Where - 9 vs 16#

Next section compares an older to a newer version of the same operator after both definition are converted into markdown text.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. Where9 → Where16 +1 -6
Where9 → Where16 RENAMED
@@ -1 +1 @@
1
1
  Return elements, either from X or Y, depending on condition.
2
2
  Where behaves like
3
3
  [numpy.where](https://docs.scipy.org/doc/numpy/reference/generated/numpy.where.html)
4
4
  with three parameters.
5
5
  This operator supports **multidirectional (i.e., Numpy-style) broadcasting**; for more details please check Broadcasting in ONNX <https://github.com/onnx/onnx/blob/master/docs/Broadcasting.md>_.
6
-
7
- **History**
8
- - Version 16 adds bfloat16 to the types allowed (for the second and third parameter).
9
6
  **Inputs**
10
7
  * **condition** (heterogeneous) - **B**:
11
8
  When True (nonzero), yield X, otherwise yield Y
12
9
  * **X** (heterogeneous) - **T**:
13
10
  values selected at indices where condition is True
14
11
  * **Y** (heterogeneous) - **T**:
15
12
  values selected at indices where condition is False
16
13
  **Outputs**
17
14
  * **output** (heterogeneous) - **T**:
18
15
  Tensor of shape equal to the broadcasted shape of condition, X, and
19
16
  Y.
20
17
  **Type Constraints**
21
18
  * **B** in (
22
19
  tensor(bool)
23
20
  ):
24
21
  Constrain to boolean tensors.
25
22
  * **T** in (
26
- tensor(bfloat16),
27
23
  tensor(bool),
28
24
  tensor(complex128),
29
25
  tensor(complex64),
30
26
  tensor(double),
31
27
  tensor(float),
32
28
  tensor(float16),
33
29
  tensor(int16),
34
30
  tensor(int32),
35
31
  tensor(int64),
36
32
  tensor(int8),
37
33
  tensor(string),
38
34
  tensor(uint16),
39
35
  tensor(uint32),
40
36
  tensor(uint64),
41
37
  tensor(uint8)
42
38
  ):
43
- Constrain input and output types to all tensor types (including
39
+ Constrain input and output types to all tensor types.? ^
44
- bfloat).