org.nd4j.linalg.api.ops.impl.loss.SoftmaxCrossEntropyWithLogitsLoss 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.nd4j.linalg.api.ops.impl.loss;
import lombok.NoArgsConstructor;
import org.nd4j.autodiff.samediff.SDVariable;
import org.nd4j.autodiff.samediff.SameDiff;
import org.nd4j.base.Preconditions;
import org.nd4j.linalg.api.buffer.DataType;
import org.nd4j.linalg.api.ops.DynamicCustomOp;
import java.util.Arrays;
import java.util.Collections;
import java.util.List;
/**
* Softmax cross entropy loss with Logits
*
* @author Max Pumperla
*/
@NoArgsConstructor
public class SoftmaxCrossEntropyWithLogitsLoss extends DynamicCustomOp {
protected int classesDim;
public SoftmaxCrossEntropyWithLogitsLoss(SameDiff sameDiff, SDVariable logits, SDVariable weights, SDVariable labels, int classesDim) {
super(null, sameDiff, new SDVariable[]{logits, weights, labels}, false);
this.classesDim = classesDim;
addIArgument(classesDim);
}
@Override
public String opName() {
return "softmax_cross_entropy_loss_with_logits";
}
@Override
public String tensorflowName() {
return "SoftmaxCrossEntropyWithLogits";
}
@Override
public List calculateOutputDataTypes(List inputDataTypes){
Preconditions.checkState(inputDataTypes != null && (inputDataTypes.size() == 2 || inputDataTypes.size() == 3),
"Expected 2 or 3 input datatypes for %s, got %s", getClass(), inputDataTypes);
return Collections.singletonList(inputDataTypes.get(0)); //Same as predictions
}
@Override
public List doDiff(List grad){
//No external gradient
//Args: logits, weigths, label
SDVariable[] grads = f().lossSoftmaxCrossEntropyWithLogitsBp(arg(2), arg(0), arg(1), classesDim);
return Arrays.asList(grads);
}
}