o Qeȱ@s@ddlmZddlZddlZddlmZddlmZm Z m Z ddl m Z ddlm Z mZmZddlmZdd lmZdd lmZdd lmZdd lmZmZd dlmZgZd/ddZd0ddZd0ddZd1ddZd2ddZd/ddZ d/ddZ!d3d d!Z"d"d#Z#d4d$d%Z$d5d'd(Z%d6d)d*Z&   d6d+d,Z'd7d-d.Z(dS)8)print_functionN) LayerHelper)check_variable_and_dtype check_type check_dtype)layers)corein_dygraph_mode_non_static_mode)_in_legacy_dygraph)convert_np_dtype_to_dtype_)Variable)VarDesc)_C_ops _legacy_C_ops) logical_notFcCstrt|||\}}|Strt|d|d|\}}|St|dgddtdit}|j|j dd}|jt j j dd}|j dd |i||d ||d d |S)aI Sorts the input along the given axis, and returns the corresponding index tensor for the sorted output values. The default sort algorithm is ascending, if you want the sort algorithm to be descending, you must set the :attr:`descending` as True. Args: x(Tensor): An input N-D Tensor with type float32, float64, int16, int32, int64, uint8. axis(int, optional): Axis to compute indices along. The effective range is [-R, R), where R is Rank(x). when axis<0, it works the same way as axis+R. Default is -1. descending(bool, optional) : Descending is a flag, if set to true, algorithm will sort by descending order, else sort by ascending order. Default is false. name (str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: Tensor: sorted indices(with the same shape as ``x`` and with data type int64). Examples: .. code-block:: python import paddle x = paddle.to_tensor([[[5,8,9,5], [0,0,1,7], [6,9,2,4]], [[5,2,4,2], [4,7,7,9], [1,7,0,6]]], dtype='float32') out1 = paddle.argsort(x, axis=-1) out2 = paddle.argsort(x, axis=0) out3 = paddle.argsort(x, axis=1) print(out1) #[[[0 3 1 2] # [0 1 2 3] # [2 3 0 1]] # [[1 3 2 0] # [0 1 2 3] # [2 0 3 1]]] print(out2) #[[[0 1 1 1] # [0 0 0 0] # [1 1 1 0]] # [[1 0 0 0] # [1 1 1 1] # [0 0 0 1]]] print(out3) #[[[1 1 1 2] # [0 0 2 0] # [2 2 0 1]] # [[2 0 2 0] # [1 1 0 2] # [0 2 1 1]]] axis descendingxfloat32float64Zint16int32int64Zuint8argsortTdtype stop_gradientr XOutZIndicesrrtypeinputsoutputsattrsN)r)r rrr rrrlocals"create_variable_for_type_inferencerrVarTypeINT64 append_op)rrrname_idshelperoutr5DD:\Projects\ConvertPro\env\Lib\site-packages\paddle/tensor/search.pyr#s:<    rrc Cs|durt|ttfstdt||durtdt|}d}|dur)d}d}tr5t |||||St rGt |d|d|d |d | }|St dit}t|d gd dt|dddgdi} ||}|| d <|| d<|| d <|| d<|jdd|id|gi| dd|_|S)a Computes the indices of the max elements of the input tensor's element along the provided axis. Args: x(Tensor): An input N-D Tensor with type float32, float64, int16, int32, int64, uint8. axis(int, optional): Axis to compute indices along. The effective range is [-R, R), where R is x.ndim. when axis < 0, it works the same way as axis + R. Default is None, the input `x` will be into the flatten tensor, and selecting the min value index. keepdim(bool, optional): Whether to keep the given axis in output. If it is True, the dimensions will be same as input x and with size one in the axis. Otherwise the output dimentions is one fewer than x since the axis is squeezed. Default is False. dtype(str|np.dtype, optional): Data type of the output tensor which can be int32, int64. The default value is ``int64`` , and it will return the int64 indices. name (str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: Tensor, return the tensor of int32 if set :attr:`dtype` is int32, otherwise return the tensor of int64. Examples: .. code-block:: python import paddle x = paddle.to_tensor([[5,8,9,5], [0,0,1,7], [6,9,2,4]]) out1 = paddle.argmax(x) print(out1) # 2 out2 = paddle.argmax(x, axis=0) print(out2) # [2, 2, 0, 1] out3 = paddle.argmax(x, axis=-1) print(out3) # [2, 3, 1] out4 = paddle.argmax(x, axis=0, keepdim=True) print(out4) # [[2, 2, 0, 1]] NzMThe type of 'axis' must be int or Tensor or None in argmax, but received %s.zCthe value of 'dtype' in argmax could not be None, but received NoneFTrrrkeepdimsflattenargmaxrrz paddle.argmaxrrargminarg_maxr"r$r&)r9) isinstanceintr TypeErrorr' ValueErrorr r rr9r rr;rr+rrr,r/r rrkeepdimrr0Z var_dtyper8r4r3r*r5r5r6r9}sR(  r9c Cs|durt|ttfstdt||durtdt|}d}|dur)d}d}tr5t |||||St rGt |d|d|d |d | }|St dit}t|d gd dt|dddgd ||}i} || d <|| d<|| d <|| d<|jdd|id|gi| dd|_|S)a Computes the indices of the min elements of the input tensor's element along the provided axis. Args: x(Tensor): An input N-D Tensor with type float32, float64, int16, int32, int64, uint8. axis(int, optional): Axis to compute indices along. The effective range is [-R, R), where R is x.ndim. when axis < 0, it works the same way as axis + R. Default is None, the input `x` will be into the flatten tensor, and selecting the min value index. keepdim(bool, optional): Whether to keep the given axis in output. If it is True, the dimensions will be same as input x and with size one in the axis. Otherwise the output dimentions is one fewer than x since the axis is squeezed. Default is False. dtype(str, optional): Data type of the output tensor which can be int32, int64. The default value is 'int64', and it will return the int64 indices. name(str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: Tensor, return the tensor of `int32` if set :attr:`dtype` is `int32`, otherwise return the tensor of `int64`. Examples: .. code-block:: python import paddle x = paddle.to_tensor([[5,8,9,5], [0,0,1,7], [6,9,2,4]]) out1 = paddle.argmin(x) print(out1) # 4 out2 = paddle.argmin(x, axis=0) print(out2) # [1, 1, 1, 2] out3 = paddle.argmin(x, axis=-1) print(out3) # [0, 0, 2] out4 = paddle.argmin(x, axis=0, keepdim=True) print(out4) # [[1, 1, 1, 2]] NzMThe type of 'axis' must be int or Tensor or None in argmin, but received %s.zCthe value of 'dtype' in argmin could not be None, but received NoneFTrrrr7r8r:rrz paddle.argminrrarg_minr"r$r&)r:)r<r=rr>r'r?r r rr:r rrBrr+rrr,r/r r@r5r5r6r:sR(  r:cCstr t|||Strt||d|Std it}t|dgddt|dddgd||j }|j d||d d |id|id |S)a Returns a new tensor which indexes the ``input`` tensor along dimension ``axis`` using the entries in ``index`` which is a Tensor. The returned tensor has the same number of dimensions as the original ``x`` tensor. The dim-th dimension has the same size as the length of ``index``; other dimensions have the same size as in the ``x`` tensor. Args: x (Tensor): The input Tensor to be operated. The data of ``x`` can be one of float32, float64, int32, int64. index (Tensor): The 1-D Tensor containing the indices to index. The data type of ``index`` must be int32 or int64. axis (int, optional): The dimension in which we index. Default: if None, the ``axis`` is 0. name(str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: Tensor: A Tensor with same data type as ``x``. Examples: .. code-block:: python import paddle x = paddle.to_tensor([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0], [9.0, 10.0, 11.0, 12.0]]) index = paddle.to_tensor([0, 1, 1], dtype='int32') out_z1 = paddle.index_select(x=x, index=index) #[[1. 2. 3. 4.] # [5. 6. 7. 8.] # [5. 6. 7. 8.]] out_z2 = paddle.index_select(x=x, index=index, axis=1) #[[ 1. 2. 2.] # [ 5. 6. 6.] # [ 9. 10. 10.]] dim index_selectrrrrrz!paddle.tensor.search.index_selectindexrrr"Indexr$r&N)rD) r rrDr rrr+rr,rr/)rrFrr0r3r4r5r5r6rD!s($   rDc Csg}|j}t|}trt|}n(trt|}ntd it }|j t j j jd}|jdd|id|gid|s>|S|dkrGt|gSt|D]}|tj|dg|g|dgdqKt|S) a Return a tensor containing the indices of all non-zero elements of the `input` tensor. If as_tuple is True, return a tuple of 1-D tensors, one for each dimension in `input`, each containing the indices (in that dimension) of all non-zero elements of `input`. Given a n-Dimensional `input` tensor with shape [x_1, x_2, ..., x_n], If as_tuple is False, we can get a output tensor with shape [z, n], where `z` is the number of all non-zero elements in the `input` tensor. If as_tuple is True, we can get a 1-D tensor tuple of length `n`, and the shape of each 1-D tensor is [z, 1]. Args: x (Tensor): The input tensor variable. as_tuple (bool): Return type, Tensor or tuple of Tensor. Returns: Tensor. The data type is int64. Examples: .. code-block:: python import paddle x1 = paddle.to_tensor([[1.0, 0.0, 0.0], [0.0, 2.0, 0.0], [0.0, 0.0, 3.0]]) x2 = paddle.to_tensor([0.0, 1.0, 0.0, 3.0]) out_z1 = paddle.nonzero(x1) print(out_z1) #[[0 0] # [1 1] # [2 2]] out_z1_tuple = paddle.nonzero(x1, as_tuple=True) for out in out_z1_tuple: print(out) #[[0] # [1] # [2]] #[[0] # [1] # [2]] out_z2 = paddle.nonzero(x2) print(out_z2) #[[1] # [3]] out_z2_tuple = paddle.nonzero(x2, as_tuple=True) for out in out_z2_tuple: print(out) #[[1] # [3]] where_indexr Conditionr$r'r(r)r)ZaxesZstartsZendsN)rI)shapelenr rrIpaddlein_dynamic_moderrr+r,r rr-r.r/tuplerangeappendslice)ras_tupleZlist_outrMZrankoutsr3ir5r5r6nonzero]s04    rXc Cstrt|||\}}|Strt|d|d|\}}|Stdit}|j|jdd}|jt j j dd}|j dd |i||d ||d d |S)a- Sorts the input along the given axis, and returns the sorted output tensor. The default sort algorithm is ascending, if you want the sort algorithm to be descending, you must set the :attr:`descending` as True. Args: x(Tensor): An input N-D Tensor with type float32, float64, int16, int32, int64, uint8. axis(int, optional): Axis to compute indices along. The effective range is [-R, R), where R is Rank(x). when axis<0, it works the same way as axis+R. Default is -1. descending(bool, optional) : Descending is a flag, if set to true, algorithm will sort by descending order, else sort by ascending order. Default is false. name (str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: Tensor: sorted tensor(with the same shape and data type as ``x``). Examples: .. code-block:: python import paddle x = paddle.to_tensor([[[5,8,9,5], [0,0,1,7], [6,9,2,4]], [[5,2,4,2], [4,7,7,9], [1,7,0,6]]], dtype='float32') out1 = paddle.sort(x=x, axis=-1) out2 = paddle.sort(x=x, axis=0) out3 = paddle.sort(x=x, axis=1) print(out1) #[[[5. 5. 8. 9.] # [0. 0. 1. 7.] # [2. 4. 6. 9.]] # [[2. 2. 4. 5.] # [4. 7. 7. 9.] # [0. 1. 6. 7.]]] print(out2) #[[[5. 2. 4. 2.] # [0. 0. 1. 7.] # [1. 7. 0. 4.]] # [[5. 8. 9. 5.] # [4. 7. 7. 9.] # [6. 9. 2. 6.]]] print(out3) #[[[0. 0. 1. 4.] # [5. 8. 2. 5.] # [6. 9. 9. 7.]] # [[1. 2. 0. 2.] # [4. 7. 4. 6.] # [5. 7. 7. 9.]]] rrsortFrTr!rr"r#r%r&N)rY) r rrr rrr+r,rrr-r.r/) rrrr0rVr1r3r4r2r5r5r6rYs28   rYc Cstr t|||Strt|d|d|Std it}d|gi}i}||d<||d<|j|jd}|jdd}|j d||g|gd|dd |_ ||fS) a Used to find values and indices of the modes at the optional axis. Args: x(Tensor): Tensor, an input N-D Tensor with type float32, float64, int32, int64. axis(int, optional): Axis to compute indices along. The effective range is [-R, R), where R is x.ndim. when axis < 0, it works the same way as axis + R. Default is -1. keepdim(bool, optional): Whether to keep the given axis in output. If it is True, the dimensions will be same as input x and with size one in the axis. Otherwise the output dimentions is one fewer than x since the axis is squeezed. Default is False. name (str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: tuple(Tensor), return the values and indices. The value data type is the same as the input `x`. The indices data type is int64. Examples: .. code-block:: python import paddle tensor = paddle.to_tensor([[[1,2,2],[2,3,3]],[[0,5,5],[9,9,0]]], dtype=paddle.float32) res = paddle.mode(tensor, 2) print(res) # (Tensor(shape=[2, 2], dtype=float32, place=CUDAPlace(0), stop_gradient=True, # [[2., 3.], # [5., 9.]]), Tensor(shape=[2, 2], dtype=int64, place=CUDAPlace(0), stop_gradient=True, # [[1, 1], # [1, 0]])) rrAmoder"rJrr#r&TN)rZ) r rrZr rrr+r,rr/r ) rrrAr0r3r(r*valuesindicesr5r5r6rZs(  rZcCst|rtdg|t|gjj}t|r&tdg|t|gjj}|dur4|dur4t|ddS|dus<|dur@tdt s^t |ddgdt |d gd dt |d gd dt |j }t |j }t |j }||kr|||kr||}|}|} nAt |} t |} t |} t| |j} t||j} t| | }t|| }t||}t||} t| |}t|d}trt||| Strt||| Stdit}|j|jd }|jd||| d d|gid|S)a. Return a Tensor of elements selected from either :attr:`x` or :attr:`y` according to corresponding elements of :attr:`condition`. Concretely, .. math:: out_i = \begin{cases} x_i, & \text{if} \ condition_i \ \text{is} \ True \\ y_i, & \text{if} \ condition_i \ \text{is} \ False \\ \end{cases}. Notes: ``numpy.where(condition)`` is identical to ``paddle.nonzero(condition, as_tuple=True)``, please refer to :ref:`api_tensor_search_nonzero`. Args: condition (Tensor): The condition to choose x or y. When True (nonzero), yield x, otherwise yield y. x (Tensor|scalar, optional): A Tensor or scalar to choose when the condition is True with data type of float32, float64, int32 or int64. Either both or neither of x and y should be given. y (Tensor|scalar, optional): A Tensor or scalar to choose when the condition is False with data type of float32, float64, int32 or int64. Either both or neither of x and y should be given. name (str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: Tensor: A Tensor with the same shape as :attr:`condition` and same data type as :attr:`x` and :attr:`y`. Examples: .. code-block:: python import paddle x = paddle.to_tensor([0.9383, 0.1983, 3.2, 1.2]) y = paddle.to_tensor([1.0, 1.0, 1.0, 1.0]) out = paddle.where(x>1, x, y) print(out) #out: [1.0, 1.0, 3.2, 1.2] out = paddle.where(x>1) print(out) #out: (Tensor(shape=[2, 1], dtype=int64, place=CPUPlace, stop_gradient=True, # [[2], # [3]]),) rNT)rUz1either both or neither of x and y should be given conditionboolwhererrEyrJ)rKr"Yr$rL)r_)npZisscalarrOfullarrayrr0rXr?rPrlistrMZ zeros_likecastaddr rr_r rrr+r,r/)r]rr`r0Zcondition_shapeZx_shapeZy_shapeZbroadcast_conditionZ broadcast_xZ broadcast_yZ zeros_like_xZ zeros_like_yZzeros_like_conditionZ cast_condZbroadcast_zerosr3r4r5r5r6r_8sh +              r_cCstr t||Strt||Std it}t|dgddt|dddgd|j|j d}|j d||d d |id |S)at **IndexSample Layer** IndexSample OP returns the element of the specified location of X, and the location is specified by Index. .. code-block:: text Given: X = [[1, 2, 3, 4, 5], [6, 7, 8, 9, 10]] Index = [[0, 1, 3], [0, 2, 4]] Then: Out = [[1, 2, 4], [6, 8, 10]] Args: x (Tensor): The source input tensor with 2-D shape. Supported data type is int32, int64, float32, float64. index (Tensor): The index input tensor with 2-D shape, first dimension should be same with X. Data type is int32 or int64. Returns: output (Tensor): The output is a tensor with the same shape as index. Examples: .. code-block:: python import paddle x = paddle.to_tensor([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0], [9.0, 10.0, 11.0, 12.0]], dtype='float32') index = paddle.to_tensor([[0, 1, 2], [1, 2, 3], [0, 0, 0]], dtype='int32') target = paddle.to_tensor([[100, 200, 300, 400], [500, 600, 700, 800], [900, 1000, 1100, 1200]], dtype='int32') out_z1 = paddle.index_sample(x, index) print(out_z1) #[[1. 2. 3.] # [6. 7. 8.] # [9. 9. 9.]] # Use the index of the maximum value by topk op # get the value of the element of the corresponding index in other tensors top_value, top_index = paddle.topk(x, k=2) out_z2 = paddle.index_sample(target, top_index) print(top_value) #[[ 4. 3.] # [ 8. 7.] # [12. 11.]] print(top_index) #[[3 2] # [3 2] # [3 2]] print(out_z2) #[[ 400 300] # [ 800 700] # [1200 1100]] index_samplerrEz!paddle.tensor.search.index_samplerFrrrJrGr$rLN)rh) r rrhr rrr+rr,rr/)rrFr3r4r5r5r6rhs(I   rhcCstr t||Strt||Std it}t|dgddt|ddgd|j|j d}|j d||d d |id |S)a2 Returns a new 1-D tensor which indexes the input tensor according to the ``mask`` which is a tensor with data type of bool. Args: x (Tensor): The input Tensor, the data type can be int32, int64, float32, float64. mask (Tensor): The Tensor containing the binary mask to index with, it's data type is bool. name(str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: A 1-D Tensor which is the same data type as ``x``. Examples: .. code-block:: python import paddle x = paddle.to_tensor([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0], [9.0, 10.0, 11.0, 12.0]]) mask = paddle.to_tensor([[True, False, False, False], [True, True, False, False], [True, False, False, False]]) out = paddle.masked_select(x, mask) #[1.0 5.0 6.0 9.0] masked_selectrrEz paddle.tensor.search.mask_selectmaskr^z"paddle.tensor.search.masked_selectrJ)r"MaskrarLN)ri) r rrir rrr+rr,rr/)rrjr0r3r4r5r5r6ris&    riTc Cs.tr|dkr d}t|||||\}}||fStrG|dur2t|dt|d|d|\}}||fSt|dt|d|d|d| \}}||fStdit}d|gi} i} t |t ra|g| d <nd|i} || d<|| d<|duru|| d<|j |j d } |j d d }|j d| | g|gd | d d|_| |fS)a' Return values and indices of the k largest or smallest at the optional axis. If the input is a 1-D Tensor, finds the k largest or smallest values and indices. If the input is a Tensor with higher rank, this operator computes the top k values and indices along the :attr:`axis`. Args: x(Tensor): Tensor, an input N-D Tensor with type float32, float64, int32, int64. k(int, Tensor): The number of top elements to look for along the axis. axis(int, optional): Axis to compute indices along. The effective range is [-R, R), where R is x.ndim. when axis < 0, it works the same way as axis + R. Default is -1. largest(bool, optional) : largest is a flag, if set to true, algorithm will sort by descending order, otherwise sort by ascending order. Default is True. sorted(bool, optional): controls whether to return the elements in sorted order, default value is True. In gpu device, it always return the sorted value. name (str, optional): Name for the operation (optional, default is None). For more information, please refer to :ref:`api_guide_Name`. Returns: tuple(Tensor), return the values and indices. The value data type is the same as the input `x`. The indices data type is int64. Examples: .. code-block:: python import paddle data_1 = paddle.to_tensor([1, 4, 5, 7]) value_1, indices_1 = paddle.topk(data_1, k=1) print(value_1) # [7] print(indices_1) # [3] data_2 = paddle.to_tensor([[1, 4, 5, 7], [2, 6, 2, 5]]) value_2, indices_2 = paddle.topk(data_2, k=1) print(value_2) # [[7], [6]] print(indices_2) # [[3], [1]] value_3, indices_3 = paddle.topk(data_2, k=1, axis=-1) print(value_3) # [[7], [6]] print(indices_3) # [[3], [1]] value_4, indices_4 = paddle.topk(data_2, k=1, axis=0) print(value_4) # [[2, 6, 5, 7]] print(indices_4) # [[1, 1, 0, 0]] Nrklargestsortedrtop_k_v2r"KrJrr#r&T)ro)r rZtop_kr rror=rr+r<rr,rr/r ) rrlrrmrnr0r4r\r3r(r*r[r5r5r6topk5sJ0    rqcCs@t|dgdd|dkrtd|t|||||S)u This API is used to find the index of the corresponding 1D tensor `sorted_sequence` in the innermost dimension based on the given `x`. Args: x(Tensor): An input N-D tensor value with type int32, int64, float32, float64. sorted_sequence(Tensor): An input 1-D tensor with type int32, int64, float32, float64. The value of the tensor monotonically increases in the innermost dimension. out_int32(bool, optional): Data type of the output tensor which can be int32, int64. The default value is False, and it indicates that the output data type is int64. right(bool, optional): Find the upper or lower bounds of the sorted_sequence range in the innermost dimension based on the given `x`. If the value of the sorted_sequence is nan or inf, return the size of the innermost dimension. The default value is False and it shows the lower bounds. name(str, optional): The default value is None. Normally there is no need for user to set this property. For more information, please refer to :ref:`api_guide_Name`. Returns: Tensor（the same sizes of the `x`）, return the tensor of int32 if set :attr:`out_int32` is True, otherwise return the tensor of int64. Examples: .. code-block:: python import paddle sorted_sequence = paddle.to_tensor([2, 4, 8, 16], dtype='int32') x = paddle.to_tensor([[0, 8, 4, 16], [-1, 2, 8, 4]], dtype='int32') out1 = paddle.bucketize(x, sorted_sequence) print(out1) # Tensor(shape=[2, 4], dtype=int64, place=CPUPlace, stop_gradient=True, # [[0, 2, 1, 3], # [0, 0, 2, 1]]) out2 = paddle.bucketize(x, sorted_sequence, right=True) print(out2) # Tensor(shape=[2, 4], dtype=int64, place=CPUPlace, stop_gradient=True, # [[0, 3, 2, 4], # [0, 1, 3, 2]]) out3 = x.bucketize(sorted_sequence) print(out3) # Tensor(shape=[2, 4], dtype=int64, place=CPUPlace, stop_gradient=True, # [[0, 2, 1, 3], # [0, 0, 2, 1]]) out4 = x.bucketize(sorted_sequence, right=True) print(out4) # Tensor(shape=[2, 4], dtype=int64, place=CPUPlace, stop_gradient=True, # [[0, 3, 2, 4], # [0, 1, 3, 2]]) SortedSequencerEpaddle.searchsortedrz8sorted_sequence tensor must be 1 dimension, but got dim )rrCr? searchsorted)rsorted_sequence out_int32rightr0r5r5r6 bucketizes-  rxcCstr t||||Strt||d|d|St|dgddt|dgddtdit}|r6dnd }|j|d }|j d||d d |i||d d|S)uX Find the index of the corresponding `sorted_sequence` in the innermost dimension based on the given `values`. Args: sorted_sequence(Tensor): An input N-D or 1-D tensor with type int32, int64, float32, float64. The value of the tensor monotonically increases in the innermost dimension. values(Tensor): An input N-D tensor value with type int32, int64, float32, float64. out_int32(bool, optional): Data type of the output tensor which can be int32, int64. The default value is False, and it indicates that the output data type is int64. right(bool, optional): Find the upper or lower bounds of the sorted_sequence range in the innermost dimension based on the given `values`. If the value of the sorted_sequence is nan or inf, return the size of the innermost dimension. The default value is False and it shows the lower bounds. name(str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: Tensor（the same sizes of the `values`）, return the tensor of int32 if set :attr:`out_int32` is True, otherwise return the tensor of int64. Examples: .. code-block:: python import paddle sorted_sequence = paddle.to_tensor([[1, 3, 5, 7, 9, 11], [2, 4, 6, 8, 10, 12]], dtype='int32') values = paddle.to_tensor([[3, 6, 9, 10], [3, 6, 9, 10]], dtype='int32') out1 = paddle.searchsorted(sorted_sequence, values) print(out1) # Tensor(shape=[2, 4], dtype=int64, place=CUDAPlace(0), stop_gradient=True, # [[1, 3, 4, 5], # [1, 2, 4, 4]]) out2 = paddle.searchsorted(sorted_sequence, values, right=True) print(out2) # Tensor(shape=[2, 4], dtype=int64, place=CUDAPlace(0), stop_gradient=True, # [[2, 3, 5, 5], # [1, 3, 4, 5]]) sorted_sequence_1d = paddle.to_tensor([1, 3, 5, 7, 9, 11, 13]) out3 = paddle.searchsorted(sorted_sequence_1d, values) print(out3) # Tensor(shape=[2, 4], dtype=int64, place=CUDAPlace(0), stop_gradient=True, # [[1, 3, 4, 5], # [1, 3, 4, 5]]) rvrwrrrErsValuesrtrrrJ)rrryr$)rvrwr&N)rt) r rrtr rrrr+r,r/)rur[rvrwr0r3Zout_typer4r5r5r6rts6.    rtc Cstr1|durtrt|d|d|d|St||||Str)t|d|d|St||d|Std it}d|gi}d|i}|durJ||d<|j|jd}|jd d} |j d||g| gd |d d | _ || fS)a Find values and indices of the k-th smallest at the axis. Args: x(Tensor): A N-D Tensor with type float32, float64, int32, int64. k(int): The k for the k-th smallest number to look for along the axis. axis(int, optional): Axis to compute indices along. The effective range is [-R, R), where R is x.ndim. when axis < 0, it works the same way as axis + R. The default is None. And if the axis is None, it will computed as -1 by default. keepdim(bool, optional): Whether to keep the given axis in output. If it is True, the dimensions will be same as input x and with size one in the axis. Otherwise the output dimentions is one fewer than x since the axis is squeezed. Default is False. name (str, optional): For details, please refer to :ref:`api_guide_Name`. Generally, no setting is required. Default: None. Returns: tuple(Tensor), return the values and indices. The value data type is the same as the input `x`. The indices data type is int64. Examples: .. code-block:: python import paddle x = paddle.randn((2,3,2)) # Tensor(shape=[2, 3, 2], dtype=float32, place=CUDAPlace(0), stop_gradient=True, # [[[ 0.22954939, -0.01296274], # [ 1.17135799, -0.34493217], # [-0.19550551, -0.17573971]], # # [[ 0.15104349, -0.93965352], # [ 0.14745511, 0.98209465], # [ 0.10732264, -0.55859774]]]) y = paddle.kthvalue(x, 2, 1) # (Tensor(shape=[2, 2], dtype=float32, place=CUDAPlace(0), stop_gradient=True, # [[ 0.22954939, -0.17573971], # [ 0.14745511, -0.55859774]]), Tensor(shape=[2, 2], dtype=int64, place=CUDAPlace(0), stop_gradient=True, # [[0, 2], # [1, 2]])) NrlrrArkthvaluer"rJrr#r&T)rz) r r rrzrrr+r,rr/r ) rrlrrAr0r3r(r*r[r\r5r5r6rzs4&  rz)rFN)NFrN)rN)F)NNN)N)NTTN)FFN)NFN)) __future__rnumpyrbrOZ frameworkrZfluid.data_feederrrrZfluidrr r r Zfluid.frameworkr Zpaddle.common_ops_importr rrrrZlogicr__all__rr9r:rDrXrYrZr_rhrirqrxrtrzr5r5r5r6s>          Z R R < Q R 8k ` 2 Z9 M