SpaceToDepth - 1 vs 13#
Next section compares an older to a newer version of the same operator after both definition are converted into markdown text.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.
SpaceToDepth1 → SpaceToDepth13
RENAMED
|
@@ -1 +1 @@
|
|
|
1
1
|
SpaceToDepth rearranges blocks of spatial data into depth. More specifically,
|
|
2
2
|
this op outputs a copy of the input tensor where values from the height and width dimensions
|
|
3
3
|
are moved to the depth dimension.
|
|
4
4
|
**Attributes**
|
|
5
5
|
* **blocksize** (required):
|
|
6
6
|
Blocks of [blocksize, blocksize] are moved.
|
|
7
7
|
**Inputs**
|
|
8
8
|
* **input** (heterogeneous) - **T**:
|
|
9
9
|
Input tensor of [N,C,H,W], where N is the batch axis, C is the
|
|
10
10
|
channel or depth, H is the height and W is the width.
|
|
11
11
|
**Outputs**
|
|
12
12
|
* **output** (heterogeneous) - **T**:
|
|
13
13
|
Output tensor of [N, C * blocksize * blocksize, H/blocksize,
|
|
14
14
|
W/blocksize].
|
|
15
15
|
**Type Constraints**
|
|
16
16
|
* **T** in (
|
|
17
|
-
tensor(bfloat16),
|
|
18
17
|
tensor(bool),
|
|
19
18
|
tensor(complex128),
|
|
20
19
|
tensor(complex64),
|
|
21
20
|
tensor(double),
|
|
22
21
|
tensor(float),
|
|
23
22
|
tensor(float16),
|
|
24
23
|
tensor(int16),
|
|
25
24
|
tensor(int32),
|
|
26
25
|
tensor(int64),
|
|
27
26
|
tensor(int8),
|
|
28
27
|
tensor(string),
|
|
29
28
|
tensor(uint16),
|
|
30
29
|
tensor(uint32),
|
|
31
30
|
tensor(uint64),
|
|
32
31
|
tensor(uint8)
|
|
33
32
|
):
|
|
34
33
|
Constrain input and output types to all tensor types.
|