All Downloads are FREE. Search and download functionalities are using the official Maven repository.

org.nd4j.linalg.api.ops.impl.transforms.custom.TopK Maven / Gradle / Ivy

There is a newer version: 1.0.0-M2.1
Show newest version
/*******************************************************************************
 * 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.transforms.custom;

import org.nd4j.autodiff.samediff.SDVariable;
import org.nd4j.autodiff.samediff.SameDiff;
import org.nd4j.base.Preconditions;
import org.nd4j.imports.graphmapper.tf.TFGraphMapper;
import org.nd4j.linalg.api.buffer.DataType;
import org.nd4j.linalg.api.ndarray.INDArray;
import org.nd4j.linalg.api.ops.DynamicCustomOp;
import org.nd4j.linalg.util.ArrayUtil;
import org.tensorflow.framework.AttrValue;
import org.tensorflow.framework.GraphDef;
import org.tensorflow.framework.NodeDef;

import java.util.Arrays;
import java.util.Collections;
import java.util.List;
import java.util.Map;

/**
 * Top K op
 *
 * @author Alex Black
 */
public class TopK extends DynamicCustomOp {

    private boolean sorted;
    private int k;

    public TopK(){ }

    public TopK(SameDiff sd, SDVariable in, int k, boolean sorted){
        super(sd, new SDVariable[]{in}, false);
        this.k = k;
        this.sorted = sorted;
        addIArgument(ArrayUtil.fromBoolean(sorted), k);
    }

    @Override
    public String opName(){
        return "top_k";
    }

    @Override
    public String tensorflowName() {
        return "TopKV2";
    }

    @Override
    public void initFromTensorFlow(NodeDef nodeDef, SameDiff initWith, Map attributesForNode, GraphDef graph) {

        String thisName = nodeDef.getName();

        // FIXME: ????
        String inputName = thisName + "/k";
        NodeDef kNode = null;
        for(int i = 0; i < graph.getNodeCount(); i++) {
            if(graph.getNode(i).getName().equals(inputName)){
                kNode = graph.getNode(i);
                break;
            }
        }

        this.sorted = nodeDef.getAttrOrThrow("sorted").getB();

        if (kNode != null) {
            Preconditions.checkState(kNode != null, "Could not find 'k' parameter node for op: %s", thisName);

            INDArray arr = TFGraphMapper.getInstance().getNDArrayFromTensor(inputName, kNode, graph);
            this.k = arr.getInt(0);

            addIArgument(ArrayUtil.fromBoolean(sorted), k);
        } else
            addIArgument(ArrayUtil.fromBoolean(sorted));
    }

    @Override
    public List doDiff(List i_v) {
        throw new UnsupportedOperationException("Not implemented yet");
    }

    @Override
    public List calculateOutputDataTypes(List dataTypes){
        //2 outputs: values and indices
        //TODO make thit configurable
        return Arrays.asList(dataTypes.get(0), DataType.INT);
    }
}




© 2015 - 2024 Weber Informatics LLC | Privacy Policy