AutoFormerV2Block Constructor
Definition
Important
Some information relates to prerelease product that may be substantially modified before it’s released. Microsoft makes no warranties, express or implied, with respect to the information provided here.
Initializes a new instance of the AutoFormerV2Block class.
public AutoFormerV2Block (int inChannels, int numHeads, int windowSize = 7, int shiftSize = 0, double mlpRatio = 4, double dropRatio = 0, int localConvSize = 3, bool useShiftWindow = false, bool useInterpolate = false);new Microsoft.ML.TorchSharp.AutoFormerV2.AutoFormerV2Block : int * int * int * int * double * double * int * bool * bool -> Microsoft.ML.TorchSharp.AutoFormerV2.AutoFormerV2BlockPublic Sub New (inChannels As Integer, numHeads As Integer, Optional windowSize As Integer = 7, Optional shiftSize As Integer = 0, Optional mlpRatio As Double = 4, Optional dropRatio As Double = 0, Optional localConvSize As Integer = 3, Optional useShiftWindow As Boolean = false, Optional useInterpolate As Boolean = false)Parameters
- inChannels
- Int32
The input channels.
- numHeads
- Int32
The number of blocks.
- windowSize
- Int32
The size of window.
- shiftSize
- Int32
The size of shift.
- mlpRatio
- Double
The ratio of MLP.
- dropRatio
- Double
The ratio of drop.
- localConvSize
- Int32
The size of local convolution.
- useShiftWindow
- Boolean
Whether use shift window.
- useInterpolate
- Boolean
Whether use interpolation.