/nn/initializer/xavier.py
浏览文件 @
0a15b0db
...
@@ -214,24 +214,33 @@ class XavierNormal(XavierInitializer):
...
@@ -214,24 +214,33 @@ class XavierNormal(XavierInitializer):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
>>> paddle.seed(1)
data = paddle.ones(shape=[3, 1, 2], dtype='float32')
>>> data = paddle.ones(shape=[3, 1, 2], dtype='float32')
weight_attr = paddle.framework.ParamAttr(
>>> weight_attr = paddle.framework.ParamAttr(
name="linear_weight",
... name="linear_weight",
initializer=paddle.nn.initializer.XavierNormal())
... initializer=paddle.nn.initializer.XavierNormal())
bias_attr = paddle.framework.ParamAttr(
>>> bias_attr = paddle.framework.ParamAttr(
name="linear_bias",
... name="linear_bias",
initializer=paddle.nn.initializer.XavierNormal())
... initializer=paddle.nn.initializer.XavierNormal())
linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr)
>>> linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr)
# inear.weight: [[ 0.06910077 -0.18103665]
>>> print(linear.weight)
# [-0.02546741 -1.0402188 ]]
Parameter containing:
# linear.bias: [-0.5012929 0.12418364]
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=False,
[[-0.21607460, 0.08382989],
res = linear(data)
[ 0.29147008, -0.07049121]])
# res: [[[-0.4576595 -1.0970719]]
# [[-0.4576595 -1.0970719]]
>>> print(linear.bias)
# [[-0.4576595 -1.0970719]]]
Parameter containing:
Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=False,
[1.06076419, 0.87684733])
>>> res = linear(data)
>>> print(res)
Tensor(shape=[3, 1, 2], dtype=float32, place=Place(cpu), stop_gradient=False,
[[[1.13615966, 0.89018601]],
[[1.13615966, 0.89018601]],
[[1.13615966, 0.89018601]]])
"""
"""
def
__init__
(
self
,
fan_in
=
None
,
fan_out
=
None
,
name
=
None
):
def
__init__
(
self
,
fan_in
=
None
,
fan_out
=
None
,
name
=
None
):
...
@@ -266,24 +275,32 @@ class XavierUniform(XavierInitializer):
...
@@ -266,24 +275,32 @@ class XavierUniform(XavierInitializer):
Examples:
Examples:
.. code-block:: python
.. code-block:: python
import paddle
>>> import paddle
>>> paddle.seed(1)
data = paddle.ones(shape=[3, 1, 2], dtype='float32')
>>> data = paddle.ones(shape=[3, 1, 2], dtype='float32')
weight_attr = paddle.framework.ParamAttr(
>>> weight_attr = paddle.framework.ParamAttr(
name="linear_weight",
... name="linear_weight",
initializer=paddle.nn.initializer.XavierUniform())
... initializer=paddle.nn.initializer.XavierUniform())
bias_attr = paddle.framework.ParamAttr(
>>> bias_attr = paddle.framework.ParamAttr(
name="linear_bias",
... name="linear_bias",
initializer=paddle.nn.initializer.XavierUniform())
... initializer=paddle.nn.initializer.XavierUniform())
linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr)
>>> linear = paddle.nn.Linear(2, 2, weight_attr=weight_attr, bias_attr=bias_attr)
# linear.weight: [[-0.04229349 -1.1248565 ]
>>> print(linear.weight)
# [-0.10789523 -0.5938053 ]]
Parameter containing:
# linear.bias: [ 1.1983747 -0.40201235]
Tensor(shape=[2, 2], dtype=float32, place=Place(cpu), stop_gradient=False,
[[-1.18095720, 0.64892638],
res = linear(data)
[ 0.43125069, -1.11156428]])
# res: [[[ 1.0481861 -2.1206741]]
>>> print(linear.bias)
# [[ 1.0481861 -2.1206741]]
Parameter containing:
# [[ 1.0481861 -2.1206741]]]
Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=False,
[-0.27524316, 1.13808715])
>>> res = linear(data)
>>> print(res)
Tensor(shape=[3, 1, 2], dtype=float32, place=Place(cpu), stop_gradient=False,
[[[-1.02494967, 0.67544925]],
[[-1.02494967, 0.67544925]],
[[-1.02494967, 0.67544925]]])
"""
"""
def
__init__
(
self
,
fan_in
=
None
,
fan_out
=
None
,
name
=
None
):
def
__init__
(
self
,
fan_in
=
None
,
fan_out
=
None
,
name
=
None
):
...
...
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