EncodedGradientsAccumulator (deeplearning4j-nn 1.0.0-beta7 API)

java.lang.Object
- org.deeplearning4j.optimize.solvers.accumulation.EncodedGradientsAccumulator

All Implemented Interfaces:

Serializable, GradientsAccumulator, Registerable
```
public class EncodedGradientsAccumulator
extends Object
implements GradientsAccumulator, Registerable
```
This GradientsAccumulator is suited for CUDA backend.

Author:

[email protected]

See Also:

Serialized Form

Nested Class Summary

Nested Classes
Modifier and Type Class and Description

static class EncodedGradientsAccumulator.Builder

Nested Classes
Modifier and Type	Class and Description
`static class`	`EncodedGradientsAccumulator.Builder`

Field Summary

Fields
Modifier and Type	Field and Description
`protected ThreadLocal<INDArray>`	`accumulator`
`protected WorkspaceConfiguration`	`appliedConfiguration`
`protected AtomicInteger`	`barrier`
`protected Integer`	`boundary`
`protected AtomicBoolean`	`bypassMode`
`protected AtomicInteger`	`currentConsumers`
`static long`	`DEFAULT_INITIAL_MEMORY`
`protected boolean`	`encodingDebugMode`
`protected IndexedTail`	`externalSource`
`protected AtomicBoolean`	`externalUpdatesAvailable`
`protected MessageHandler`	`handler`
`protected ThreadLocal<Integer>`	`index`
`protected long`	`initialMemory`
`protected boolean`	`isDebug`
`protected AtomicBoolean`	`isDone`
`protected AtomicBoolean`	`isFirst`
`protected List<ReentrantLock>`	`locks`
`protected List<BlockingQueue<INDArray>>`	`messages`
`protected int`	`parties`
`protected int`	`queueSize`
`protected AtomicBoolean`	`registered`
`protected boolean`	`relocatable`
`protected AtomicInteger`	`secondary`
`protected AtomicThrowable`	`throwable`
`protected ThreadLocal<AtomicLong>`	`updatesApplied`
`protected AtomicInteger`	`workersCounter`
`protected List<MemoryWorkspace>`	`workspaces`

Constructor Summary

Constructors
Constructor and Description
`EncodedGradientsAccumulator(int parties, double threshold)`
`EncodedGradientsAccumulator(int parties, @NonNull MessageHandler handler, long initialMemory, int queueSize, Integer boundary, boolean encodingDebugMode)`
`EncodedGradientsAccumulator(int parties, ThresholdAlgorithm thresholdAlgorithm, ResidualPostProcessor residualPostProcessor, boolean encodingDebugMode)`

Method Summary

All Methods Static Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`void`	`applyUpdate(StepFunction function, INDArray params, INDArray updates, boolean isFinalStep)` This method applies accumulated updates via given StepFunction
`void`	`applyUpdate(StepFunction function, INDArray params, INDArray updates, double alpha)` This method applies accumulated updates via given StepFunction
`void`	`fallbackToSingleConsumerMode(boolean reallyFallback)` This method enables/disables bypass mode
`IndexedTail`	`getExternalSource()`
`static long`	`getOptimalBufferSize(long paramsLength, int numWorkers, int queueSize)` This method returns optimal bufferSize for a given model We know, that updates are guaranteed to have MAX size of params / 16.
`static long`	`getOptimalBufferSize(Model model, int numWorkers, int queueSize)`
`boolean`	`hasAnything()` This method checks if there are any (probably external) updates available
`void`	`markExternalUpdates(boolean updatesAvailable)` This method allows to highlight early availability of updates
`void`	`receiveUpdate(INDArray array)` This method accepts updates suitable for StepFunction and puts them to the queue, which is used in backpropagation loop
`void`	`registerConsumers(int numConsumers)` This method notifies producer about number of consumers for the current consumption cycle
`void`	`reset()` This method resets all accumulated updates (if any)
`void`	`setExternalSource(IndexedTail source)` This method allows to pass external updates to accumulator, they will be populated across all workers using this GradientsAccumulator instance
`void`	`storeUpdate(INDArray array, int iterationNumber, int epochNumber)` This method accepts updates suitable for StepFunction, and accumulates/propagates it across all workers
`protected void`	`synchronize(int consumers)`
`protected void`	`synchronize(int consumers, boolean finalLock)`
`void`	`touch()` This method does initialization of given worker wrt Thread-Device Affinity

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

DEFAULT_INITIAL_MEMORY
```
public static final long DEFAULT_INITIAL_MEMORY
```
See Also:

Constant Field Values

accumulator

protected ThreadLocal<INDArray> accumulator

parties
```
protected int parties
```

handler
```
protected MessageHandler handler
```

messages

protected List<BlockingQueue<INDArray>> messages

workspaces

protected List<MemoryWorkspace> workspaces

locks
```
protected List<ReentrantLock> locks
```

workersCounter
```
protected AtomicInteger workersCounter
```

index
```
protected ThreadLocal<Integer> index
```

initialMemory
```
protected long initialMemory
```

queueSize
```
protected int queueSize
```

boundary
```
protected Integer boundary
```

encodingDebugMode
```
protected boolean encodingDebugMode
```

externalSource
```
protected IndexedTail externalSource
```

isFirst
```
protected AtomicBoolean isFirst
```

isDone
```
protected AtomicBoolean isDone
```

barrier
```
protected AtomicInteger barrier
```

secondary
```
protected AtomicInteger secondary
```

registered
```
protected AtomicBoolean registered
```

bypassMode
```
protected AtomicBoolean bypassMode
```

currentConsumers

protected final AtomicInteger currentConsumers

throwable

protected final AtomicThrowable throwable

isDebug
```
protected boolean isDebug
```

relocatable
```
protected final boolean relocatable
```

updatesApplied

protected ThreadLocal<AtomicLong> updatesApplied

externalUpdatesAvailable

protected AtomicBoolean externalUpdatesAvailable

appliedConfiguration

protected WorkspaceConfiguration appliedConfiguration

Constructor Detail

EncodedGradientsAccumulator

public EncodedGradientsAccumulator(int parties,
                                   double threshold)

EncodedGradientsAccumulator

public EncodedGradientsAccumulator(int parties,
                                   ThresholdAlgorithm thresholdAlgorithm,
                                   ResidualPostProcessor residualPostProcessor,
                                   boolean encodingDebugMode)

EncodedGradientsAccumulator

public EncodedGradientsAccumulator(int parties,
                                   @NonNull
                                   @NonNull MessageHandler handler,
                                   long initialMemory,
                                   int queueSize,
                                   Integer boundary,
                                   boolean encodingDebugMode)

Method Detail
- getOptimalBufferSize
```
public static long getOptimalBufferSize(long paramsLength,
                                        int numWorkers,
                                        int queueSize)
```
  This method returns optimal bufferSize for a given model We know, that updates are guaranteed to have MAX size of params / 16. So, here we go. I.e. for model with 100m params, that's 400m of floats (or 800m of doubles) The worst case for us is bitmap encoding, that takes 2 bits to encode each gradient value so, for float in worst case we'll have (100m / 16) int elements. So, our buffer size will be 6.25m * queueSize * 4 bytes per int
  
  Parameters:
  
  paramsLength -
  
  numWorkers -
  
  queueSize -
  
  Returns:
- getOptimalBufferSize
```
public static long getOptimalBufferSize(Model model,
                                        int numWorkers,
                                        int queueSize)
```
- fallbackToSingleConsumerMode
```
public void fallbackToSingleConsumerMode(boolean reallyFallback)
```
  Description copied from interface: Registerable
  
  This method enables/disables bypass mode
  
  Specified by:
  
  fallbackToSingleConsumerMode in interface Registerable
- registerConsumers
```
public void registerConsumers(int numConsumers)
```
  Description copied from interface: Registerable
  
  This method notifies producer about number of consumers for the current consumption cycle
  
  Specified by:
  
  registerConsumers in interface Registerable
- getExternalSource
```
public IndexedTail getExternalSource()
```
  Specified by:
  
  getExternalSource in interface GradientsAccumulator
- markExternalUpdates
```
public void markExternalUpdates(boolean updatesAvailable)
```
  Description copied from interface: GradientsAccumulator
  
  This method allows to highlight early availability of updates
  
  Specified by:
  
  markExternalUpdates in interface GradientsAccumulator
- synchronize
```
protected void synchronize(int consumers)
```
- synchronize
```
protected void synchronize(int consumers,
                           boolean finalLock)
```
- applyUpdate
```
public void applyUpdate(StepFunction function,
                        INDArray params,
                        INDArray updates,
                        boolean isFinalStep)
```
  This method applies accumulated updates via given StepFunction
  
  Specified by:
  
  applyUpdate in interface GradientsAccumulator
  
  Parameters:
  
  function -
  
  params -
- applyUpdate
```
public void applyUpdate(StepFunction function,
                        INDArray params,
                        INDArray updates,
                        double alpha)
```
  This method applies accumulated updates via given StepFunction
  
  Specified by:
  
  applyUpdate in interface GradientsAccumulator
  
  Parameters:
  
  function -
  
  params -
  
  alpha -
- setExternalSource
```
public void setExternalSource(IndexedTail source)
```
  This method allows to pass external updates to accumulator, they will be populated across all workers using this GradientsAccumulator instance
  
  Specified by:
  
  setExternalSource in interface GradientsAccumulator
  
  Parameters:
  
  source -
- touch
```
public void touch()
```
  This method does initialization of given worker wrt Thread-Device Affinity
  
  Specified by:
  
  touch in interface GradientsAccumulator
- storeUpdate
```
public void storeUpdate(INDArray array,
                        int iterationNumber,
                        int epochNumber)
```
  This method accepts updates suitable for StepFunction, and accumulates/propagates it across all workers
  
  Specified by:
  
  storeUpdate in interface GradientsAccumulator
  
  Parameters:
  
  array -
- receiveUpdate
```
public void receiveUpdate(INDArray array)
```
  This method accepts updates suitable for StepFunction and puts them to the queue, which is used in backpropagation loop
  PLEASE NOTE: array is expected to be ready for use and match params dimensionality
  
  Specified by:
  
  receiveUpdate in interface GradientsAccumulator
  
  Parameters:
  
  array -
- reset
```
public void reset()
```
  This method resets all accumulated updates (if any)
  
  Specified by:
  
  reset in interface GradientsAccumulator
- hasAnything
```
public boolean hasAnything()
```
  Description copied from interface: GradientsAccumulator
  
  This method checks if there are any (probably external) updates available
  
  Specified by:
  
  hasAnything in interface GradientsAccumulator
  
  Returns:

Class EncodedGradientsAccumulator

Nested Class Summary

Field Summary

Constructor Summary

Method Summary

Methods inherited from class java.lang.Object

Field Detail

DEFAULT_INITIAL_MEMORY

accumulator

parties

handler

messages

workspaces

locks

workersCounter

index

initialMemory

queueSize

boundary

encodingDebugMode

externalSource

isFirst

isDone

barrier

secondary

registered

bypassMode

currentConsumers

throwable

isDebug

relocatable

updatesApplied

externalUpdatesAvailable

appliedConfiguration

Constructor Detail

EncodedGradientsAccumulator

EncodedGradientsAccumulator

EncodedGradientsAccumulator

Method Detail

getOptimalBufferSize

getOptimalBufferSize

fallbackToSingleConsumerMode

registerConsumers

getExternalSource

markExternalUpdates

synchronize

synchronize

applyUpdate

applyUpdate

setExternalSource

touch

storeUpdate

receiveUpdate

reset

hasAnything