Linux and UNIX Man Pages

Linux & Unix Commands - Search Man Pages

mpsnnstatenode(3) [mojave man page]

MPSNNStateNode(3)					 MetalPerformanceShaders.framework					 MPSNNStateNode(3)

NAME
MPSNNStateNode SYNOPSIS
#import <MPSNNGraphNodes.h> Inherits NSObject. Inherited by MPSNNBinaryGradientStateNode, MPSNNGradientStateNode, and MPSNNLabelsNode. Instance Methods (nonnull instancetype) - init Properties id< MPSHandle > handle BOOL exportFromGraph BOOL synchronizeResource Detailed Description A placeholder node denoting the position in the graph of a MPSState object Some filters need additional information about an image in order to function. For example a max-pooling gradient filter needs to know which position the max result came from in the original pooling filter in order to select the right data for gradient computation. In other cases, state may be moved into a MPSState object in order to keep the filter itself immutable. The MPSState object typically encapsulates one or more MTLResource objects. Method Documentation - (nonnull instancetype) init Property Documentation - (BOOL) exportFromGraph [read], [write], [nonatomic], [assign] Tag a state node for view later Most state nodes are private to the graph. These alias memory heavily and consequently generally have invalid state when the graph exits. When exportFromGraph = YES, the image is preserved and made available through the [MPSNNGraph encode... resultStates:... list. CAUTION: exporting an state from a graph prevents MPS from recycling memory. It will nearly always cause the amount of memory used by the graph to increase by the size of the state. There will probably be a performance regression accordingly. This feature should generally be used only when the node is needed as an input for further work and recomputing it is prohibitively costly. Default: NO - (id<MPSHandle>) handle [read], [write], [nonatomic], [retain] MPS resource identification See MPSHandle protocol reference. Default: nil - (BOOL) synchronizeResource [read], [write], [nonatomic], [assign] Set to true to cause the resource to be synchronized with the CPU Ignored on non-MacOS. Author Generated automatically by Doxygen for MetalPerformanceShaders.framework from the source code. Version MetalPerformanceShaders-100 Thu Feb 8 2018 MPSNNStateNode(3)

Check Out this Related Man Page

MPSCNNLossNode(3)					 MetalPerformanceShaders.framework					 MPSCNNLossNode(3)

NAME
MPSCNNLossNode SYNOPSIS
#import <MPSNNGraphNodes.h> Inherits MPSNNFilterNode. Instance Methods (nonnull instancetype) - initWithSource:lossDescriptor: (MPSNNGradientFilterNode *__nonnull) - gradientFilterWithSources: Class Methods (nonnull instancetype) + nodeWithSource:lossDescriptor: Properties MPSNNLabelsNode * inputLabels Detailed Description This node calculates loss information during training typically immediately after the inference portion of network evaluation is performed. The result image of the loss operations is typically the first gradient image to be comsumed by the gradient passes that work their way back up the graph. In addition, the node will update the loss image in the MPSNNLabels with the desired estimate of correctness. Method Documentation - (MPSNNGradientFilterNode*__nonnull) gradientFilterWithSources: (NSArray< MPSNNImageNode * > *__nonnull) gradientImages The loss filter is its own gradient filter and doesn't provide a corresponding gradient node. The image returned by the loss filter is the gradient image to be consumed by the gradient filters corresponding to preceeding inference nodes. Reimplemented from MPSNNFilterNode. - (nonnull instancetype) initWithSource: (MPSNNImageNode *__nonnull) source(MPSCNNLossDescriptor *__nonnull) descriptor + (nonnull instancetype) nodeWithSource: (MPSNNImageNode *__nonnull) source(MPSCNNLossDescriptor *__nonnull) descriptor Property Documentation - (MPSNNLabelsNode*) inputLabels [read], [nonatomic], [retain] Get the input node for labes and weights, for example to set the handle Author Generated automatically by Doxygen for MetalPerformanceShaders.framework from the source code. Version MetalPerformanceShaders-100 Thu Feb 8 2018 MPSCNNLossNode(3)
Man Page