Upsample - 1 vs 9#

Next section compares an older to a newer version of the same operator after both definition are converted into markdown text. Green means an addition to the newer version, red means a deletion. Anything else is unchanged.

Files changed (1) hide show
  1. Upsample1 → Upsample9 +30 -21
Upsample1 → Upsample9 RENAMED
@@ -1 +1 @@
1
1
  Upsample the input tensor.
2
- Each dimension value of the output tensor is:
2
+ The width and height of the output tensor are:
3
- output_dimension = floor(input_dimension * scale).
3
+ output_width = floor(input_width * width_scale),
4
+ output_height = floor(input_height * height_scale).
5
+ Example:
6
+ Given data tensor, width_scale, height_scale, mode,
7
+ Upsample the input 4-D tensor in nearest mode:
8
+ data = [[[
9
+ [1, 2],
10
+ [3, 4]
11
+ ]]]
12
+ width_scale = 2
13
+ height_scale = 2
14
+ mode = "nearest"
15
+ output = [[[
16
+ [1, 1, 2, 2],
17
+ [1, 1, 2, 2],
18
+ [3, 3, 4, 4],
19
+ [3, 3, 4, 4]
20
+ ]]]
4
21
  **Attributes**
22
+ * **height_scale** (required):
23
+ The scale along height dimension. It takes value greater than or
24
+ equal to 1.
5
25
  * **mode**:
6
- Two interpolation modes: nearest (default), and linear (including
26
+ Two interpolation modes: nearest(default), bilinear
7
- bilinear, trilinear, etc)
27
+ * **width_scale** (required):
28
+ The scale along width dimension. It takes value greater than or
29
+ equal to 1.
8
30
  **Inputs**
9
31
  * **X** (heterogeneous) - **T**:
32
+ 4-D tensor, [N,C,H,W]
10
- N-D tensor
11
- * **scales** (heterogeneous) - **tensor(float)**:
12
- The scale array along each dimension. It takes value greater than or
13
- equal to 1. The number of elements of 'scales' should be the same as
14
- the rank of input 'X'.
15
33
  **Outputs**
16
34
  * **Y** (heterogeneous) - **T**:
17
- N-D tensor after resizing
35
+ 4-D tensor after resizing, [N,C,H,W]
18
36
  **Type Constraints**
19
37
  * **T** in (
20
38
  tensor(bool),
21
- tensor(complex128),
22
- tensor(complex64),
23
39
  tensor(double),
24
40
  tensor(float),
25
41
  tensor(float16),
26
- tensor(int16),
27
42
  tensor(int32),
28
- tensor(int64),
43
+ tensor(int64)
29
- tensor(int8),
30
- tensor(string),
31
- tensor(uint16),
32
- tensor(uint32),
33
- tensor(uint64),
34
- tensor(uint8)
35
44
  ):
36
- Constrain input 'X' and output 'Y' to all tensor types.+ Constrain output types to bool, int32, int64, float16, float, double
45
+ tensors.