GlobalLpPool - 1 vs 2

Files changed (1) hide show
  1. GlobalLpPool1 → GlobalLpPool2 +7 -6
GlobalLpPool1 → GlobalLpPool2 RENAMED
@@ -1 +1 @@
1
- GlobalLpPool consumes an input tensor X and applies lp pool pooling across the
1
+ GlobalLpPool consumes an input tensor X and applies lp pool pooling across
2
2
  the values in the same channel. This is equivalent to LpPool with kernel size
3
3
  equal to the spatial dimension of input tensor.
4
4
  **Attributes**
5
5
  * **p**:
6
- p value of the Lp norm used to pool over the input data, default is
6
+ p value of the Lp norm used to pool over the input data.
7
- 2.0.
8
7
  **Inputs**
9
8
  * **X** (heterogeneous) - **T**:
10
9
  Input data tensor from the previous operator; dimensions for image
11
10
  case are (N x C x H x W), where N is the batch size, C is the number
12
11
  of channels, and H and W are the height and the width of the data.
13
- For non image case, the dimension are in the form of (N x C x D1 x
12
+ For non image case, the dimensions are in the form of (N x C x D1 x
14
13
  D2 ... Dn), where N is the batch size.
15
14
  **Outputs**
16
15
  * **Y** (heterogeneous) - **T**:
17
- Output data tensor from pooling across the input tensor. Dimensions
16
+ Output data tensor from pooling across the input tensor. The output
17
+ tensor has the same rank as the input. The first two dimensions of
18
+ output shape are the same as the input (N x C), while the other
18
- will be N x C x 1 x 1
19
+ dimensions are all 1.
19
20
  **Type Constraints**
20
21
  * **T** in (
21
22
  tensor(double),
22
23
  tensor(float),
23
24
  tensor(float16)
24
25
  ):
25
26
  Constrain input and output types to float tensors.