org.deeplearning4j.nn.conf.layers.GravesLSTM Maven / Gradle / Ivy
/*******************************************************************************
* Copyright (c) 2015-2018 Skymind, Inc.
*
* This program and the accompanying materials are made available under the
* terms of the Apache License, Version 2.0 which is available at
* https://www.apache.org/licenses/LICENSE-2.0.
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
* SPDX-License-Identifier: Apache-2.0
******************************************************************************/
package org.deeplearning4j.nn.conf.layers;
import lombok.*;
import org.deeplearning4j.nn.api.Layer;
import org.deeplearning4j.nn.api.ParamInitializer;
import org.deeplearning4j.nn.api.layers.LayerConstraint;
import org.deeplearning4j.nn.conf.NeuralNetConfiguration;
import org.deeplearning4j.nn.conf.inputs.InputType;
import org.deeplearning4j.nn.conf.memory.LayerMemoryReport;
import org.deeplearning4j.nn.layers.recurrent.LSTMHelpers;
import org.deeplearning4j.nn.params.GravesLSTMParamInitializer;
import org.deeplearning4j.optimize.api.TrainingListener;
import org.nd4j.linalg.activations.IActivation;
import org.nd4j.linalg.activations.impl.ActivationSigmoid;
import org.nd4j.linalg.api.buffer.DataType;
import org.nd4j.linalg.api.ndarray.INDArray;
import java.util.ArrayList;
import java.util.Collection;
import java.util.Collections;
import java.util.Map;
/**
* LSTM recurrent net, based on Graves: Supervised Sequence Labelling with Recurrent Neural Networks
* http://www.cs.toronto.edu/~graves/phd.pdf
*
* @author Alex Black
* @see LSTM LSTM class, for the version without peephole connections
* @deprecated Will be eventually removed. Use {@link LSTM} instead, which has similar prediction accuracy, but supports
* CuDNN for faster network training on CUDA (Nvidia) GPUs
*/
@Deprecated
@Data
@NoArgsConstructor
@ToString(callSuper = true)
@EqualsAndHashCode(callSuper = true)
public class GravesLSTM extends AbstractLSTM {
private double forgetGateBiasInit;
private IActivation gateActivationFn = new ActivationSigmoid();
private GravesLSTM(Builder builder) {
super(builder);
this.forgetGateBiasInit = builder.forgetGateBiasInit;
this.gateActivationFn = builder.gateActivationFn;
initializeConstraints(builder);
}
@Override
protected void initializeConstraints(org.deeplearning4j.nn.conf.layers.Layer.Builder builder) {
super.initializeConstraints(builder);
if (((Builder) builder).recurrentConstraints != null) {
if (constraints == null) {
constraints = new ArrayList<>();
}
for (LayerConstraint c : ((Builder) builder).recurrentConstraints) {
LayerConstraint c2 = c.clone();
c2.setParams(Collections.singleton(GravesLSTMParamInitializer.RECURRENT_WEIGHT_KEY));
constraints.add(c2);
}
}
}
@Override
public Layer instantiate(NeuralNetConfiguration conf, Collection trainingListeners,
int layerIndex, INDArray layerParamsView, boolean initializeParams, DataType networkDataType) {
LayerValidation.assertNInNOutSet("GravesLSTM", getLayerName(), layerIndex, getNIn(), getNOut());
org.deeplearning4j.nn.layers.recurrent.GravesLSTM ret =
new org.deeplearning4j.nn.layers.recurrent.GravesLSTM(conf, networkDataType);
ret.setListeners(trainingListeners);
ret.setIndex(layerIndex);
ret.setParamsViewArray(layerParamsView);
Map paramTable = initializer().init(conf, layerParamsView, initializeParams);
ret.setParamTable(paramTable);
ret.setConf(conf);
return ret;
}
@Override
public ParamInitializer initializer() {
return GravesLSTMParamInitializer.getInstance();
}
@Override
public LayerMemoryReport getMemoryReport(InputType inputType) {
//TODO - CuDNN etc
return LSTMHelpers.getMemoryReport(this, inputType);
}
@AllArgsConstructor
public static class Builder extends AbstractLSTM.Builder {
@SuppressWarnings("unchecked")
public GravesLSTM build() {
return new GravesLSTM(this);
}
}
}