apple.mlcompute

Class MLCGraph

java.lang.Object

org.moe.natj.general.NativeObject

org.moe.natj.objc.ObjCObject

apple.NSObject

apple.mlcompute.MLCGraph

All Implemented Interfaces:

NSObject

Direct Known Subclasses:

MLCInferenceGraph, MLCTrainingGraph

public class MLCGraph
extends NSObject

MLCGraph

A graph of layers that can be used to build a training or inference graph

Nested Class Summary

Nested classes/interfaces inherited from class apple.NSObject

NSObject.Function_instanceMethodForSelector_ret, NSObject.Function_methodForSelector_ret

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	MLCGraph(org.moe.natj.general.Pointer peer)

Method Summary

Modifier and Type	Method and Description
static boolean	accessInstanceVariablesDirectly()
static MLCGraph	alloc()
static MLCGraph	allocWithZone(org.moe.natj.general.ptr.VoidPtr zone)
static boolean	automaticallyNotifiesObserversForKey(java.lang.String key)
boolean	bindAndWriteDataForInputsToDeviceBatchSizeSynchronous(NSDictionary<java.lang.String,? extends MLCTensorData> inputsData, NSDictionary<java.lang.String,? extends MLCTensor> inputTensors, MLCDevice device, long batchSize, boolean synchronous) Associates data with input tensors.
boolean	bindAndWriteDataForInputsToDeviceSynchronous(NSDictionary<java.lang.String,? extends MLCTensorData> inputsData, NSDictionary<java.lang.String,? extends MLCTensor> inputTensors, MLCDevice device, boolean synchronous) Associates data with input tensors.
static void	cancelPreviousPerformRequestsWithTarget(java.lang.Object aTarget)
static void	cancelPreviousPerformRequestsWithTargetSelectorObject(java.lang.Object aTarget, org.moe.natj.objc.SEL aSelector, java.lang.Object anArgument)
static NSArray<java.lang.String>	classFallbacksForKeyedArchiver()
static org.moe.natj.objc.Class	classForKeyedUnarchiver()
MLCTensor	concatenateWithSourcesDimension(NSArray<? extends MLCTensor> sources, long dimension) Add a concat layer to the graph
static java.lang.String	debugDescription_static()
static java.lang.String	description_static()
MLCDevice	device() The device to be used when compiling and executing a graph
MLCTensor	gatherWithDimensionSourceIndices(long dimension, MLCTensor source, MLCTensor indices) Add a gather layer to the graph
static MLCGraph	graph() Creates a new graph.
static long	hash_static()
MLCGraph	init()
static NSObject.Function_instanceMethodForSelector_ret	instanceMethodForSelector(org.moe.natj.objc.SEL aSelector)
static NSMethodSignature	instanceMethodSignatureForSelector(org.moe.natj.objc.SEL aSelector)
static boolean	instancesRespondToSelector(org.moe.natj.objc.SEL aSelector)
static boolean	isSubclassOfClass(org.moe.natj.objc.Class aClass)
static NSSet<java.lang.String>	keyPathsForValuesAffectingValueForKey(java.lang.String key)
NSArray<? extends MLCLayer>	layers() Layers in the graph
static MLCGraph	new_objc()
MLCTensor	nodeWithLayerSource(MLCLayer layer, MLCTensor source) Add a layer to the graph
MLCTensor	nodeWithLayerSources(MLCLayer layer, NSArray<? extends MLCTensor> sources) Add a layer to the graph
MLCTensor	nodeWithLayerSourcesDisableUpdate(MLCLayer layer, NSArray<? extends MLCTensor> sources, boolean disableUpdate) Add a layer to the graph
MLCTensor	nodeWithLayerSourcesLossLabels(MLCLayer layer, NSArray<? extends MLCTensor> sources, NSArray<? extends MLCTensor> lossLabels) Add a loss layer to the graph
MLCTensor	reshapeWithShapeSource(NSArray<? extends NSNumber> shape, MLCTensor source) Add a reshape layer to the graph
static boolean	resolveClassMethod(org.moe.natj.objc.SEL sel)
static boolean	resolveInstanceMethod(org.moe.natj.objc.SEL sel)
NSArray<? extends MLCTensor>	resultTensorsForLayer(MLCLayer layer) Get the result tensors for a layer in the training graph
MLCTensor	scatterWithDimensionSourceIndicesCopyFromReductionType(long dimension, MLCTensor source, MLCTensor indices, MLCTensor copyFrom, int reductionType) Add a scatter layer to the graph
MLCTensor	selectWithSourcesCondition(NSArray<? extends MLCTensor> sources, MLCTensor condition) Add a select layer to the graph
static void	setVersion_static(long aVersion)
NSArray<? extends MLCTensor>	sourceTensorsForLayer(MLCLayer layer) Get the source tensors for a layer in the training graph
NSArray<? extends MLCTensor>	splitWithSourceSplitCountDimension(MLCTensor source, long splitCount, long dimension) Add a split layer to the graph
NSArray<? extends MLCTensor>	splitWithSourceSplitSectionLengthsDimension(MLCTensor source, NSArray<? extends NSNumber> splitSectionLengths, long dimension) Add a split layer to the graph
java.lang.String	summarizedDOTDescription() A DOT representation of the graph.
static org.moe.natj.objc.Class	superclass_static()
MLCTensor	transposeWithDimensionsSource(NSArray<? extends NSNumber> dimensions, MLCTensor source) Add a transpose layer to the graph
static long	version_static()

Methods inherited from class apple.NSObject

accessibilityActivate, accessibilityActivationPoint, accessibilityAssistiveTechnologyFocusedIdentifiers, accessibilityAttributedHint, accessibilityAttributedLabel, accessibilityAttributedUserInputLabels, accessibilityAttributedValue, accessibilityContainerType, accessibilityCustomActions, accessibilityCustomRotors, accessibilityDecrement, accessibilityDragSourceDescriptors, accessibilityDropPointDescriptors, accessibilityElementAtIndex, accessibilityElementCount, accessibilityElementDidBecomeFocused, accessibilityElementDidLoseFocus, accessibilityElementIsFocused, accessibilityElements, accessibilityElementsHidden, accessibilityFrame, accessibilityHint, accessibilityIncrement, accessibilityLabel, accessibilityLanguage, accessibilityNavigationStyle, accessibilityPath, accessibilityPerformEscape, accessibilityPerformMagicTap, accessibilityRespondsToUserInteraction, accessibilityScroll, accessibilityTextualContext, accessibilityTraits, accessibilityUserInputLabels, accessibilityValue, accessibilityViewIsModal, addObserverForKeyPathOptionsContext, attemptRecoveryFromErrorOptionIndex, attemptRecoveryFromErrorOptionIndexDelegateDidRecoverSelectorContextInfo, autoContentAccessingProxy, awakeAfterUsingCoder, awakeFromNib, class_objc, classForCoder, classForKeyedArchiver, copy, dealloc, debugDescription, description, dictionaryWithValuesForKeys, didChangeValueForKey, didChangeValueForKeyWithSetMutationUsingObjects, didChangeValuesAtIndexesForKey, doesNotRecognizeSelector, fileManagerShouldProceedAfterError, fileManagerWillProcessPath, finalize_objc, forwardingTargetForSelector, forwardInvocation, hash, indexOfAccessibilityElement, isAccessibilityElement, isEqual, isKindOfClass, isMemberOfClass, isProxy, methodForSelector, methodSignatureForSelector, mutableArrayValueForKey, mutableArrayValueForKeyPath, mutableCopy, mutableOrderedSetValueForKey, mutableOrderedSetValueForKeyPath, mutableSetValueForKey, mutableSetValueForKeyPath, observationInfo, observeValueForKeyPathOfObjectChangeContext, performSelector, performSelectorInBackgroundWithObject, performSelectorOnMainThreadWithObjectWaitUntilDone, performSelectorOnMainThreadWithObjectWaitUntilDoneModes, performSelectorOnThreadWithObjectWaitUntilDone, performSelectorOnThreadWithObjectWaitUntilDoneModes, performSelectorWithObject, performSelectorWithObjectAfterDelay, performSelectorWithObjectAfterDelayInModes, performSelectorWithObjectWithObject, prepareForInterfaceBuilder, provideImageDataBytesPerRowOrigin_Size_UserInfo, removeObserverForKeyPath, removeObserverForKeyPathContext, replacementObjectForCoder, replacementObjectForKeyedArchiver, respondsToSelector, self, setAccessibilityActivationPoint, setAccessibilityAttributedHint, setAccessibilityAttributedLabel, setAccessibilityAttributedUserInputLabels, setAccessibilityAttributedValue, setAccessibilityContainerType, setAccessibilityCustomActions, setAccessibilityCustomRotors, setAccessibilityDragSourceDescriptors, setAccessibilityDropPointDescriptors, setAccessibilityElements, setAccessibilityElementsHidden, setAccessibilityFrame, setAccessibilityHint, setAccessibilityLabel, setAccessibilityLanguage, setAccessibilityNavigationStyle, setAccessibilityPath, setAccessibilityRespondsToUserInteraction, setAccessibilityTextualContext, setAccessibilityTraits, setAccessibilityUserInputLabels, setAccessibilityValue, setAccessibilityViewIsModal, setIsAccessibilityElement, setNilValueForKey, setObservationInfo, setShouldGroupAccessibilityChildren, setValueForKey, setValueForKeyPath, setValueForUndefinedKey, setValuesForKeysWithDictionary, shouldGroupAccessibilityChildren, superclass, validateValueForKeyError, validateValueForKeyPathError, valueForKey, valueForKeyPath, valueForUndefinedKey, willChangeValueForKey, willChangeValueForKeyWithSetMutationUsingObjects, willChangeValuesAtIndexesForKey

Methods inherited from class org.moe.natj.objc.ObjCObject

equals, toString

Methods inherited from class org.moe.natj.general.NativeObject

finalize, getPeer, getPeerPointer

Methods inherited from class java.lang.Object

clone, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

MLCGraph

protected MLCGraph(org.moe.natj.general.Pointer peer)

Method Detail

accessInstanceVariablesDirectly

public static boolean accessInstanceVariablesDirectly()

alloc

public static MLCGraph alloc()

allocWithZone

public static MLCGraph allocWithZone(org.moe.natj.general.ptr.VoidPtr zone)

automaticallyNotifiesObserversForKey

public static boolean automaticallyNotifiesObserversForKey(java.lang.String key)

bindAndWriteDataForInputsToDeviceBatchSizeSynchronous

public boolean bindAndWriteDataForInputsToDeviceBatchSizeSynchronous(NSDictionary<java.lang.String,? extends MLCTensorData> inputsData,
                                                                     NSDictionary<java.lang.String,? extends MLCTensor> inputTensors,
                                                                     MLCDevice device,
                                                                     long batchSize,
                                                                     boolean synchronous)

Associates data with input tensors. If the device is GPU, also copies the data to the device memory. Returns true if the data is successfully associated with input tensors.

This function should be used if you execute the forward, gradient and optimizer updates independently. Before the forward pass is executed, the inputs should be written to device memory. Similarly, before the gradient pass is executed, the inputs (typically the initial gradient tensor) should be written to device memory. The caller must guarantee the lifetime of the underlying memory of each value of \p inputsData for the entirety of each corresponding input tensor's lifetime.

Parameters:

inputsData - The input data to use to write to device memory

inputTensors - The list of tensors to perform writes on

device - The device

batchSize - The batch size. This should be set to the actual batch size that may be used when we execute the graph and can be a value less than or equal to the batch size specified in the tensor. If set to 0, we use batch size specified in the tensor.

synchronous - Whether to execute the copy to the device synchronously. For performance, asynchronous execution is recommended.

Returns:

A Boolean value indicating whether the data is successfully associated with the tensor.

bindAndWriteDataForInputsToDeviceSynchronous

public boolean bindAndWriteDataForInputsToDeviceSynchronous(NSDictionary<java.lang.String,? extends MLCTensorData> inputsData,
                                                            NSDictionary<java.lang.String,? extends MLCTensor> inputTensors,
                                                            MLCDevice device,
                                                            boolean synchronous)

Associates data with input tensors. If the device is GPU, also copies the data to the device memory. Returns true if the data is successfully associated with input tensors.

This function should be used if you execute the forward, gradient and optimizer updates independently. Before the forward pass is executed, the inputs should be written to device memory. Similarly, before the gradient pass is executed, the inputs (typically the initial gradient tensor) should be written to device memory. The caller must guarantee the lifetime of the underlying memory of each value of \p inputsData for the entirety of each corresponding input tensor's lifetime.

Parameters:

inputsData - The input data to use to write to device memory

inputTensors - The list of tensors to perform writes on

device - The device

synchronous - Whether to execute the copy to the device synchronously. For performance, asynchronous execution is recommended.

Returns:

A Boolean value indicating whether the data is successfully associated with the tensor.

cancelPreviousPerformRequestsWithTarget

public static void cancelPreviousPerformRequestsWithTarget(java.lang.Object aTarget)

cancelPreviousPerformRequestsWithTargetSelectorObject

public static void cancelPreviousPerformRequestsWithTargetSelectorObject(java.lang.Object aTarget,
                                                                         org.moe.natj.objc.SEL aSelector,
                                                                         java.lang.Object anArgument)

classFallbacksForKeyedArchiver

public static NSArray<java.lang.String> classFallbacksForKeyedArchiver()

classForKeyedUnarchiver

public static org.moe.natj.objc.Class classForKeyedUnarchiver()

concatenateWithSourcesDimension

public MLCTensor concatenateWithSourcesDimension(NSArray<? extends MLCTensor> sources,
                                                 long dimension)

Add a concat layer to the graph

Parameters:

sources - The source tensors to concatenate

dimension - The concatenation dimension

Returns:

A result tensor

debugDescription_static

public static java.lang.String debugDescription_static()

description_static

public static java.lang.String description_static()

device

public MLCDevice device()

The device to be used when compiling and executing a graph

gatherWithDimensionSourceIndices

public MLCTensor gatherWithDimensionSourceIndices(long dimension,
                                                  MLCTensor source,
                                                  MLCTensor indices)

Add a gather layer to the graph

Parameters:

dimension - The dimension along which to index

source - The source tensor

indices - The index of elements to gather

Returns:

A result tensor

graph

public static MLCGraph graph()

Creates a new graph.

Returns:

A new graph.

hash_static

public static long hash_static()

init

public MLCGraph init()

Overrides:

init in class NSObject

instanceMethodForSelector

public static NSObject.Function_instanceMethodForSelector_ret instanceMethodForSelector(org.moe.natj.objc.SEL aSelector)

instanceMethodSignatureForSelector

public static NSMethodSignature instanceMethodSignatureForSelector(org.moe.natj.objc.SEL aSelector)

instancesRespondToSelector

public static boolean instancesRespondToSelector(org.moe.natj.objc.SEL aSelector)

isSubclassOfClass

public static boolean isSubclassOfClass(org.moe.natj.objc.Class aClass)

keyPathsForValuesAffectingValueForKey

public static NSSet<java.lang.String> keyPathsForValuesAffectingValueForKey(java.lang.String key)

layers

public NSArray<? extends MLCLayer> layers()

Layers in the graph

new_objc

public static MLCGraph new_objc()

nodeWithLayerSource

public MLCTensor nodeWithLayerSource(MLCLayer layer,
                                     MLCTensor source)

Add a layer to the graph

Parameters:

layer - The layer

source - The source tensor

Returns:

A result tensor

nodeWithLayerSources

public MLCTensor nodeWithLayerSources(MLCLayer layer,
                                      NSArray<? extends MLCTensor> sources)

Add a layer to the graph

For variable length sequences of LSTMs/RNNs layers, create an MLCTensor of sortedSequenceLengths and pass it as the last index (i.e. index 2 or 4) of sources. This tensor must of be type MLCDataTypeInt32.

Parameters:

layer - The layer

sources - A list of source tensors

Returns:

A result tensor

nodeWithLayerSourcesDisableUpdate

public MLCTensor nodeWithLayerSourcesDisableUpdate(MLCLayer layer,
                                                   NSArray<? extends MLCTensor> sources,
                                                   boolean disableUpdate)

Add a layer to the graph

For variable length sequences of LSTMs/RNNs layers, create an MLCTensor of sortedSequenceLengths and pass it as the last index (i.e. index 2 or 4) of sources. This tensor must of be type MLCDataTypeInt32.

Parameters:

layer - The layer

sources - A list of source tensors

disableUpdate - A flag to indicate if optimizer update should be disabled for this layer

Returns:

A result tensor

nodeWithLayerSourcesLossLabels

public MLCTensor nodeWithLayerSourcesLossLabels(MLCLayer layer,
                                                NSArray<? extends MLCTensor> sources,
                                                NSArray<? extends MLCTensor> lossLabels)

Add a loss layer to the graph

For variable length sequences of LSTMs/RNNs layers, create an MLCTensor of sortedSequenceLengths and pass it as the last index (i.e. index 2 or 4) of sources. This tensor must of be type MLCDataTypeInt32.

Parameters:

layer - The loss layer

lossLabels - The loss labels tensor

Returns:

A result tensor

reshapeWithShapeSource

public MLCTensor reshapeWithShapeSource(NSArray<? extends NSNumber> shape,
                                        MLCTensor source)

Add a reshape layer to the graph

Parameters:

shape - An array representing the shape of result tensor

source - The source tensor

Returns:

A result tensor

resolveClassMethod

public static boolean resolveClassMethod(org.moe.natj.objc.SEL sel)

resolveInstanceMethod

public static boolean resolveInstanceMethod(org.moe.natj.objc.SEL sel)

resultTensorsForLayer

public NSArray<? extends MLCTensor> resultTensorsForLayer(MLCLayer layer)

Get the result tensors for a layer in the training graph

Parameters:

layer - A layer in the training graph

Returns:

A list of tensors

scatterWithDimensionSourceIndicesCopyFromReductionType

public MLCTensor scatterWithDimensionSourceIndicesCopyFromReductionType(long dimension,
                                                                        MLCTensor source,
                                                                        MLCTensor indices,
                                                                        MLCTensor copyFrom,
                                                                        int reductionType)

Add a scatter layer to the graph

Parameters:

dimension - The dimension along which to index

source - The updates to use with scattering with index positions specified in indices to result tensor

indices - The index of elements to scatter

copyFrom - The source tensor whose data is to be first copied to the result tensor

reductionType - The reduction type applied for all values in source tensor that are scattered to a specific location in the result tensor. Must be: MLCReductionTypeNone or MLCReductionTypeSum.

Returns:

A result tensor

selectWithSourcesCondition

public MLCTensor selectWithSourcesCondition(NSArray<? extends MLCTensor> sources,
                                            MLCTensor condition)

Add a select layer to the graph

Parameters:

sources - The source tensors

condition - The condition mask

Returns:

A result tensor

setVersion_static

public static void setVersion_static(long aVersion)

sourceTensorsForLayer

public NSArray<? extends MLCTensor> sourceTensorsForLayer(MLCLayer layer)

Get the source tensors for a layer in the training graph

Parameters:

layer - A layer in the training graph

Returns:

A list of tensors

splitWithSourceSplitCountDimension

public NSArray<? extends MLCTensor> splitWithSourceSplitCountDimension(MLCTensor source,
                                                                       long splitCount,
                                                                       long dimension)

Add a split layer to the graph

Parameters:

source - The source tensor

splitCount - The number of splits

dimension - The dimension to split the source tensor

Returns:

A result tensor

splitWithSourceSplitSectionLengthsDimension

public NSArray<? extends MLCTensor> splitWithSourceSplitSectionLengthsDimension(MLCTensor source,
                                                                                NSArray<? extends NSNumber> splitSectionLengths,
                                                                                long dimension)

Add a split layer to the graph

Parameters:

source - The source tensor

splitSectionLengths - The lengths of each split section

dimension - The dimension to split the source tensor

Returns:

A result tensor

summarizedDOTDescription

public java.lang.String summarizedDOTDescription()

A DOT representation of the graph.

For more info on the DOT language, refer to https://en.wikipedia.org/wiki/DOT_(graph_description_language). Edges that have a dashed lines are those that have stop gradients, while those with solid lines don't.

superclass_static

public static org.moe.natj.objc.Class superclass_static()

transposeWithDimensionsSource

public MLCTensor transposeWithDimensionsSource(NSArray<? extends NSNumber> dimensions,
                                               MLCTensor source)

Add a transpose layer to the graph

Parameters:

dimensions - NSArray representing the desired ordering of dimensions The dimensions array specifies the input axis source for each output axis, such that the K'th element in the dimensions array specifies the input axis source for the K'th axis in the output. The batch dimension which is typically axis 0 cannot be transposed.

Returns:

A result tensor

version_static

public static long version_static()