Brutex
/
cavis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Various ND4J/DL4J fixes and improvements (#87) 

* Reshape and reallocate - small fixes

Signed-off-by: AlexDBlack <blacka101@gmail.com>

* Reshape and reallocate - small fixes

Signed-off-by: AlexDBlack <blacka101@gmail.com>

* #6488 ElementWiseVertex broadcast support

Signed-off-by: AlexDBlack <blacka101@gmail.com>

* Constructors and broadcast supported it Transforms.max/min

Signed-off-by: AlexDBlack <blacka101@gmail.com>

* #8054 ElementWiseVertex now supports broadcast inputs

Signed-off-by: AlexDBlack <blacka101@gmail.com>

* #8057 Nd4j.create overload dtype fix

Signed-off-by: AlexDBlack <blacka101@gmail.com>

* #7551 ND4J Shape validation fix

Signed-off-by: AlexDBlack <blacka101@gmail.com>
master
Alex Black 2019-07-30 00:27:38 +10:00 committed by AlexDBlack
parent aa4af2c36d
commit b95417f7c5
13 changed files with 346 additions and 48 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

55
deeplearning4j/deeplearning4j-core/src/test/java/org/deeplearning4j/gradientcheck/GradientCheckTestsComputationGraph.java
View File

@ -16,6 +16,7 @@
package org.deeplearning4j.gradientcheck;
import lombok.extern.slf4j.Slf4j;
import org.deeplearning4j.BaseDL4JTest;
import org.deeplearning4j.TestUtils;
import org.deeplearning4j.datasets.iterator.impl.IrisDataSetIterator;
@ -53,9 +54,9 @@ import java.util.Arrays;
import java.util.Map;
import java.util.Random;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertTrue;
import static org.junit.Assert.*;
@Slf4j
public class GradientCheckTestsComputationGraph extends BaseDL4JTest {
public static final boolean PRINT_RESULTS = true;
@ -287,6 +288,56 @@ public class GradientCheckTestsComputationGraph extends BaseDL4JTest {
}
}
@Test
public void testElementWiseVertexBroadcast(){
ElementWiseVertex.Op[] ops =
new ElementWiseVertex.Op[] {ElementWiseVertex.Op.Add, ElementWiseVertex.Op.Average,
ElementWiseVertex.Op.Subtract, ElementWiseVertex.Op.Max, ElementWiseVertex.Op.Product};
for(boolean firstSmaller : new boolean[]{false, true}) {
for (ElementWiseVertex.Op op : ops) {
ComputationGraphConfiguration conf = new NeuralNetConfiguration.Builder()
.updater(new NoOp())
.dataType(DataType.DOUBLE)
.activation(Activation.TANH)
.seed(12345)
.graphBuilder()
.addInputs("in")
.setOutputs("out")
.layer("l1", new DenseLayer.Builder().nIn(3).nOut(firstSmaller ? 1 : 3).build(), "in") //[mb,3]
.layer("l2", new DenseLayer.Builder().nIn(3).nOut(firstSmaller ? 3 : 1).build(), "in") //[mb,1]
.addVertex("ew", new ElementWiseVertex(op), "l1", "l2")
.layer("out", new OutputLayer.Builder().nIn(3).nOut(2).lossFunction(LossFunctions.LossFunction.MCXENT).activation(Activation.SOFTMAX).build(), "ew")
.build();
ComputationGraph graph = new ComputationGraph(conf);
graph.init();
for (int mb : new int[]{1, 5}) {
String msg = (firstSmaller ? "first smaller, " : "second smaller, ") + "mb=" + mb + ", op=" + op;
log.info("Test: {}", msg);
INDArray in = Nd4j.rand(DataType.FLOAT, mb, 3);
INDArray out = graph.outputSingle(in);
assertArrayEquals(new long[]{mb, 2}, out.shape());
INDArray labels = TestUtils.randomOneHot(mb, 2);
graph.fit(new DataSet(in, labels));
boolean gradOK = GradientCheckUtil.checkGradients(graph, DEFAULT_EPS, DEFAULT_MAX_REL_ERROR,
DEFAULT_MIN_ABS_ERROR, PRINT_RESULTS, RETURN_ON_FIRST_FAILURE, new INDArray[]{in},
new INDArray[]{labels});
assertTrue(msg, gradOK);
TestUtils.testModelSerialization(graph);
}
}
}
}
@Test
public void testCnnDepthMerge() {

5
deeplearning4j/deeplearning4j-core/src/test/java/org/deeplearning4j/nn/conf/graph/ElementWiseVertexTest.java
View File

@ -17,6 +17,7 @@
package org.deeplearning4j.nn.conf.graph;
import org.deeplearning4j.BaseDL4JTest;
import org.deeplearning4j.TestUtils;
import org.deeplearning4j.nn.api.OptimizationAlgorithm;
import org.deeplearning4j.nn.conf.ComputationGraphConfiguration;
import org.deeplearning4j.nn.conf.NeuralNetConfiguration;
@ -34,6 +35,7 @@ import org.nd4j.linalg.activations.impl.ActivationTanH;
import org.nd4j.linalg.api.buffer.DataType;
import org.nd4j.linalg.api.ndarray.INDArray;
import org.nd4j.linalg.api.rng.distribution.impl.UniformDistribution;
import org.nd4j.linalg.dataset.DataSet;
import org.nd4j.linalg.factory.Nd4j;
import org.nd4j.linalg.learning.config.Sgd;
import org.nd4j.linalg.lossfunctions.LossFunctions.LossFunction;
@ -42,6 +44,8 @@ import org.nd4j.linalg.primitives.Pair;
import java.util.Map;
import static org.junit.Assert.assertArrayEquals;
/**
* Created by binesh on 6/14/2017.
*/
@ -690,6 +694,7 @@ public class ElementWiseVertexTest extends BaseDL4JTest {
Assert.assertEquals(0, mse(nullsafe(gradients.get("dense2_b")), dEdb2), this.epsilon);
}
private static double mse(INDArray output, INDArray target) {
double mse_expect = Transforms.pow(output.sub(target), 2.0).sumNumber().doubleValue()
/ (output.columns() * output.rows());

2
deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/conf/ComputationGraphConfiguration.java
View File

@ -350,7 +350,7 @@ public class ComputationGraphConfiguration implements Serializable, Cloneable {
"Use .addInputs(String...) to label (and give an ordering to) the network inputs");
}
if ((networkOutputs == null || networkOutputs.isEmpty()) && !allowNoOutput) {
throw new IllegalStateException("Invalid configuration: network has no outputs." +
throw new IllegalStateException("Invalid configuration: network has no outputs. " +
"Use .setOutput(String...) to specify (and give an ordering to) the output vertices, " +
"or use allowNoOutputs(true) to disable this check");
}

197
deeplearning4j/deeplearning4j-nn/src/main/java/org/deeplearning4j/nn/graph/vertex/impl/ElementWiseVertex.java
View File

@ -27,15 +27,20 @@ import org.nd4j.linalg.api.memory.MemoryWorkspace;
import org.nd4j.linalg.api.ndarray.INDArray;
import org.nd4j.linalg.api.ops.CustomOp;
import org.nd4j.linalg.api.ops.DynamicCustomOp;
import org.nd4j.linalg.api.ops.impl.broadcast.BroadcastTo;
import org.nd4j.linalg.api.ops.impl.transforms.bool.MatchConditionTransform;
import org.nd4j.linalg.api.ops.impl.transforms.pairwise.arithmetic.SubOp;
import org.nd4j.linalg.api.ops.impl.transforms.pairwise.bool.Or;
import org.nd4j.linalg.api.ops.impl.transforms.pairwise.arithmetic.OldSubOp;
import org.nd4j.linalg.api.shape.Shape;
import org.nd4j.linalg.factory.Nd4j;
import org.nd4j.linalg.indexing.conditions.Conditions;
import org.nd4j.linalg.ops.transforms.Transforms;
import org.nd4j.linalg.primitives.Pair;
import org.deeplearning4j.nn.workspace.ArrayType;
import org.deeplearning4j.nn.workspace.LayerWorkspaceMgr;
import java.util.Arrays;
/** An ElementWiseVertex is used to combine the activations of two or more layer in an element-wise manner<br>
* For example, the activations may be combined by addition, subtraction or multiplication or by selecting the maximum.
* Addition, Average, Product and Max may use an arbitrary number of input arrays. Note that in the case of subtraction, only two inputs may be used.
@ -80,17 +85,44 @@ public class ElementWiseVertex extends BaseGraphVertex {
if (inputs.length == 1)
return workspaceMgr.dup(ArrayType.ACTIVATIONS, inputs[0]);
boolean isBc = false;
for(int i=1; i<inputs.length; i++ ){
if(!inputs[0].equalShapes(inputs[i])){
isBc = true;
break;
}
}
long[] outShape;
if(!isBc){
outShape = inputs[0].shape();
} else {
outShape = Shape.broadcastOutputShape(inputs[0].shape(), inputs[1].shape());
for( int i=2; i<inputs.length; i++ ){
outShape = Shape.broadcastOutputShape(outShape, inputs[i].shape());
}
}
switch (op) {
case Add:
INDArray sum = workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, dataType, inputs[0].shape());
sum.assign(inputs[0]);
INDArray sum = workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, dataType, outShape);
if(isBc && !Arrays.equals(outShape, inputs[0].shape())){
Nd4j.exec(new BroadcastTo(inputs[0], outShape, sum));
} else {
sum.assign(inputs[0]);
}
for (int i = 1; i < inputs.length; i++) {
sum.addi(inputs[i].castTo(dataType));
}
return sum;
case Average:
INDArray average = workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, dataType, inputs[0].shape());
average.assign(inputs[0]);
INDArray average = workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, dataType, outShape);
if(isBc && !Arrays.equals(outShape, inputs[0].shape())){
Nd4j.exec(new BroadcastTo(inputs[0], outShape, average));
} else {
average.assign(inputs[0]);
}
for (int i = 1; i < inputs.length; i++) {
average.addi(inputs[i].castTo(dataType));
}
@ -98,23 +130,49 @@ public class ElementWiseVertex extends BaseGraphVertex {
case Subtract:
if (inputs.length != 2)
throw new IllegalArgumentException("ElementWise subtraction only supports 2 inputs");
return Nd4j.getExecutioner().exec(new OldSubOp(inputs[0], inputs[1], workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, inputs[0].dataType(), inputs[0].shape())));
return Nd4j.exec(new SubOp(inputs, new INDArray[]{workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, inputs[0].dataType(), outShape)}))[0];
case Product:
INDArray product = workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, dataType, inputs[0].shape());
product.assign(inputs[0]);
INDArray product = workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, dataType, outShape);
if(isBc && !Arrays.equals(outShape, inputs[0].shape())){
Nd4j.exec(new BroadcastTo(inputs[0], outShape, product));
} else {
product.assign(inputs[0]);
}
for (int i = 1; i < inputs.length; i++) {
product.muli(inputs[i].castTo(dataType));
}
return product;
case Max:
INDArray max = workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, inputs[0].dataType(), inputs[0].shape(), inputs[0].ordering());
CustomOp op = DynamicCustomOp.builder("mergemax")
.addInputs(inputs)
.addOutputs(max)
.callInplace(false)
.build();
Nd4j.getExecutioner().exec(op);
return max;
boolean isBroadcast = false;
for(int i=1; i<inputs.length; i++ ){
isBroadcast |= !inputs[0].equalShapes(inputs[i]);
if(isBroadcast)
break;
}
if(!isBroadcast) {
INDArray max = workspaceMgr.createUninitialized(ArrayType.ACTIVATIONS, inputs[0].dataType(), inputs[0].shape(), inputs[0].ordering());
CustomOp op = DynamicCustomOp.builder("mergemax")
.addInputs(inputs)
.addOutputs(max)
.callInplace(false)
.build();
Nd4j.getExecutioner().exec(op);
return max;
} else {
//AB 20190729 mergemax doesn't support broadcast at this point
if(inputs.length == 1){
return workspaceMgr.leverageTo(ArrayType.ACTIVATIONS, inputs[0]);
} else {
INDArray max = Transforms.max(inputs[0], inputs[1], true);
for( int i=2; i<inputs.length; i++ ){
max = Transforms.max(max, inputs[i], false);
}
return workspaceMgr.leverageTo(ArrayType.ACTIVATIONS, max);
}
}
default:
throw new UnsupportedOperationException("Unknown op: " + this.op);
}
@ -128,40 +186,119 @@ public class ElementWiseVertex extends BaseGraphVertex {
if (nInForwardPass == 1)
return new Pair<>(null, new INDArray[] {workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon)});
boolean broadcastCase = false;
for( int i=1; i<nInForwardPass; i++ ){
broadcastCase |= !inputs[0].equalShapes(inputs[i]);
}
switch (op) {
case Add:
//If x=sum_i a_i then dL/da_i = dL/dx * dx/da_i = dL/dx
INDArray[] out = new INDArray[nInForwardPass];
for (int i = 0; i < nInForwardPass; i++)
out[i] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon);
for (int i = 0; i < nInForwardPass; i++) {
if(!broadcastCase) {
//Standard case
out[i] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon);
} else {
//For broadcast case, we need to sum along the broadcast dimensions
//So if [mb,3]+[mb,1] -> input 0 backprops epsilon, input 1 backprops epsilon.sum(1,keepDim=true)
if(inputs[i].equalShapes(epsilon)){
out[i] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon);
} else {
int[] bcDim = Shape.getBroadcastDimensions(inputs[i].shape(), epsilon.shape());
try(MemoryWorkspace ws = workspaceMgr.notifyScopeBorrowed(ArrayType.ACTIVATION_GRAD)){
out[i] = epsilon.sum(true, bcDim);
}
}
}
}
return new Pair<>(null, out);
case Average:
INDArray[] outAverage = new INDArray[nInForwardPass];
try(MemoryWorkspace ws = workspaceMgr.notifyScopeBorrowed(ArrayType.ACTIVATION_GRAD)){
for (int i = 0; i < nInForwardPass; i++)
outAverage[i] = epsilon.div(nInForwardPass);
for (int i = 0; i < nInForwardPass; i++) {
if(inputs[i].equalShapes(epsilon)){
outAverage[i] = epsilon.div(nInForwardPass);
} else {
int[] bcDim = Shape.getBroadcastDimensions(inputs[i].shape(), epsilon.shape());
outAverage[i] = epsilon.div(nInForwardPass).sum(true, bcDim);
}
}
}
return new Pair<>(null, outAverage);
case Subtract:
INDArray[] out2 = new INDArray[2];
out2[0] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon);
out2[1] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon).negi();
if(!broadcastCase){
out2[0] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon);
out2[1] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon).negi();
} else {
if(inputs[0].equalShapes(epsilon)){
//Second input is smaller/broadcast
out2[0] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon);
int[] bcDim = Shape.getBroadcastDimensions(inputs[1].shape(), epsilon.shape());
try(MemoryWorkspace ws = workspaceMgr.notifyScopeBorrowed(ArrayType.ACTIVATION_GRAD)) {
out2[1] = epsilon.sum(true, bcDim).negi();
}
} else {
//First input is smaller/broadcast
int[] bcDim = Shape.getBroadcastDimensions(inputs[0].shape(), epsilon.shape());
try(MemoryWorkspace ws = workspaceMgr.notifyScopeBorrowed(ArrayType.ACTIVATION_GRAD)) {
out2[0] = epsilon.sum(true, bcDim);
}
out2[1] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon).negi();
}
}
return new Pair<>(null, out2);
case Product:
INDArray[] out_product = new INDArray[nInForwardPass];
INDArray[] inBc = inputs;
if(broadcastCase){
inBc = new INDArray[inputs.length];
for( int i=0; i<inputs.length; i++ ){
if(inputs[i].equalShapes(epsilon)){
inBc[i] = inputs[i];
} else {
inBc[i] = epsilon.ulike();
Nd4j.exec(new BroadcastTo(inputs[i], epsilon.shape(), inBc[i]));
}
}
}
for (int i = 0; i < nInForwardPass; i++) {
out_product[i] = workspaceMgr.dup(ArrayType.ACTIVATION_GRAD, epsilon);
for (int j = 0; j < nInForwardPass; ++j) {
if (i != j)
out_product[i].muli(inputs[j]);
out_product[i].muli(inBc[j]);
}
if(!inputs[i].equalShapes(epsilon)){
int[] bcDim = Shape.getBroadcastDimensions(inputs[i].shape(), epsilon.shape());
try(MemoryWorkspace ws = workspaceMgr.notifyScopeBorrowed(ArrayType.ACTIVATION_GRAD)) {
out_product[i] = out_product[i].sum(true, bcDim);
}
}
}
return new Pair<>(null, out_product);
case Max:
INDArray[] outMax = new INDArray[nInForwardPass];
INDArray maxIndices = workspaceMgr.createUninitialized(ArrayType.BP_WORKING_MEM, DataType.INT, epsilon.shape(), epsilon.ordering());
INDArray[] bcIn = inputs;
if(broadcastCase){
//Broadcast to right shape...
bcIn = new INDArray[inputs.length];
for( int i=0; i<inputs.length; i++ ){
if(inputs[i].equalShapes(epsilon)){
bcIn[i] = inputs[i];
} else {
bcIn[i] = epsilon.ulike();
Nd4j.exec(new BroadcastTo(inputs[i], epsilon.shape(), bcIn[i]));
}
}
}
CustomOp op = DynamicCustomOp.builder("mergemaxindex")
.addInputs(inputs)
.addInputs(bcIn)
.addOutputs(maxIndices)
.callInplace(false)
.build();
@ -172,7 +309,17 @@ public class ElementWiseVertex extends BaseGraphVertex {
//generate a mask with 1s and 0s in the right places and muli with epsilon
MatchConditionTransform nd4jop = new MatchConditionTransform(maxIndices, outMax[i], Conditions.equals(i));
Nd4j.getExecutioner().exec(nd4jop);
outMax[i] = workspaceMgr.leverageTo(ArrayType.ACTIVATION_GRAD, outMax[i].castTo(Nd4j.defaultFloatingPointType())).muli(epsilon);
if(broadcastCase && !epsilon.equalShapes(inputs[i])){
//Broadcast for ths input
outMax[i] = outMax[i].castTo(epsilon.dataType()).mul(epsilon);
int[] bcDim = Shape.getBroadcastDimensions(inputs[i].shape(), epsilon.shape());
try(MemoryWorkspace ws = workspaceMgr.notifyScopeBorrowed(ArrayType.ACTIVATION_GRAD)) {
outMax[i] = outMax[i].sum(true, bcDim);
}
} else {
//Standard case
outMax[i] = workspaceMgr.leverageTo(ArrayType.ACTIVATION_GRAD, outMax[i].castTo(epsilon.dataType()).muli(epsilon));
}
}
return new Pair<>(null, outMax);
default:

3
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/api/ndarray/BaseNDArray.java
View File

@ -2476,6 +2476,7 @@ public abstract class BaseNDArray implements INDArray, Iterable {
// length/data.length can be different in case of Threshold conversion
if(isEmpty() || isS())
return false;
return Shape.offset(jvmShapeInfo.javaShapeInformation) > 0
|| (length() < data().length() && data.dataType() != DataType.INT)
|| data().originalDataBuffer() != null;
@ -4577,7 +4578,7 @@ public abstract class BaseNDArray implements INDArray, Iterable {
return ret;
} else {
INDArray ret = this.dup(order);
return ret.reshape(order, shape);
return Nd4j.create(ret.data(), shape);
}
}

4
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/api/ops/impl/transforms/custom/Max.java
View File

@ -44,6 +44,10 @@ public class Max extends BaseDynamicTransformOp {
super(sameDiff, args, inPlace);
}
public Max( INDArray first, INDArray second, INDArray out){
super(new INDArray[]{first, second}, out == null ? null : new INDArray[]{out});
}
public Max( INDArray[] inputs, INDArray[] outputs) {
super(inputs, outputs);
}

4
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/api/ops/impl/transforms/custom/Min.java
View File

@ -44,6 +44,10 @@ public class Min extends BaseDynamicTransformOp {
super(sameDiff, args, inPlace);
}
public Min( INDArray first, INDArray second, INDArray out){
super(new INDArray[]{first, second}, out == null ? null : new INDArray[]{out});
}
public Min( INDArray[] inputs, INDArray[] outputs) {
super(inputs, outputs);
}

2
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/api/ops/impl/transforms/same/Max.java
View File

@ -26,7 +26,7 @@ import java.util.Collections;
import java.util.List;
/**
* Calculate the absolute minimum over a vector
* Calculate the maximum value between two arrays in an elementwise fashion, broadcasting if required
*
* @author raver119@gmail.com
*/

2
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/api/ops/impl/transforms/same/Min.java
View File

@ -26,7 +26,7 @@ import java.util.Collections;
import java.util.List;
/**
* Calculate the absolute minimum over a vector
* Calculate the minimum value between two arrays in an elementwise fashion, broadcasting if required
*
* @author raver119@gmail.com
*/

19
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/factory/Nd4j.java
View File

@ -538,7 +538,7 @@ public class Nd4j {
public static INDArray create(int[] sliceShape, float[]... arrays) {
//TODO: Remove duplicate code.
int slices = arrays.length;
INDArray ret = Nd4j.create(ArrayUtil.combine(new int[] {slices}, sliceShape));
INDArray ret = Nd4j.createUninitialized(DataType.FLOAT, ArrayUtil.toLongArray(ArrayUtil.combine(new int[] {slices}, sliceShape)));
for (int i = 0; i < ret.slices(); i++)
ret.putSlice(i, Nd4j.create(arrays[i]).reshape(ArrayUtil.toLongArray(sliceShape)));
return ret;
@ -572,7 +572,7 @@ public class Nd4j {
*/
public static INDArray create(int[] sliceShape, double[]... arrays) {
int slices = arrays.length;
INDArray ret = Nd4j.create(ArrayUtil.combine(new int[] {slices}, sliceShape));
INDArray ret = Nd4j.createUninitialized(DataType.DOUBLE, ArrayUtil.toLongArray(ArrayUtil.combine(new int[] {slices}, sliceShape)));
for (int i = 0; i < ret.slices(); i++)
ret.putSlice(i, Nd4j.create(arrays[i]).reshape(ArrayUtil.toLongArray(sliceShape)));
return ret;
@ -3984,6 +3984,7 @@ public class Nd4j {
* @return the created ndarray.
*/
public static INDArray create(int[] data, long[] shape, DataType type) {
checkShapeValues(data.length, shape);
return INSTANCE.create(data, shape, Nd4j.getStrides(shape), type, Nd4j.getMemoryManager().getCurrentWorkspace());
}
@ -3991,6 +3992,7 @@ public class Nd4j {
* See {@link #create(int[], long[], DataType)}
*/
public static INDArray create(long[] data, long[] shape, DataType type) {
checkShapeValues(data.length, shape);
return INSTANCE.create(data, shape, Nd4j.getStrides(shape), type, Nd4j.getMemoryManager().getCurrentWorkspace());
}
@ -3998,6 +4000,7 @@ public class Nd4j {
* See {@link #create(int[], long[], DataType)}
*/
public static INDArray create(double[] data, long[] shape, DataType type) {
checkShapeValues(data.length, shape);
return INSTANCE.create(data, shape, Nd4j.getStrides(shape), type, Nd4j.getMemoryManager().getCurrentWorkspace());
}
@ -4005,6 +4008,7 @@ public class Nd4j {
* See {@link #create(int[], long[], DataType)}
*/
public static INDArray create(float[] data, long[] shape, DataType type) {
checkShapeValues(data.length, shape);
return INSTANCE.create(data, shape, Nd4j.getStrides(shape), type, Nd4j.getMemoryManager().getCurrentWorkspace());
}
@ -4012,6 +4016,7 @@ public class Nd4j {
* See {@link #create(int[], long[], DataType)}
*/
public static INDArray create(short[] data, long[] shape, DataType type) {
checkShapeValues(data.length, shape);
return INSTANCE.create(data, shape, Nd4j.getStrides(shape), type, Nd4j.getMemoryManager().getCurrentWorkspace());
}
@ -4019,6 +4024,7 @@ public class Nd4j {
* See {@link #create(int[], long[], DataType)}
*/
public static INDArray create(byte[] data, long[] shape, DataType type) {
checkShapeValues(data.length, shape);
return INSTANCE.create(data, shape, Nd4j.getStrides(shape), type, Nd4j.getMemoryManager().getCurrentWorkspace());
}
@ -4026,6 +4032,7 @@ public class Nd4j {
* See {@link #create(int[], long[], DataType)}
*/
public static INDArray create(boolean[] data, long[] shape, DataType type) {
checkShapeValues(data.length, shape);
return INSTANCE.create(data, shape, Nd4j.getStrides(shape), type, Nd4j.getMemoryManager().getCurrentWorkspace());
}
@ -5165,17 +5172,17 @@ public class Nd4j {
protected static void checkShapeValues(int length, int... shape) {
checkShapeValues(shape);
if (ArrayUtil.prodLong(shape) > length)
if (ArrayUtil.prodLong(shape) != length && !(length == 1 && shape.length == 0))
throw new ND4JIllegalStateException("Shape of the new array " + Arrays.toString(shape)
+ " doesn't match data length: " + length);
+ " doesn't match data length: " + length + " - prod(shape) must equal the number of values provided");
}
protected static void checkShapeValues(int length, long... shape) {
checkShapeValues(shape);
if (ArrayUtil.prodLong(shape) > length)
if (ArrayUtil.prodLong(shape) != length && !(length == 1 && shape.length == 0))
throw new ND4JIllegalStateException("Shape of the new array " + Arrays.toString(shape)
+ " doesn't match data length: " + length);
+ " doesn't match data length: " + length + " - prod(shape) must equal the number of values provided");
}

32
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/ops/transforms/Transforms.java
View File

@ -45,9 +45,11 @@ import org.nd4j.linalg.api.ops.impl.transforms.pairwise.bool.Xor;
import org.nd4j.linalg.api.ops.impl.transforms.same.*;
import org.nd4j.linalg.api.ops.impl.transforms.strict.*;
import org.nd4j.linalg.api.shape.LongShapeDescriptor;
import org.nd4j.linalg.api.shape.Shape;
import org.nd4j.linalg.factory.Nd4j;
import org.nd4j.linalg.inverse.InvertMatrix;
import java.util.Arrays;
import java.util.List;
/**
@ -858,11 +860,11 @@ public class Transforms {
* @return
*/
public static INDArray max(INDArray first, INDArray second, boolean dup) {
INDArray result = first;
if (dup) {
result = first.ulike();
}
return exec(new OldMax(first, second, result));
long[] outShape = broadcastResultShape(first, second); //Also validates
Preconditions.checkState(dup || Arrays.equals(outShape, first.shape()), "Cannot do inplace max operation when first input is not equal to result shape (%ndShape vs. result %s)",
first, outShape);
INDArray out = dup ? Nd4j.create(first.dataType(), outShape) : first;
return Nd4j.exec(new org.nd4j.linalg.api.ops.impl.transforms.custom.Max(first, second, out))[0];
}
/**
@ -908,10 +910,11 @@ public class Transforms {
* @return
*/
public static INDArray min(INDArray first, INDArray second, boolean dup) {
if (dup) {
first = first.dup();
}
return exec(new OldMin(second, first, first));
long[] outShape = broadcastResultShape(first, second); //Also validates
Preconditions.checkState(dup || Arrays.equals(outShape, first.shape()), "Cannot do inplace min operation when first input is not equal to result shape (%ndShape vs. result %s)",
first, outShape);
INDArray out = dup ? Nd4j.create(first.dataType(), outShape) : first;
return Nd4j.exec(new org.nd4j.linalg.api.ops.impl.transforms.custom.Min(first, second, out))[0];
}
/**
@ -1179,4 +1182,15 @@ public class Transforms {
}
}
protected static long[] broadcastResultShape(INDArray first, INDArray second){
if(first.equalShapes(second)){
return first.shape();
} else if(Shape.areShapesBroadcastable(first.shape(), second.shape())){
return Shape.broadcastOutputShape(first.shape(), second.shape());
} else {
throw new IllegalStateException("Array shapes are not broadcastable: " + Arrays.toString(first.shape()) +
" vs. " + Arrays.toString(second.shape()));
}
}
}

66
nd4j/nd4j-backends/nd4j-tests/src/test/java/org/nd4j/linalg/Nd4jTestsC.java
View File

@ -2699,6 +2699,21 @@ public class Nd4jTestsC extends BaseNd4jTest {
assertEquals(expected, actual);
}
@Test
public void testBroadcastDiv2(){
INDArray arr = Nd4j.ones(DataType.DOUBLE, 1, 64, 125, 125).muli(2);
INDArray vec = Nd4j.ones(DataType.DOUBLE, 64).muli(2);
INDArray exp = Nd4j.ones(DataType.DOUBLE, 1, 64, 125, 125);
INDArray out = arr.like();
for( int i=0; i<10; i++ ) {
out.assign(0.0);
Nd4j.getExecutioner().exec(new BroadcastDivOp(arr, vec, out, 1));
assertEquals(exp, out);
}
}
@Test
public void testBroadcastMult() {
@ -7417,7 +7432,8 @@ public class Nd4jTestsC extends BaseNd4jTest {
INDArray arr1a = Nd4j.create(new long[]{2,3}, 'c').get(NDArrayIndex.all(), NDArrayIndex.interval(0,2));
INDArray arr3 = arr1a.reshape('c', false, 4,1);
assertFalse(arr3.isView()); //Should be copy
boolean isView = arr3.isView();
assertFalse(isView); //Should be copy
try{
INDArray arr4 = arr1a.reshape('c', true, 4,1);
@ -7861,6 +7877,54 @@ public class Nd4jTestsC extends BaseNd4jTest {
final INDArray arr2 = arr1.reshape(3,1);
assertEquals("Incorrect type!", DataType.FLOAT, arr1.mmul(arr2).dataType());
}
@Test
public void testCreateDtypes() {
int[] sliceShape = new int[] {9};
float[] arrays = new float[] {1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f, 7.0f, 8.0f, 9.0f};
double [] arrays_double = new double[] {1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0};
INDArray x = Nd4j.create( sliceShape, arrays, arrays );
assertEquals(DataType.FLOAT, x.dataType());
INDArray xd = Nd4j.create( sliceShape, arrays_double, arrays_double );
assertEquals(DataType.DOUBLE, xd.dataType());
}
@Test
public void testCreateShapeValidation(){
try {
Nd4j.create(new double[]{1, 2, 3}, new int[]{1, 1});
fail();
} catch (Exception t){
assertTrue(t.getMessage().contains("length"));
}
try {
Nd4j.create(new float[]{1, 2, 3}, new int[]{1, 1});
fail();
} catch (Exception t){
assertTrue(t.getMessage().contains("length"));
}
try {
Nd4j.create(new byte[]{1, 2, 3}, new long[]{1, 1}, DataType.BYTE);
fail();
} catch (Exception t){
assertTrue(t.getMessage().contains("length"));
}
try {
Nd4j.create(new double[]{1, 2, 3}, new int[]{1, 1}, 'c');
fail();
} catch (Exception t){
assertTrue(t.getMessage().contains("length"));
}
}
///////////////////////////////////////////////////////
protected static void fillJvmArray3D(float[][][] arr) {
int cnt = 1;

3
nd4j/nd4j-buffer/src/main/java/org/nd4j/linalg/api/buffer/BaseDataBuffer.java
View File

@ -2601,7 +2601,8 @@ public abstract class BaseDataBuffer implements DataBuffer {
}
Pointer.memcpy(pointer, oldPointer, this.length() * getElementSize());
//this.underlyingLength = length;
this.underlyingLength = length;
this.length = length;
return this;
}