Change NDArray.toString() output

api/src/main/java/ai/djl/ndarray/NDArray.java

@@ -4397,6 +4397,15 @@ default NDArray countNonzero(int axis) {
      */
     NDArrayEx getNDArrayInternal();
 
+    /**
+     * Runs the debug string representation of this {@code NDArray}.
+     *
+     * @return the debug string representation of this {@code NDArray}
+     */
+    default String toDebugString() {
+        return toDebugString(100, 10, 10, 20);
+    }
+
     /**
      * Runs the debug string representation of this {@code NDArray}.
      *

api/src/main/java/ai/djl/ndarray/internal/NDFormat.java

@@ -16,6 +16,8 @@
 import ai.djl.ndarray.types.DataType;
 import ai.djl.ndarray.types.Shape;
 import ai.djl.util.Utils;
+import java.lang.management.ManagementFactory;
+import java.util.Arrays;
 import java.util.Locale;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
@@ -26,6 +28,11 @@ public abstract class NDFormat {
     private static final int PRECISION = 8;
     private static final String LF = System.getProperty("line.separator");
     private static final Pattern PATTERN = Pattern.compile("\\s*\\d\\.(\\d*?)0*e[+-](\\d+)");
+    private static final boolean DEBUG =
+            ManagementFactory.getRuntimeMXBean()
+                    .getInputArguments()
+                    .stream()
+                    .anyMatch(arg -> arg.startsWith("-agentlib:jdwp"));
 
     /**
      * Formats the contents of an array as a pretty printable string.
@@ -39,24 +46,6 @@ public abstract class NDFormat {
      */
     public static String format(
             NDArray array, int maxSize, int maxDepth, int maxRows, int maxColumns) {
-        NDFormat format;
-        DataType dataType = array.getDataType();
-
-        if (dataType == DataType.UINT8) {
-            format = new HexFormat();
-        } else if (dataType == DataType.BOOLEAN) {
-            format = new BooleanFormat();
-        } else if (dataType.isInteger()) {
-            format = new IntFormat(array);
-        } else {
-            format = new FloatFormat(array);
-        }
-        return format.dump(array, maxSize, maxDepth, maxRows, maxColumns);
-    }
-
-    protected abstract CharSequence format(Number value);
-
-    private String dump(NDArray array, int maxSize, int maxDepth, int maxRows, int maxColumns) {
         StringBuilder sb = new StringBuilder(1000);
         String name = array.getName();
         if (name != null) {
@@ -72,6 +61,39 @@ private String dump(NDArray array, int maxSize, int maxDepth, int maxRows, int m
         if (array.hasGradient()) {
             sb.append(" hasGradient");
         }
+        if (DEBUG) {
+            return sb.toString();
+        }
+
+        NDFormat format;
+        DataType dataType = array.getDataType();
+
+        if (dataType == DataType.UINT8) {
+            format = new HexFormat();
+        } else if (dataType == DataType.BOOLEAN) {
+            format = new BooleanFormat();
+        } else if (dataType == DataType.STRING) {
+            format = new StringFormat();
+        } else if (dataType.isInteger()) {
+            format = new IntFormat();
+        } else {
+            format = new FloatFormat();
+        }
+        return format.dump(sb, array, maxSize, maxDepth, maxRows, maxColumns);
+    }
+
+    protected abstract CharSequence format(Number value);
+
+    protected void init(NDArray array) {}
+
+    protected String dump(
+            StringBuilder sb,
+            NDArray array,
+            int maxSize,
+            int maxDepth,
+            int maxRows,
+            int maxColumns) {
+        init(array);
         sb.append(LF);
 
         long size = array.size();
@@ -152,7 +174,9 @@ private static final class FloatFormat extends NDFormat {
         private int precision;
         private int totalLength;
 
-        public FloatFormat(NDArray array) {
+        /** {@inheritDoc} */
+        @Override
+        public void init(NDArray array) {
             Number[] values = array.toArray();
             int maxIntPartLen = 0;
             int maxFractionLen = 0;
@@ -290,7 +314,9 @@ private static final class IntFormat extends NDFormat {
         private int precision;
         private int totalLength;
 
-        public IntFormat(NDArray array) {
+        /** {@inheritDoc} */
+        @Override
+        public void init(NDArray array) {
             Number[] values = array.toArray();
             // scalar case
             if (values.length == 1) {
@@ -338,4 +364,25 @@ public CharSequence format(Number value) {
             return value.byteValue() != 0 ? " true" : "false";
         }
     }
+
+    private static final class StringFormat extends NDFormat {
+
+        /** {@inheritDoc} */
+        @Override
+        public CharSequence format(Number value) {
+            return null;
+        }
+
+        /** {@inheritDoc} */
+        @Override
+        protected String dump(
+                StringBuilder sb,
+                NDArray array,
+                int maxSize,
+                int maxDepth,
+                int maxRows,
+                int maxColumns) {
+            return Arrays.toString(array.toStringArray());
+        }
+    }
 }

dlr/dlr-engine/src/main/java/ai/djl/dlr/engine/DlrNDArray.java

@@ -19,7 +19,6 @@
 import ai.djl.ndarray.types.DataType;
 import ai.djl.ndarray.types.Shape;
 import java.nio.ByteBuffer;
-import java.util.Arrays;
 import java.util.UUID;
 
 /** {@code DlrNDArray} is the DLR implementation of {@link NDArray}. */
@@ -129,14 +128,7 @@ public String toString() {
         if (isClosed) {
             return "This array is already closed";
         }
-        return "ND: "
-                + getShape()
-                + ' '
-                + getDevice()
-                + ' '
-                + getDataType()
-                + '\n'
-                + Arrays.toString(toArray());
+        return toDebugString();
     }
 
     /** {@inheritDoc} */

extensions/benchmark/build.gradle

@@ -14,6 +14,8 @@ dependencies {
         implementation "ai.djl:model-zoo"
         runtimeOnly "ai.djl.pytorch:pytorch-model-zoo"
         runtimeOnly "ai.djl.pytorch:pytorch-native-auto"
+
+        runtimeOnly "ai.djl.tensorflow:tensorflow-model-zoo"
         runtimeOnly "ai.djl.tensorflow:tensorflow-native-auto"
 
         runtimeOnly "ai.djl.mxnet:mxnet-model-zoo"

ml/xgboost/src/main/java/ai/djl/ml/xgboost/XgbNDArray.java

@@ -20,7 +20,6 @@
 import ai.djl.ndarray.types.Shape;
 import ai.djl.ndarray.types.SparseFormat;
 import java.nio.ByteBuffer;
-import java.util.Arrays;
 import java.util.UUID;
 import java.util.concurrent.atomic.AtomicLong;
 import ml.dmlc.xgboost4j.java.JniUtils;
@@ -147,14 +146,7 @@ public String toString() {
         if (isClosed) {
             return "This array is already closed";
         }
-        return "ND: "
-                + getShape()
-                + ' '
-                + getDevice()
-                + ' '
-                + getDataType()
-                + '\n'
-                + Arrays.toString(toArray());
+        return toDebugString();
     }
 
     /** {@inheritDoc} */

mxnet/mxnet-engine/src/main/java/ai/djl/mxnet/engine/MxNDArray.java

@@ -37,11 +37,6 @@
 /** {@code MxNDArray} is the MXNet implementation of {@link NDArray}. */
 public class MxNDArray extends NativeResource<Pointer> implements LazyNDArray {
 
-    private static final int MAX_SIZE = 100;
-    private static final int MAX_DEPTH = 10;
-    private static final int MAX_ROWS = 10;
-    private static final int MAX_COLUMNS = 20;
-
     private String name;
     private Device device;
     private SparseFormat sparseFormat;
@@ -1617,7 +1612,7 @@ public String toString() {
         if (isReleased()) {
             return "This array is already closed";
         }
-        return toDebugString(MAX_SIZE, MAX_DEPTH, MAX_ROWS, MAX_COLUMNS);
+        return toDebugString();
     }
 
     /** {@inheritDoc} */

onnxruntime/onnxruntime-engine/src/main/java/ai/djl/onnxruntime/engine/OrtNDArray.java

@@ -23,7 +23,6 @@
 import ai.onnxruntime.OrtException;
 import java.nio.ByteBuffer;
 import java.nio.ByteOrder;
-import java.util.Arrays;
 import java.util.UUID;
 
 /** {@code OrtNDArray} is the ONNX Runtime implementation of {@link NDArray}. */
@@ -150,13 +149,7 @@ public String toString() {
         if (isClosed) {
             return "This array is already closed";
         }
-        String arrStr;
-        if (getDataType() == DataType.STRING) {
-            arrStr = Arrays.toString(toStringArray());
-        } else {
-            arrStr = Arrays.toString(toArray());
-        }
-        return "ND: " + getShape() + ' ' + getDevice() + ' ' + getDataType() + '\n' + arrStr;
+        return toDebugString();
     }
 
     /** {@inheritDoc} */

paddlepaddle/paddlepaddle-engine/src/main/java/ai/djl/paddlepaddle/engine/PpNDArray.java

@@ -21,7 +21,6 @@
 import ai.djl.paddlepaddle.jni.JniUtils;
 import ai.djl.util.NativeResource;
 import java.nio.ByteBuffer;
-import java.util.Arrays;
 
 /** {@code PpNDArray} is the PaddlePaddle implementation of {@link NDArray}. */
 public class PpNDArray extends NativeResource<Long> implements NDArrayAdapter {
@@ -140,14 +139,7 @@ public String toString() {
         if (isReleased()) {
             return "This array is already closed";
         }
-        return "ND: "
-                + getShape()
-                + ' '
-                + getDevice()
-                + ' '
-                + getDataType()
-                + '\n'
-                + Arrays.toString(toArray());
+        return toDebugString();
     }
 
     /** {@inheritDoc} */

pytorch/pytorch-engine/src/main/java/ai/djl/pytorch/engine/PtNDArray.java

@@ -16,7 +16,6 @@
 import ai.djl.ndarray.NDArray;
 import ai.djl.ndarray.NDList;
 import ai.djl.ndarray.NDManager;
-import ai.djl.ndarray.internal.NDFormat;
 import ai.djl.ndarray.types.DataType;
 import ai.djl.ndarray.types.Shape;
 import ai.djl.ndarray.types.SparseFormat;
@@ -39,11 +38,6 @@
 /** {@code PtNDArray} is the PyTorch implementation of {@link NDArray}. */
 public class PtNDArray extends NativeResource<Long> implements NDArray {
 
-    private static final int MAX_SIZE = 100;
-    private static final int MAX_DEPTH = 10;
-    private static final int MAX_ROWS = 10;
-    private static final int MAX_COLUMNS = 20;
-
     private String name;
     private Device device;
     private DataType dataType;
@@ -328,9 +322,9 @@ public PtNDArray booleanMask(NDArray index, int axis) {
         } else {
             throw new UnsupportedOperationException(
                     "Not supported for shape not broadcastable "
-                            + indexShape.toString()
+                            + indexShape
                             + " vs "
-                            + getShape().toString());
+                            + getShape());
         }
     }
 
@@ -1447,10 +1441,10 @@ public String toString() {
         // index operator in toDebugString is not supported for MKLDNN & Sparse layout
         if (JniUtils.getLayout(this) != 0) {
             try (NDArray tmp = toDense()) {
-                return NDFormat.format(tmp, MAX_SIZE, MAX_DEPTH, MAX_ROWS, MAX_COLUMNS);
+                return tmp.toDebugString();
             }
         }
-        return toDebugString(MAX_SIZE, MAX_DEPTH, MAX_ROWS, MAX_COLUMNS);
+        return toDebugString();
     }
 
     /** {@inheritDoc} */

tensorflow/tensorflow-engine/src/main/java/ai/djl/tensorflow/engine/TfNDArray.java

@@ -39,11 +39,6 @@
 @SuppressWarnings("PMD.UseTryWithResources")
 public class TfNDArray extends NativeResource<TFE_TensorHandle> implements NDArray {
 
-    private static final int MAX_SIZE = 100;
-    private static final int MAX_DEPTH = 10;
-    private static final int MAX_ROWS = 10;
-    private static final int MAX_COLUMNS = 20;
-
     private Shape shape;
     private Device device;
     private TfNDManager manager;
@@ -1611,7 +1606,7 @@ public String toString() {
         if (isReleased()) {
             return "This array is already closed";
         }
-        return toDebugString(MAX_SIZE, MAX_DEPTH, MAX_ROWS, MAX_COLUMNS);
+        return toDebugString();
     }
 
     /** {@inheritDoc} */

tflite/tflite-engine/src/main/java/ai/djl/tflite/engine/TfLiteNDArray.java

@@ -161,8 +161,7 @@ public NDArray toType(DataType dataType, boolean copy) {
                 return manager.create(booleanResult).reshape(shape);
             default:
                 throw new UnsupportedOperationException(
-                        "Type conversion is not supported for TFLite for data type "
-                                + dataType.toString());
+                        "Type conversion is not supported for TFLite for data type " + dataType);
         }
     }
 
@@ -196,14 +195,7 @@ public String toString() {
         if (isClosed) {
             return "This array is already closed";
         }
-        return "ND: "
-                + getShape()
-                + ' '
-                + getDevice()
-                + ' '
-                + getDataType()
-                + '\n'
-                + Arrays.toString(toArray());
+        return toDebugString();
     }
 
     /** {@inheritDoc} */