Squashed commit of the following:

commit 0092f8e Author: Aziz Zayed <azayed01@gmail.com> Date: Tue Jun 15 08:22:51 2021 -0700 Fixed truncated-normal bug commit a6ded8c Author: Aziz Zayed <azayed01@gmail.com> Date: Mon Jun 14 13:33:30 2021 -0700 [pytorch] Add BigGAN demo commit f145614 Merge: a8a1a9b ec8405b Author: Abd-El-Aziz Zayed <48853777+AzizZayed@users.noreply.github.com> Date: Fri Jun 11 20:45:34 2021 -0700 Merge branch 'deepjavalibrary:master' into master commit ec8405b Author: Abd-El-Aziz Zayed <48853777+AzizZayed@users.noreply.github.com> Date: Fri Jun 11 14:53:59 2021 -0700 [pytorch] Add oneHot operator (deepjavalibrary#1014) [tensoflow] Add truncated normal operation commit 50600fd Author: Frank Liu <frankfliu2000@gmail.com> Date: Fri Jun 11 14:53:43 2021 -0700 upgrade dependencies version (deepjavalibrary#1012) Change-Id: I709938f69f21096bc5cd29a24191f0f282dcbc97 commit 3379fd2 Author: Frank Liu <frankfliu2000@gmail.com> Date: Fri Jun 11 14:53:29 2021 -0700 [serving] Fix flaky test (deepjavalibrary#1013) Change-Id: I13b89e04516c59a3d28ecafd49f4f808630b22fb commit 23157fd Author: Frank Liu <frankfliu2000@gmail.com> Date: Thu Jun 10 16:31:03 2021 -0700 Enable spotbugs for java 11+ (deepjavalibrary#1010) Change-Id: I74effbf45492a5cf50e09ba8af0223d2b1bcb5a5 commit 4f38708 Author: Frank Liu <frankfliu2000@gmail.com> Date: Thu Jun 10 16:30:50 2021 -0700 Fix model zoo test typo (deepjavalibrary#1009) Change-Id: I7c0109c6e5fc0ece16288082fd830718f20ad489 commit a8a1a9b Merge: 77809f4 30b03f4 Author: Aziz Zayed <azayed01@gmail.com> Date: Thu Jun 10 15:16:05 2021 -0700 Merge Truncated-Normal branch commit 77809f4 Author: Frank Liu <frankfliu2000@gmail.com> Date: Thu Jun 10 14:07:43 2021 -0700 Make model zoo test weekly (deepjavalibrary#1004) Change-Id: I1c73df17cb077b9ce8905fcc2fc8bbb37b9688d8 commit 0aec8ca Author: Abd-El-Aziz Zayed <48853777+AzizZayed@users.noreply.github.com> Date: Thu Jun 10 12:46:16 2021 -0700 [tensoflow] Add truncated normal operation (deepjavalibrary#1005) commit 30b03f4 Author: Aziz Zayed <azayed01@gmail.com> Date: Wed Jun 9 01:40:33 2021 -0700 [tensoflow] Add truncated normal operation commit d8e7e1d Author: Frank Liu <frankfliu2000@gmail.com> Date: Wed Jun 9 07:55:15 2021 -0700 Fixes deepjavalibrary#999, hanlde UTF16 surrogate charactors properly. (deepjavalibrary#1003) Change-Id: I19e77cf5a8282bea901434041806eb102549ec0f commit b0fe73a Author: Frank Liu <frankfliu2000@gmail.com> Date: Tue Jun 8 17:56:19 2021 -0700 [pytorch] Update load model jupyter notebook (deepjavalibrary#1002) Change-Id: I1889aa93d2002e6ce02c740d2d1d3517bf586760 commit 8286930 Author: Frank Liu <frankfliu2000@gmail.com> Date: Tue Jun 8 15:29:27 2021 -0700 [tensorflow] fix optOption usage document (deepjavalibrary#1001) Change-Id: Ie044839cf082d63010a5c26d3f2f8833447919c6 commit a26f5b2 Author: Abd-El-Aziz Zayed <48853777+AzizZayed@users.noreply.github.com> Date: Tue Jun 8 15:29:10 2021 -0700 Updated PyTorch Docs (deepjavalibrary#1000) * Added auto softmax metadata for action_recognition * Update PyTorch Docs commit e6890f9 Author: Lanking <qingla@amazon.com> Date: Mon Jun 7 18:25:19 2021 -0700 upgrade xgboost (deepjavalibrary#993) commit a0dcf3a Author: Lanking <qingla@amazon.com> Date: Mon Jun 7 18:25:12 2021 -0700 bump up onnx runtime version (deepjavalibrary#992)
AzizZayed · Jun 15, 2021 · 697655b · 697655b
1 parent 2fc24f2
commit 697655b
Show file tree

Hide file tree

Showing 29 changed files with 1,657 additions and 69 deletions.
diff --git a/api/src/main/java/ai/djl/ndarray/BaseNDManager.java b/api/src/main/java/ai/djl/ndarray/BaseNDManager.java
@@ -16,6 +16,7 @@
 import ai.djl.ndarray.types.DataType;
 import ai.djl.ndarray.types.Shape;
 import ai.djl.util.PairList;
+import ai.djl.util.RandomUtils;
 import java.nio.Buffer;
 import java.nio.file.Path;
 import java.util.UUID;
@@ -153,6 +154,24 @@ public NDArray randomNormal(float loc, float scale, Shape shape, DataType dataTy
         throw new UnsupportedOperationException("Not supported!");
     }
 
+    /** {@inheritDoc} */
+    @Override
+    public NDArray truncatedNormal(float loc, float scale, Shape shape, DataType dataType) {
+        int sampleSize = (int) shape.size();
+        double[] dist = new double[sampleSize];
+
+        for (int i = 0; i < sampleSize; i++) {
+            double sample = RandomUtils.nextGaussian();
+            while (sample < -2 || sample > 2) {
+                sample = RandomUtils.nextGaussian();
+            }
+
+            dist[i] = sample;
+        }
+
+        return create(dist).muli(scale).addi(loc).reshape(shape).toType(dataType, false);
+    }
+
     /** {@inheritDoc} */
     @Override
     public NDArray randomMultinomial(int n, NDArray pValues) {

diff --git a/api/src/main/java/ai/djl/ndarray/NDArray.java b/api/src/main/java/ai/djl/ndarray/NDArray.java
@@ -4592,6 +4592,57 @@ default NDArray oneHot(int depth) {
         return oneHot(depth, 1f, 0f, DataType.FLOAT32);
     }
 
+    /**
+     * Returns a one-hot {@code NDArray}.
+     *
+     * <ul>
+     *   <li>The locations represented by indices take value 1, while all other locations take value
+     *       0.
+     *   <li>If the input {@code NDArray} is rank N, the output will have rank N+1. The new axis is
+     *       appended at the end.
+     *   <li>If {@code NDArray} is a scalar the output shape will be a vector of length depth.
+     *   <li>If {@code NDArray} is a vector of length features, the output shape will be features x
+     *       depth.
+     *   <li>If {@code NDArray} is a matrix with shape [batch, features], the output shape will be
+     *       batch x features x depth.
+     * </ul>
+     *
+     * <p>Examples
+     *
+     * <pre>
+     * jshell&gt; NDArray array = manager.create(new int[] {1, 0, 2, 0});
+     * jshell&gt; array.oneHot(3);
+     * ND: (4, 3) cpu() float32
+     * [[0., 1., 0.],
+     *  [1., 0., 0.],
+     *  [0., 0., 1.],
+     *  [1., 0., 0.],
+     * ]
+     * jshell&gt; NDArray array = manager.create(new int[][] {{1, 0}, {1, 0}, {2, 0}});
+     * jshell&gt; array.oneHot(3);
+     * ND: (3, 2, 3) cpu() float32
+     * [[[0., 1., 0.],
+     *   [1., 0., 0.],
+     *  ],
+     *  [[0., 1., 0.],
+     *   [1., 0., 0.],
+     *  ],
+     *  [[0., 0., 1.],
+     *   [1., 0., 0.],
+     *  ],
+     * ]
+     * </pre>
+     *
+     * @param depth Depth of the one hot dimension.
+     * @param dataType dataType of the output.
+     * @return one-hot encoding of this {@code NDArray}
+     * @see <a
+     *     href=https://d2l.djl.ai/chapter_linear-networks/softmax-regression.html#classification-problems>Classification-problems</a>
+     */
+    default NDArray oneHot(int depth, DataType dataType) {
+        return oneHot(depth, 0f, 1f, dataType);
+    }
+
     /**
      * Returns a one-hot {@code NDArray}.
      *

diff --git a/api/src/main/java/ai/djl/ndarray/NDManager.java b/api/src/main/java/ai/djl/ndarray/NDManager.java
@@ -1232,6 +1232,65 @@ default NDArray randomNormal(
         return newSubManager(device).randomNormal(loc, scale, shape, dataType);
     }
 
+    /**
+     * Draws random samples from a normal (Gaussian) distribution with mean 0 and standard deviation
+     * 1, discarding and re-drawing any samples that are more than two standard deviations from the
+     * mean.
+     *
+     * <p>Samples are distributed according to a normal distribution parametrized by mean = 0 and
+     * standard deviation = 1.
+     *
+     * @param shape the output {@link Shape}
+     * @return the drawn samples {@link NDArray}
+     */
+    default NDArray truncatedNormal(Shape shape) {
+        return truncatedNormal(0f, 1f, shape, DataType.FLOAT32);
+    }
+
+    /**
+     * Draws random samples from a normal (Gaussian) distribution with mean 0 and standard deviation
+     * 1, discarding and re-drawing any samples that are more than two standard deviations from the
+     * mean.
+     *
+     * @param shape the output {@link Shape}
+     * @param dataType the {@link DataType} of the {@link NDArray}
+     * @return the drawn samples {@link NDArray}
+     */
+    default NDArray truncatedNormal(Shape shape, DataType dataType) {
+        return truncatedNormal(0.0f, 1.0f, shape, dataType);
+    }
+
+    /**
+     * Draws random samples from a normal (Gaussian) distribution, discarding and re-drawing any
+     * samples that are more than two standard deviations from the mean.
+     *
+     * @param loc the mean (centre) of the distribution
+     * @param scale the standard deviation (spread or "width") of the distribution
+     * @param shape the output {@link Shape}
+     * @param dataType the {@link DataType} of the {@link NDArray}
+     * @return the drawn samples {@link NDArray}
+     */
+    NDArray truncatedNormal(float loc, float scale, Shape shape, DataType dataType);
+
+    /**
+     * Draws random samples from a normal (Gaussian) distribution, discarding and re-drawing any
+     * samples that are more than two standard deviations from the mean.
+     *
+     * @param loc the mean (centre) of the distribution
+     * @param scale the standard deviation (spread or "width") of the distribution
+     * @param shape the output {@link Shape}
+     * @param dataType the {@link DataType} of the {@link NDArray}
+     * @param device the {@link Device} of the {@link NDArray}
+     * @return the drawn samples {@link NDArray}
+     */
+    default NDArray truncatedNormal(
+            float loc, float scale, Shape shape, DataType dataType, Device device) {
+        if (device == null || device.equals(getDevice())) {
+            return truncatedNormal(loc, scale, shape, dataType);
+        }
+        return newSubManager(device).truncatedNormal(loc, scale, shape, dataType);
+    }
+
     /**
      * Draw samples from a multinomial distribution.
      *

diff --git a/api/src/main/native/djl/utils.h b/api/src/main/native/djl/utils.h
@@ -29,9 +29,21 @@ inline std::string GetStringFromJString(JNIEnv* env, jstring jstr) {
   if (jstr == nullptr) {
     return std::string();
   }
-  const char* c_str = env->GetStringUTFChars(jstr, JNI_FALSE);
-  std::string str = std::string(c_str);
-  env->ReleaseStringUTFChars(jstr, c_str);
+
+  // TODO: cache reflection to improve performance
+  const jclass string_class = env->GetObjectClass(jstr);
+  const jmethodID getbytes_method = env->GetMethodID(string_class, "getBytes", "(Ljava/lang/String;)[B");
+
+  const jstring charset = env->NewStringUTF("UTF-8");
+  const jbyteArray jbytes = (jbyteArray) env->CallObjectMethod(jstr, getbytes_method, charset);
+  env->DeleteLocalRef(charset);
+
+  const jsize length = env->GetArrayLength(jbytes);
+  jbyte* c_str = env->GetByteArrayElements(jbytes, NULL);
+  std::string str = std::string(reinterpret_cast<const char *>(c_str), length);
+
+  env->ReleaseByteArrayElements(jbytes, c_str, RELEASE_MODE);
+  env->DeleteLocalRef(jbytes);
   return str;
 }
 
@@ -100,9 +112,23 @@ inline std::vector<std::string> GetVecFromJStringArray(JNIEnv* env, jobjectArray
 // String[]
 inline jobjectArray GetStringArrayFromVec(JNIEnv* env, const std::vector <std::string> &vec) {
   jobjectArray array = env->NewObjectArray(vec.size(), env->FindClass("Ljava/lang/String;"), nullptr);
+
+  // TODO: cache reflection to improve performance
+  const jclass string_class = env->FindClass("java/lang/String");
+  const jmethodID ctor = env->GetMethodID(string_class, "<init>", "([BLjava/lang/String;)V");
+  const jstring charset = env->NewStringUTF("UTF-8");
+
   for (int i = 0; i < vec.size(); ++i) {
-    env->SetObjectArrayElement(array, i, env->NewStringUTF(vec[i].c_str()));
+    const char* c_str = vec[i].c_str();
+    int len = vec[i].length();
+    auto jbytes = env->NewByteArray(len);
+    env->SetByteArrayRegion(jbytes, 0, len, reinterpret_cast<const jbyte*>(c_str));
+    jobject jstr = env->NewObject(string_class, ctor, jbytes, charset);
+    env->DeleteLocalRef(jbytes);
+    env->SetObjectArrayElement(array, i, jstr);
   }
+
+  env->DeleteLocalRef(charset);
   return array;
 }
 

diff --git a/docs/tensorflow/how_to_import_tensorflow_models_in_DJL.md b/docs/tensorflow/how_to_import_tensorflow_models_in_DJL.md
@@ -121,7 +121,7 @@ Criteria<Image, DetectedObjects> criteria =
                         .setTypes(Image.class, DetectedObjects.class)
                         .optFilter("backbone", "mobilenet_v2")
                         .optEngine("TensorFlow")
-                        .optOption("Tags", new String[] {})
+                        .optOption("Tags", "")
                         .optOption("SignatureDefKey", "default")
                         .optProgress(new ProgressBar())
                         .build();

diff --git a/examples/build.gradle b/examples/build.gradle
@@ -44,7 +44,7 @@ dependencies {
 }
 
 application {
-    mainClassName = System.getProperty("main", "ai.djl.examples.inference.ObjectDetection")
+    mainClassName = System.getProperty("main", "ai.djl.examples.inference.biggan.Generator")
 }
 
 run {

diff --git a/examples/src/main/java/ai/djl/examples/inference/biggan/BigGANCategory.java b/examples/src/main/java/ai/djl/examples/inference/biggan/BigGANCategory.java
@@ -0,0 +1,98 @@
+/*
+ * Copyright 2019 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"). You may not use this file except in compliance
+ * with the License. A copy of the License is located at
+ *
+ * http://aws.amazon.com/apache2.0/
+ *
+ * or in the "license" file accompanying this file. This file 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.
+ */
+package ai.djl.examples.inference.biggan;
+
+import java.io.IOException;
+import java.nio.file.Files;
+import java.nio.file.Paths;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Map;
+import java.util.concurrent.ConcurrentHashMap;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+public final class BigGANCategory {
+    private static final Logger logger = LoggerFactory.getLogger(BigGANCategory.class);
+
+    public static final int NUMBER_OF_CATEGORIES = 1000;
+    private static final Map<String, BigGANCategory> CATEGORIES_BY_NAME =
+            new ConcurrentHashMap<>(NUMBER_OF_CATEGORIES);
+    private static String[] categoriesById;
+
+    private int id;
+    private String[] names;
+
+    static {
+        try {
+            parseCategories();
+        } catch (IOException e) {
+            logger.error("Error parsing the ImageNet categories: {}", e);
+        }
+        createCategoriesByName();
+    }
+
+    private BigGANCategory(int id, String[] names) {
+        this.id = id;
+        this.names = names;
+    }
+
+    public int getId() {
+        return id;
+    }
+
+    public String[] getNames() {
+        return names.clone();
+    }
+
+    public static BigGANCategory id(int id) {
+        String names = categoriesById[id];
+        int index = names.indexOf(',');
+        if (index < 0) {
+            return of(names);
+        } else {
+            return of(names.substring(0, index));
+        }
+    }
+
+    public static BigGANCategory of(String name) {
+        if (!CATEGORIES_BY_NAME.containsKey(name)) {
+            throw new IllegalArgumentException(name + " is not a valid category.");
+        }
+        return CATEGORIES_BY_NAME.get(name);
+    }
+
+    private static void createCategoriesByName() {
+        for (int i = 0; i < NUMBER_OF_CATEGORIES; i++) {
+            String[] categoryNames = categoriesById[i].split(", ");
+            BigGANCategory category = new BigGANCategory(i, categoryNames);
+
+            for (String name : categoryNames) {
+                CATEGORIES_BY_NAME.put(name, category);
+            }
+        }
+    }
+
+    private static void parseCategories() throws IOException {
+        String filePath = "src/main/resources/categories.txt";
+
+        List<String> fileLines = Files.readAllLines(Paths.get(filePath));
+        List<String> categories = new ArrayList<>(NUMBER_OF_CATEGORIES);
+        for (String line : fileLines) {
+            int nameIndex = line.indexOf(':') + 2;
+            categories.add(line.substring(nameIndex));
+        }
+
+        categoriesById = categories.toArray(new String[] {});
+    }
+}
-Original file line number
+Diff line change
@@ Expand Up / @@ -44,7 +44,7 @@ dependencies { @@
     }
     application {
-        mainClassName = System.getProperty("main", "ai.djl.examples.inference.ObjectDetection")
+        mainClassName = System.getProperty("main", "ai.djl.examples.inference.biggan.Generator")
     }
     run {
@@ Expand Down @@