Package org.tribuo.interop.tensorflow

package org.tribuo.interop.tensorflow

Provides an interface to TensorFlow, allowing the training of non-sequential models using any supported Tribuo output type. Sequential model support is found in org.tribuo.interop.tensorflow.sequence.

Tribuo's TensorFlow support operates in Graph mode, as in v0.3.1 that is the only way to access gradients. The set of supported gradients is determined by TensorFlow, and not all gradients are available in TensorFlow Java in v0.3.1. Unsupported gradients will trigger an exception when the train method is called.

Models can store their trained parameters in two ways, either inside the Tribuo serialized model file (using TensorFlowTrainer.TFModelFormat.TRIBUO_NATIVE) or as a TensorFlow checkpoint folder on disk (using TensorFlowTrainer.TFModelFormat.CHECKPOINT). The choice is made at training time, as they result in slightly different TF graph structures.

Similarly there are two supported kinds of ExternalModel for TensorFlow, TensorFlowSavedModelExternalModel which loads a SavedModelBundle and always reads from that path, and TensorFlowFrozenExternalModel which loads a TensorFlow v1 frozen graph and stores the graph inside the Tribuo serialized object.

There are two main interfaces for interacting with TensorFlow Tensors, FeatureConverter and OutputConverter, with provide conversions to and from Tribuo's features and outputs respectively. There are implementations of a dense feature transformation and one for images as 3d arrays, along with output converters for Label Regressor, and MultiLabel. The loss function and output transformation used is controlled by the OutputConverter, if a different one is desired then users are recommended to implement that interface separately.

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

Related Packages

Package	Description
org.tribuo.interop	This package contains the abstract implementation of an external model trained by something outside of Tribuo.
org.tribuo.interop.tensorflow.example	Example architectures for use with Tribuo's TF interface.
org.tribuo.interop.tensorflow.sequence	Provides an interface for working with TensorFlow sequence models, using Tribuo's SequenceModel abstraction.
org.tribuo.interop.oci	Code for uploading models to Oracle Cloud Infrastructure Data Science, and also for scoring models deployed in Oracle Cloud Infrastructure Data Science.
org.tribuo.interop.onnx	This package contains a Tribuo wrapper around the ONNX Runtime.

Class	Description
DenseFeatureConverter	Converts a sparse example into a dense float vector, then wraps it in a TFloat32.
FeatureConverter	Transforms an Example or SGDVector, extracting the features from it as a TensorMap.
GradientOptimiser	An enum for the gradient optimisers exposed by TensorFlow-Java.
ImageConverter	Image converter.
LabelConverter	Can convert a Label into a Tensor containing one hot encoding of the label and can convert a TFloat16 or TFloat32 into a Prediction or a Label.
MultiLabelConverter	Can convert a MultiLabel into a Tensor containing a binary encoding of the label vector and can convert a TFloat16 or TFloat32 into a Prediction or a MultiLabel.
OutputConverter<T extends Output<T>>	Converts the Output into a Tensor and vice versa.
RegressorConverter	Can convert a Regressor to a TFloat32 vector and a TFloat32 into a Prediction or Regressor.
TensorFlowCheckpointModel<T extends Output<T>>	This model encapsulates a simple model with an input feed dict, and produces a single output tensor.
TensorFlowFrozenExternalModel<T extends Output<T>>	A Tribuo wrapper around a TensorFlow frozen model.
TensorFlowModel<T extends Output<T>>	Base class for a TensorFlow model that operates on Examples.
TensorFlowNativeModel<T extends Output<T>>	This model encapsulates a TensorFlow model running in graph mode with a single tensor output.
TensorFlowSavedModelExternalModel<T extends Output<T>>	A Tribuo wrapper around a TensorFlow saved model bundle.
TensorFlowTrainer<T extends Output<T>>	Trainer for TensorFlow.
TensorFlowTrainer.TensorFlowTrainerProvenance	Provenance for TensorFlowTrainer.
TensorFlowTrainer.TFModelFormat	The model format to emit.
TensorFlowUtil	Helper functions for working with TensorFlow.
TensorFlowUtil.TensorTuple	A serializable tuple containing the tensor class name, the shape and the data.
TensorMap	A map of names and tensors to feed into a session.
TrainTest	Build and run a Tensorflow multi-class classifier for a standard dataset.
TrainTest.InputType	Type of feature extractor.
TrainTest.TensorflowOptions	Options for training a model in TensorFlow.