cc.mallet.grmm.learning
Class ACRFEvaluator
java.lang.Object
cc.mallet.grmm.learning.ACRFEvaluator
- Direct Known Subclasses:
- AcrfSerialEvaluator, DefaultAcrfTrainer.FileEvaluator, DefaultAcrfTrainer.LogEvaluator, MultiSegmentationEvaluatorACRF
public abstract class ACRFEvaluator
- extends java.lang.Object
Created: Sun Jan 25 23:28:45 2004
- Version:
- $Id: ACRFEvaluator.java,v 1.1 2007/10/22 21:37:43 mccallum Exp $
- Author:
- Charles Sutton
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
ACRFEvaluator
public ACRFEvaluator()
setNumIterToSkip
public void setNumIterToSkip(int n)
setNumIterToWait
public void setNumIterToWait(int n)
evaluate
public abstract boolean evaluate(ACRF acrf,
int iter,
InstanceList training,
InstanceList validation,
InstanceList testing)
- Evalutes the model in the middle of training.
- Parameters:
acrf
- Model tha is being trained.iter
- How many iterations have been completed.training
- Training set.validation
- Validation set; may be null.testing
- Testing set; maybe null.
- Returns:
- Whether to continue training. If return is false, training should be be stopped.
test
public abstract void test(InstanceList gold,
java.util.List returned,
java.lang.String description)
test
public void test(ACRF acrf,
InstanceList data,
java.lang.String description)
setOutputPrefix
public void setOutputPrefix(java.io.File file)
makeOutputFile
protected java.io.File makeOutputFile()
shouldDoEvaluate
protected boolean shouldDoEvaluate(int iter)