org.tribuo.interop.tensorflow.sequence

Interface SequenceOutputTransformer<T extends Output<T>>

All Superinterfaces:

com.oracle.labs.mlrg.olcut.config.Configurable, com.oracle.labs.mlrg.olcut.provenance.Provenancable<com.oracle.labs.mlrg.olcut.provenance.ConfiguredObjectProvenance>, Serializable

public interface SequenceOutputTransformer<T extends Output<T>>
extends com.oracle.labs.mlrg.olcut.config.Configurable, com.oracle.labs.mlrg.olcut.provenance.Provenancable<com.oracle.labs.mlrg.olcut.provenance.ConfiguredObjectProvenance>, Serializable

Converts a Tensorflow output tensor into a list of predictions, and a Tribuo sequence example into a Tensorflow tensor suitable for training.

Method Summary

Modifier and Type	Method and Description
List<List<Prediction<T>>>	decode(org.tensorflow.Tensor<?> outputs, List<SequenceExample<T>> inputBatch, ImmutableOutputInfo<T> labelMap) Decode graph output tensors corresponding to a batch of input sequences.
List<Prediction<T>>	decode(org.tensorflow.Tensor<?> output, SequenceExample<T> input, ImmutableOutputInfo<T> labelMap) Decode a tensor of graph output into a list of predictions for the input sequence.
Map<String,org.tensorflow.Tensor<?>>	encode(List<SequenceExample<T>> batch, ImmutableOutputInfo<T> labelMap) Encodes a batch of labels as a feed dict.
Map<String,org.tensorflow.Tensor<?>>	encode(SequenceExample<T> example, ImmutableOutputInfo<T> labelMap) Encodes an example's label as a feed dict.

Methods inherited from interface com.oracle.labs.mlrg.olcut.config.Configurable

postConfig

Methods inherited from interface com.oracle.labs.mlrg.olcut.provenance.Provenancable

getProvenance

Method Detail

decode

List<Prediction<T>> decode(org.tensorflow.Tensor<?> output,
                           SequenceExample<T> input,
                           ImmutableOutputInfo<T> labelMap)

Decode a tensor of graph output into a list of predictions for the input sequence.

Parameters:

output - graph output

input - original input sequence example

labelMap - label domain

Returns:

the model's decoded prediction for the input sequence.

decode

List<List<Prediction<T>>> decode(org.tensorflow.Tensor<?> outputs,
                                 List<SequenceExample<T>> inputBatch,
                                 ImmutableOutputInfo<T> labelMap)

Decode graph output tensors corresponding to a batch of input sequences.

Parameters:

outputs - a tensor corresponding to a batch of outputs.

inputBatch - the original input batch.

labelMap - label domain

Returns:

the model's decoded predictions, one for each example in the input batch.

encode

Map<String,org.tensorflow.Tensor<?>> encode(SequenceExample<T> example,
                                            ImmutableOutputInfo<T> labelMap)

Encodes an example's label as a feed dict.

Parameters:

example - the input example

labelMap - label domain

Returns:

a map from graph placeholder names to their fed-in values.

encode

Map<String,org.tensorflow.Tensor<?>> encode(List<SequenceExample<T>> batch,
                                            ImmutableOutputInfo<T> labelMap)

Encodes a batch of labels as a feed dict.

Parameters:

batch - a batch of examples.

labelMap - label domain

Returns:

a map from graph placeholder names to their fed-in values.