LogSoftmax#
LogSoftmax operation applies the \(\log(softmax(src))\) function to an n-dimensional input Tensor. The formulation can be simplified as:
\[\dst_i = \log\Big( \frac{e^(\src_i)}{\sum_{j}^{ } e^(\src_j)} \Big)\]
Operation Attributes#
|
Description |
Value Type |
|
|
|---|---|---|---|---|
Represents the axis of which the LogSoftmax is calculated. |
s64 |
Arbitrary
s64 value
( |
Optional |
Execution Arguments#
The inputs and outputs must be provided according to the below index order when constructing an operation.
Inputs#
Index |
Argument Name |
Required or Optional |
|---|---|---|
0 |
|
Required |
Outputs#
Index |
Argument Name |
Required or Optional |
|---|---|---|
0 |
|
Required |
Supported Data Types#
LogSoftmax operation supports the following data type combinations.
Src |
Dst |
|---|---|
f32 |
f32 |
bf16 |
bf16 |
f16 |
f16 |