GraphXlaDequantize Method |
https://www.emgu.com/wiki/index.php/Emgu_TF
Takes the packed uint32 input and unpacks the input to uint8 to do
Namespace:
Emgu.TF
Assembly:
Emgu.TF.Netstandard (in Emgu.TF.Netstandard.dll) Version: 2.6.0.1294
Syntaxpublic Operation XlaDequantize(
Output input,
float min_range,
float max_range,
string mode,
bool transpose_output,
string opName = "XlaDequantize"
)
Public Function XlaDequantize (
input As Output,
min_range As Single,
max_range As Single,
mode As String,
transpose_output As Boolean,
Optional opName As String = "XlaDequantize"
) As Operation
public:
Operation^ XlaDequantize(
Output^ input,
float min_range,
float max_range,
String^ mode,
bool transpose_output,
String^ opName = L"XlaDequantize"
)
member XlaDequantize :
input : Output *
min_range : float32 *
max_range : float32 *
mode : string *
transpose_output : bool *
?opName : string
(* Defaults:
let _opName = defaultArg opName "XlaDequantize"
*)
-> Operation
Parameters
- input
- Type: Emgu.TFOutput
Input to the operation: Input tensors whose types is uint32, shape is [d0, ..., dn]. - min_range
- Type: SystemSingle
The minimum scalar value possibly produced for the input. - max_range
- Type: SystemSingle
The maximum scalar value possibly produced for the input. - mode
- Type: SystemString
String to determine the dequantize mode in {"MIN_COMBINED", "MIN_FIRST", "SCALED"}. - transpose_output
- Type: SystemBoolean
Boolean to determine if output is transposed. transpose_output is faster when input is large and rank of input is higher than 1. - opName (Optional)
- Type: SystemString
The name of the operation
Return Value
Type:
Operation
The operation, where:
[0] output(type: DtBfloat16): Output tensors whose types is bloat16. If transpose_output is true, output shape is [dn * 4, dn-1, ..., d1, d0]. If transpose_output is false, output shape is [d0,..., dn * 4].
See Also