Skip to content

P
Paddle

PaddlePaddle / Paddle
大约 2 年 前同步成功

通知 2325
Star 20933
Fork 5424
P
Paddle
未验证 提交 84fe0454 编写于 作者: G gouzil 提交者： GitHub
浏览文件
操作

[xdoctest] No.44-47 and No.50-59 doc style (#55813)

上级 30a02d27
隐藏空白更改
内联 并排
Showing with 538 addition and 519 deletion
python/paddle/distribution/cauchy.py
浏览文件 @ 84fe0454
...@@ -39,20 +39,20 @@ class Cauchy(distribution.Distribution): ...@@ -39,20 +39,20 @@ class Cauchy(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Cauchy >>> from paddle.distribution import Cauchy
# init Cauchy with float >>> # init Cauchy with float
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.entropy()) >>> print(rv.entropy())
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 2.71334577) 2.71334577)
# init Cauchy with N-Dim tensor >>> # init Cauchy with N-Dim tensor
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.entropy()) >>> print(rv.entropy())
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [2.53102422, 3.22417140]) [2.53102422, 3.22417140])
""" """
def __init__(self, loc, scale, name=None): def __init__(self, loc, scale, name=None):
...@@ -114,32 +114,32 @@ class Cauchy(distribution.Distribution): ...@@ -114,32 +114,32 @@ class Cauchy(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Cauchy >>> from paddle.distribution import Cauchy
# init Cauchy with float >>> # init Cauchy with float
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.sample([10]).shape) >>> print(rv.sample([10]).shape)
# [10] [10]
# init Cauchy with 0-Dim tensor >>> # init Cauchy with 0-Dim tensor
rv = Cauchy(loc=paddle.full((), 0.1), scale=paddle.full((), 1.2)) >>> rv = Cauchy(loc=paddle.full((), 0.1), scale=paddle.full((), 1.2))
print(rv.sample([10]).shape) >>> print(rv.sample([10]).shape)
# [10] [10]
# init Cauchy with N-Dim tensor >>> # init Cauchy with N-Dim tensor
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.sample([10]).shape) >>> print(rv.sample([10]).shape)
# [10, 2] [10, 2]
# sample 2-Dim data >>> # sample 2-Dim data
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.sample([10, 2]).shape) >>> print(rv.sample([10, 2]).shape)
# [10, 2] [10, 2]
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.sample([10, 2]).shape) >>> print(rv.sample([10, 2]).shape)
# [10, 2, 2] [10, 2, 2]
""" """
name = name if name is not None else (self.name + '_sample') name = name if name is not None else (self.name + '_sample')
with paddle.no_grad(): with paddle.no_grad():
...@@ -159,32 +159,32 @@ class Cauchy(distribution.Distribution): ...@@ -159,32 +159,32 @@ class Cauchy(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Cauchy >>> from paddle.distribution import Cauchy
# init Cauchy with float >>> # init Cauchy with float
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.rsample([10]).shape) >>> print(rv.rsample([10]).shape)
# [10] [10]
# init Cauchy with 0-Dim tensor >>> # init Cauchy with 0-Dim tensor
rv = Cauchy(loc=paddle.full((), 0.1), scale=paddle.full((), 1.2)) >>> rv = Cauchy(loc=paddle.full((), 0.1), scale=paddle.full((), 1.2))
print(rv.rsample([10]).shape) >>> print(rv.rsample([10]).shape)
# [10] [10]
# init Cauchy with N-Dim tensor >>> # init Cauchy with N-Dim tensor
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.rsample([10]).shape) >>> print(rv.rsample([10]).shape)
# [10, 2] [10, 2]
# sample 2-Dim data >>> # sample 2-Dim data
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.rsample([10, 2]).shape) >>> print(rv.rsample([10, 2]).shape)
# [10, 2] [10, 2]
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.rsample([10, 2]).shape) >>> print(rv.rsample([10, 2]).shape)
# [10, 2, 2] [10, 2, 2]
""" """
name = name if name is not None else (self.name + '_rsample') name = name if name is not None else (self.name + '_rsample')
...@@ -222,32 +222,32 @@ class Cauchy(distribution.Distribution): ...@@ -222,32 +222,32 @@ class Cauchy(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Cauchy >>> from paddle.distribution import Cauchy
# init Cauchy with float >>> # init Cauchy with float
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.prob(paddle.to_tensor(1.5))) >>> print(rv.prob(paddle.to_tensor(1.5)))
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.11234467) 0.11234467)
# broadcast to value >>> # broadcast to value
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.prob(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.prob(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.11234467, 0.01444674]) [0.11234467, 0.01444674])
# init Cauchy with N-Dim tensor >>> # init Cauchy with N-Dim tensor
rv = Cauchy(loc=paddle.to_tensor([0.1, 0.1]), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor([0.1, 0.1]), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.prob(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.prob(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.10753712, 0.02195240]) [0.10753712, 0.02195240])
# init Cauchy with N-Dim tensor with broadcast >>> # init Cauchy with N-Dim tensor with broadcast
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.prob(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.prob(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.10753712, 0.02195240]) [0.10753712, 0.02195240])
""" """
name = self.name + '_prob' name = self.name + '_prob'
...@@ -271,32 +271,32 @@ class Cauchy(distribution.Distribution): ...@@ -271,32 +271,32 @@ class Cauchy(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Cauchy >>> from paddle.distribution import Cauchy
# init Cauchy with float >>> # init Cauchy with float
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.log_prob(paddle.to_tensor(1.5))) >>> print(rv.log_prob(paddle.to_tensor(1.5)))
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# -2.18618369) -2.18618369)
# broadcast to value >>> # broadcast to value
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.log_prob(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.log_prob(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [-2.18618369, -4.23728657]) [-2.18618369, -4.23728657])
# init Cauchy with N-Dim tensor >>> # init Cauchy with N-Dim tensor
rv = Cauchy(loc=paddle.to_tensor([0.1, 0.1]), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor([0.1, 0.1]), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.log_prob(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.log_prob(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [-2.22991920, -3.81887865]) [-2.22991920, -3.81887865])
# init Cauchy with N-Dim tensor with broadcast >>> # init Cauchy with N-Dim tensor with broadcast
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.log_prob(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.log_prob(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [-2.22991920, -3.81887865]) [-2.22991920, -3.81887865])
""" """
name = self.name + '_log_prob' name = self.name + '_log_prob'
...@@ -338,32 +338,32 @@ class Cauchy(distribution.Distribution): ...@@ -338,32 +338,32 @@ class Cauchy(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Cauchy >>> from paddle.distribution import Cauchy
# init Cauchy with float >>> # init Cauchy with float
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.cdf(paddle.to_tensor(1.5))) >>> print(rv.cdf(paddle.to_tensor(1.5)))
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.77443725) 0.77443725)
# broadcast to value >>> # broadcast to value
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.cdf(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.cdf(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.77443725, 0.92502367]) [0.77443725, 0.92502367])
# init Cauchy with N-Dim tensor >>> # init Cauchy with N-Dim tensor
rv = Cauchy(loc=paddle.to_tensor([0.1, 0.1]), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor([0.1, 0.1]), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.cdf(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.cdf(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.80256844, 0.87888104]) [0.80256844, 0.87888104])
# init Cauchy with N-Dim tensor with broadcast >>> # init Cauchy with N-Dim tensor with broadcast
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.cdf(paddle.to_tensor([1.5, 5.1]))) >>> print(rv.cdf(paddle.to_tensor([1.5, 5.1])))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.80256844, 0.87888104]) [0.80256844, 0.87888104])
""" """
name = self.name + '_cdf' name = self.name + '_cdf'
...@@ -399,20 +399,20 @@ class Cauchy(distribution.Distribution): ...@@ -399,20 +399,20 @@ class Cauchy(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Cauchy >>> from paddle.distribution import Cauchy
# init Cauchy with float >>> # init Cauchy with float
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
print(rv.entropy()) >>> print(rv.entropy())
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 2.71334577) 2.71334577)
# init Cauchy with N-Dim tensor >>> # init Cauchy with N-Dim tensor
rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0])) >>> rv = Cauchy(loc=paddle.to_tensor(0.1), scale=paddle.to_tensor([1.0, 2.0]))
print(rv.entropy()) >>> print(rv.entropy())
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [2.53102422, 3.22417140]) [2.53102422, 3.22417140])
""" """
name = self.name + '_entropy' name = self.name + '_entropy'
...@@ -438,14 +438,14 @@ class Cauchy(distribution.Distribution): ...@@ -438,14 +438,14 @@ class Cauchy(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Cauchy >>> from paddle.distribution import Cauchy
rv = Cauchy(loc=0.1, scale=1.2) >>> rv = Cauchy(loc=0.1, scale=1.2)
rv_other = Cauchy(loc=paddle.to_tensor(1.2), scale=paddle.to_tensor([2.3, 3.4])) >>> rv_other = Cauchy(loc=paddle.to_tensor(1.2), scale=paddle.to_tensor([2.3, 3.4]))
print(rv.kl_divergence(rv_other)) >>> print(rv.kl_divergence(rv_other))
# Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.19819736, 0.31532931]) [0.19819736, 0.31532931])
""" """
name = self.name + '_kl_divergence' name = self.name + '_kl_divergence'
......
python/paddle/distribution/independent.py
浏览文件 @ 84fe0454
...@@ -31,21 +31,21 @@ class Independent(distribution.Distribution): ...@@ -31,21 +31,21 @@ class Independent(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import independent >>> from paddle.distribution import independent
beta = paddle.distribution.Beta(paddle.to_tensor([0.5, 0.5]), paddle.to_tensor([0.5, 0.5])) >>> beta = paddle.distribution.Beta(paddle.to_tensor([0.5, 0.5]), paddle.to_tensor([0.5, 0.5]))
print(beta.batch_shape, beta.event_shape) >>> print(beta.batch_shape, beta.event_shape)
# (2,) () (2,) ()
print(beta.log_prob(paddle.to_tensor(0.2))) >>> print(beta.log_prob(paddle.to_tensor(0.2)))
# Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True,
# [-0.22843921, -0.22843921]) [-0.22843921, -0.22843921])
reinterpreted_beta = independent.Independent(beta, 1) >>> reinterpreted_beta = independent.Independent(beta, 1)
print(reinterpreted_beta.batch_shape, reinterpreted_beta.event_shape) >>> print(reinterpreted_beta.batch_shape, reinterpreted_beta.event_shape)
# () (2,) () (2,)
print(reinterpreted_beta.log_prob(paddle.to_tensor([0.2, 0.2]))) >>> print(reinterpreted_beta.log_prob(paddle.to_tensor([0.2, 0.2])))
# Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True,
# -0.45687842) -0.45687842)
""" """
def __init__(self, base, reinterpreted_batch_rank): def __init__(self, base, reinterpreted_batch_rank):
......
python/paddle/distribution/kl.py
浏览文件 @ 84fe0454
...@@ -53,14 +53,14 @@ def kl_divergence(p, q): ...@@ -53,14 +53,14 @@ def kl_divergence(p, q):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
p = paddle.distribution.Beta(alpha=0.5, beta=0.5) >>> p = paddle.distribution.Beta(alpha=0.5, beta=0.5)
q = paddle.distribution.Beta(alpha=0.3, beta=0.7) >>> q = paddle.distribution.Beta(alpha=0.3, beta=0.7)
print(paddle.distribution.kl_divergence(p, q)) >>> print(paddle.distribution.kl_divergence(p, q))
# Tensor(shape=[], dtype=float32, place=CUDAPlace(0), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.21193528) 0.21193528)
""" """
return _dispatch(type(p), type(q))(p, q) return _dispatch(type(p), type(q))(p, q)
...@@ -82,11 +82,11 @@ def register_kl(cls_p, cls_q): ...@@ -82,11 +82,11 @@ def register_kl(cls_p, cls_q):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
@paddle.distribution.register_kl(paddle.distribution.Beta, paddle.distribution.Beta) >>> @paddle.distribution.register_kl(paddle.distribution.Beta, paddle.distribution.Beta)
def kl_beta_beta(): >>> def kl_beta_beta():
pass # insert implementation here ... pass # insert implementation here
""" """
if not issubclass(cls_p, Distribution) or not issubclass( if not issubclass(cls_p, Distribution) or not issubclass(
cls_q, Distribution cls_q, Distribution
......
python/paddle/distribution/laplace.py
浏览文件 @ 84fe0454
...@@ -44,12 +44,12 @@ class Laplace(distribution.Distribution): ...@@ -44,12 +44,12 @@ class Laplace(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
>>> paddle.seed(2023)
m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0)) >>> m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0))
m.sample() # Laplace distributed with loc=0, scale=1 >>> m.sample() # Laplace distributed with loc=0, scale=1
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 3.68546247) 1.31554604)
""" """
...@@ -173,13 +173,13 @@ class Laplace(distribution.Distribution): ...@@ -173,13 +173,13 @@ class Laplace(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0)) >>> m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0))
value = paddle.to_tensor(0.1) >>> value = paddle.to_tensor(0.1)
m.log_prob(value) >>> m.log_prob(value)
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# -0.79314721) -0.79314721)
""" """
loc, scale, value = self._validate_value(value) loc, scale, value = self._validate_value(value)
...@@ -205,12 +205,12 @@ class Laplace(distribution.Distribution): ...@@ -205,12 +205,12 @@ class Laplace(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0)) >>> m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0))
m.entropy() >>> m.entropy()
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 1.69314718) 1.69314718)
""" """
return 1 + paddle.log(2 * self.scale) return 1 + paddle.log(2 * self.scale)
...@@ -236,13 +236,13 @@ class Laplace(distribution.Distribution): ...@@ -236,13 +236,13 @@ class Laplace(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0)) >>> m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0))
value = paddle.to_tensor(0.1) >>> value = paddle.to_tensor(0.1)
m.cdf(value) >>> m.cdf(value)
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.54758132) 0.54758132)
""" """
loc, scale, value = self._validate_value(value) loc, scale, value = self._validate_value(value)
iterm = ( iterm = (
...@@ -275,13 +275,12 @@ class Laplace(distribution.Distribution): ...@@ -275,13 +275,12 @@ class Laplace(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
>>> m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0))
m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0)) >>> value = paddle.to_tensor(0.1)
value = paddle.to_tensor(0.1) >>> m.icdf(value)
m.icdf(value) Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, -1.60943794)
# -1.60943794)
""" """
loc, scale, value = self._validate_value(value) loc, scale, value = self._validate_value(value)
term = value - 0.5 term = value - 0.5
...@@ -300,12 +299,12 @@ class Laplace(distribution.Distribution): ...@@ -300,12 +299,12 @@ class Laplace(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
>>> paddle.seed(2023)
m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0)) >>> m = paddle.distribution.Laplace(paddle.to_tensor(0.0), paddle.to_tensor(1.0))
m.sample() # Laplace distributed with loc=0, scale=1 >>> m.sample() # Laplace distributed with loc=0, scale=1
# Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 3.68546247) 1.31554604)
""" """
shape = shape if isinstance(shape, tuple) else tuple(shape) shape = shape if isinstance(shape, tuple) else tuple(shape)
with paddle.no_grad(): with paddle.no_grad():
...@@ -323,12 +322,12 @@ class Laplace(distribution.Distribution): ...@@ -323,12 +322,12 @@ class Laplace(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
>>> paddle.seed(2023)
m = paddle.distribution.Laplace(paddle.to_tensor([0.0]), paddle.to_tensor([1.0])) >>> m = paddle.distribution.Laplace(paddle.to_tensor([0.0]), paddle.to_tensor([1.0]))
m.rsample((1,)) # Laplace distributed with loc=0, scale=1 >>> m.rsample((1,)) # Laplace distributed with loc=0, scale=1
# Tensor(shape=[1, 1], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[1, 1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[0.04337667]]) [[1.31554604]])
""" """
eps = self._get_eps() eps = self._get_eps()
...@@ -395,13 +394,13 @@ class Laplace(distribution.Distribution): ...@@ -395,13 +394,13 @@ class Laplace(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
m1 = paddle.distribution.Laplace(paddle.to_tensor([0.0]), paddle.to_tensor([1.0])) >>> m1 = paddle.distribution.Laplace(paddle.to_tensor([0.0]), paddle.to_tensor([1.0]))
m2 = paddle.distribution.Laplace(paddle.to_tensor([1.0]), paddle.to_tensor([0.5])) >>> m2 = paddle.distribution.Laplace(paddle.to_tensor([1.0]), paddle.to_tensor([0.5]))
m1.kl_divergence(m2) >>> m1.kl_divergence(m2)
# Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True, Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1.04261160]) [1.04261160])
""" """
var_ratio = other.scale / self.scale var_ratio = other.scale / self.scale
......
python/paddle/distribution/lognormal.py
浏览文件 @ 84fe0454
...@@ -49,36 +49,44 @@ class LogNormal(TransformedDistribution): ...@@ -49,36 +49,44 @@ class LogNormal(TransformedDistribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import LogNormal >>> from paddle.distribution import LogNormal
# Define a single scalar LogNormal distribution. >>> # Define a single scalar LogNormal distribution.
dist = LogNormal(loc=0., scale=3.) >>> dist = LogNormal(loc=0., scale=3.)
# Define a batch of two scalar valued LogNormals. >>> # Define a batch of two scalar valued LogNormals.
# The underlying Normal of first has mean 1 and standard deviation 11, the underlying Normal of second 2 and 22. >>> # The underlying Normal of first has mean 1 and standard deviation 11, the underlying Normal of second 2 and 22.
dist = LogNormal(loc=[1., 2.], scale=[11., 22.]) >>> dist = LogNormal(loc=[1., 2.], scale=[11., 22.])
# Get 3 samples, returning a 3 x 2 tensor. >>> # Get 3 samples, returning a 3 x 2 tensor.
dist.sample((3, )) >>> dist.sample((3, ))
# Define a batch of two scalar valued LogNormals. >>> # Define a batch of two scalar valued LogNormals.
# Their underlying Normal have mean 1, but different standard deviations. >>> # Their underlying Normal have mean 1, but different standard deviations.
dist = LogNormal(loc=1., scale=[11., 22.]) >>> dist = LogNormal(loc=1., scale=[11., 22.])
# Complete example >>> # Complete example
value_tensor = paddle.to_tensor([0.8], dtype="float32") >>> value_tensor = paddle.to_tensor([0.8], dtype="float32")
lognormal_a = LogNormal([0.], [1.]) >>> lognormal_a = LogNormal([0.], [1.])
lognormal_b = LogNormal([0.5], [2.]) >>> lognormal_b = LogNormal([0.5], [2.])
sample = lognormal_a.sample((2, )) >>> sample = lognormal_a.sample((2, ))
# a random tensor created by lognormal distribution with shape: [2, 1] >>> # a random tensor created by lognormal distribution with shape: [2, 1]
entropy = lognormal_a.entropy() >>> entropy = lognormal_a.entropy()
# [1.4189385] with shape: [1] >>> print(entropy)
lp = lognormal_a.log_prob(value_tensor) Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [-0.72069150] with shape: [1] [1.41893852])
p = lognormal_a.probs(value_tensor) >>> lp = lognormal_a.log_prob(value_tensor)
# [0.48641577] with shape: [1] >>> print(lp)
kl = lognormal_a.kl_divergence(lognormal_b) Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.34939718] with shape: [1] [-0.72069150])
>>> p = lognormal_a.probs(value_tensor)
>>> print(p)
Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
[0.48641577])
>>> kl = lognormal_a.kl_divergence(lognormal_b)
>>> print(kl)
Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
[0.34939718])
""" """
def __init__(self, loc, scale): def __init__(self, loc, scale):
......
python/paddle/distribution/multinomial.py
浏览文件 @ 84fe0454
...@@ -53,18 +53,17 @@ class Multinomial(distribution.Distribution): ...@@ -53,18 +53,17 @@ class Multinomial(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
>>> paddle.seed(2023)
multinomial = paddle.distribution.Multinomial(10, paddle.to_tensor([0.2, 0.3, 0.5])) >>> multinomial = paddle.distribution.Multinomial(10, paddle.to_tensor([0.2, 0.3, 0.5]))
print(multinomial.sample((2, 3))) >>> print(multinomial.sample((2, 3)))
# Tensor(shape=[2, 3, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2, 3, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[[1., 4., 5.], [[[1., 5., 4.],
# [0., 2., 8.], [0., 4., 6.],
# [2., 4., 4.]], [1., 3., 6.]],
[[2., 2., 6.],
# [[1., 6., 3.], [0., 6., 4.],
# [3., 3., 4.], [3., 3., 4.]]])
# [3., 4., 3.]]])
""" """
def __init__(self, total_count, probs): def __init__(self, total_count, probs):
......
python/paddle/distribution/normal.py
浏览文件 @ 84fe0454
...@@ -54,36 +54,44 @@ class Normal(distribution.Distribution): ...@@ -54,36 +54,44 @@ class Normal(distribution.Distribution):
Examples: Examples:
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import Normal >>> from paddle.distribution import Normal
# Define a single scalar Normal distribution. >>> # Define a single scalar Normal distribution.
dist = Normal(loc=0., scale=3.) >>> dist = Normal(loc=0., scale=3.)
# Define a batch of two scalar valued Normals. >>> # Define a batch of two scalar valued Normals.
# The first has mean 1 and standard deviation 11, the second 2 and 22. >>> # The first has mean 1 and standard deviation 11, the second 2 and 22.
dist = Normal(loc=[1., 2.], scale=[11., 22.]) >>> dist = Normal(loc=[1., 2.], scale=[11., 22.])
# Get 3 samples, returning a 3 x 2 tensor. >>> # Get 3 samples, returning a 3 x 2 tensor.
dist.sample([3]) >>> dist.sample([3])
# Define a batch of two scalar valued Normals. >>> # Define a batch of two scalar valued Normals.
# Both have mean 1, but different standard deviations. >>> # Both have mean 1, but different standard deviations.
dist = Normal(loc=1., scale=[11., 22.]) >>> dist = Normal(loc=1., scale=[11., 22.])
# Complete example >>> # Complete example
value_tensor = paddle.to_tensor([0.8], dtype="float32") >>> value_tensor = paddle.to_tensor([0.8], dtype="float32")
normal_a = Normal([0.], [1.]) >>> normal_a = Normal([0.], [1.])
normal_b = Normal([0.5], [2.]) >>> normal_b = Normal([0.5], [2.])
sample = normal_a.sample([2]) >>> sample = normal_a.sample([2])
# a random tensor created by normal distribution with shape: [2, 1] >>> # a random tensor created by normal distribution with shape: [2, 1]
entropy = normal_a.entropy() >>> entropy = normal_a.entropy()
# [1.4189385] with shape: [1] >>> print(entropy)
lp = normal_a.log_prob(value_tensor) Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [-1.2389386] with shape: [1] [1.41893852])
p = normal_a.probs(value_tensor) >>> lp = normal_a.log_prob(value_tensor)
# [0.28969154] with shape: [1] >>> print(lp)
kl = normal_a.kl_divergence(normal_b) Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.34939718] with shape: [1] [-1.23893857])
>>> p = normal_a.probs(value_tensor)
>>> print(p)
Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
[0.28969154])
>>> kl = normal_a.kl_divergence(normal_b)
>>> print(kl)
Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
[0.34939718])
""" """
def __init__(self, loc, scale, name=None): def __init__(self, loc, scale, name=None):
......
python/paddle/distribution/transform.py
浏览文件 @ 84fe0454
...@@ -360,34 +360,34 @@ class AbsTransform(Transform): ...@@ -360,34 +360,34 @@ class AbsTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
abs = paddle.distribution.AbsTransform() >>> abs = paddle.distribution.AbsTransform()
print(abs.forward(paddle.to_tensor([-1., 0., 1.]))) >>> print(abs.forward(paddle.to_tensor([-1., 0., 1.])))
# Tensor(shape=[3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1., 0., 1.]) [1., 0., 1.])
print(abs.inverse(paddle.to_tensor([1.]))) >>> print(abs.inverse(paddle.to_tensor([1.])))
# (Tensor(shape=[1], dtype=float32, place=Place(gpu:0), stop_gradient=True, (Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [-1.]), Tensor(shape=[1], dtype=float32, place=Place(gpu:0), stop_gradient=True, [-1.]), Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1.])) [1.]))
# The |dX/dY| is constant 1. So Log|dX/dY| == 0 >>> # The |dX/dY| is constant 1. So Log|dX/dY| == 0
print(abs.inverse_log_det_jacobian(paddle.to_tensor(1.))) >>> print(abs.inverse_log_det_jacobian(paddle.to_tensor(1.)))
# (Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, (Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.), Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, 0.), Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.)) 0.))
#Special case handling of 0. >>> #Special case handling of 0.
print(abs.inverse(paddle.to_tensor([0.]))) >>> print(abs.inverse(paddle.to_tensor([0.])))
# (Tensor(shape=[1], dtype=float32, place=Place(gpu:0), stop_gradient=True, (Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.]), Tensor(shape=[1], dtype=float32, place=Place(gpu:0), stop_gradient=True, [0.]), Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.])) [0.]))
print(abs.inverse_log_det_jacobian(paddle.to_tensor(0.))) >>> print(abs.inverse_log_det_jacobian(paddle.to_tensor(0.)))
# (Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, (Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.), Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, 0.), Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.)) 0.))
""" """
_type = Type.SURJECTION _type = Type.SURJECTION
...@@ -423,20 +423,20 @@ class AffineTransform(Transform): ...@@ -423,20 +423,20 @@ class AffineTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.to_tensor([1., 2.]) >>> x = paddle.to_tensor([1., 2.])
affine = paddle.distribution.AffineTransform(paddle.to_tensor(0.), paddle.to_tensor(1.)) >>> affine = paddle.distribution.AffineTransform(paddle.to_tensor(0.), paddle.to_tensor(1.))
print(affine.forward(x)) >>> print(affine.forward(x))
# Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1., 2.]) [1., 2.])
print(affine.inverse(affine.forward(x))) >>> print(affine.inverse(affine.forward(x)))
# Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1., 2.]) [1., 2.])
print(affine.forward_log_det_jacobian(x)) >>> print(affine.forward_log_det_jacobian(x))
# Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.) 0.)
""" """
_type = Type.BIJECTION _type = Type.BIJECTION
...@@ -503,28 +503,28 @@ class ChainTransform(Transform): ...@@ -503,28 +503,28 @@ class ChainTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.to_tensor([0., 1., 2., 3.]) >>> x = paddle.to_tensor([0., 1., 2., 3.])
chain = paddle.distribution.ChainTransform(( >>> chain = paddle.distribution.ChainTransform((
paddle.distribution.AffineTransform( ... paddle.distribution.AffineTransform(
paddle.to_tensor(0.), paddle.to_tensor(1.)), ... paddle.to_tensor(0.), paddle.to_tensor(1.)),
paddle.distribution.ExpTransform() ... paddle.distribution.ExpTransform()
)) >>> ))
print(chain.forward(x)) >>> print(chain.forward(x))
# Tensor(shape=[4], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1. , 2.71828175 , 7.38905621 , 20.08553696]) [1. , 2.71828175 , 7.38905621 , 20.08553696])
print(chain.inverse(chain.forward(x))) >>> print(chain.inverse(chain.forward(x)))
# Tensor(shape=[4], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0., 1., 2., 3.]) [0., 1., 2., 3.])
print(chain.forward_log_det_jacobian(x)) >>> print(chain.forward_log_det_jacobian(x))
# Tensor(shape=[4], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0., 1., 2., 3.]) [0., 1., 2., 3.])
print(chain.inverse_log_det_jacobian(chain.forward(x))) >>> print(chain.inverse_log_det_jacobian(chain.forward(x)))
# Tensor(shape=[4], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [ 0., -1., -2., -3.]) [ 0., -1., -2., -3.])
""" """
def __init__(self, transforms): def __init__(self, transforms):
...@@ -625,24 +625,24 @@ class ExpTransform(Transform): ...@@ -625,24 +625,24 @@ class ExpTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
exp = paddle.distribution.ExpTransform() >>> exp = paddle.distribution.ExpTransform()
print(exp.forward(paddle.to_tensor([1., 2., 3.]))) >>> print(exp.forward(paddle.to_tensor([1., 2., 3.])))
# Tensor(shape=[3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [2.71828175 , 7.38905621 , 20.08553696]) [2.71828175 , 7.38905621 , 20.08553696])
print(exp.inverse(paddle.to_tensor([1., 2., 3.]))) >>> print(exp.inverse(paddle.to_tensor([1., 2., 3.])))
# Tensor(shape=[3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0. , 0.69314718, 1.09861231]) [0. , 0.69314718, 1.09861231])
print(exp.forward_log_det_jacobian(paddle.to_tensor([1., 2., 3.]))) >>> print(exp.forward_log_det_jacobian(paddle.to_tensor([1., 2., 3.])))
# Tensor(shape=[3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1., 2., 3.]) [1., 2., 3.])
print(exp.inverse_log_det_jacobian(paddle.to_tensor([1., 2., 3.]))) >>> print(exp.inverse_log_det_jacobian(paddle.to_tensor([1., 2., 3.])))
# Tensor(shape=[3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [ 0. , -0.69314718, -1.09861231]) [ 0. , -0.69314718, -1.09861231])
""" """
_type = Type.BIJECTION _type = Type.BIJECTION
...@@ -695,20 +695,20 @@ class IndependentTransform(Transform): ...@@ -695,20 +695,20 @@ class IndependentTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.to_tensor([[1., 2., 3.], [4., 5., 6.]]) >>> x = paddle.to_tensor([[1., 2., 3.], [4., 5., 6.]])
# Exponential transform with event_rank = 1 >>> # Exponential transform with event_rank = 1
multi_exp = paddle.distribution.IndependentTransform( >>> multi_exp = paddle.distribution.IndependentTransform(
paddle.distribution.ExpTransform(), 1) ... paddle.distribution.ExpTransform(), 1)
print(multi_exp.forward(x)) >>> print(multi_exp.forward(x))
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[2.71828175 , 7.38905621 , 20.08553696 ], [[2.71828175 , 7.38905621 , 20.08553696 ],
# [54.59814835 , 148.41316223, 403.42880249]]) [54.59814835 , 148.41316223, 403.42880249]])
print(multi_exp.forward_log_det_jacobian(x)) >>> print(multi_exp.forward_log_det_jacobian(x))
# Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [6. , 15.]) [6. , 15.])
""" """
def __init__(self, base, reinterpreted_batch_rank): def __init__(self, base, reinterpreted_batch_rank):
...@@ -773,20 +773,20 @@ class PowerTransform(Transform): ...@@ -773,20 +773,20 @@ class PowerTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.to_tensor([1., 2.]) >>> x = paddle.to_tensor([1., 2.])
power = paddle.distribution.PowerTransform(paddle.to_tensor(2.)) >>> power = paddle.distribution.PowerTransform(paddle.to_tensor(2.))
print(power.forward(x)) >>> print(power.forward(x))
# Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1., 4.]) [1., 4.])
print(power.inverse(power.forward(x))) >>> print(power.inverse(power.forward(x)))
# Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [1., 2.]) [1., 2.])
print(power.forward_log_det_jacobian(x)) >>> print(power.forward_log_det_jacobian(x))
# Tensor(shape=[2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.69314718, 1.38629436]) [0.69314718, 1.38629436])
""" """
_type = Type.BIJECTION _type = Type.BIJECTION
...@@ -840,24 +840,24 @@ class ReshapeTransform(Transform): ...@@ -840,24 +840,24 @@ class ReshapeTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.ones((1,2,3)) >>> x = paddle.ones((1,2,3))
reshape_transform = paddle.distribution.ReshapeTransform((2, 3), (3, 2)) >>> reshape_transform = paddle.distribution.ReshapeTransform((2, 3), (3, 2))
print(reshape_transform.forward_shape((1,2,3))) >>> print(reshape_transform.forward_shape((1,2,3)))
# (5, 2, 6) (1, 3, 2)
print(reshape_transform.forward(x)) >>> print(reshape_transform.forward(x))
# Tensor(shape=[1, 3, 2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[1, 3, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[[1., 1.], [[[1., 1.],
# [1., 1.], [1., 1.],
# [1., 1.]]]) [1., 1.]]])
print(reshape_transform.inverse(reshape_transform.forward(x))) >>> print(reshape_transform.inverse(reshape_transform.forward(x)))
# Tensor(shape=[1, 2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[1, 2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[[1., 1., 1.], [[[1., 1., 1.],
# [1., 1., 1.]]]) [1., 1., 1.]]])
print(reshape_transform.forward_log_det_jacobian(x)) >>> print(reshape_transform.forward_log_det_jacobian(x))
# Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=True,
# 0.) [0.])
""" """
_type = Type.BIJECTION _type = Type.BIJECTION
...@@ -956,22 +956,22 @@ class SigmoidTransform(Transform): ...@@ -956,22 +956,22 @@ class SigmoidTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.ones((2,3)) >>> x = paddle.ones((2,3))
t = paddle.distribution.SigmoidTransform() >>> t = paddle.distribution.SigmoidTransform()
print(t.forward(x)) >>> print(t.forward(x))
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[0.73105860, 0.73105860, 0.73105860], [[0.73105860, 0.73105860, 0.73105860],
# [0.73105860, 0.73105860, 0.73105860]]) [0.73105860, 0.73105860, 0.73105860]])
print(t.inverse(t.forward(x))) >>> print(t.inverse(t.forward(x)))
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[1.00000012, 1.00000012, 1.00000012], [[1.00000012, 1.00000012, 1.00000012],
# [1.00000012, 1.00000012, 1.00000012]]) [1.00000012, 1.00000012, 1.00000012]])
print(t.forward_log_det_jacobian(x)) >>> print(t.forward_log_det_jacobian(x))
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[-1.62652326, -1.62652326, -1.62652326], [[-1.62652326, -1.62652326, -1.62652326],
# [-1.62652326, -1.62652326, -1.62652326]]) [-1.62652326, -1.62652326, -1.62652326]])
""" """
@property @property
...@@ -1003,18 +1003,18 @@ class SoftmaxTransform(Transform): ...@@ -1003,18 +1003,18 @@ class SoftmaxTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.ones((2,3)) >>> x = paddle.ones((2,3))
t = paddle.distribution.SoftmaxTransform() >>> t = paddle.distribution.SoftmaxTransform()
print(t.forward(x)) >>> print(t.forward(x))
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[0.33333334, 0.33333334, 0.33333334], [[0.33333334, 0.33333334, 0.33333334],
# [0.33333334, 0.33333334, 0.33333334]]) [0.33333334, 0.33333334, 0.33333334]])
print(t.inverse(t.forward(x))) >>> print(t.inverse(t.forward(x)))
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[-1.09861231, -1.09861231, -1.09861231], [[-1.09861231, -1.09861231, -1.09861231],
# [-1.09861231, -1.09861231, -1.09861231]]) [-1.09861231, -1.09861231, -1.09861231]])
""" """
_type = Type.OTHER _type = Type.OTHER
...@@ -1061,32 +1061,32 @@ class StackTransform(Transform): ...@@ -1061,32 +1061,32 @@ class StackTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.stack( >>> x = paddle.stack(
(paddle.to_tensor([1., 2., 3.]), paddle.to_tensor([1, 2., 3.])), 1) ... (paddle.to_tensor([1., 2., 3.]), paddle.to_tensor([1, 2., 3.])), 1)
t = paddle.distribution.StackTransform( >>> t = paddle.distribution.StackTransform(
(paddle.distribution.ExpTransform(), ... (paddle.distribution.ExpTransform(),
paddle.distribution.PowerTransform(paddle.to_tensor(2.))), ... paddle.distribution.PowerTransform(paddle.to_tensor(2.))),
1 ... 1
) >>> )
print(t.forward(x)) >>> print(t.forward(x))
# Tensor(shape=[3, 2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[2.71828175 , 1. ], [[2.71828175 , 1. ],
# [7.38905621 , 4. ], [7.38905621 , 4. ],
# [20.08553696, 9. ]]) [20.08553696, 9. ]])
print(t.inverse(t.forward(x))) >>> print(t.inverse(t.forward(x)))
# Tensor(shape=[3, 2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[1., 1.], [[1., 1.],
# [2., 2.], [2., 2.],
# [3., 3.]]) [3., 3.]])
print(t.forward_log_det_jacobian(x)) >>> print(t.forward_log_det_jacobian(x))
# Tensor(shape=[3, 2], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3, 2], dtype=float32, place=Place(cpu), stop_gradient=True,
# [[1. , 0.69314718], [[1. , 0.69314718],
# [2. , 1.38629436], [2. , 1.38629436],
# [3. , 1.79175949]]) [3. , 1.79175949]])
""" """
def __init__(self, transforms, axis=0): def __init__(self, transforms, axis=0):
...@@ -1176,20 +1176,20 @@ class StickBreakingTransform(Transform): ...@@ -1176,20 +1176,20 @@ class StickBreakingTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
x = paddle.to_tensor([1.,2.,3.]) >>> x = paddle.to_tensor([1.,2.,3.])
t = paddle.distribution.StickBreakingTransform() >>> t = paddle.distribution.StickBreakingTransform()
print(t.forward(x)) >>> print(t.forward(x))
# Tensor(shape=[4], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[4], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.47536686, 0.41287899, 0.10645414, 0.00530004]) [0.47536686, 0.41287899, 0.10645414, 0.00530004])
print(t.inverse(t.forward(x))) >>> print(t.inverse(t.forward(x)))
# Tensor(shape=[3], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.99999988, 2. , 2.99999881]) [0.99999988, 2. , 2.99999881])
print(t.forward_log_det_jacobian(x)) >>> print(t.forward_log_det_jacobian(x))
# Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
# -9.10835075) -9.10835075)
""" """
_type = Type.BIJECTION _type = Type.BIJECTION
...@@ -1241,28 +1241,30 @@ class TanhTransform(Transform): ...@@ -1241,28 +1241,30 @@ class TanhTransform(Transform):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
tanh = paddle.distribution.TanhTransform() >>> tanh = paddle.distribution.TanhTransform()
x = paddle.to_tensor([[1., 2., 3.], [4., 5., 6.]]) >>> x = paddle.to_tensor([[1., 2., 3.], [4., 5., 6.]])
print(tanh.forward(x)) >>> # doctest: +SKIP('random sample')
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, >>> print(tanh.forward(x))
# [[0.76159418, 0.96402758, 0.99505478], Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [0.99932933, 0.99990922, 0.99998772]]) [[0.76159418, 0.96402758, 0.99505472],
print(tanh.inverse(tanh.forward(x))) [0.99932921, 0.99990916, 0.99998784]])
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, >>> print(tanh.inverse(tanh.forward(x)))
# [[1.00000012, 2. , 3.00000286], Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [4.00002146, 5.00009823, 6.00039864]]) [[1. , 2. , 2.99999666],
print(tanh.forward_log_det_jacobian(x)) [3.99993253, 4.99977016, 6.00527668]])
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, >>> print(tanh.forward_log_det_jacobian(x))
# [[-0.86756170 , -2.65000558 , -4.61865711 ], Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [-6.61437654 , -8.61379623 , -10.61371803]]) [[-0.86756170 , -2.65000558 , -4.61865711 ],
print(tanh.inverse_log_det_jacobian(tanh.forward(x))) [-6.61437654 , -8.61379623 , -10.61371803]])
# Tensor(shape=[2, 3], dtype=float32, place=Place(gpu:0), stop_gradient=True, >>> print(tanh.inverse_log_det_jacobian(tanh.forward(x)))
# [[0.86756176 , 2.65000558 , 4.61866283 ], Tensor(shape=[2, 3], dtype=float32, place=Place(cpu), stop_gradient=True,
# [6.61441946 , 8.61399269 , 10.61451530]]) [[0.86756176 , 2.65000558 , 4.61866283 ],
[6.61441946 , 8.61399269 , 10.61451530]])
>>> # doctest: -SKIP
""" """
_type = Type.BIJECTION _type = Type.BIJECTION
......
python/paddle/distribution/transformed_distribution.py
浏览文件 @ 84fe0454
...@@ -29,21 +29,24 @@ class TransformedDistribution(distribution.Distribution): ...@@ -29,21 +29,24 @@ class TransformedDistribution(distribution.Distribution):
.. code-block:: python .. code-block:: python
import paddle >>> import paddle
from paddle.distribution import transformed_distribution >>> paddle.seed(2023)
>>> from paddle.distribution import transformed_distribution
d = transformed_distribution.TransformedDistribution(
paddle.distribution.Normal(0., 1.), >>> d = transformed_distribution.TransformedDistribution(
[paddle.distribution.AffineTransform(paddle.to_tensor(1.), paddle.to_tensor(2.))] ... paddle.distribution.Normal(0., 1.),
) ... [paddle.distribution.AffineTransform(paddle.to_tensor(1.), paddle.to_tensor(2.))]
... )
print(d.sample([10]))
# Tensor(shape=[10], dtype=float32, place=Place(gpu:0), stop_gradient=True, >>> # doctest: +SKIP('random sample')
# [-0.10697651, 3.33609009, -0.86234951, 5.07457638, 0.75925219, >>> print(d.sample([10]))
# -4.17087793, 2.22579336, -0.93845034, 0.66054249, 1.50957513]) Tensor(shape=[10], dtype=float32, place=Place(cpu), stop_gradient=True,
print(d.log_prob(paddle.to_tensor(0.5))) [ 3.22699189, 1.12264419, 0.50283587, 1.83812487, -2.00740123,
# Tensor(shape=[], dtype=float32, place=Place(gpu:0), stop_gradient=True, -2.70338631, 1.26663208, 4.47909021, -0.11529565, 4.32719326])
# -1.64333570) >>> print(d.log_prob(paddle.to_tensor(0.5)))
Tensor(shape=[], dtype=float32, place=Place(cpu), stop_gradient=True,
-1.64333570)
>>> # doctest: -SKIP
""" """
def __init__(self, base, transforms): def __init__(self, base, transforms):
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或