PRelu - 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Next section compares an older to a newer version of the same operator after both definition are converted into markdown text.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.
- PRelu9 → PRelu16 +0 -4
PRelu9 → PRelu16
RENAMED
|
@@ -1 +1 @@
|
|
|
1
1
|
PRelu takes input data (Tensor<T>) and slope tensor as input, and produces one
|
|
2
2
|
output data (Tensor<T>) where the function f(x) = slope * x for x < 0,
|
|
3
3
|
f(x) = x for x >= 0., is applied to the data tensor elementwise.
|
|
4
|
-
|
|
5
|
-
**History**
|
|
6
|
-
- Version 16 adds bfloat16 to the types allowed.
|
|
7
4
|
This operator supports **unidirectional broadcasting** (tensor slope should be unidirectional broadcastable to input tensor X); for more details please check Broadcasting in ONNX <https://github.com/onnx/onnx/blob/master/docs/Broadcasting.md>_.
|
|
8
5
|
**Inputs**
|
|
9
6
|
* **X** (heterogeneous) - **T**:
|
|
10
7
|
Input tensor
|
|
11
8
|
* **slope** (heterogeneous) - **T**:
|
|
12
9
|
Slope tensor. The shape of slope can be smaller then first input X;
|
|
13
10
|
if so, its shape must be unidirectional broadcastable to X
|
|
14
11
|
**Outputs**
|
|
15
12
|
* **Y** (heterogeneous) - **T**:
|
|
16
13
|
Output tensor (same size as X)
|
|
17
14
|
**Type Constraints**
|
|
18
15
|
* **T** in (
|
|
19
|
-
tensor(bfloat16),
|
|
20
16
|
tensor(double),
|
|
21
17
|
tensor(float),
|
|
22
18
|
tensor(float16),
|
|
23
19
|
tensor(int32),
|
|
24
20
|
tensor(int64),
|
|
25
21
|
tensor(uint32),
|
|
26
22
|
tensor(uint64)
|
|
27
23
|
):
|
|
28
24
|
Constrain input and output types to float/int tensors.
|