mindspore.ops.SplitV

class mindspore.ops.SplitV(size_splits, split_dim, num_split)[source]

Splits the input tensor into num_split tensors along the given dimension.

The input_x tensor will be split into sub-tensors with individual shapes given by size_splits along the split dimension. This requires that input_x.shape(split_dim) is equal to the sum of size_splits.

The shape of input_x is $(x_{1}, x_{2}, . . ., x_{M}, . . ., x_{R})$ . The rank of input_x is R. Set the given split_dim as M, and $- R \leq M < R$ . Set the given num_split as N, the given size_splits as $(x_{m_{1}}, x_{m_{2}}, . . ., x_{m_{N}})$ , $x_{M} = \sum_{i = 1}^{N} x_{m_{i}}$ . The output is a list of tensor objects, for the $i$ -th tensor, it has the shape of $(x_{1}, x_{2}, . . ., x_{m_{i}}, . . ., x_{R})$ . $x_{m_{i}}$ is the $M$ -th dimension of the $i$ -th tensor. Then, the shape of the output tensor is

((x_{1}, x_{2}, . . ., x_{m_{1}}, . . ., x_{R}), (x_{1}, x_{2}, . . ., x_{m_{2}}, . . ., x_{R}), . . ., (x_{1}, x_{2}, . . ., x_{m_{N}}, . . ., x_{R}))

Parameters

size_splits (Union[tuple, list]) – The list containing the sizes of each output tensor along the split dimension. Must sum to the dimension of value along split_dim. Can contain one -1 indicating that dimension is to be inferred.
split_dim (int) – The dimension along which to split. Must be in the range [-len(input_x.shape), len(input_x.shape)).
num_split (int) – The number of output tensors. Must be positive int.

Inputs:

input_x (Tensor) - The shape of tensor is $(x_{1}, x_{2}, . . ., x_{M} . . ., x_{R})$ .

Outputs:

Tensor, a list of num_split Tensor objects with the shape $((x_{1}, x_{2}, . . ., x_{m_{1}}, . . ., x_{R}), (x_{1}, x_{2}, . . ., x_{m_{2}}, . . ., x_{R}), . . ., (x_{1}, x_{2}, . . ., x_{m_{N}}, . . ., x_{R}))$ , $x_{M} = \sum_{i = 1}^{N} x_{m_{i}}$ . The data type is the same with input_x.

Raises

TypeError – If input_x is not a Tensor.
TypeError – If size_splits is not a tuple or a list.
TypeError – If element of size_splits is not an int.
TypeError – If split_dim or num_split is not an int.
ValueError – If rank of the size_splits is not equal to num_split.
ValueError – If sum of the size_splits is not equal to the dimension of value along split_dim.
ValueError – If split_dim is out of the range [-len(input_x.shape), len(input_x.shape)).
ValueError – If the num_split is less than or equal to 0.

Supported Platforms:: Ascend

Examples

>>> input_x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), mindspore.int32)
>>> op = ops.SplitV(size_splits=[1, -1], split_dim=1, num_split=2)
>>> output = op(input_x)
>>> print(output)
(Tensor(shape=[3, 1], dtype=Int32, value=
[[1],
 [4],
 [7]]), Tensor(shape=[3, 2], dtype=Int32, value=
[[2, 3],
 [5, 6],
 [8, 9]]))
>>> input_x = Tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), mindspore.int32)
>>> op = ops.SplitV(size_splits=[2, 1], split_dim=0, num_split=2)
>>> output = op(input_x)
>>> print(output)
(Tensor(shape=[2, 3], dtype=Int32, value=
[[1, 2, 3],
 [4, 5, 6]]), Tensor(shape=[1, 3], dtype=Int32, value=
[[7, 8, 9]]))