MPSNNOptimizerAdam 构造函数

定义

重载

MPSNNOptimizerAdam(NSCoder)
MPSNNOptimizerAdam(NSObjectFlag)
MPSNNOptimizerAdam(IntPtr)
MPSNNOptimizerAdam(IMTLDevice, Single)
MPSNNOptimizerAdam(IMTLDevice, Double, Double, Single, nuint, MPSNNOptimizerDescriptor)

MPSNNOptimizerAdam(NSCoder)

[Foundation.Export("initWithCoder:")]
[ObjCRuntime.DesignatedInitializer]
public MPSNNOptimizerAdam (Foundation.NSCoder coder);
new MetalPerformanceShaders.MPSNNOptimizerAdam : Foundation.NSCoder -> MetalPerformanceShaders.MPSNNOptimizerAdam

参数

coder
NSCoder
属性

适用于

MPSNNOptimizerAdam(NSObjectFlag)

protected MPSNNOptimizerAdam (Foundation.NSObjectFlag t);
new MetalPerformanceShaders.MPSNNOptimizerAdam : Foundation.NSObjectFlag -> MetalPerformanceShaders.MPSNNOptimizerAdam

参数

适用于

MPSNNOptimizerAdam(IntPtr)

protected internal MPSNNOptimizerAdam (IntPtr handle);
new MetalPerformanceShaders.MPSNNOptimizerAdam : nativeint -> MetalPerformanceShaders.MPSNNOptimizerAdam

参数

handle
IntPtr

nativeint

适用于

MPSNNOptimizerAdam(IMTLDevice, Single)

[Foundation.Export("initWithDevice:learningRate:")]
public MPSNNOptimizerAdam (Metal.IMTLDevice device, float learningRate);
new MetalPerformanceShaders.MPSNNOptimizerAdam : Metal.IMTLDevice * single -> MetalPerformanceShaders.MPSNNOptimizerAdam

参数

device
IMTLDevice
learningRate
Single
属性

适用于

MPSNNOptimizerAdam(IMTLDevice, Double, Double, Single, nuint, MPSNNOptimizerDescriptor)

[Foundation.Export("initWithDevice:beta1:beta2:epsilon:timeStep:optimizerDescriptor:")]
public MPSNNOptimizerAdam (Metal.IMTLDevice device, double beta1, double beta2, float epsilon, nuint timeStep, MetalPerformanceShaders.MPSNNOptimizerDescriptor optimizerDescriptor);
new MetalPerformanceShaders.MPSNNOptimizerAdam : Metal.IMTLDevice * double * double * single * nuint * MetalPerformanceShaders.MPSNNOptimizerDescriptor -> MetalPerformanceShaders.MPSNNOptimizerAdam

参数

device
IMTLDevice
beta1
Double
beta2
Double
epsilon
Single
timeStep
System.System.UIntPtr System.unativeint
optimizerDescriptor
MPSNNOptimizerDescriptor
属性

适用于