Fix metric test failures (#414)

* Migrate metric tests from randomUniform to statelessRandomUniform

* pom updates.

* Spotless changes.

Author: Craigacp

pom.xml

@@ -46,7 +46,7 @@
     <maven.javadoc.skip>true</maven.javadoc.skip>
     <maven.source.skip>true</maven.source.skip>
     <gpg.skip>true</gpg.skip>
-    <spotless.version>2.11.1</spotless.version>
+    <spotless.version>2.20.2</spotless.version>
   </properties>
 
   <repositories>
@@ -371,7 +371,9 @@
 
           <lineEndings/>
           <java>
-            <googleJavaFormat/>
+            <googleJavaFormat>
+              <version>1.14.0</version>
+            </googleJavaFormat>
 
             <removeUnusedImports/>
           </java>

tensorflow-core/pom.xml

@@ -43,7 +43,7 @@
     
     Bumped to newer version to patch a CVE only present in protobuf-java
     -->
-    <protobuf.version>3.19.2</protobuf.version>
+    <protobuf.version>3.19.4</protobuf.version>
 
     <native.classifier>${javacpp.platform}${javacpp.platform.extension}</native.classifier>
     <javacpp.build.skip>false</javacpp.build.skip>       <!-- To skip execution of build.sh: -Djavacpp.build.skip=true -->

tensorflow-framework/pom.xml

@@ -93,7 +93,7 @@
         <configuration>
           <forkCount>1</forkCount>
           <reuseForks>false</reuseForks>
-          <argLine>-Xmx2G -XX:MaxPermSize=256m</argLine>
+          <argLine>-Xmx2G</argLine>
           <includes>
             <include>**/*Test.java</include>
           </includes>

tensorflow-framework/src/test/java/org/tensorflow/framework/metrics/PrecisionAtRecallTest.java

@@ -24,7 +24,6 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Ops;
-import org.tensorflow.op.random.RandomUniform;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
 
@@ -39,11 +38,11 @@ public void testValueIsIdempotent() {
       PrecisionAtRecall<TFloat32> instance = new PrecisionAtRecall<>(0.7f, 1001L, TFloat32.class);
 
       Operand<TFloat32> predictions =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
       Operand<TFloat32> labels =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
 
       Op update = instance.updateState(tf, labels, predictions, null);
 

tensorflow-framework/src/test/java/org/tensorflow/framework/metrics/PrecisionTest.java

@@ -22,7 +22,6 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Ops;
-import org.tensorflow.op.random.RandomUniform;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TFloat64;
 import org.tensorflow.types.TInt32;
@@ -39,11 +38,11 @@ public void testValueIsIdempotent() {
       Precision<TFloat64> instance =
           new Precision<>(new float[] {0.3f, 0.72f}, 1001L, TFloat64.class);
       Operand<TFloat32> predictions =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1001L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1001L, 0L}), TFloat32.class);
       Operand<TFloat32> labels =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1001L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1001L, 0L}), TFloat32.class);
 
       Op update = instance.updateState(tf, labels, predictions, null);
 
@@ -81,7 +80,11 @@ public void testUnweightedAllIncorrect() {
       Precision<TFloat32> instance = new Precision<>(0.5f, 1001L, TFloat32.class);
 
       Operand<TInt32> predictions =
-          tf.random.randomUniformInt(tf.constant(Shape.of(100, 1)), tf.constant(0), tf.constant(2));
+          tf.random.statelessMultinomial(
+              tf.constant(new float[][] {{0.5f, 0.5f}}),
+              tf.constant(100),
+              tf.constant(new long[] {1001L, 0L}),
+              TInt32.class);
       Operand<TInt32> labels = tf.math.sub(tf.constant(1), predictions);
       Op update = instance.updateState(tf, labels, predictions, null);
       session.run(update);

tensorflow-framework/src/test/java/org/tensorflow/framework/metrics/RecallAtPrecisionTest.java

@@ -24,7 +24,6 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Ops;
-import org.tensorflow.op.random.RandomUniform;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
 
@@ -39,11 +38,11 @@ public void testValueIsIdempotent() {
       RecallAtPrecision<TFloat32> instance = new RecallAtPrecision<>(0.7f, 1001L, TFloat32.class);
 
       Operand<TFloat32> predictions =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
       Operand<TFloat32> labels =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
       labels = tf.math.mul(labels, tf.constant(2.0f));
 
       Op update = instance.updateState(tf, labels, predictions);

tensorflow-framework/src/test/java/org/tensorflow/framework/metrics/RecallTest.java

@@ -36,9 +36,11 @@ public void testValueIsIdempotent() {
       Recall<TFloat32> instance = new Recall<>(new float[] {0.3f, 0.72f}, 1001L, TFloat32.class);
 
       Operand<TFloat32> predictions =
-          tf.random.randomUniform(tf.constant(Shape.of(10, 3)), TFloat32.class);
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
       Operand<TFloat32> labels =
-          tf.random.randomUniform(tf.constant(Shape.of(10, 3)), TFloat32.class);
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
       Op update = instance.updateState(tf, labels, predictions, null);
 
       for (int i = 0; i < 10; i++) session.run(update);

tensorflow-framework/src/test/java/org/tensorflow/framework/metrics/SensitivityAtSpecificityTest.java

@@ -24,7 +24,6 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Ops;
-import org.tensorflow.op.random.RandomUniform;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TFloat64;
 import org.tensorflow.types.TInt64;
@@ -40,11 +39,11 @@ public void testValueIsIdempotent() {
       SensitivityAtSpecificity<TFloat32> instance =
           new SensitivityAtSpecificity<>(0.7f, 1001L, TFloat32.class);
       Operand<TFloat32> predictions =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
       Operand<TFloat32> labels =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
       labels = tf.math.mul(labels, tf.constant(2.0f));
 
       // instance.setDebug(session.getGraphSession());

tensorflow-framework/src/test/java/org/tensorflow/framework/metrics/SpecificityAtSensitivityTest.java

@@ -24,7 +24,6 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Ops;
-import org.tensorflow.op.random.RandomUniform;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TFloat64;
 import org.tensorflow.types.TInt32;
@@ -42,11 +41,11 @@ public void testValueIsIdempotent() {
           new SpecificityAtSensitivity<>(0.7f, 1001L, TFloat32.class);
 
       Operand<TFloat32> predictions =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
       Operand<TFloat32> labels =
-          tf.random.randomUniform(
-              tf.constant(Shape.of(10, 3)), TFloat32.class, RandomUniform.seed(1L));
+          tf.random.statelessRandomUniform(
+              tf.constant(Shape.of(10, 3)), tf.constant(new long[] {1L, 0L}), TFloat32.class);
 
       // instance.setDebug(session.getGraphSession());
       Op update = instance.updateState(tf, labels, predictions, null);