Splits input tensor across all dimensions.
An op which slices the input tensor based on the given num_splits attribute, pads slices optionally, and returned the slices. Slices are returned in row-major order.
This op may be generated via the TPU bridge.
For example, with `input` tensor:
[[0, 1, 2],
[3, 4, 5],
[6, 7, 8]]
`num_splits`:
[2, 2]
and `paddings`:
[1, 1]
the expected `outputs` is:
[[0, 1],
[3, 4]]
[[2, 0],
[5, 0]]
[[6, 7],
[0, 0]]
[[8, 0],
[0, 0]]
Nested Classes
class | XlaSplitND.Options | Optional attributes for XlaSplitND
|
Public Methods
static <T> XlaSplitND<T> |
create(Scope scope, Operand<T> input, Long N, List<Long> numSplits, Options... options)
Factory method to create a class wrapping a new XlaSplitND operation.
|
Iterator<Operand<T>> |
iterator()
|
List<Output<T>> |
outputs()
|
static XlaSplitND.Options |
paddings(List<Long> paddings)
|
Inherited Methods
Public Methods
public static XlaSplitND<T> create (Scope scope, Operand<T> input, Long N, List<Long> numSplits, Options... options)
Factory method to create a class wrapping a new XlaSplitND operation.
Parameters
scope | current scope |
---|---|
input | Input tensor to split across all dimensions.
}
out_arg {
name: "outputs"
description: < |
numSplits | Number of ways to split per dimension. Shape dimensions must be evenly divisible. |
options | carries optional attributes values |
Returns
- a new instance of XlaSplitND
public static XlaSplitND.Options paddings (List<Long> paddings)
Parameters
paddings | Optional list of right paddings per dimension of input tensor to apply before splitting. This can be used to make a dimension evenly divisible. |
---|