dequantize(_:scaleTensor:zeroPointTensor:dataType:axis:name:)
Creates a dequantize operation and returns the result tensor.
Declaration
func dequantize(_ tensor: MPSGraphTensor, scaleTensor: MPSGraphTensor, zeroPointTensor: MPSGraphTensor, dataType: MPSDataType, axis: Int, name: String?) -> MPSGraphTensorParameters
- tensor:
Input tensor to be dequantized
- scaleTensor:
Scale scalar or 1D Tensor parameter with size == tensor.shape[axis]
- zeroPointTensor:
Bias scalar or 1D Tensor parameter with size == tensor.shape[axis]
- dataType:
Float data type of the result tensor.
- axis:
Axis on which the scale 1D value is being broadcasted
- name:
The name for the operation.
Return Value
A valid MPSGraphTensor array of datatype dataType
Discussion
Convert the quantized tensor to a float tensor by applying a scale + bias transform: result = scaleTensor * (tensor - zeroPointTensor)
Supported input types: MPSDataTypeInt4, MPSDataTypeUInt4, MPSDataTypeInt8, MPSDataTypeUInt8, MPSDataTypeFloat8E4M3, MPSDataTypeFloat8E5M2. The scaleTensor type and dataType must match (MPSDataTypeFloat16, MPSDataTypeFloat32, or MPSDataTypeBFloat16).