ConcatOperation

Applicable Products

Hardware Model	Supported
Atlas 350 accelerator cards	√
Atlas A3 inference products/Atlas A3 training products	√
Atlas A2 training products/Atlas A2 inference products	√
Atlas training products	√
Atlas inference products	√
Atlas 200I/500 A2 inference products	√

Description

Combines two input tensors into an output tensor at a specified dimension.

Figure 1 Implementation principle of the Concat operator

Operator Context

Figure 2 ConcatOperation

Definition

struct ConcatParam {
    int concatDim = 0;
    uint8_t rsv[12] = {0};
};

Parameters

Member	Type	Default Value	Description
concatDim	int	0	Dimension index for concat. Assume that the dimension index for concatenation is k and the number of dimensions of x is dimNum: When concatDim is a negative number, the access starts from the highest dimension. k = dimNum + concatDim. For example, if concatDim is -1, the concatenation dimension k is dimNum - 1. Otherwise, k = concatDim.
rsv[12]	uint8_t	{0}	Reserved

Member

Type

Default Value

Description

concatDim

int

Dimension index for concat.

Assume that the dimension index for concatenation is k and the number of dimensions of x is dimNum:

When concatDim is a negative number, the access starts from the highest dimension. k = dimNum + concatDim. For example, if concatDim is -1, the concatenation dimension k is dimNum - 1.
Otherwise, k = concatDim.

rsv[12]

uint8_t

{0}

Reserved

Input

Parameter	Dimension	Data Type	Format	Description
x	[x_dim_0, x_dim_1, ... , x_dim_n]	float16/bf16	ND	Tensor to be concatenated.
y	[y_dim_0, y_dim_1, ... , y_dim_n]	float16/bf16	ND	Tensor to be concatenated. The dimensions except y_dim_k are the same as those of x.

Output

Parameter	Dimension	Data Type	Format	Description
output	[dim_0, dim_1, ..., dim_n]	float16/bf16	ND	Concatenated tensor.

Constraints

Restrictions on the parameters:
- dimNum: number of dimensions of the concatenated tensor.
- -dimNum ≤ concatDim ≤ dimNum - 1.

Input constraints
- The inputs x and y have the same dimensions.
- The dimension size of the inputs x and y must be the same except the concatDim dimension.

API Calling Example

Input

x shape = [3, 2, 3]:
[[[1.0, 1.0, 1.0],
  [1.0, 1.0, 1.0]],
 [[1.0, 1.0, 1.0],
  [1.0, 1.0, 1.0]],
 [[1.0, 1.0, 1.0],
  [1.0, 1.0, 1.0]]]
y shape = [3, 1, 3]:
[[[2.0, 2.0, 2.0]],
 [[2.0, 2.0, 2.0]],
 [[2.0, 2.0, 2.0]]]
param.concatDim = 1 (or -2)

Output

output shape = [3, 3, 3]
[[[1.0, 1.0, 1.0],
  [1.0, 1.0, 1.0],
  [2.0, 2.0, 2.0]],
 [[1.0, 1.0, 1.0],
  [1.0, 1.0, 1.0],
  [2.0, 2.0, 2.0]],
 [[1.0, 1.0, 1.0],
  [1.0, 1.0, 1.0],
  [2.0, 2.0, 2.0]]]

Parent topic: atb/infer_op_params.h