ThresholdedRelu - 10 vs 22¶
Next section compares an older to a newer version of the same operator after both definition are converted into markdown text.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.
ThresholdedRelu10 → ThresholdedRelu22
RENAMED
|
@@ -1 +1 @@
|
|
|
1
1
|
ThresholdedRelu takes one input data (Tensor<T>) and produces one output data
|
|
2
2
|
(Tensor<T>) where the rectified linear function, y = x for x > alpha, y = 0 otherwise,
|
|
3
3
|
is applied to the tensor elementwise.
|
|
4
4
|
#### Function Body
|
|
5
5
|
The function definition for this operator.
|
|
6
6
|
<
|
|
7
7
|
domain: "",
|
|
8
8
|
opset_import: ["" : 18]
|
|
9
9
|
>
|
|
10
10
|
ThresholdedRelu <alpha>(X) => (Y)
|
|
11
11
|
{
|
|
12
12
|
Alpha = Constant <value_float: float = @alpha> ()
|
|
13
13
|
AlphaCast = CastLike (Alpha, X)
|
|
14
14
|
Zero = Constant <value: tensor = float {0}> ()
|
|
15
15
|
ZeroCast = CastLike (Zero, X)
|
|
16
16
|
AlphaLessThanX = Less (AlphaCast, X)
|
|
17
17
|
Y = Where (AlphaLessThanX, X, ZeroCast)
|
|
18
18
|
}
|
|
19
19
|
### Attributes
|
|
20
20
|
* **alpha - FLOAT** (default is '1.0'):
|
|
21
21
|
Threshold value
|
|
22
22
|
### Inputs
|
|
23
23
|
- **X** (heterogeneous) - **T**:
|
|
24
24
|
Input tensor
|
|
25
25
|
### Outputs
|
|
26
26
|
- **Y** (heterogeneous) - **T**:
|
|
27
27
|
Output tensor
|
|
28
28
|
### Type Constraints
|
|
29
|
-
* **T** in ( tensor(double), tensor(float), tensor(float16) ):
|
|
29
|
+
* **T** in ( tensor(bfloat16), tensor(double), tensor(float), tensor(float16) ):
|
|
30
30
|
Constrain input and output types to float tensors.
|