caffe2::AbstractSortedSegmentDef< T, SIndex, Context, ReducerDef > Struct Template Reference

Public Types
using	OpDef = ReducerDef
using	Reducer = typename ReducerDef::template Reducer< T, Context >
using	ReducerGradient = typename ReducerDef::template ReducerGradient< T, Context >
using	ForwardOp = AbstractSortedSegmentOp< T, SIndex, Context, Reducer, false >
using	BackwardOp = AbstractSortedSegmentGradientOp< T, SIndex, Context, ReducerGradient >
using	GetGradient = SegmentOpGetGradient< ForwardOp, ReducerDef, ReducerGradient, true, false >

Static Public Member Functions
static void	PopulateSchema (OpSchema &schema)

Static Public Attributes
static constexpr const char *	basename = "SortedSegment"
static constexpr const char *	doc

Detailed Description

template<typename T, typename SIndex, typename Context, typename ReducerDef>
struct caffe2::AbstractSortedSegmentDef< T, SIndex, Context, ReducerDef >

Definition at line 847 of file segment_reduction_op.h.

Field Documentation

template<typename T , typename SIndex , typename Context , typename ReducerDef >

constexpr const char* caffe2::AbstractSortedSegmentDef< T, SIndex, Context, ReducerDef >::doc

static

Initial value:

= R"DOC(
Applies '{op}' to each segment of input tensor. Segments need to be sorted and
contiguous. See also UnsortedSegment{op} that doesn't have this requirement.

SEGMENT_IDS is a vector that maps each of the first dimension slices of the
DATA to a particular group (segment). Values belonging to the same segment are
aggregated together.

The first dimension of the output is equal to the number of input segments,
i.e. `SEGMENT_IDS[-1]+1`. Other dimensions are inherited from the input tensor.

{op_doc}
  )DOC"

Definition at line 850 of file segment_reduction_op.h.

---

The documentation for this struct was generated from the following file:

• caffe2/operators/segment_reduction_op.h