Package org.tribuo.interop.tensorflow

Class TensorFlowCheckpointModel<T extends Output<T>>

java.lang.Object

org.tribuo.Model<T>

org.tribuo.interop.tensorflow.TensorFlowModel<T>

org.tribuo.interop.tensorflow.TensorFlowCheckpointModel<T>

All Implemented Interfaces:

com.oracle.labs.mlrg.olcut.provenance.Provenancable<ModelProvenance>, Closeable, Serializable, AutoCloseable, ProtoSerializable<org.tribuo.protos.core.ModelProto>

public final class TensorFlowCheckpointModel<T extends Output<T>>
extends TensorFlowModel<T>
implements Closeable

This model encapsulates a simple model with an input feed dict, and produces a single output tensor.

If the checkpoint is not available on construction or after deserialisation then the model is uninitialised. Models can be initialised by calling initialize() after calling setCheckpointDirectory(java.lang.String) and setCheckpointName(java.lang.String) with the right directory and name respectively.

This model's serialized form stores the weights in the specified model checkpoint directory. If you wish to convert it into a model which stores the weights inside the model then call convertToNativeModel().

It accepts an FeatureConverter that converts an example's features into a TensorMap, and an OutputConverter that converts a Tensor into a Prediction.

The model's serialVersionUID is set to the major TensorFlow version number times 100.

N.B. TensorFlow support is experimental and may change without a major version bump.

See Also:

• Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
static final int	CURRENT_VERSION	Protobuf serialization version.

Fields inherited from class org.tribuo.interop.tensorflow.TensorFlowModel

batchSize, closed, featureConverter, modelGraph, outputConverter, outputName, session

Fields inherited from class org.tribuo.Model

ALL_OUTPUTS, BIAS_FEATURE, featureIDMap, generatesProbabilities, name, outputIDInfo, provenance, provenanceOutput

Fields inherited from interface org.tribuo.protos.ProtoSerializable

DESERIALIZATION_METHOD_NAME, PROVENANCE_SERIALIZER

Method Summary

Modifier and Type	Method	Description
TensorFlowNativeModel<T>	convertToNativeModel()	Creates a TensorFlowNativeModel version of this model.
protected TensorFlowCheckpointModel<T>	copy(String newName, ModelProvenance newProvenance)	Copies a model, replacing its provenance and name with the supplied values.
static TensorFlowCheckpointModel<?>	deserializeFromProto(int version, String className, com.google.protobuf.Any message)	Deserialization factory.
String	getCheckpointDirectory()	Gets the checkpoint directory this model loads from.
String	getCheckpointName()	Gets the checkpoint name this model loads from.
final void	initialize()	Initializes the model.
boolean	isInitialized()	Is this model initialized?
org.tribuo.protos.core.ModelProto	serialize()	Serializes this object to a protobuf.
void	setCheckpointDirectory(String newCheckpointDirectory)	Sets the checkpoint directory.
void	setCheckpointName(String newCheckpointName)	Sets the checkpoint name.

Methods inherited from class org.tribuo.interop.tensorflow.TensorFlowModel

close, exportModel, getBatchSize, getExcuse, getOutputName, getTopFeatures, innerPredict, predict, setBatchSize

Methods inherited from class org.tribuo.Model

castModel, copy, createDataCarrier, deserialize, deserializeFromFile, deserializeFromStream, generatesProbabilities, getExcuses, getFeatureIDMap, getName, getOutputIDInfo, getProvenance, predict, predict, serializeToFile, serializeToStream, setName, toString, validate

Methods inherited from class java.lang.Object

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

Methods inherited from interface java.io.Closeable

close

Field Details

CURRENT_VERSION

public static final int CURRENT_VERSION

Protobuf serialization version.

See Also:

• Constant Field Values

Method Details

deserializeFromProto

public static TensorFlowCheckpointModel<?> deserializeFromProto(int version,
 String className,
 com.google.protobuf.Any message)
                                                         throws com.google.protobuf.InvalidProtocolBufferException

Deserialization factory.

Parameters:

version - The serialized object version.

className - The class name.

message - The serialized data.

Returns:

The deserialized object.

Throws:

com.google.protobuf.InvalidProtocolBufferException - If the protobuf could not be parsed from the message.

isInitialized

public boolean isInitialized()

Is this model initialized?

Returns:

True if the model is ready to make predictions.

initialize

public final void initialize()

Initializes the model.

This call closes the old session (if it exists) and creates a fresh session from the current checkpoint path.

Throws TensorFlowException if it failed to read the checkpoint.

setCheckpointDirectory

public void setCheckpointDirectory(String newCheckpointDirectory)

Sets the checkpoint directory.

The model likely needs re-initializing after this call.

Parameters:

newCheckpointDirectory - The new checkpoint directory.

getCheckpointDirectory

public String getCheckpointDirectory()

Gets the checkpoint directory this model loads from.

Returns:

The checkpoint directory.

setCheckpointName

public void setCheckpointName(String newCheckpointName)

Sets the checkpoint name.

The model likely needs re-initializing after this call.

Parameters:

newCheckpointName - The new checkpoint name.

getCheckpointName

public String getCheckpointName()

Gets the checkpoint name this model loads from.

Returns:

The checkpoint name.

convertToNativeModel

public TensorFlowNativeModel<T> convertToNativeModel()

Creates a TensorFlowNativeModel version of this model.

Returns:

A version of this model using Tribuo's native serialization mechanism.

copy

protected TensorFlowCheckpointModel<T> copy(String newName,
 ModelProvenance newProvenance)

Description copied from class: Model

Copies a model, replacing its provenance and name with the supplied values.

Used to provide the provenance removal functionality.

Specified by:

copy in class Model<T extends Output<T>>

Parameters:

newName - The new name.

newProvenance - The new provenance.

Returns:

A copy of the model.

serialize

public org.tribuo.protos.core.ModelProto serialize()

Description copied from interface: ProtoSerializable

Serializes this object to a protobuf.

Specified by:

serialize in interface ProtoSerializable<T extends Output<T>>

Overrides:

serialize in class Model<T extends Output<T>>

Returns:

The protobuf.