Package de.uni_mannheim.informatik.dws.melt.matching_ml.python.nlptransformers

Class TransformersBaseFineTuner

java.lang.Object

eu.sealsproject.platform.res.tool.impl.AbstractPlugin

de.uni_mannheim.informatik.dws.melt.matching_base.MatcherURL

de.uni_mannheim.informatik.dws.melt.matching_base.MatcherFile

de.uni_mannheim.informatik.dws.melt.matching_jena.MatcherYAAA

de.uni_mannheim.informatik.dws.melt.matching_jena.MatcherYAAAJena

de.uni_mannheim.informatik.dws.melt.matching_ml.python.nlptransformers.TransformersBase

de.uni_mannheim.informatik.dws.melt.matching_ml.python.nlptransformers.TransformersBaseFineTuner

All Implemented Interfaces:

IMatcher<org.apache.jena.ontology.OntModel,Alignment,Properties>, eu.sealsproject.platform.res.domain.omt.IOntologyMatchingToolBridge, eu.sealsproject.platform.res.tool.api.IPlugin, eu.sealsproject.platform.res.tool.api.IToolBridge

Direct Known Subclasses:

SentenceTransformersFineTuner, TransformersFineTuner

public abstract class TransformersBaseFineTuner
extends TransformersBase

This is a base class for all Transformers fine tuners. It just contains some variables and getter and setters.

Field Summary

Fields

Modifier and Type	Field	Description
protected boolean	additionallySwitchSourceTarget
private static final org.slf4j.Logger	LOGGER
protected static final String	NEWLINE
protected File	resultingModelLocation
protected File	trainingFile

Fields inherited from class de.uni_mannheim.informatik.dws.melt.matching_ml.python.nlptransformers.TransformersBase

cudaVisibleDevices, extractor, modelName, multipleTextsToMultipleExamples, multiProcessing, trainingArguments, transformersCache, usingTensorflow

Fields inherited from class de.uni_mannheim.informatik.dws.melt.matching_base.MatcherFile

FILE_PREFIX, FILE_SUFFIX

Constructor Summary

Constructors

Constructor	Description
TransformersBaseFineTuner(TextExtractorMap extractor, String initialModelName, File resultingModelLocation)	Run the training of a NLP transformer.
TransformersBaseFineTuner(TextExtractor extractor, String initialModelName, File resultingModelLocation)	Run the training of a NLP transformer.

Method Summary

Modifier and Type	Method	Description
void	clearTrainingData()	Removes the training data
File	createTrainingFile(org.apache.jena.ontology.OntModel source, org.apache.jena.ontology.OntModel target, Alignment trainingAlignment)	Creates a new file and writes all correspondences as textual data to it.
File	finetuneModel()	This method should only be used when appendOnlyToFile is set to true.
abstract File	finetuneModel(File trainingFile)	Finetune a given model with the provided text in the csv file (three columns: first text, second text, label(0/1))
File	getResultingModelLocation()	Returns the final location where the finetuned model should be stored
File	getTrainingFile()	Returns the training file generated during multiple calls of the match method.
boolean	isAdditionallySwitchSourceTarget()	Return the boolean value if training examples are additionally changed in their order.
Alignment	match(org.apache.jena.ontology.OntModel source, org.apache.jena.ontology.OntModel target, Alignment inputAlignment, Properties properties)	This method does only fine tune the model and does not match any entities.
void	setAdditionallySwitchSourceTarget(boolean additionallySwitchSourceTarget)	If set to true, the training examples not only contain e.g.
void	setResultingModelLocation(File resultingModelLocation)	Sets the final location where the finetuned model should be stored.
protected int	writeClassificationFormat(org.apache.jena.ontology.OntModel source, org.apache.jena.ontology.OntModel target, Alignment trainingAlignment, File trainFile, boolean append)
int	writeTrainingFile(org.apache.jena.ontology.OntModel source, org.apache.jena.ontology.OntModel target, Alignment trainingAlignment, File trainFile, boolean append)	Writes the correspondences to a file (append or not can be chosen by a parameter).

Methods inherited from class de.uni_mannheim.informatik.dws.melt.matching_ml.python.nlptransformers.TransformersBase

addTrainingArgument, getCudaVisibleDevices, getCudaVisibleDevicesButOnlyOneGPU, getExamplesForBatchSizeOptimization, getExtractor, getExtractorMap, getModelName, getMultiProcessing, getTextualRepresentation, getTrainingArguments, getTransformersCache, isMultipleTextsToMultipleExamples, isOptimizeForMixedPrecisionTraining, isUsingTensorflow, setCudaVisibleDevices, setCudaVisibleDevices, setExtractor, setExtractorMap, setModelName, setMultipleTextsToMultipleExamples, setMultiProcessing, setOptimizeForMixedPrecisionTraining, setTrainingArguments, setTransformersCache, setUsingTensorflow, writeExamplesToFile

Methods inherited from class de.uni_mannheim.informatik.dws.melt.matching_jena.MatcherYAAAJena

getModelSpec, match, readOntology

Methods inherited from class de.uni_mannheim.informatik.dws.melt.matching_jena.MatcherYAAA

match

Methods inherited from class de.uni_mannheim.informatik.dws.melt.matching_base.MatcherFile

match

Methods inherited from class de.uni_mannheim.informatik.dws.melt.matching_base.MatcherURL

align, align, canExecute, getType

Methods inherited from class eu.sealsproject.platform.res.tool.impl.AbstractPlugin

getId, getVersion, setId, setVersion

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface eu.sealsproject.platform.res.tool.api.IPlugin

getId, getVersion

Field Details

LOGGER

private static final org.slf4j.Logger LOGGER

NEWLINE

protected static final String NEWLINE

resultingModelLocation

protected File resultingModelLocation

trainingFile

protected File trainingFile

additionallySwitchSourceTarget

protected boolean additionallySwitchSourceTarget

Constructor Details

TransformersBaseFineTuner

public TransformersBaseFineTuner(TextExtractorMap extractor,
 String initialModelName,
 File resultingModelLocation)

Run the training of a NLP transformer.

Parameters:

extractor - used to extract text from a given resource. This is the text which represents a resource.

initialModelName - the initial model name for fine tuning which can be downloaded or a path to a directory containing model weights ( see first parameter pretrained_model_name_or_path of the from_pretrained function in huggingface library). This value can be also changed by TransformersBase.setModelName(java.lang.String).

resultingModelLocation - the final location where the fine-tuned model should be stored.

TransformersBaseFineTuner

public TransformersBaseFineTuner(TextExtractor extractor,
 String initialModelName,
 File resultingModelLocation)

Run the training of a NLP transformer.

Parameters:

extractor - used to extract text from a given resource. This is the text which represents a resource.

initialModelName - the initial model name for fine tuning which can be downloaded or a path to a directory containing model weights ( see first parameter pretrained_model_name_or_path of the from_pretrained function in huggingface library). This value can be also changed by TransformersBase.setModelName(java.lang.String).

resultingModelLocation - the final location where the fine-tuned model should be stored.

Method Details

match

public Alignment match(org.apache.jena.ontology.OntModel source,
 org.apache.jena.ontology.OntModel target,
 Alignment inputAlignment,
 Properties properties)
                throws Exception

This method does only fine tune the model and does not match any entities.

Specified by:

match in interface IMatcher<org.apache.jena.ontology.OntModel,Alignment,Properties>

Specified by:

match in class MatcherYAAAJena

Parameters:

source - This OntModel represents the source ontology.

target - This OntModel represents the target ontology.

inputAlignment - This mapping represents the input alignment.

properties - Additional properties.

Returns:

the resulting alignment - in this special case the unmodified input alignment.

Throws:

Exception - in case something goes wrong

createTrainingFile

public File createTrainingFile(org.apache.jena.ontology.OntModel source,
 org.apache.jena.ontology.OntModel target,
 Alignment trainingAlignment)
                        throws IOException

Creates a new file and writes all correspondences as textual data to it. Can also be used for creating a validation file.

Parameters:

source - the source model

target - the target model

trainingAlignment - the training alignment to be written to file.

Returns:

the training file.

Throws:

IOException - in case the writing fails

writeTrainingFile

public int writeTrainingFile(org.apache.jena.ontology.OntModel source,
 org.apache.jena.ontology.OntModel target,
 Alignment trainingAlignment,
 File trainFile,
 boolean append)
                      throws IOException

Writes the correspondences to a file (append or not can be chosen by a parameter).

Parameters:

source - the source model

target - the target model

trainingAlignment - the training alignment to be written to file.

trainFile - the file to write all texts

append - true if all content should be appended to the file.

Returns:

how many correspondences were written to the file.

Throws:

IOException - in case the writing fails

writeClassificationFormat

protected int writeClassificationFormat(org.apache.jena.ontology.OntModel source,
 org.apache.jena.ontology.OntModel target,
 Alignment trainingAlignment,
 File trainFile,
 boolean append)
                                 throws IOException

Throws:

IOException

finetuneModel

public abstract File finetuneModel(File trainingFile)
                            throws Exception

Finetune a given model with the provided text in the csv file (three columns: first text, second text, label(0/1))

Parameters:

trainingFile - csv file with three columns: first text, second text, label(0/1) (can be generated with createTrainingFile(OntModel, OntModel, Alignment) )

Returns:

the final location (directory) of the finetuned model (which is also given in the constructor)

Throws:

Exception - in case of any error

finetuneModel

public File finetuneModel()
                   throws Exception

This method should only be used when appendOnlyToFile is set to true. This will train a transformers model on the file generated by the match method during possibley multiple calls.

Returns:

the final location (directory) of the fine-tuned model (which is also given in the constructor)

Throws:

Exception - in case of any error

clearTrainingData

public void clearTrainingData()

Removes the training data

getTrainingFile

public File getTrainingFile()

Returns the training file generated during multiple calls of the match method. If the match method is not called yet, this will return a non existent file.

Returns:

the training file

getResultingModelLocation

public File getResultingModelLocation()

Returns the final location where the finetuned model should be stored

Returns:

the location for the trained model.

setResultingModelLocation

public void setResultingModelLocation(File resultingModelLocation)

Sets the final location where the finetuned model should be stored.

Parameters:

resultingModelLocation - the model location as a file.

isAdditionallySwitchSourceTarget

public boolean isAdditionallySwitchSourceTarget()

Return the boolean value if training examples are additionally changed in their order. If true, the training examples not only contain e.g. A,B but also B,A because positive and negative examples still hold even when the order is changed. This will double the number of training examples.

Returns:

true, training examples are additionally changed in their order

setAdditionallySwitchSourceTarget

public void setAdditionallySwitchSourceTarget(boolean additionallySwitchSourceTarget)

If set to true, the training examples not only contain e.g. A,B but also B,A because positive and negative example still hold even when the order is changed. This will double the number of training examples. The default is false.

Parameters:

additionallySwitchSourceTarget - true, if source and target should be changed.