forked from mindspore-Ecosystem/mindspore
Gitee
commit
106411e71e
|
|
@ -18,5 +18,5 @@ mindspore.get_grad
|
|||
identifier所指定的求导输入位置的索引所对应的梯度值，或者网络变量所对应的梯度值。
|
||||
|
||||
异常：
|
||||
- **ValueError** - 无法找到identifier所对应的梯度值。
|
||||
- **RuntimeError** - 无法找到identifier所对应的梯度值。
|
||||
- **TypeError** - 入参类型不符合要求。
|
||||
|
|
|
|||
|
|
@ -7,7 +7,7 @@ mindspore.nn.probability.distribution.StudentT
|
|||
连续随机分布，取值范围为 :math:`(-\inf, \inf)` ，概率密度函数为
|
||||
|
||||
.. math::
|
||||
f(x, \nu, \mu, \sigma) = (1 + y^2 / \nu)^(-0.5*(\nu + 1)) / Z
|
||||
f(x, \nu, \mu, \sigma) = (1 + y^2 / \nu)^{(-0.5*(\nu + 1))} / Z
|
||||
|
||||
|
||||
其中 :math:`y = (x - \mu) / \sigma`, :math:`Z = abs(\sigma) * \sqrt(\nu * \pi) * \Gamma(0.5 * \nu) / \Gamma(0.5 * (\nu + 1))`, :math:`\nu, \mu, \sigma` 为分别为StudentT分布的自由度，期望与标准差。
|
||||
|
|
|
|||
|
|
@ -0,0 +1,26 @@
|
|||
mindspore.ops.PSROIPooling
|
||||
==========================
|
||||
|
||||
.. py:class:: mindspore.ops.PSROIPooling(spatial_scale, group_size, output_dim)
|
||||
|
||||
对输入Tensor应用Position Sensitive ROI-Pooling。
|
||||
|
||||
参数：
|
||||
- **spatial_scale** (float) - 将框坐标映射到输入坐标的比例因子。例如，如果你的框定义在224x224的图像上，并且你的输入是112x112的特征图（由原始图像的0.5倍缩放产生），此时需要将其设置为0.5。
|
||||
- **group_size** (int) - 执行池化后输出的大小（以像素为单位），以（高度，宽度）的格式输出。
|
||||
- **output_dim** (int) - 执行池化后输出的维度。
|
||||
|
||||
输入：
|
||||
- **features** (Tensor) - 输入特征Tensor，其shape必须为 :math:`(N, C, H, W)` 。 各维度的值应满足： :math:`(C == output\_dim * group\_size * group\_size)` 。数据类型为float16或者float32。
|
||||
- **rois** (Tensor) - 其shape为 :math:`(batch, 5, rois_n)` ，数据类型为float16或者float32。第一个维度的batch为批处理大小。第二个维度的大小必须为5。第三维度rois_n是rois的数量。rois_n的值格式为：(index, x1, y1, x2, y2)。其中第一个元素是rois的索引。方框坐标格式为(x1、y1、x2、y2)，之后将把这些方框的选中的区域提取出来。区域坐标必须满足0 <= x1 < x2和0 <= y1 < y2。
|
||||
|
||||
输出：
|
||||
- **out** (Tensor) - 池化后的输出。其shape为 :math:`(rois.shape[0] * rois.shape[2], output\_dim, group\_size, group\_size)` 。
|
||||
|
||||
异常：
|
||||
- **TypeError** - `spatial_scale` 不是float类型。
|
||||
- **TypeError** - `group_size` 或者 `output_dim` 不是 int类型。
|
||||
- **TypeError** - `features` 或者 `rois` 不是Tensor。
|
||||
- **TypeError** - `rois` 数据类型不是float16或者float32。
|
||||
- **ValueError** - `features` 的shape不满足 :math:`(C == output\_dim * group\_size * group\_size)` 。
|
||||
- **ValueError** - `spatial_scale` 为负数。
|
||||
|
|
@ -6,7 +6,7 @@ mindspore.ops.rand_like
|
|||
返回一个Tensor，shape和dtype由输入决定，其元素为服从均匀分布的 :math:`[0, 1)` 区间的数字。
|
||||
|
||||
参数：
|
||||
- **x** (Tensor)：输入的Tensor，用来决定输出Tensor的shape和默认的dtype。
|
||||
- **x** (Tensor) - 输入的Tensor，用来决定输出Tensor的shape和默认的dtype。
|
||||
- **seed** (int，可选) - 随机种子，必须大于或等于0。默认值：None，值将取0。
|
||||
|
||||
关键字参数：
|
||||
|
|
|
|||
|
|
@ -31,7 +31,7 @@ class StudentT(Distribution):
|
|||
and the probability density function:
|
||||
|
||||
.. math::
|
||||
f(x, \nu, \mu, \sigma) = (1 + y^2 / \nu)^(-0.5*(\nu + 1)) / Z
|
||||
f(x, \nu, \mu, \sigma) = (1 + y^2 / \nu)^{(-0.5*(\nu + 1))} / Z
|
||||
|
||||
where :math:`y = (x-\mu)/\sigma`, :math:`Z = abs(\sigma)*\sqrt(\nu * \pi)*\Gamma(0.5 * \nu)/\Gamma(0.5*(\nu + 1))`,
|
||||
:math:`\nu, \mu, \sigma` are the degrees of freedom , mean and scale of the laplace distribution respectively.
|
||||
|
|
|
|||
|
|
@ -847,7 +847,7 @@ def rand(*size, dtype=None, seed=None):
|
|||
interval :math:`[0, 1)`.
|
||||
|
||||
Args:
|
||||
*size (Union[int, tuple(int), list(int)]): Shape of the new tensor, e.g. :math:`(2, 3)` or :math:`2`.
|
||||
size (Union[int, tuple(int), list(int)]): Shape of the new tensor, e.g. :math:`(2, 3)` or :math:`2`.
|
||||
|
||||
Keyword Args:
|
||||
dtype (:class:`mindspore.dtype`, optional): Designated tensor dtype, it must be float type. If None,
|
||||
|
|
@ -856,7 +856,7 @@ def rand(*size, dtype=None, seed=None):
|
|||
|
||||
Returns:
|
||||
Tensor, with the designated shape and dtype, filled with random numbers from the uniform distribution on
|
||||
the interval :math:`[0, 1)`.
|
||||
the interval :math:`[0, 1)`.
|
||||
|
||||
Raises:
|
||||
TypeError: `seed` is not a non-negative integer.
|
||||
|
|
@ -936,16 +936,16 @@ def randn(*size, dtype=None, seed=None):
|
|||
from the standard normal distribution.
|
||||
|
||||
Args:
|
||||
*size (Union[int, tuple(int), list(int)]): Shape of the new tensor, e.g., :math:`(2, 3)` or :math:`2`.
|
||||
size (Union[int, tuple(int), list(int)]): Shape of the new tensor, e.g., :math:`(2, 3)` or :math:`2`.
|
||||
|
||||
Keyword Args:
|
||||
dtype (:class:`mindspore.dtype`, optional): Designated tensor dtype, it must be float type. If None,
|
||||
:class:`mindspore.float32` will be used. Default: None.
|
||||
`mindspore.float32` will be used. Default: None.
|
||||
seed (int, optional): Random seed, must be greater or equal to 0. Default: None, and 0 will be used.
|
||||
|
||||
Returns:
|
||||
Tensor, with the designated shape and dtype, filled with a sample (or samples) from the
|
||||
"standard normal" distribution.
|
||||
"standard normal" distribution.
|
||||
|
||||
Raises:
|
||||
TypeError: `seed` is not a non-negative integer.
|
||||
|
|
@ -985,11 +985,11 @@ def randn_like(x, seed=None, *, dtype=None):
|
|||
|
||||
Keyword Args:
|
||||
dtype (:class:`mindspore.dtype`, optional): Designated tensor dtype, it must be float type. If None,
|
||||
:class:`mindspore.float32` will be used. Default: None.
|
||||
`mindspore.float32` will be used. Default: None.
|
||||
|
||||
Returns:
|
||||
Tensor, with the designated shape and dtype, filled with a sample (or samples) from the
|
||||
"standard normal" distribution.
|
||||
"standard normal" distribution.
|
||||
|
||||
Raises:
|
||||
TypeError: `seed` is not a non-negative integer.
|
||||
|
|
@ -1035,7 +1035,7 @@ def randint(low, high, size, seed=None, *, dtype=None):
|
|||
|
||||
Returns:
|
||||
Tensor, with the designated shape and dtype, filled with random integers from low (inclusive)
|
||||
to high (exclusive).
|
||||
to high (exclusive).
|
||||
|
||||
Raises:
|
||||
TypeError: `seed` is not a non-negative integer.
|
||||
|
|
@ -1084,11 +1084,11 @@ def randint_like(x, low, high, seed=None, *, dtype=None):
|
|||
|
||||
Keyword Args:
|
||||
dtype (:class:`mindspore.dtype`, optional): Designated tensor dtype, it must be int type. If None,
|
||||
:class:`mindspore.int64` will be used. Default is :class:`mindspore.int64`.
|
||||
`mindspore.int64` will be used. Default is `mindspore.int64`.
|
||||
|
||||
Returns:
|
||||
Tensor, with the designated shape and dtype, filled with random integers from low (inclusive)
|
||||
to high (exclusive).
|
||||
to high (exclusive).
|
||||
|
||||
Raises:
|
||||
TypeError: `seed` is not a non-negative integer.
|
||||
|
|
|
|||
|
|
@ -9069,11 +9069,9 @@ class FractionalMaxPool3DWithFixedKsize(Primitive):
|
|||
ksize (Union[float, tuple]): The target ksize is D x H x W.
|
||||
ksize can be a tuple, or a single K for K x K x K.
|
||||
specifying the window size (D, H, W) of the input tensor.
|
||||
|
||||
output_shape (Union[int, tuple]): The target output_shape is D x H x W.
|
||||
output_shape can be a tuple, or a single H for H x H x H.
|
||||
specifying the size (D, H, W) of the output tensor.
|
||||
|
||||
data_format (str, optional) : The optional value for data format.
|
||||
Currently support 'NCDHW' and 'NHDWC'. Default: 'NCDHW'.
|
||||
|
||||
|
|
@ -9081,7 +9079,6 @@ class FractionalMaxPool3DWithFixedKsize(Primitive):
|
|||
- **x** (Tensor) - The input of FractionalMaxPool3DWithFixedKsize, which is a 4D or 5D tensor.
|
||||
Tensor of data type : float16, float32, double, int32, int64.
|
||||
Supported shape :math:`(N, C, D_{in}, H_{in}, W_{in})` or :math:`(N, D_{in}, H_{in}, W_{in}, C)`.
|
||||
|
||||
- **random_samples** (Tensor) - The random step of FractionalMaxPool3DWithFixedKsize, which is a 3D tensor.
|
||||
Tensor of data type : float16, float32, double, and value is between (0, 1).
|
||||
Supported shape :math:`(N, C, 3)`
|
||||
|
|
@ -9090,7 +9087,6 @@ class FractionalMaxPool3DWithFixedKsize(Primitive):
|
|||
- **y** (Tensor) - A tensor, the output of FractionalMaxPool3DWithFixedKsize.
|
||||
Has the same data type with `x`.
|
||||
Tensor of shape :math:`(N, C, D_{out}, H_{out}, W_{out})` or :math:`(N, D_{out}, H_{out}, W_{out}, C)`.
|
||||
|
||||
- **argmax** (Tensor) - A tensor, the indices along with the outputs.
|
||||
Has the same shape as the `y` and int32 or int64 data type.
|
||||
|
||||
|
|
@ -9284,7 +9280,7 @@ class PSROIPooling(Primitive):
|
|||
spatial_scale (float): a scaling factor that maps the box coordinates to the input coordinates.
|
||||
For example, if your boxes are defined on the scale of a 224x224 image and
|
||||
your input is a 112x112 feature map (resulting from a 0.5x scaling of the original
|
||||
image), you’ll want to set this to 0.5.
|
||||
image), you'll want to set this to 0.5.
|
||||
group_size (int): the size of the output (in pixels) after the pooling is performed, as (height, width).
|
||||
output_dim (int): the dim of the output after the pooling is performed.
|
||||
|
||||
|
|
@ -9344,9 +9340,9 @@ class PSROIPooling(Primitive):
|
|||
... group_size=7)
|
||||
>>> out = psROIPooling(features, rois)
|
||||
>>> print(out.shape)
|
||||
(4, 3, 7, 7)
|
||||
(4, 3, 7, 7)
|
||||
>>> print(out.dtype)
|
||||
Float32
|
||||
Float32
|
||||
"""
|
||||
|
||||
@prim_attr_register
|
||||
|
|
|
|||
Loading…
Reference in New Issue