o + iC2@sddlmZddlmZmZddlZddlZddlmZm Z ddl m Z ddl m Z mZmZmZmZddlmZddlmZmZer^ddlmZdd lmZmZmZmZdd lmZmZm Z d d gZ!d dZ"ddZ#ddZ$ddZ%    d+d,d$d Z&d%d&Z'e    d+d-d*d Z(dS).) annotations) TYPE_CHECKINGAnyN)_C_opsin_dynamic_mode) convert_dtype)_current_expected_place_get_paddle_placecore dygraph_only in_pir_mode) LayerHelper)max to_tensor)CPUPlaceCUDAPinnedPlace CUDAPlaceTensor) DTypeLikeNumericSequence ShapeLikesparse_coo_tensorsparse_csr_tensorcCs(|rt|t|jkr|t|S|S)N)rdtypeZastype)datarrb/home/app/PaddleOCR-VL-test/.venv_paddleocr/lib/python3.10/site-packages/paddle/sparse/creation.py _handle_dtype.srcCsZt|jdks Jt|dd}|d}|}t|jdkr)t||jdd}t|S)N)Zaxis)lenshapernumpynpappendlist)indicesvaluesZlensrrr_infer_dense_shape5s r(cCs@t|}|dur t}|St|tjtjtjtjfstd|S)Nz^'place' must be any of paddle.Place, paddle.CPUPlace, paddle.CUDAPinnedPlace, paddle.CUDAPlace) r r isinstancer ZPlacerrr ValueError)placerrr _get_place?sr,cCs$|tjtjtjtjfvrtddS)NzDthe dtype of indices must be 'int8' or 'int16' or 'int32' or 'int64')paddleZint8Zint16Zint32Zint64 TypeError)rrrr_check_indices_dtypeLs r/Tr&Mlist[list[int]] | tuple[tuple[int, ...], ...] | npt.NDArray[np.int_] | Tensorr'+NumericSequence | npt.NDArray[Any] | Tensorr!ShapeLike | NonerDTypeLike | Noner+3CPUPlace | CUDAPinnedPlace | CUDAPlace | str | None stop_gradientboolreturnrcCstrt|}t|tjjst|d|dd}t|tjjs$t||||}t|jdkr/t d|jd}|jd}t |j ||jdkrRt d|d |jdt|jd}|j |se||d }|j |sq||d }t||}||_t||} |dur| }n)t|}|| krt d | d |t|||krt d |d|dt|t|||Strt|||Sd} ||d} |ddurd|d<d|i} t| } | |}| j| | d|i| d|S)a5 Constructs a sparse ``paddle.Tensor`` in coordinate format according to the indices and values of the specified non-zero elements. Args: indices(list|tuple|ndarray|Tensor): the indices of non-zero elements. Can be a list, tuple, numpy\.ndarray, paddle\.Tensor. The indices must be 2-D. values(list|tuple|ndarray|Tensor): Initial values for the tensor. Can be a scalar, list, tuple, numpy\.ndarray, paddle\.Tensor. shape(list|tuple|None, optional): The shape of the sparse tensor also represents the shape of original dense tensor. If not provided the smallest shape will be inferred to hold all elements. dtype(str|np.dtype|None, optional): The desired data type of returned tensor. Can be 'bool' , 'float16' , 'float32' , 'float64' , 'int8' , 'int16' , 'int32' , 'int64' , 'uint8', 'complex64' , 'complex128'. Default: None, infers dtype from ``data`` except for python float number which gets dtype from ``get_default_type`` . place(CPUPlace|CUDAPinnedPlace|CUDAPlace|str|None, optional): The place to allocate Tensor. Can be CPUPlace, CUDAPinnedPlace, CUDAPlace. Default: None, means global place. If ``place`` is string, It can be ``cpu``, ``gpu:x`` and ``gpu_pinned``, where ``x`` is the index of the GPUs. stop_gradient(bool, optional): Whether to block the gradient propagation of Autograd. Default: True. Returns: Tensor: A Tensor constructed from ``indices`` and ``values`` . Examples: .. code-block:: python >>> import paddle >>> indices = [[0, 1, 2], [1, 2, 0]] >>> values = [1.0, 2.0, 3.0] >>> dense_shape = [3, 3] >>> coo = paddle.sparse.sparse_coo_tensor(indices, values, dense_shape) >>> print(coo) Tensor(shape=[3, 3], dtype=paddle.float32, place=Place(cpu), stop_gradient=True, indices=[[0, 1, 2], [1, 2, 0]], values=[1., 2., 3.]) NTrr+r5rz'indices' must be 2-D.rrzBthe indices and values must have same number of non-zero, but get z and Fzthe minimum shape required is z , but get z7the number of dimensions(len(shape) must be sparse_dim(z) + dense_dim(z ), but get sparse_sparse_coo_tensor)r'r&r!out)typeinputsZoutputsattrs)rr,r)r eagerrrr r!r*r/rr+_equals_copy_torr5r(r%rr:r r Z)create_sparse_variable_for_type_inferenceZ append_op)r&r'r!rr+r5ZnnzZ sparse_dimZ dense_dimZ min_shapeZop_typer=r>helperr;rrrrSsd6             cCs|}|}t|dd|ddkd}tt|||t|kr/td|d|d}tt||d}|dkrG||gS|||gS)Nr9rz-The calculated original matrix batch size is za, but it cannot correctly split the row data. Please carefully check the data or the input shape.)r"r#sumintr r*r)crowscolsZ crows_numpyZ cols_numpyZbatchscolrowrrr_infer_dense_csr_shapes"   rIrE;list[int] | tuple[int, ...] | npt.NDArray[np.int_] | TensorrFcCst|}t|tjjst|d|dd}t|tjjs"t|d|dd}t|tjjs0t||||}t|jt|j|durRt|dkrQt|dkrQt d|nt ||}|t|d}|j |sk| |d}|j |sw| |d}|j |s| |d}t||}||_t|jdkst|jdkst|jdkrt d t|t|krt d t|dkr|jd |dkrt d |jd d |d|d d krt d|d|jd krt dn|jd |dd krt d |jd d|dtj|||||S)aG Constructs a sparse ``paddle.Tensor`` in CSR(Compressed Sparse Row) format according to the ``crows``, ``cols`` and ``values``. Currently, the crows and cols of each batch must be incremented. Args: crows(list|tuple|ndarray|Tensor): 1-D array, each element in the rows represents the starting position of the first non-zero element of each row in values. Can be a list, tuple, numpy\.ndarray, paddle\.Tensor. cols(list|tuple|ndarray|Tensor): 1-D array, the column of non-zero elements. Can be a list, tuple, numpy\.ndarray, paddle\.Tensor. values(list|tuple|ndarray|Tensor): 1-D array, the non-zero elements. Can be a scalar, list, tuple, numpy\.ndarray, paddle\.Tensor. shape(list|tuple, optional): The shape of the sparse tensor also represents the shape of original dense tensor. hold all elements. dtype(str|np.dtype|None, optional): The desired data type of returned tensor. Can be 'bool' , 'float16' , 'float32' , 'float64' , 'int8' , 'int16' , 'int32' , 'int64' , 'uint8', 'complex64' , 'complex128'. Default: None, infers dtype from ``data`` except for python float number which gets dtype from ``get_default_type`` . place(CPUPlace|CUDAPinnedPlace|CUDAPlace|str|None, optional): The place to allocate Tensor. Can be CPUPlace, CUDAPinnedPlace, CUDAPlace. Default: None, means global place. If ``place`` is string, It can be ``cpu``, ``gpu:x`` and ``gpu_pinned``, where ``x`` is the index of the GPUs. stop_gradient(bool, optional): Whether to block the gradient propagation of Autograd. Default: True. Returns: Tensor: A Tensor constructed from ``crows``, ``cols`` and ``values`` . Examples: .. code-block:: python >>> import paddle >>> crows = [0, 2, 3, 5] >>> cols = [1, 3, 2, 0, 1] >>> values = [1, 2, 3, 4, 5] >>> dense_shape = [3, 4] >>> csr = paddle.sparse.sparse_csr_tensor(crows, cols, values, dense_shape) >>> print(csr) Tensor(shape=[3, 4], dtype=paddle.int64, place=Place(cpu), stop_gradient=True, crows=[0, 2, 3, 5], cols=[1, 3, 2, 0, 1], values=[1, 2, 3, 4, 5]) NTr8rz>SparseCsrTensor only support 2-D or 3-D matrix. but get shape Frz-The 'crows', 'cols' and 'values' must be 1-D.z3the length of cols must be same as length of valuesrz The length(z%) of crows must be equal to the rows(z)+1 of matrix.z the 0th value of crows must be 0r9zs@       w