CheckpointListener (deeplearning4j-nn 1.0.0-beta API)

java.lang.Object
- org.deeplearning4j.optimize.api.BaseTrainingListener
- - org.deeplearning4j.optimize.listeners.checkpoint.CheckpointListener

All Implemented Interfaces:

Serializable, TrainingListener
```
public class CheckpointListener
extends BaseTrainingListener
implements Serializable
```
CheckpointListener: The goal of this listener is to periodically save a copy of the model during training..
Model saving may be done:
1. Every N epochs
2. Every N iterations
3. Every T time units (every 15 minutes, for example)
Or some combination of the 3.

Models can be restored using loadCheckpointMLN(Checkpoint) and loadCheckpointCG(Checkpoint). Model files can be obtained using getFileForCheckpoint(Checkpoint)
Checkpoints can be obtained using lastCheckpoint() and availableCheckpoints()
Example 1: Saving a checkpoint every 2 epochs, keep all model files
```
  CheckpointListener l = new CheckpointListener.Builder("/save/directory")
       .keepAll() //Don't delete any models
       .saveEveryNEpochs(2)
       .build()
 
 
```
Example 2: Saving a checkpoint every 1000 iterations, but keeping only the last 3 models (all older model files will be automatically deleted)
```
  CheckpointListener l = new CheckpointListener.Builder(new File("/save/directory"))
          .keepLast(3)
          .saveEveryNIterations(1000)
          .build();
 
 
```
Example 3: Saving a checkpoint every 15 minutes, keeping the most recent 3 and otherwise every 4th checkpoint file:
```
  CheckpointListener l = new CheckpointListener.Builder(new File("/save/directory"))
          .keepLastAndEvery(3, 4)
          .saveEvery(15, TimeUnit.MINUTES)
          .build();
 
 
```
Note that you can mix these: for example, to save every epoch and every 15 minutes (independent of last save time):
.saveEveryEpoch().saveEvery(15, TimeUnit.MINUTES)
To save every epoch, and every 15 minutes, since the last model save use:
.saveEveryEpoch().saveEvery(15, TimeUnit.MINUTES, true)
Note that is this last example, the sinceLast parameter is true. This means the 15-minute counter will be reset any time a model is saved.
Author:

Alex Black

See Also:

Serialized Form

Nested Class Summary

Nested Classes
Modifier and Type Class and Description

static class CheckpointListener.Builder

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

Method Summary

All Methods Static Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`List<Checkpoint>`	`availableCheckpoints()` List all available checkpoints.
`protected static int`	`getEpoch(Model model)`
`File`	`getFileForCheckpoint(Checkpoint checkpoint)` Get the model file for the given checkpoint.
`File`	`getFileForCheckpoint(int checkpointNum)` Get the model file for the given checkpoint number.
`protected static int`	`getIter(Model model)`
`protected static String`	`getModelType(Model model)`
`void`	`iterationDone(Model model, int iteration, int epoch)` Event listener for each iteration.
`Checkpoint`	`lastCheckpoint()` Return the most recent checkpoint, if one exists - otherwise returns null
`ComputationGraph`	`loadCheckpointCG(Checkpoint checkpoint)` Load a ComputationGraph for the given checkpoint
`ComputationGraph`	`loadCheckpointCG(int checkpointNum)` Load a ComputationGraph for the given checkpoint
`MultiLayerNetwork`	`loadCheckpointMLN(Checkpoint checkpoint)` Load a MultiLayerNetwork for the given checkpoint
`MultiLayerNetwork`	`loadCheckpointMLN(int checkpointNum)` Load a MultiLayerNetwork for the given checkpoint number
`void`	`onEpochEnd(Model model)` Called once at the end of each epoch, when using methods such as `MultiLayerNetwork.fit(DataSetIterator)`, `ComputationGraph.fit(DataSetIterator)` or `ComputationGraph.fit(MultiDataSetIterator)`

Methods inherited from class org.deeplearning4j.optimize.api.BaseTrainingListener
onBackwardPass, onEpochStart, onForwardPass, onForwardPass, onGradientCalculation

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

- Method Detail
  - onEpochEnd
```
public void onEpochEnd(Model model)
```
    Description copied from interface: TrainingListener
    
    Called once at the end of each epoch, when using methods such as MultiLayerNetwork.fit(DataSetIterator), ComputationGraph.fit(DataSetIterator) or ComputationGraph.fit(MultiDataSetIterator)
    
    Specified by:
    
    onEpochEnd in interface TrainingListener
    
    Overrides:
    
    onEpochEnd in class BaseTrainingListener
  - iterationDone
```
public void iterationDone(Model model,
                          int iteration,
                          int epoch)
```
    Description copied from interface: TrainingListener
    
    Event listener for each iteration. Called once, after each parameter update has ocurred while training the network
    
    Specified by:
    
    iterationDone in interface TrainingListener
    
    Overrides:
    
    iterationDone in class BaseTrainingListener
    
    Parameters:
    
    model - the model iterating
    
    iteration - the iteration
  - getIter
```
protected static int getIter(Model model)
```
  - getEpoch
```
protected static int getEpoch(Model model)
```
  - getModelType
```
protected static String getModelType(Model model)
```
  - availableCheckpoints
```
public List<Checkpoint> availableCheckpoints()
```
    List all available checkpoints. A checkpoint is 'available' if the file can be loaded. Any checkpoint files that have been automatically deleted (given the configuration) will not be returned here.
    
    Returns:
    
    List of checkpoint files that can be loaded
  - lastCheckpoint
```
public Checkpoint lastCheckpoint()
```
    Return the most recent checkpoint, if one exists - otherwise returns null
    
    Returns:
    
    Checkpoint
  - getFileForCheckpoint
```
public File getFileForCheckpoint(Checkpoint checkpoint)
```
    Get the model file for the given checkpoint. Checkpoint model file must exist
    
    Parameters:
    
    checkpoint - Checkpoint to get the model file for
    
    Returns:
    
    Model file for the checkpoint
  - getFileForCheckpoint
```
public File getFileForCheckpoint(int checkpointNum)
```
    Get the model file for the given checkpoint number. Checkpoint model file must exist
    
    Parameters:
    
    checkpointNum - Checkpoint number to get the model file for
    
    Returns:
    
    Model file for the checkpoint
  - loadCheckpointMLN
```
public MultiLayerNetwork loadCheckpointMLN(Checkpoint checkpoint)
```
    Load a MultiLayerNetwork for the given checkpoint
    
    Parameters:
    
    checkpoint - Checkpoint model to load
    
    Returns:
    
    The loaded model
  - loadCheckpointMLN
```
public MultiLayerNetwork loadCheckpointMLN(int checkpointNum)
```
    Load a MultiLayerNetwork for the given checkpoint number
    
    Parameters:
    
    checkpointNum - Checkpoint model to load
    
    Returns:
    
    The loaded model
  - loadCheckpointCG
```
public ComputationGraph loadCheckpointCG(Checkpoint checkpoint)
```
    Load a ComputationGraph for the given checkpoint
    
    Parameters:
    
    checkpoint - Checkpoint model to load
    
    Returns:
    
    The loaded model
  - loadCheckpointCG
```
public ComputationGraph loadCheckpointCG(int checkpointNum)
```
    Load a ComputationGraph for the given checkpoint
    
    Parameters:
    
    checkpointNum - Checkpoint model number to load
    
    Returns:
    
    The loaded model

Class CheckpointListener

Nested Class Summary

Method Summary

Methods inherited from class org.deeplearning4j.optimize.api.BaseTrainingListener

Methods inherited from class java.lang.Object

Method Detail

onEpochEnd

iterationDone

getIter

getEpoch

getModelType

availableCheckpoints

lastCheckpoint

getFileForCheckpoint

getFileForCheckpoint

loadCheckpointMLN

loadCheckpointMLN

loadCheckpointCG

loadCheckpointCG