mindspore 2.5.0__cp311-cp311-win_amd64.whl → 2.6.0__cp311-cp311-win_amd64.whl

mindspore/ops/functional_overload.py

@@ -13,20 +13,150 @@
 # limitations under the License.
 # ============================================================================
 """Holding mint APIs"""
+from mindspore._c_expression import _add_instance
+from mindspore._c_expression import _addcdiv_instance
 from mindspore._c_expression import _all_gather_matmul_instance
 from mindspore._c_expression import _bitwise_not_instance
 from mindspore._c_expression import _clamp_instance
 from mindspore._c_expression import _div_instance
 from mindspore._c_expression import _empty_instance
+from mindspore._c_expression import _floor_divide_instance
 from mindspore._c_expression import _fmod_instance
+from mindspore._c_expression import _gelu_instance
+from mindspore._c_expression import _gmm_instance
+from mindspore._c_expression import _gmm_backward_instance
+from mindspore._c_expression import _gmm_backward_fusion_instance
+from mindspore._c_expression import _greater_equal_instance
+from mindspore._c_expression import _kthvalue_instance
 from mindspore._c_expression import _lerp_instance
 from mindspore._c_expression import _matmul_reduce_scatter_instance
 from mindspore._c_expression import _max_instance
 from mindspore._c_expression import _min_instance
 from mindspore._c_expression import _nansum_instance
+from mindspore._c_expression import _pixel_shuffle_instance
 from mindspore._c_expression import _remainder_instance
 from mindspore._c_expression import _repeat_interleave_instance
+from mindspore._c_expression import _sub_instance
 from mindspore._c_expression import _where_instance
+from mindspore._c_expression import _xlogy_instance
+
+def add(*args, **kwargs):
+    r"""
+    add(input, other, *, alpha=1) -> Tensor
+
+    Adds scaled other value to `self`.
+
+    .. math::
+
+        out_{i} = self_{i} + alpha \times other_{i}
+
+    Note:
+        - When `self` and `other` have different shapes,
+          they must be able to broadcast to a common shape.
+        - `self`, `other` and `alpha` comply with the implicit type conversion rules to make the data types
+          consistent.
+
+    Args:
+        input (Union[Tensor, number.Number, bool]): `input` is a number.Number or a bool or a tensor whose data type is
+            `number <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_ or
+            `bool_ <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_.
+        other (Union[Tensor, number.Number, bool]): `other` is a number.Number or a bool or a tensor whose data type is
+            `number <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_ or
+            `bool_ <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_.
+
+    Keyword Args:
+        alpha (number.Number, optional): A scaling factor applied to `other`, default ``1``.
+
+    Returns:
+        Tensor with a shape that is the same as the broadcasted shape of the `self` and `other`,
+        and the data type is the one with higher precision or higher digits among `self`, `other` and `alpha`.
+
+    Raises:
+        TypeError: If the type of `other` or `alpha` is not one of the following: Tensor, number.Number, bool.
+        TypeError: If `alpha` is of type float but `self` and `other` are not of type float.
+        TypeError: If `alpha` is of type bool but `self` and `other` are not of type bool.
+
+    Supported Platforms:
+        ``Ascend`` ``GPU`` ``CPU``
+
+    Examples:
+        >>> import numpy as np
+        >>> import mindspore
+        >>> from mindspore import Tensor, mint
+        >>> x = Tensor(1, mindspore.int32)
+        >>> y = Tensor(np.array([4, 5, 6]).astype(np.float32))
+        >>> alpha = 0.5
+        >>> output = mint.add(x, y, alpha=alpha)  # x.add(y, alpha=alpha)
+        >>> print(output)
+        [3. 3.5 4.]
+        >>> # the data type of x is int32, the data type of y is float32,
+        >>> # alpha is a float, and the output is the data format of higher precision float32.
+        >>> print(output.dtype)
+        Float32
+    """
+    return _add_instance(*args, **kwargs)
+
+
+def __add__(*args, **kwargs):
+    r"""
+    __add__(input, other, *, alpha=1) -> Tensor
+
+    Alias for :func:`mindspore.mint.add`.
+
+    .. method:: mint.__add__(input, other, *, alpha=1) -> Tensor
+        :noindex:
+
+    Alias for overload function of :func:`mindspore.mint.add`.
+    """
+    return _add_instance(*args, **kwargs)
+
+
+def addcdiv(*args, **kwargs):
+    r"""
+    addcdiv_ext(input, tensor1, tensor2, *, value=1) -> Tensor
+
+    Performs the element-wise division of tensor tensor1 by tensor tensor2,
+    multiply the result by the scalar value and add it to input data.
+
+    .. math::
+        y[i] = input[i] + value * (tensor1[i] / tensor2[i])
+
+    .. warning::
+        This is an experimental API that is subject to change or deletion.
+
+    Args:
+        input (Tensor): The tensor to be added.
+        tensor1 (Tensor): The numerator tensor.
+        tensor2 (Tensor): The denominator tensor.
+
+    Keyword Args:
+        value (Number, optional): The multiplier for tensor1/tensor2. Default: ``1`` .
+
+    Returns:
+        Tensor, has the same shape and dtype as tensor1/tensor2.
+
+    Raises:
+        TypeError: If dtype of `tensor1`, `tensor2`, or `input` is not tensor.
+        ValueError: If `tensor1` could not be broadcast to a tensor with shape of `tensor2`.
+        ValueError: If `value` could not be broadcast to tensors with shapes of `tensor1/tensor2`.
+        ValueError: If `input` could not be broadcast to tensors with shapes of `value*(tensor1/tensor2)`.
+
+    Supported Platforms:
+        ``Ascend``
+
+    Examples:
+        >>> import mindspore
+        >>> import numpy as np
+        >>> from mindspore import Tensor, ops
+        >>> input_data = Tensor(np.array([1, 1, 1, 1]), mindspore.float32)
+        >>> x1 = Tensor(np.array([1, 2, 3, 4]), mindspore.float32)
+        >>> x2 = Tensor(np.array([4, 3, 2, 1]), mindspore.float32)
+        >>> y = ops.addcdiv_ext(input_data, x1, x2, value=1)
+        >>> print(y)
+        [1.25      1.6666667 2.5       5.       ]
+    """
+    return _addcdiv_instance(*args, **kwargs)
+
 
 def all_gather_matmul(*args, **kwargs):
     r"""
@@ -99,7 +229,7 @@ def all_gather_matmul(*args, **kwargs):
             Before running the following examples, you need to configure the communication environment variables.
 
             For Ascend/GPU/CPU devices, it is recommended to use the msrun startup method without any third-party or
-            configuration file dependencies. Please see the `msrun start up <https://www.mindspore.cn/docs/en/master/model_train/parallel/msrun_launcher.html>`_
+            configuration file dependencies. Please see the `msrun start up <https://www.mindspore.cn/tutorials/en/master/parallel/msrun_launcher.html>`_
             for more details.
 
             This example should be run with 2 devices.
@@ -351,6 +481,60 @@ def empty(*args, **kwargs):
     return _empty_instance(*args, **kwargs)
 
 
+def floor_divide(*args, **kwargs):
+    r"""
+    Divides the first input tensor by the second input tensor element-wise and round down to the closest integer.
+
+    Inputs of `input` and `other` comply with the implicit type conversion rules to make the data types consistent.
+    Inputs must be two tensors or one tensor and one scalar.
+    When the inputs are two tensors,
+    dtypes of them cannot be bool at the same time, and the shapes of them could be broadcast.
+    When the inputs are one tensor and one scalar,
+    the scalar could only be a constant.
+
+    .. math::
+        out_{i} = \text{floor}( \frac{input_i}{other_i})
+
+    where the :math:`floor` indicates the Floor operator. For more details,
+    please refer to the :class:`mindspore.mint.floor` operator.
+
+    .. warning::
+        This is an experimental API that is subject to change or deletion.
+
+    Args:
+        input (Union[Tensor, Number, bool]): The first input is a number or
+            a bool or a tensor whose data type is number or bool.
+        other (Union[Tensor, Number, bool]): The second input is a number or
+            a bool or a tensor whose data type is number or bool.
+
+    Returns:
+        Tensor, the shape is the same as the one after broadcasting,
+        and the data type is the one with higher precision or higher digits among the two inputs.
+
+    Raises:
+        TypeError: If `input` and `other` are not the following: Tensor, number.Number or bool.
+
+    Supported Platforms:
+        ``Ascend`` ``GPU`` ``CPU``
+
+    Examples:
+        >>> import mindspore
+        >>> from mindspore import Tensor, mint
+        >>> import numpy as np
+        >>> input = Tensor(np.array([2, 4, -1]), mindspore.int32)
+        >>> other = Tensor(np.array([3, 3, 3]), mindspore.int32)
+        >>> output = mint.floor_divide(input, other)
+        >>> print(output)
+        [ 0  1 -1]
+        >>> input = Tensor(2.0, mindspore.float32)
+        >>> other = Tensor(2.0, mindspore.float32)
+        >>> output = mint.floor_divide(input, other)
+        >>> print(output)
+        1.0
+    """
+    return _floor_divide_instance(*args, **kwargs)
+
+
 def fmod(*args, **kwargs):
     r"""
     fmod(input, other) -> Tensor
@@ -393,6 +577,294 @@ def fmod(*args, **kwargs):
     return _fmod_instance(*args, **kwargs)
 
 
+def gelu(*args, **kwargs):
+    r"""
+    gelu(input, *, approximate='none') -> Tensor
+
+    Gaussian Error Linear Units activation function.
+
+    GeLU is described in the paper `Gaussian Error Linear Units (GELUs) <https://arxiv.org/abs/1606.08415>`_.
+    And also please refer to `BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
+    <https://arxiv.org/abs/1810.04805>`_.
+
+    When `approximate` argument is `none`, GELU is defined as follows:
+
+    .. math::
+        GELU(x_i) = x_i*P(X < x_i),
+
+    where :math:`P` is the cumulative distribution function of the standard Gaussian distribution,
+    :math:`x_i` is the input element.
+
+    When `approximate` argument is `tanh`, GELU is estimated with:
+
+    .. math::
+        GELU(x_i) = 0.5 * x_i * (1 + \tanh(\sqrt(2 / \pi) * (x_i + 0.044715 * x_i^3)))
+
+    GELU Activation Function Graph:
+
+    .. image:: ../images/GELU.png
+        :align: center
+
+    .. note::
+        On the Ascend platform, when `input` is -inf, its gradient is 0,
+        and when `input` is inf, its gradient is `dout`.
+
+    Args:
+        input (Tensor): The input of the activation function GeLU, the data type is float16, float32 or float64.
+
+    Keyword Args:
+        approximate (str, optional): the gelu approximation algorithm to use. Acceptable vaslues are ``'none'`` and ``'tanh'`` .
+            Default: ``'none'`` .
+
+    Returns:
+        Tensor, with the same type and shape as `input`.
+
+    Raises:
+        TypeError: If `input` is not a Tensor.
+        TypeError: If dtype of `input` is not bfloat16, float16, float32 or float64.
+        ValueError: If `approximate` value is neither `none` nor `tanh`.
+
+    Supported Platforms:
+        ``Ascend``
+
+    Examples:
+        >>> import mindspore
+        >>> import numpy as np
+        >>> from mindspore import Tensor, mint
+        >>> input = Tensor(np.array([[-1.0, 4.0, -8.0], [2.0, -5.0, 9.0]]), mindspore.float32)
+        >>> result = mint.nn.functional.gelu(input)
+        >>> print(result)
+        [[-1.58655241e-01  3.99987316e+00 -0.00000000e+00]
+         [ 1.95449972e+00 -1.41860323e-06  9.0000000e+00]]
+        >>> result = mint.nn.functional.gelu(input, approximate="tanh")
+        >>> print(result)
+        [[-1.58808023e-01  3.99992990e+00 -3.10779147e-21]
+         [ 1.95459759e+00 -2.29180174e-07  9.0000000e+00]]
+    """
+    return _gelu_instance(*args, **kwargs)
+
+
+def gmm(*args, **kwargs):
+    r"""
+    gmm(x, weight, bias=None, group_list=None, group_type=0, group_list_type=0) -> tuple[Tensor]
+
+    Grouping matrix multiplication.
+
+    .. warning::
+        - This is an experimental API that is subject to change or deletion.
+        - `group_type` must be a constant.
+        - Only support on Atlas A2 training series.
+        - When the type of `group_list` is tuple[int] or list[int], it should a non-negative non-decreasing sequence,
+          indicating indexes of each group along the split axis. In this scenario, the arg `group_list_type` is useless.
+
+    .. note::
+        - When `group_type` is 2, the tensors in `x` must be non-continuous tensors which has
+          been transposed.
+        - Only when `group_type` is 0 and `bias` is None, the reverse derivative is supported,
+          which is implemented by ops.function.math_func.gmm_backward or through automatic differentiation.
+
+    Args:
+        x (tuple[Tensor]): The first tensors to be multiplied, whose num should be 1.
+        weight (tuple[Tensor]): The second tensors to be multiplied, whose num should be 1.
+        bias (tuple[Tensor], optional): Biases added to outputs, whose num should be 1.
+            The shape of each tensor in `bias` should be :math: `(group_list.shape[0], n)`
+            or :math: `(len(group_list), n)`. In the training scenario, the bias only supports None.
+            Default: ``None`` .
+        group_list (Union[Tensor, list[int], tuple[int]], optional): 1-D Tensor, list[int]
+            or tuple[int], indicating indexes or sizes of each group along the split axis.
+            When `group_list` is list[int] or tuple[int], it's length should be less than or equal to 128.
+            When `group_list` is a Tensor, it's size should be less than or equal to 1024.
+            Supported dtypes: int64.
+            Default: ``None`` .
+
+            - If `group_list_type` is 0, it must be  a non-negative non-decreasing sequence.
+              And when `group_type` is 0, the last element in `group_list` should be equal to
+              the first dimension of the tensor in `x` . When `group_type` is 2, the last element
+              in `group_list` should be equal to the second dimension of the tensor in `x` .
+
+            - If `group_list_type` is 1, the value in `group_list` are the sizes of each group.
+        group_type (int, optional): Represents the axes that need to be grouped. For example,
+            :math: `C[m,n] = A[m,k] \times B[k,n]`. Default: ``0`` .
+
+            - If `group_type` is 0, it means that the m-axis is grouped, meaning that the shape
+              of each tensor in `x` should be :math: `(m, k)` , the shape of each tensor in `weight`
+              should be :math: `(group_list.shape[0], k, n)` or :math: `(len(group_list), k, n)`,
+              and the shape of each tensor in result would be :math: `(m, n)` .
+
+            - If `group_type` is 2, it means that the k-axis is grouped, meaning that
+              the shape of each tensor in `x` should be :math: `(m, k)`, the shape of each
+              tensor in `weight` should be :math: `(k, n)`, and the shape of each tensor
+              in result would be :math: `(group_list.shape[0], m, n)` or :math: `(len(group_list), m, n)`.
+        group_list_type (int, optional): If it's 0, the value in `group_list` are the cumsum
+            result of the size of each group. If it's 1, the value in `group_list` are the size
+            of each group. Default: ``0`` .
+
+    `x` , `weight` and `bias` only support the following 3 type combinations:
+
+    - x: float16, weight: float16, bias: float16
+    - x: bfloat16, weight: bfloat16, bias: float32
+    - x: float32, weight: float32, bias: float32
+
+    Returns:
+        tuple[Tensor], the results of grouping matrix multiplication.
+
+    Supported Platforms:
+        ``Ascend``
+
+    Examples:
+        >>> import numpy as np
+        >>> from mindspore import Tensor, ops
+        >>> x = Tensor(np.random.uniform(0,1, (10, 20)).astype(np.float32))
+        >>> weight = Tensor(np.random.uniform(0,1, (4, 20, 8)).astype(np.float32))
+        >>> group_list = Tensor([2, 4, 2, 2])
+        >>> y = ops.function.math_func.gmm([x,], [weight,], group_list=group_list, group_list_type=1)
+        >>> print(y[0].shape)
+        >>> (10, 8)
+        >>> group_list = [2, 6, 8, 10]
+        >>> y = ops.function.math_func.gmm([x,], [weight,], group_list=group_list, group_list_type=0)
+        >>> print(y[0].shape)
+        >>> (10, 8)
+    """
+    return _gmm_instance(*args, **kwargs)
+
+
+def gmm_backward(*args, **kwargs):
+    r"""
+    gmm_backward(grad, x, weight, *, group_list=None, group_list_type=0) -> tuple[tuple[Tensor]]
+
+    the grad of ops.function.math_func.gmm
+    """
+    return _gmm_backward_instance(*args, **kwargs)
+
+
+def gmm_backward_fusion(*args, **kwargs):
+    r"""
+    gmm_backward_fusion(grad, weight, *, group_list=None, group_list_type=0) -> tuple[tuple[Tensor]]
+
+    the grad of ops.function.math_func.gmm, only dx
+    """
+    return _gmm_backward_fusion_instance(*args, **kwargs)
+
+
+def greater_equal(*args, **kwargs):
+    r"""
+    greater_equal(input, other) -> Tensor
+
+    Computes the boolean value of :math:`input >= other` element-wise.
+
+    .. math::
+
+        out_{i} =\begin{cases}
+            & \text{True,    if } input_{i}>=other_{i} \\
+            & \text{False,   if } input_{i}<other_{i}
+            \end{cases}
+
+    Note:
+        - Inputs of `input` and `other` comply with the implicit type conversion rules to make the data types
+          consistent.
+        - The inputs must be two tensors or one tensor and one scalar.
+        - When the inputs are two tensors, dtypes of them cannot be bool at the same time,
+          and the shapes of them can be broadcast.
+        - When the inputs are one tensor and one scalar, the scalar could only be a constant.
+        - Broadcasting is supported.
+        - If the input Tensor can be broadcast, the low dimension will be extended to the corresponding high dimension
+          in another input by copying the value of the dimension.
+
+    Args:
+        input (Union[Tensor, Number]): The first input is a number
+            or a tensor whose data type is `number <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html#mindspore.dtype>`_ or `bool_ <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html#mindspore.dtype>`_.
+        other (Union[Tensor, Number]): Second input. When the first input is a Tensor, the second input should be a Number,
+            or a Tensor of the number or bool_ data type. When the first input is a Scalar,
+            the second input must be a Tensor of number or bool_ data type.
+
+    Returns:
+        Tensor, the shape is the same as the one after broadcasting, and the data type is bool.
+
+    Raises:
+        TypeError: If neither `input` nor `other` is a Tensor.
+
+    Supported Platforms:
+        ``Ascend``
+
+    Examples:
+        >>> import mindspore
+        >>> import numpy as np
+        >>> from mindspore import Tensor, mint
+        >>> input = Tensor(np.array([1, 2, 3]), mindspore.int32)
+        >>> other = Tensor(np.array([1, 1, 4]), mindspore.int32)
+        >>> output = mint.greater_equal(input, other)
+        >>> print(output)
+        [True True False]
+        >>> y = 2.1
+        >>> output = mint.greater_equal(input, y)
+        >>> print(output)
+        [False False True]
+    """
+    return _greater_equal_instance(*args, **kwargs)
+
+
+def ge(*args, **kwargs):
+    r"""
+    ge(input, other) -> Tensor
+
+    Alias for :func:`mindspore.mint.greater_equal`.
+    """
+    return _greater_equal_instance(*args, **kwargs)
+
+
+def kthvalue(*args, **kwargs):
+    r"""
+    Calculates the kth smallest value along given dim specified by `dim` of the input
+    tensor, and returns a tuple of (`values`, `indices`) where `values` contains the k-th smallest element
+    and `indices` provides the index of each corresponding element.
+
+    Args:
+        input (Tensor): The input tensor, can be any dimension. Set the shape of input tensor as
+            :math:`(input_1, input_2, ..., input_N)`.
+        k (int): Specifies the k-th smallest element to retrieve.
+        dim (int, optional): The dimension along which to find the k-th smallest value. Default: ``-1`` .
+        keepdim (bool, optional): Whether to reduce dimension, if ``True`` , the output will keep same dimension with the
+            input, the output will reduce dimension if ``False`` . Default: ``False`` .
+
+    Returns:
+        A tuple consisting of `values` and `indices`.
+
+        - **values** (Tensor) - The k-th smallest value of input tensor, with the same dtype as `input`.
+
+          -If `keepdim` is ``True`` , the shape of output tensors is :math:`(input_1, input_2, ..., input_{dim-1}, 1, input_{dim+1}, ..., input_N)`.
+          -If `keepdim` is ``False`` , the shape is :math:`(input_1, input_2, ..., input_{dim-1}, input_{dim+1}, ..., input_N)` .
+
+        - **indices** (Tensor) - The `indices` for the k-th smallest value of the input tensor, it has the same shape as `values` with dtype of int64.
+        
+    Raises:
+        TypeError: If `k` or `dim` is not an int.
+        TypeError: If `keepdim` is not a bool.
+        TypeError: If dtype of `input` is not supported.
+        ValueError: If `input` is an empty Tensor.
+        RuntimeError: If `k` is not in the proper range.
+
+    Supported Platforms:
+        ``Ascend``
+
+    Examples:
+        >>> import mindspore
+        >>> import numpy as np
+        >>> from mindspore import Tensor, ops
+        >>> input_x = Tensor(np.array([[1.01, 2.02, 3.03], [1.04, 2.05, 3.06]]), mindspore.float32)
+        >>> out = ops.auto_generate.kthvalue(input_x, 2, 1, False)
+        >>> print(out)
+        (Tensor(shape=[2], dtype=Float32, value= [ 2.01999998e+00,  2.04999995e+00]), Tensor(shape=[2], dtype=Int64, value= [1, 1]))
+        >>> out1 = ops.auto_generate.kthvalue(input_x, 2, 1, True)
+        >>> print(out1)
+        (Tensor(shape=[2, 1], dtype=Float32, value=
+        [[ 2.01999998e+00],
+         [ 2.04999995e+00]]), Tensor(shape=[2, 1], dtype=Int64, value=
+        [[1],
+         [1]]))
+    """
+    return _kthvalue_instance(*args, **kwargs)
+
+
 def lerp(*args, **kwargs):
     r"""
     lerp(input, end, weight) -> Tensor
@@ -510,7 +982,7 @@ def matmul_reduce_scatter(*args, **kwargs):
             Before running the following examples, you need to configure the communication environment variables.
 
             For Ascend/GPU/CPU devices, it is recommended to use the msrun startup method without any third-party or
-            configuration file dependencies. Please see the `msrun start up <https://www.mindspore.cn/docs/en/master/model_train/parallel/msrun_launcher.html>`_
+            configuration file dependencies. Please see the `msrun start up <https://www.mindspore.cn/tutorials/en/master/parallel/msrun_launcher.html>`_
             for more details.
 
             This example should be run with 2 devices.
@@ -578,7 +1050,7 @@ def max(*args, **kwargs):
             :math:`(input_1, input_2, ..., input_N)` , Complex tensor is not supported.
         dim (int): The dimension to reduce.
         keepdim (bool, optional): Whether to reduce dimension, if ``True`` the output will keep the same dimension as the
-            `input` , the output will reduce dimension if ``false``. Default: ``False``.
+            `input` , the output will reduce dimension if ``False``. Default: ``False``.
 
     Returns:
         tuple (Tensor), tuple of 2 tensors, containing the maximum value of the self tensor along the given
@@ -649,7 +1121,7 @@ def min(*args, **kwargs):
             :math:`(input_1, input_2, ..., input_N)` , Complex tensor is not supported.
         dim (int): The dimension to reduce.
         keepdim (bool, optional): Whether to reduce dimension, if ``True`` the output will keep the same dimension as the
-            input, the output will reduce dimension if ``false``. Default: ``False``.
+            input, the output will reduce dimension if ``False``. Default: ``False``.
 
     Returns:
         tuple (Tensor), tuple of 2 tensors, containing the minimum value of the self tensor along the given
@@ -739,6 +1211,51 @@ def nansum(*args, **kwargs):
     return _nansum_instance(*args, **kwargs)
 
 
+def pixel_shuffle(*args, **kwargs):
+    r"""
+    pixel_shuffle(input, upscale_factor) -> Tensor
+
+    Rearrange elements in a tensor according to an upscaling factor.
+
+    Rearranges elements in a tensor of shape :math:`(*, C \times r^2, H, W)`
+    to a tensor of shape :math:`(*, C, H \times r, W \times r)`, where r is an upscale factor.
+
+    This is useful for implementing efficient sub-pixel convolution
+    with a stride of :math:`1/r`.
+
+    For detailed introduction to the pixel_shuffle algorithm, refer to
+    `Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network <https://arxiv.org/abs/1609.05158>`_ .
+
+    .. warning::
+        This is an experimental API that is subject to change or deletion.
+
+    Args:
+        input (Tensor): Tensor of shape :math:`(*, C \times r^2, H, W)` . The dimension of `input` is larger than 2,
+            and the length of third to last dimension can be divisible by the square of `upscale_factor`.
+        upscale_factor (int): factor to shuffle the input Tensor, and is a positive integer.
+            `upscale_factor` is the above-mentioned :math:`r`.
+
+    Returns:
+        - **output** (Tensor) - Tensor of shape :math:`(*, C, H \times r, W \times r)` .
+
+    Raises:
+        ValueError: If `upscale_factor` is not a positive integer.
+        ValueError: If the length of third to last dimension is not divisible by the square of `upscale_factor`.
+        ValueError: If the dimension of `input` is less than 3.
+
+    Supported Platforms:
+        ``Ascend``
+
+    Examples:
+        >>> from mindspore import mint
+        >>> input = mint.randn(1, 9, 4, 4)
+        >>> output = mint.nn.functional.pixel_shuffle(input, 3)
+        >>> print(output.shape)
+            (1, 1, 12, 12)
+    """
+    return _pixel_shuffle_instance(*args, **kwargs)
+
+
 def remainder(*args, **kwargs):
     r"""
     remainder(input, other) -> Tensor
@@ -830,6 +1347,77 @@ def repeat_interleave(*args, **kwargs):
     return _repeat_interleave_instance(*args, **kwargs)
 
 
+def sub(*args, **kwargs):
+    r"""
+    sub(input, other, *, alpha=1) -> Tensor
+
+    Subtracts scaled other value from self Tensor.
+
+    .. math::
+
+        out_{i} = self_{i} - alpha \times other_{i}
+
+    Note:
+        - When the two inputs have different shapes,
+          they must be able to broadcast to a common shape.
+        - The two inputs and alpha comply with the implicit type conversion rules to make the data types
+          consistent.
+
+    Args:
+        input (Union[Tensor, number.Number, bool]): `input` is a number.Number or a bool or a tensor whose data type is
+            `number <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_ or
+            `bool_ <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_.
+        other (Union[Tensor, number.Number, bool]): `other` is a number.Number or a bool or a tensor whose data type is
+            `number <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_ or
+            `bool_ <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_.
+
+    Keyword Args:
+        alpha (number.Number, optional): A scaling factor applied to `other`, default ``1``.
+
+    Returns:
+        Tensor with a shape that is the same as the broadcasted shape of the self `self` and `other`,
+        and the data type is the one with higher precision or higher digits among the two inputs and alpha.
+
+    Raises:
+        TypeError: If the type of `other` or `alpha` is not one of the following: Tensor, number.Number, bool.
+        TypeError: If `alpha` is of type float but `input` and `other` are not of type float.
+        TypeError: If `alpha` is of type bool but `input` and `other` are not of type bool.
+
+    Supported Platforms:
+        ``Ascend`` ``GPU`` ``CPU``
+
+    Examples:
+        >>> import numpy as np
+        >>> import mindspore
+        >>> from mindspore import Tensor, mint
+        >>> x = Tensor(np.array([4, 5, 6]).astype(np.float32))
+        >>> y = Tensor(1, mindspore.int32)
+        >>> alpha = 0.5
+        >>> output = mint.sub(x, y, alpha=alpha)
+        >>> print(output)
+        [3.5 4.5 5.5]
+        >>> # the data type of x is float32, the data type of y is int32,
+        >>> # alpha is a float, and the output is the data format of higher precision float32.
+        >>> print(output.dtype)
+        Float32
+    """
+    return _sub_instance(*args, **kwargs)
+
+
+def __sub__(*args, **kwargs):
+    r"""
+    __sub__(input, other, *, alpha=1) -> Tensor
+
+    Alias for :func:`mindspore.mint.sub`.
+
+    .. method:: mint.__sub__(input, other, *, alpha=1) -> Tensor
+        :noindex:
+
+    Alias for overload function of :func:`mindspore.mint.sub`.
+    """
+    return _sub_instance(*args, **kwargs)
+
+
 def where(*args, **kwargs):
     r"""
     where(condition, input, other) -> Tensor
@@ -877,7 +1465,58 @@ def where(*args, **kwargs):
     """
     return _where_instance(*args, **kwargs)
 
+
+def xlogy(*args, **kwargs):
+    r"""
+    xlogy(input, other) -> Tensor
+
+    Computes the first input multiplied by the logarithm of second input element-wise.
+    Returns zero when `input` is zero.
+
+    .. math::
+
+        out_i = input_{i}\log{other_{i}}
+
+    Inputs of `input` and `other` comply with the implicit type conversion rules to make the data types consistent.
+    The inputs must be two tensors or one tensor and one scalar.
+    When the inputs are two tensors, the shapes of them could be broadcast.
+
+    Args:
+        input (Union[Tensor, numbers.Number, bool]): The first input is a numbers.Number or
+            a bool or a tensor whose data type is
+            `number <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_ or
+            `bool_ <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.dtype.html>`_.
+        other (Union[Tensor, numbers.Number, bool]): The second input is a numbers.Number or
+            a bool or a tensor whose data type is number or bool when the first input is a tensor.
+            When the first input is Scalar, the second input must be a Tensor whose data type is number or bool.
+
+    Returns:
+        Tensor, the shape is the same as the one after broadcasting,
+        and the data type is the one with higher precision or higher digits among the two inputs.
+
+    Raises:
+        TypeError: If `input` and `other` is not a numbers.Number or a bool or a Tensor.
+        ValueError: If `input` could not be broadcast to a tensor with shape of `other`.
+
+    Supported Platforms:
+        ``Ascend`` ``GPU`` ``CPU``
+
+    Examples:
+        >>> import mindspore
+        >>> import numpy as np
+        >>> from mindspore import Tensor, ops
+        >>> input = Tensor(np.array([-5, 0, 4]), mindspore.float32)
+        >>> other = Tensor(np.array([2, 2, 2]), mindspore.float32)
+        >>> output = ops.xlogy(input, other)
+        >>> print(output)
+        [-3.465736   0.        2.7725887]
+    """
+    return _xlogy_instance(*args, **kwargs)
+
 __all__ = [
+    "add",
+    "__add__",
+    "addcdiv",
     "all_gather_matmul",
     "bitwise_not",
     "clamp",
@@ -885,13 +1524,25 @@ __all__ = [
     "div",
     "divide",
     "empty",
+    "floor_divide",
     "fmod",
+    "gelu",
+    "gmm",
+    "gmm_backward",
+    "gmm_backward_fusion",
+    "greater_equal",
+    "ge",
+    "kthvalue",
     "lerp",
     "matmul_reduce_scatter",
     "max",
     "min",
     "nansum",
+    "pixel_shuffle",
     "remainder",
     "repeat_interleave",
+    "sub",
+    "__sub__",
     "where",
+    "xlogy",
 ]