diff --git a/pom.xml b/pom.xml
index f9b94e8164a..c76ba81cf3f 100644
--- a/pom.xml
+++ b/pom.xml
@@ -1,4 +1,6 @@
-<project xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd" xmlns="http://maven.apache.org/POM/4.0.0"
+<project
+  xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd"
+  xmlns="http://maven.apache.org/POM/4.0.0"
   xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
 
   <modelVersion>4.0.0</modelVersion>
@@ -86,7 +88,9 @@
     </snapshotRepository>
     <repository>
       <id>ossrh</id>
-      <url>https://oss.sonatype.org/service/local/staging/deployByRepositoryId/${stagingRepositoryId}/</url>
+      <url>
+        https://oss.sonatype.org/service/local/staging/deployByRepositoryId/${stagingRepositoryId}/
+      </url>
     </repository>
   </distributionManagement>
 
@@ -156,7 +160,9 @@
         <repository>
           <id>ossrh-staging</id>
           <name>OSSRH Sonatype Staging</name>
-          <url>https://oss.sonatype.org/service/local/staging/deployByRepositoryId/${stagingRepositoryId}/</url>
+          <url>
+            https://oss.sonatype.org/service/local/staging/deployByRepositoryId/${stagingRepositoryId}/
+          </url>
           <releases>
             <enabled>true</enabled>
           </releases>
@@ -230,6 +236,33 @@
       </build>
     </profile>
 
+    <!--
+      Profile to run spotless:apply on builds.  Will run before format's check.
+    -->
+    <profile>
+      <id>apply-format</id>
+      <build>
+        <plugins>
+          <plugin>
+            <groupId>com.diffplug.spotless</groupId>
+            <artifactId>spotless-maven-plugin</artifactId>
+            <version>${spotless.version}</version>
+
+            <executions>
+              <execution>
+                <!-- Runs in initialize phase to fail fast in case of formatting issues (should be before codegen).-->
+                <id>spotless-apply</id>
+                <phase>initialize</phase>
+                <goals>
+                  <goal>apply</goal>
+                </goals>
+              </execution>
+            </executions>
+          </plugin>
+        </plugins>
+      </build>
+    </profile>
+
     <!--
       Profile that enables format checks on compilation
       Can auto-format using spotless:apply.
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AssertCardinalityDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AssertCardinalityDataset.pbtxt
index 7f8ad1c4491..fc93c13b627 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AssertCardinalityDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AssertCardinalityDataset.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "AssertCardinalityDataset"
+  visibility: VISIBLE
   endpoint {
-    name: "data.experimental.AssertCardinalityDataset"
+    name: "data.AssertCardinalityDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AssertNextDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AssertNextDataset.pbtxt
index 5991221a549..d85694ae56e 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AssertNextDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AssertNextDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "AssertNextDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.AssertNextDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AutoShardDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AutoShardDataset.pbtxt
index 7a5039c6744..75488eb84eb 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AutoShardDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_AutoShardDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "AutoShardDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.AutoShardDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_BytesProducedStatsDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_BytesProducedStatsDataset.pbtxt
index cd7f24d9614..b9d81b54105 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_BytesProducedStatsDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_BytesProducedStatsDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "BytesProducedStatsDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.BytesProducedStatsDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CSVDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CSVDataset.pbtxt
index 5b2950c34fb..b4637f34b40 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CSVDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CSVDataset.pbtxt
@@ -1,6 +1,4 @@
 op {
   graph_op_name: "CSVDataset"
-  endpoint {
-    name: "data.CSVDataset"
-  }
+  visibility: SKIP
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CSVDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CSVDatasetV2.pbtxt
index bba2cafbdb7..2c2848e9a74 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CSVDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CSVDatasetV2.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "CSVDatasetV2"
+  visibility: VISIBLE
   endpoint {
-    name: "data.CSVDatasetV2"
+    name: "data.CSVDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CacheDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CacheDataset.pbtxt
index 11c26c1dfc5..4b4b9bb3b53 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CacheDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CacheDataset.pbtxt
@@ -1,6 +1,4 @@
 op {
   graph_op_name: "CacheDataset"
-  endpoint {
-    name: "data.CacheDataset"
-  }
+  visibility: SKIP
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CacheDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CacheDatasetV2.pbtxt
index 2992d9ae109..8c5b58383c3 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CacheDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CacheDatasetV2.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "CacheDatasetV2"
+  visibility: VISIBLE
   endpoint {
-    name: "data.CacheDatasetV2"
+    name: "data.CacheDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ChooseFastestBranchDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ChooseFastestBranchDataset.pbtxt
index ee776c9b105..b3be1a9cd3b 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ChooseFastestBranchDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ChooseFastestBranchDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ChooseFastestBranchDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.ChooseFastestBranchDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ChooseFastestDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ChooseFastestDataset.pbtxt
index 7f6aadeca93..a25508efebc 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ChooseFastestDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ChooseFastestDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ChooseFastestDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.ChooseFastestDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CompressElement.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CompressElement.pbtxt
index f84051aa746..08158de13a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CompressElement.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_CompressElement.pbtxt
@@ -1,6 +1,6 @@
 op {
   graph_op_name: "CompressElement"
   endpoint {
-    name: "data.experimental.CompressElement"
+    name: "data.CompressElement"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DataServiceDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DataServiceDataset.pbtxt
index de2818df181..cfe0fc978fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DataServiceDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DataServiceDataset.pbtxt
@@ -1,6 +1,4 @@
 op {
   graph_op_name: "DataServiceDataset"
-  endpoint {
-    name: "data.experimental.DataServiceDataset"
-  }
+  visibility: SKIP
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DataServiceDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DataServiceDatasetV2.pbtxt
index da39be5c1c1..083caab7391 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DataServiceDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DataServiceDatasetV2.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "DataServiceDatasetV2"
+  visibility: VISIBLE
   endpoint {
-    name: "rawops.DataServiceDatasetV2"
+    name: "data.DataServiceDatasetV2"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetCardinality.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetCardinality.pbtxt
index efb67649b45..6a699fc214e 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetCardinality.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetCardinality.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "DatasetCardinality"
+  visibility: VISIBLE
   endpoint {
     name: "data.DatasetCardinality"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetFromGraph.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetFromGraph.pbtxt
index 5cf72797f85..80ea2a57196 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetFromGraph.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetFromGraph.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "DatasetFromGraph"
+  visibility: VISIBLE
   endpoint {
     name: "data.DatasetFromGraph"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToGraphV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToGraphV2.pbtxt
index 8e5eceabbfa..99be66c5e54 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToGraphV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToGraphV2.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "DatasetToGraphV2"
+  visibility: VISIBLE
   endpoint {
     name: "data.DatasetToGraph"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToSingleElement.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToSingleElement.pbtxt
index 0ac42e0e936..0f0407914d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToSingleElement.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToSingleElement.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "DatasetToSingleElement"
+  visibility: VISIBLE
   endpoint {
     name: "data.DatasetToSingleElement"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToTFRecord.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToTFRecord.pbtxt
index 3d388570630..c4d256969f7 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToTFRecord.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DatasetToTFRecord.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "DatasetToTFRecord"
+  visibility: VISIBLE
   endpoint {
     name: "data.DatasetToTfRecord"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DenseToSparseBatchDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DenseToSparseBatchDataset.pbtxt
index 76f6ba0b8ac..106059e48f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DenseToSparseBatchDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DenseToSparseBatchDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "DenseToSparseBatchDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.DenseToSparseBatchDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DirectedInterleaveDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DirectedInterleaveDataset.pbtxt
index 24ada998b23..60c9729704e 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DirectedInterleaveDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DirectedInterleaveDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "DirectedInterleaveDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.DirectedInterleaveDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DummyIterationCounter.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DummyIterationCounter.pbtxt
index 4f770897717..12b6ffbf5fa 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DummyIterationCounter.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DummyIterationCounter.pbtxt
@@ -1,6 +1,6 @@
 op {
   graph_op_name: "DummyIterationCounter"
   endpoint {
-    name: "data.experimental.DummyIterationCounter"
+    name: "data.DummyIterationCounter"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DummySeedGenerator.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DummySeedGenerator.pbtxt
index 24e723ea75c..4d550787d03 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DummySeedGenerator.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_DummySeedGenerator.pbtxt
@@ -1,6 +1,6 @@
 op {
   graph_op_name: "DummySeedGenerator"
   endpoint {
-    name: "random.experimental.DummySeedGenerator"
+    name: "random.DummySeedGenerator"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_EnqueueInQueueDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_EnqueueInQueueDataset.pbtxt
index 26051ab446f..804c2fc317e 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_EnqueueInQueueDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_EnqueueInQueueDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "EnqueueInQueueDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.EnqueueInQueueDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FilterByLastComponentDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FilterByLastComponentDataset.pbtxt
index b7111f48fa9..4ad74385bce 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FilterByLastComponentDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FilterByLastComponentDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "FilterByLastComponentDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.FilterByLastComponentDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FinalizeDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FinalizeDataset.pbtxt
index ab2a5fa846a..78b37455e28 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FinalizeDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FinalizeDataset.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "FinalizeDataset"
+  visibility: VISIBLE
   endpoint {
-    name: "rawops.FinalizeDataset"
+    name: "data.FinalizeDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FixedLengthRecordDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FixedLengthRecordDatasetV2.pbtxt
index b8012bbe168..b0f66ca1642 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FixedLengthRecordDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_FixedLengthRecordDatasetV2.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "FixedLengthRecordDatasetV2"
+  visibility: VISIBLE
   endpoint {
     name: "data.FixedLengthRecordDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GeneratorDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GeneratorDataset.pbtxt
index b1719005e99..9ff9c28330a 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GeneratorDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GeneratorDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "GeneratorDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.GeneratorDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GroupByReducerDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GroupByReducerDataset.pbtxt
index 1bd2c8f531b..57b781c5a0a 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GroupByReducerDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GroupByReducerDataset.pbtxt
@@ -1,3 +1,7 @@
 op {
   graph_op_name: "GroupByReducerDataset"
+  visibility: VISIBLE
+  endpoint {
+    name: "data.GroupByReducerDataset"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GroupByWindowDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GroupByWindowDataset.pbtxt
index 9e4c4cd4ff2..529ca47d86e 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GroupByWindowDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_GroupByWindowDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "GroupByWindowDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.GroupByWindowDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_IgnoreErrorsDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_IgnoreErrorsDataset.pbtxt
index 86ff4bd8b87..a8812f70e1a 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_IgnoreErrorsDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_IgnoreErrorsDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "IgnoreErrorsDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.IgnoreErrorsDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_InitializeTableFromDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_InitializeTableFromDataset.pbtxt
index dd2c8233b20..bfd98dd2d1c 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_InitializeTableFromDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_InitializeTableFromDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "InitializeTableFromDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.InitializeTableFromDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_KafkaDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_KafkaDataset.pbtxt
index 5f0da216cbb..6055a53c894 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_KafkaDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_KafkaDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "KafkaDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.KafkaDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LMDBDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LMDBDataset.pbtxt
index 9c4d0119d88..68de33fde0b 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LMDBDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LMDBDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "LMDBDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.LMDBDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LatencyStatsDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LatencyStatsDataset.pbtxt
index bf0bf2a5ed7..8447fb7e287 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LatencyStatsDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LatencyStatsDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "LatencyStatsDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.LatencyStatsDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LegacyParallelInterleaveDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LegacyParallelInterleaveDatasetV2.pbtxt
index a07ccaeaa1f..4046256a580 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LegacyParallelInterleaveDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LegacyParallelInterleaveDatasetV2.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "LegacyParallelInterleaveDatasetV2"
+  visibility: VISIBLE
   endpoint {
-    name: "data.experimental.LegacyParallelInterleaveDataset"
+    name: "data.LegacyParallelInterleaveDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LoadDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LoadDataset.pbtxt
index 88dce6062a3..7bbd800c889 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LoadDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_LoadDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "LoadDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.LoadDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MapAndBatchDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MapAndBatchDataset.pbtxt
index cb96bf63d8f..bbf61d0f31e 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MapAndBatchDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MapAndBatchDataset.pbtxt
@@ -1,4 +1,7 @@
 op {
   graph_op_name: "MapAndBatchDataset"
-  visibility: SKIP
+  visibility: VISIBLE
+  endpoint {
+    name: "data.MapAndBatchDataset"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MapAndBatchDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MapAndBatchDatasetV2.pbtxt
deleted file mode 100644
index b29c21888fa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MapAndBatchDatasetV2.pbtxt
+++ /dev/null
@@ -1,6 +0,0 @@
-op {
-  graph_op_name: "MapAndBatchDatasetV2"
-  endpoint {
-    name: "data.MapAndBatchDataset"
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MatchingFilesDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MatchingFilesDataset.pbtxt
index 749257c37b5..2bee85539f8 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MatchingFilesDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MatchingFilesDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "MatchingFilesDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.MatchingFilesDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MaxIntraOpParallelismDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MaxIntraOpParallelismDataset.pbtxt
index 2e90a1cbbd4..20a2e55ba80 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MaxIntraOpParallelismDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_MaxIntraOpParallelismDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "MaxIntraOpParallelismDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.MaxIntraOpParallelismDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ModelDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ModelDataset.pbtxt
index 143c7afd720..8549b4582cc 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ModelDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ModelDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ModelDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.ModelDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_NonSerializableDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_NonSerializableDataset.pbtxt
index a88d14e65c9..4efa2ec9978 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_NonSerializableDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_NonSerializableDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "NonSerializableDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.NonSerializableDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptimizeDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptimizeDataset.pbtxt
index e7ddf97d1ab..4220b306071 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptimizeDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptimizeDataset.pbtxt
@@ -1,6 +1,4 @@
 op {
   graph_op_name: "OptimizeDataset"
-  endpoint {
-    name: "data.OptimizeDataset"
-  }
+  visibility: SKIP
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptimizeDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptimizeDatasetV2.pbtxt
index 7f04d4d27a0..4b0214740a9 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptimizeDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptimizeDatasetV2.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "OptimizeDatasetV2"
+  visibility: VISIBLE
   endpoint {
-    name: "data.OptimizeDatasetV2"
+    name: "data.OptimizeDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptionsDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptionsDataset.pbtxt
index e90dfd7bd04..0ccf9d12546 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptionsDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_OptionsDataset.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "OptionsDataset"
+  visibility: VISIBLE
   endpoint {
-    name: "rawops.OptionsDataset"
+    name: "data.OptionsDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PaddedBatchDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PaddedBatchDatasetV2.pbtxt
index 22dfe84f0ca..0999120d361 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PaddedBatchDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PaddedBatchDatasetV2.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "PaddedBatchDatasetV2"
+  visibility: VISIBLE
   endpoint {
     name: "data.PaddedBatchDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelBatchDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelBatchDataset.pbtxt
index f05138a1bd8..3fd9b1a423b 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelBatchDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelBatchDataset.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "ParallelBatchDataset"
+  visibility: VISIBLE
   endpoint {
-    name: "rawops.ParallelBatchDataset"
+    name: "data.ParallelBatchDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelInterleaveDatasetV4.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelInterleaveDatasetV4.pbtxt
index 56cfd4f9429..7dddfc15a66 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelInterleaveDatasetV4.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelInterleaveDatasetV4.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "ParallelInterleaveDatasetV4"
+  visibility: VISIBLE
   endpoint {
-    name: "data.experimental.ParallelInterleaveDataset"
+    name: "data.ParallelInterleaveDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelMapDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelMapDatasetV2.pbtxt
index 8439c4ebc0e..9135d8f5c99 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelMapDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParallelMapDatasetV2.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ParallelMapDatasetV2"
+  visibility: VISIBLE
   endpoint {
     name: "data.ParallelMapDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParseExampleDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParseExampleDatasetV2.pbtxt
index e0d99d55539..4ce175177f6 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParseExampleDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ParseExampleDatasetV2.pbtxt
@@ -1,6 +1,7 @@
 op {
   graph_op_name: "ParseExampleDatasetV2"
+  visibility: VISIBLE
   endpoint {
-    name: "data.experimental.ParseExampleDataset"
+    name: "data.ParseExampleDataset"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrefetchDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrefetchDataset.pbtxt
index beaad84d153..a8a3423123d 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrefetchDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrefetchDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "PrefetchDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.PrefetchDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrependFromQueueAndPaddedBatchDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrependFromQueueAndPaddedBatchDataset.pbtxt
index 7c9d509b163..a5e8f99de23 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrependFromQueueAndPaddedBatchDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrependFromQueueAndPaddedBatchDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "PrependFromQueueAndPaddedBatchDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.PrependFromQueueAndPaddedBatchDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrivateThreadPoolDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrivateThreadPoolDataset.pbtxt
index 13490a99de5..4c0cfbe6698 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrivateThreadPoolDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_PrivateThreadPoolDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "PrivateThreadPoolDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.PrivateThreadPoolDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RandomDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RandomDataset.pbtxt
index 43921e6eafe..19975d0a927 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RandomDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RandomDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "RandomDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.RandomDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RebatchDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RebatchDataset.pbtxt
index 7fac4c9c8d0..61d010ff033 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RebatchDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RebatchDataset.pbtxt
@@ -1,6 +1,4 @@
 op {
   graph_op_name: "RebatchDataset"
-  endpoint {
-    name: "data.RebatchDataset"
-  }
+  visibility: SKIP
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RebatchDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RebatchDatasetV2.pbtxt
index f287e867d2d..42ab3bc1a6e 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RebatchDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RebatchDatasetV2.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "RebatchDatasetV2"
+  visibility: VISIBLE
   endpoint {
     name: "data.RebatchDatasetV2"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ReduceDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ReduceDataset.pbtxt
index b16c5dbb96c..4493b8a20ac 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ReduceDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ReduceDataset.pbtxt
@@ -1,3 +1,7 @@
 op {
   graph_op_name: "ReduceDataset"
+  visibility: VISIBLE
+  endpoint {
+    name: "data.ReduceDataset"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RegisterDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RegisterDataset.pbtxt
index b628dfdcf10..dca405191e5 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RegisterDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_RegisterDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "RegisterDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.RegisterDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SamplingDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SamplingDataset.pbtxt
index 23a5447e523..6b33a96d9e8 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SamplingDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SamplingDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "SamplingDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.SamplingDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SaveDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SaveDataset.pbtxt
index d670168929d..20b5562a385 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SaveDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SaveDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "SaveDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.SaveDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ScanDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ScanDataset.pbtxt
index 89b63c53f70..838de863044 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ScanDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ScanDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ScanDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.ScanDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SetStatsAggregatorDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SetStatsAggregatorDataset.pbtxt
index f57abe5a667..6b2b18d0b0b 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SetStatsAggregatorDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SetStatsAggregatorDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "SetStatsAggregatorDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.SetStatsAggregatorDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShardDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShardDataset.pbtxt
index 6bd05dae2fb..530d56293ca 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShardDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShardDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ShardDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.ShardDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleAndRepeatDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleAndRepeatDatasetV2.pbtxt
index 2c38ba7607a..b04b2ba824d 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleAndRepeatDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleAndRepeatDatasetV2.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ShuffleAndRepeatDatasetV2"
+  visibility: VISIBLE
   endpoint {
     name: "data.ShuffleAndRepeatDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleDataset.pbtxt
index 40b2a3f53d0..612510fd386 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleDataset.pbtxt
@@ -1,7 +1,4 @@
 op {
   graph_op_name: "ShuffleDataset"
-  visibility: VISIBLE
-  endpoint {
-    name: "data.ShuffleDataset"
-  }
+  visibility: SKIP
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleDatasetV3.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleDatasetV3.pbtxt
index 677f0d2c1e9..a57f540ef0a 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleDatasetV3.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ShuffleDatasetV3.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ShuffleDatasetV3"
+  visibility: VISIBLE
   endpoint {
     name: "data.ShuffleDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SleepDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SleepDataset.pbtxt
index f8abf22d64c..d43d54dca60 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SleepDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SleepDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "SleepDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.SleepDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SlidingWindowDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SlidingWindowDataset.pbtxt
index bada002ca92..332518c4cf1 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SlidingWindowDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SlidingWindowDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "SlidingWindowDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.SlidingWindowDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SnapshotDatasetV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SnapshotDatasetV2.pbtxt
index d6f7188631c..d275c27201c 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SnapshotDatasetV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SnapshotDatasetV2.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "SnapshotDatasetV2"
+  visibility: VISIBLE
   endpoint {
     name: "data.SnapshotDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SparseTensorSliceDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SparseTensorSliceDataset.pbtxt
index bb0d1d7a949..7cb0d05fba2 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SparseTensorSliceDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SparseTensorSliceDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "SparseTensorSliceDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.SparseTensorSliceDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SqlDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SqlDataset.pbtxt
index 8764e81af25..c4723512e3b 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SqlDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_SqlDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "SqlDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.SqlDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_StatsAggregatorHandleV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_StatsAggregatorHandleV2.pbtxt
index 3985b578962..358f535edb9 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_StatsAggregatorHandleV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_StatsAggregatorHandleV2.pbtxt
@@ -1,6 +1,6 @@
 op {
   graph_op_name: "StatsAggregatorHandleV2"
   endpoint {
-    name: "data.experimental.StatsAggregatorHandle"
+    name: "data.StatsAggregatorHandle"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_StatsAggregatorSetSummaryWriter.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_StatsAggregatorSetSummaryWriter.pbtxt
index 71f71d8d877..a0c6b1f833d 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_StatsAggregatorSetSummaryWriter.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_StatsAggregatorSetSummaryWriter.pbtxt
@@ -1,6 +1,6 @@
 op {
   graph_op_name: "StatsAggregatorSetSummaryWriter"
   endpoint {
-    name: "data.experimental.StatsAggregatorSetSummaryWriter"
+    name: "data.StatsAggregatorSetSummaryWriter"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_TakeWhileDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_TakeWhileDataset.pbtxt
index 5571c0a1d00..9d6bf0f5a7b 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_TakeWhileDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_TakeWhileDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "TakeWhileDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.TakeWhileDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_TensorDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_TensorDataset.pbtxt
index ed0ead6e7ab..30b95c2d9c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_TensorDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_TensorDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "TensorDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.TensorDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ThreadPoolDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ThreadPoolDataset.pbtxt
index 1f8f25a9b57..93a1a33ceeb 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ThreadPoolDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_ThreadPoolDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "ThreadPoolDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.ThreadPoolDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UnbatchDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UnbatchDataset.pbtxt
index 24907c804b0..a084a2db4ab 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UnbatchDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UnbatchDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "UnbatchDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.UnbatchDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UncompressElement.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UncompressElement.pbtxt
index 137fb6d75be..46225e53cdd 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UncompressElement.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UncompressElement.pbtxt
@@ -1,6 +1,6 @@
 op {
   graph_op_name: "UncompressElement"
   endpoint {
-    name: "data.experimental.UncompressElement"
+    name: "data.UncompressElement"
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UniqueDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UniqueDataset.pbtxt
index e761f775b51..c6313c6fea8 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UniqueDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UniqueDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "UniqueDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.UniqueDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UnwrapDatasetVariant.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UnwrapDatasetVariant.pbtxt
index 10e80a520dc..7520cc17855 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UnwrapDatasetVariant.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_UnwrapDatasetVariant.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "UnwrapDatasetVariant"
+  visibility: VISIBLE
   endpoint {
     name: "data.UnwrapDatasetVariant"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_WindowDataset.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_WindowDataset.pbtxt
index 69f12c55e1d..285d8387eb8 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_WindowDataset.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_WindowDataset.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "WindowDataset"
+  visibility: VISIBLE
   endpoint {
     name: "data.WindowDataset"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_WrapDatasetVariant.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_WrapDatasetVariant.pbtxt
index 49b2f1c4409..7fe9c7ebad3 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_WrapDatasetVariant.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_WrapDatasetVariant.pbtxt
@@ -1,5 +1,6 @@
 op {
   graph_op_name: "WrapDatasetVariant"
+  visibility: VISIBLE
   endpoint {
     name: "data.WrapDatasetVariant"
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaConv.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaConv.pbtxt
index 3f36b3e7442..0f5f916d30f 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaConv.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaConv.pbtxt
@@ -1,6 +1,4 @@
 op {
   graph_op_name: "XlaConv"
-  endpoint {
-    name: "xla.Conv"
-  }
+  visibility: SKIP
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaConvV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaConvV2.pbtxt
index d2c9637c0ba..bd999901f47 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaConvV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaConvV2.pbtxt
@@ -1,3 +1,6 @@
 op {
   graph_op_name: "XlaConvV2"
+  endpoint {
+    name: "xla.Conv"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaDot.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaDot.pbtxt
index 9489678d9f6..9af5cd4caaa 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaDot.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaDot.pbtxt
@@ -1,6 +1,4 @@
 op {
   graph_op_name: "XlaDot"
-  endpoint {
-    name: "xla.Dot"
-  }
+  visibility: SKIP
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaDotV2.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaDotV2.pbtxt
index 357866b27ac..77e09ad7491 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaDotV2.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaDotV2.pbtxt
@@ -1,3 +1,6 @@
 op {
   graph_op_name: "XlaDotV2"
+  endpoint {
+    name: "xla.Dot"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSetDynamicDimensionSize.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSetDynamicDimensionSize.pbtxt
index aeaeb87d701..724d91eab72 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSetDynamicDimensionSize.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSetDynamicDimensionSize.pbtxt
@@ -1,3 +1,6 @@
 op {
   graph_op_name: "XlaSetDynamicDimensionSize"
+  endpoint {
+    name: "xla.SetDynamicDimensionSize"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSpmdFullToShardShape.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSpmdFullToShardShape.pbtxt
index ea7bf6f9458..560b18cebcb 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSpmdFullToShardShape.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSpmdFullToShardShape.pbtxt
@@ -1,3 +1,6 @@
 op {
   graph_op_name: "XlaSpmdFullToShardShape"
+  endpoint {
+    name: "xla.SpmdFullToShardShape"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSpmdShardToFullShape.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSpmdShardToFullShape.pbtxt
index 0e5842e82d0..e960c39d326 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSpmdShardToFullShape.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaSpmdShardToFullShape.pbtxt
@@ -1,3 +1,6 @@
 op {
   graph_op_name: "XlaSpmdShardToFullShape"
+  endpoint {
+    name: "xla.SpmdShardToFullShape"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaVariadicSort.pbtxt b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaVariadicSort.pbtxt
index 5ae24c7686a..3fa661cd3c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaVariadicSort.pbtxt
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/api_def/api_def_XlaVariadicSort.pbtxt
@@ -1,3 +1,6 @@
 op {
   graph_op_name: "XlaVariadicSort"
+  endpoint {
+    name: "xla.XlaVariadicSort"
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
index 4197dac5fee..5f1765d1f59 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
@@ -18,35 +18,109 @@
 package org.tensorflow.op;
 
 import java.util.List;
+import org.tensorflow.ConcreteFunction;
 import org.tensorflow.Operand;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.data.AnonymousIterator;
+import org.tensorflow.op.data.AssertCardinalityDataset;
+import org.tensorflow.op.data.AssertNextDataset;
+import org.tensorflow.op.data.AutoShardDataset;
 import org.tensorflow.op.data.BatchDataset;
+import org.tensorflow.op.data.BytesProducedStatsDataset;
+import org.tensorflow.op.data.CSVDataset;
+import org.tensorflow.op.data.CacheDataset;
+import org.tensorflow.op.data.ChooseFastestBranchDataset;
+import org.tensorflow.op.data.ChooseFastestDataset;
 import org.tensorflow.op.data.ConcatenateDataset;
+import org.tensorflow.op.data.DataServiceDatasetV2;
+import org.tensorflow.op.data.DatasetCardinality;
+import org.tensorflow.op.data.DatasetFromGraph;
+import org.tensorflow.op.data.DatasetToGraph;
+import org.tensorflow.op.data.DatasetToSingleElement;
+import org.tensorflow.op.data.DatasetToTfRecord;
 import org.tensorflow.op.data.DeleteIterator;
+import org.tensorflow.op.data.DenseToSparseBatchDataset;
 import org.tensorflow.op.data.DeserializeIterator;
+import org.tensorflow.op.data.DirectedInterleaveDataset;
+import org.tensorflow.op.data.FilterByLastComponentDataset;
+import org.tensorflow.op.data.FilterDataset;
+import org.tensorflow.op.data.FinalizeDataset;
+import org.tensorflow.op.data.FixedLengthRecordDataset;
+import org.tensorflow.op.data.FlatMapDataset;
+import org.tensorflow.op.data.GeneratorDataset;
+import org.tensorflow.op.data.GroupByReducerDataset;
+import org.tensorflow.op.data.GroupByWindowDataset;
+import org.tensorflow.op.data.IgnoreErrorsDataset;
+import org.tensorflow.op.data.InitializeTableFromDataset;
+import org.tensorflow.op.data.InterleaveDataset;
 import org.tensorflow.op.data.Iterator;
 import org.tensorflow.op.data.IteratorGetNext;
 import org.tensorflow.op.data.IteratorGetNextAsOptional;
 import org.tensorflow.op.data.IteratorGetNextSync;
 import org.tensorflow.op.data.IteratorToStringHandle;
+import org.tensorflow.op.data.LMDBDataset;
+import org.tensorflow.op.data.LatencyStatsDataset;
+import org.tensorflow.op.data.LegacyParallelInterleaveDataset;
+import org.tensorflow.op.data.LoadDataset;
 import org.tensorflow.op.data.MakeIterator;
+import org.tensorflow.op.data.MapAndBatchDataset;
+import org.tensorflow.op.data.MapDataset;
+import org.tensorflow.op.data.MatchingFilesDataset;
+import org.tensorflow.op.data.MaxIntraOpParallelismDataset;
+import org.tensorflow.op.data.ModelDataset;
+import org.tensorflow.op.data.NonSerializableDataset;
+import org.tensorflow.op.data.OneShotIterator;
+import org.tensorflow.op.data.OptimizeDataset;
 import org.tensorflow.op.data.OptionalFromValue;
 import org.tensorflow.op.data.OptionalGetValue;
 import org.tensorflow.op.data.OptionalHasValue;
 import org.tensorflow.op.data.OptionalNone;
+import org.tensorflow.op.data.OptionsDataset;
+import org.tensorflow.op.data.PaddedBatchDataset;
+import org.tensorflow.op.data.ParallelBatchDataset;
+import org.tensorflow.op.data.ParallelInterleaveDataset;
+import org.tensorflow.op.data.ParallelMapDataset;
+import org.tensorflow.op.data.ParseExampleDataset;
+import org.tensorflow.op.data.PrefetchDataset;
+import org.tensorflow.op.data.PrivateThreadPoolDataset;
+import org.tensorflow.op.data.RandomDataset;
 import org.tensorflow.op.data.RangeDataset;
+import org.tensorflow.op.data.RebatchDatasetV2;
+import org.tensorflow.op.data.ReduceDataset;
+import org.tensorflow.op.data.RegisterDataset;
 import org.tensorflow.op.data.RepeatDataset;
+import org.tensorflow.op.data.SamplingDataset;
+import org.tensorflow.op.data.SaveDataset;
+import org.tensorflow.op.data.ScanDataset;
 import org.tensorflow.op.data.SerializeIterator;
+import org.tensorflow.op.data.SetStatsAggregatorDataset;
+import org.tensorflow.op.data.ShardDataset;
+import org.tensorflow.op.data.ShuffleAndRepeatDataset;
+import org.tensorflow.op.data.ShuffleDataset;
 import org.tensorflow.op.data.SkipDataset;
+import org.tensorflow.op.data.SleepDataset;
+import org.tensorflow.op.data.SlidingWindowDataset;
+import org.tensorflow.op.data.SnapshotDataset;
+import org.tensorflow.op.data.SparseTensorSliceDataset;
+import org.tensorflow.op.data.SqlDataset;
 import org.tensorflow.op.data.TakeDataset;
+import org.tensorflow.op.data.TakeWhileDataset;
+import org.tensorflow.op.data.TensorDataset;
 import org.tensorflow.op.data.TensorSliceDataset;
 import org.tensorflow.op.data.TextLineDataset;
 import org.tensorflow.op.data.TfRecordDataset;
+import org.tensorflow.op.data.ThreadPoolDataset;
+import org.tensorflow.op.data.UnbatchDataset;
+import org.tensorflow.op.data.UniqueDataset;
+import org.tensorflow.op.data.UnwrapDatasetVariant;
+import org.tensorflow.op.data.WindowDataset;
+import org.tensorflow.op.data.WrapDatasetVariant;
 import org.tensorflow.op.data.ZipDataset;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -76,6 +150,68 @@ public AnonymousIterator anonymousIterator(List<Class<? extends TType>> outputTy
     return AnonymousIterator.create(scope, outputTypes, outputShapes);
   }
 
+  /**
+   * The AssertCardinalityDataset operation
+   *
+   * @param inputDataset the inputDataset value
+   * @param cardinality the cardinality value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of AssertCardinalityDataset
+   */
+  public AssertCardinalityDataset assertCardinalityDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> cardinality, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return AssertCardinalityDataset.create(scope, inputDataset, cardinality, outputTypes, outputShapes);
+  }
+
+  /**
+   * A transformation that asserts which transformations happen next.
+   *  This transformation checks whether the camel-case names (i.e. &quot;FlatMap&quot;, not
+   *  &quot;flat_map&quot;) of the transformations following this transformation match the list
+   *  of names in the {@code transformations} argument. If there is a mismatch, the
+   *  transformation raises an exception.
+   *  <p>The check occurs when iterating over the contents of the dataset, which
+   *  means that the check happens <em>after</em> any static optimizations are applied
+   *  to the dataset graph.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   *  {@code data.AssertNextDataset} passes through the outputs of its input dataset.
+   * @param transformations A {@code tf.string} vector {@code tf.Tensor} identifying the transformations that are
+   *  expected to happen next.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of AssertNextDataset
+   */
+  public AssertNextDataset assertNextDataset(Operand<? extends TType> inputDataset,
+      Operand<TString> transformations, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return AssertNextDataset.create(scope, inputDataset, transformations, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that shards the input dataset.
+   *  Creates a dataset that shards the input dataset by num_workers, returning a
+   *  sharded dataset for the index-th worker. This attempts to automatically shard
+   *  a dataset by examining the Dataset graph and inserting a shard op before the
+   *  inputs to a reader Dataset (e.g. CSVDataset, TFRecordDataset).
+   *  <p>This dataset will throw a NotFound error if we cannot shard the dataset
+   *  automatically.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param numWorkers A scalar representing the number of workers to distribute this dataset across.
+   * @param index A scalar representing the index of the current worker out of num_workers.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of AutoShardDataset
+   */
+  public AutoShardDataset autoShardDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> numWorkers, Operand<TInt64> index, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, AutoShardDataset.Options... options) {
+    return AutoShardDataset.create(scope, inputDataset, numWorkers, index, outputTypes, outputShapes, options);
+  }
+
   /**
    * Creates a dataset that batches {@code batch_size} elements from {@code input_dataset}.
    *
@@ -95,223 +231,1398 @@ public BatchDataset batchDataset(Operand<? extends TType> inputDataset, Operand<
   }
 
   /**
-   * Creates a dataset that concatenates {@code input_dataset} with {@code another_dataset}.
+   * Records the bytes size of each element of {@code input_dataset} in a StatsAggregator.
+   *
+   * @param inputDataset the inputDataset value
+   * @param tag the tag value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of BytesProducedStatsDataset
+   */
+  public BytesProducedStatsDataset bytesProducedStatsDataset(Operand<? extends TType> inputDataset,
+      Operand<TString> tag, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return BytesProducedStatsDataset.create(scope, inputDataset, tag, outputTypes, outputShapes);
+  }
+
+  /**
+   * The CSVDatasetV2 operation
+   *
+   * @param filenames the filenames value
+   * @param compressionType the compressionType value
+   * @param bufferSize the bufferSize value
+   * @param header the header value
+   * @param fieldDelim the fieldDelim value
+   * @param useQuoteDelim the useQuoteDelim value
+   * @param naValue the naValue value
+   * @param selectCols the selectCols value
+   * @param recordDefaults the recordDefaults value
+   * @param excludeCols the excludeCols value
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of CSVDataset
+   */
+  public CSVDataset cSVDataset(Operand<TString> filenames, Operand<TString> compressionType,
+      Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim,
+      Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols,
+      Iterable<Operand<?>> recordDefaults, Operand<TInt64> excludeCols, List<Shape> outputShapes) {
+    return CSVDataset.create(scope, filenames, compressionType, bufferSize, header, fieldDelim, useQuoteDelim, naValue, selectCols, recordDefaults, excludeCols, outputShapes);
+  }
+
+  /**
+   * The CacheDatasetV2 operation
+   *
+   * @param inputDataset the inputDataset value
+   * @param filename the filename value
+   * @param cache the cache value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of CacheDataset
+   */
+  public CacheDataset cacheDataset(Operand<? extends TType> inputDataset, Operand<TString> filename,
+      Operand<? extends TType> cache, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return CacheDataset.create(scope, inputDataset, filename, cache, outputTypes, outputShapes);
+  }
+
+  /**
+   * The ChooseFastestBranchDataset operation
+   *
+   * @param inputDataset the inputDataset value
+   * @param ratioNumerator the ratioNumerator value
+   * @param ratioDenominator the ratioDenominator value
+   * @param otherArguments the otherArguments value
+   * @param numElementsPerBranch the value of the numElementsPerBranch property
+   * @param branches the value of the branches property
+   * @param otherArgumentsLengths the value of the otherArgumentsLengths property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of ChooseFastestBranchDataset
+   */
+  public ChooseFastestBranchDataset chooseFastestBranchDataset(
+      Operand<? extends TType> inputDataset, Operand<TInt64> ratioNumerator,
+      Operand<TInt64> ratioDenominator, Iterable<Operand<?>> otherArguments,
+      Long numElementsPerBranch, List<ConcreteFunction> branches, List<Long> otherArgumentsLengths,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return ChooseFastestBranchDataset.create(scope, inputDataset, ratioNumerator, ratioDenominator, otherArguments, numElementsPerBranch, branches, otherArgumentsLengths, outputTypes, outputShapes);
+  }
+
+  /**
+   * The ChooseFastestDataset operation
+   *
+   * @param inputDatasets the inputDatasets value
+   * @param numExperiments the value of the numExperiments property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of ChooseFastestDataset
+   */
+  public ChooseFastestDataset chooseFastestDataset(Iterable<Operand<? extends TType>> inputDatasets,
+      Long numExperiments, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return ChooseFastestDataset.create(scope, inputDatasets, numExperiments, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that concatenates {@code input_dataset} with {@code another_dataset}.
+   *
+   * @param inputDataset the inputDataset value
+   * @param anotherDataset the anotherDataset value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of ConcatenateDataset
+   */
+  public ConcatenateDataset concatenateDataset(Operand<? extends TType> inputDataset,
+      Operand<? extends TType> anotherDataset, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return ConcatenateDataset.create(scope, inputDataset, anotherDataset, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that reads data from the tf.data service.
+   *
+   * @param datasetId the datasetId value
+   * @param processingMode the processingMode value
+   * @param address the address value
+   * @param protocol the protocol value
+   * @param jobName the jobName value
+   * @param consumerIndex the consumerIndex value
+   * @param numConsumers the numConsumers value
+   * @param maxOutstandingRequests the maxOutstandingRequests value
+   * @param iterationCounter the iterationCounter value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of DataServiceDatasetV2
+   */
+  public DataServiceDatasetV2 dataServiceDatasetV2(Operand<TInt64> datasetId,
+      Operand<TString> processingMode, Operand<TString> address, Operand<TString> protocol,
+      Operand<TString> jobName, Operand<TInt64> consumerIndex, Operand<TInt64> numConsumers,
+      Operand<TInt64> maxOutstandingRequests, Operand<? extends TType> iterationCounter,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      DataServiceDatasetV2.Options... options) {
+    return DataServiceDatasetV2.create(scope, datasetId, processingMode, address, protocol, jobName, consumerIndex, numConsumers, maxOutstandingRequests, iterationCounter, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * Returns the cardinality of {@code input_dataset}.
+   *  Returns the cardinality of {@code input_dataset}.
+   *
+   * @param inputDataset A variant tensor representing the dataset to return cardinality for.
+   * @return a new instance of DatasetCardinality
+   */
+  public DatasetCardinality datasetCardinality(Operand<? extends TType> inputDataset) {
+    return DatasetCardinality.create(scope, inputDataset);
+  }
+
+  /**
+   * Creates a dataset from the given {@code graph_def}.
+   *  Creates a dataset from the provided {@code graph_def}.
+   *
+   * @param graphDef The graph representation of the dataset (as serialized GraphDef).
+   * @return a new instance of DatasetFromGraph
+   */
+  public DatasetFromGraph datasetFromGraph(Operand<TString> graphDef) {
+    return DatasetFromGraph.create(scope, graphDef);
+  }
+
+  /**
+   * Returns a serialized GraphDef representing {@code input_dataset}.
+   *  Returns a graph representation for {@code input_dataset}.
+   *
+   * @param inputDataset A variant tensor representing the dataset to return the graph representation for.
+   * @param options carries optional attribute values
+   * @return a new instance of DatasetToGraph
+   */
+  public DatasetToGraph datasetToGraph(Operand<? extends TType> inputDataset,
+      DatasetToGraph.Options... options) {
+    return DatasetToGraph.create(scope, inputDataset, options);
+  }
+
+  /**
+   * Outputs the single element from the given dataset.
+   *
+   * @param dataset A handle to a dataset that contains a single element.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of DatasetToSingleElement
+   */
+  public DatasetToSingleElement datasetToSingleElement(Operand<? extends TType> dataset,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return DatasetToSingleElement.create(scope, dataset, outputTypes, outputShapes);
+  }
+
+  /**
+   * Writes the given dataset to the given file using the TFRecord format.
+   *
+   * @param inputDataset A variant tensor representing the dataset to write.
+   * @param filename A scalar string tensor representing the filename to use.
+   * @param compressionType A scalar string tensor containing either (i) the empty string (no
+   *  compression), (ii) &quot;ZLIB&quot;, or (iii) &quot;GZIP&quot;.
+   * @return a new instance of DatasetToTfRecord
+   */
+  public DatasetToTfRecord datasetToTfRecord(Operand<? extends TType> inputDataset,
+      Operand<TString> filename, Operand<TString> compressionType) {
+    return DatasetToTfRecord.create(scope, inputDataset, filename, compressionType);
+  }
+
+  /**
+   * A container for an iterator resource.
+   *
+   * @param handle A handle to the iterator to delete.
+   * @param deleter A variant deleter.
+   * @return a new instance of DeleteIterator
+   */
+  public DeleteIterator deleteIterator(Operand<? extends TType> handle,
+      Operand<? extends TType> deleter) {
+    return DeleteIterator.create(scope, handle, deleter);
+  }
+
+  /**
+   * Creates a dataset that batches input elements into a SparseTensor.
+   *
+   * @param inputDataset A handle to an input dataset. Must have a single component.
+   * @param batchSize A scalar representing the number of elements to accumulate in a
+   *  batch.
+   * @param rowShape A vector representing the dense shape of each row in the produced
+   *  SparseTensor. The shape may be partially specified, using {@code -1} to indicate
+   *  that a particular dimension should use the maximum size of all batch elements.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of DenseToSparseBatchDataset
+   */
+  public DenseToSparseBatchDataset denseToSparseBatchDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return DenseToSparseBatchDataset.create(scope, inputDataset, batchSize, rowShape, outputTypes, outputShapes);
+  }
+
+  /**
+   * Converts the given variant tensor to an iterator and stores it in the given resource.
+   *
+   * @param resourceHandle A handle to an iterator resource.
+   * @param serialized A variant tensor storing the state of the iterator contained in the
+   *  resource.
+   * @return a new instance of DeserializeIterator
+   */
+  public DeserializeIterator deserializeIterator(Operand<? extends TType> resourceHandle,
+      Operand<? extends TType> serialized) {
+    return DeserializeIterator.create(scope, resourceHandle, serialized);
+  }
+
+  /**
+   * A substitute for {@code InterleaveDataset} on a fixed list of {@code N} datasets.
+   *
+   * @param selectorInputDataset A dataset of scalar {@code DT_INT64} elements that determines which of the
+   *  {@code N} data inputs should produce the next output element.
+   * @param dataInputDatasets {@code N} datasets with the same type that will be interleaved according to
+   *  the values of {@code selector_input_dataset}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of DirectedInterleaveDataset
+   */
+  public DirectedInterleaveDataset directedInterleaveDataset(
+      Operand<? extends TType> selectorInputDataset,
+      Iterable<Operand<? extends TType>> dataInputDatasets,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return DirectedInterleaveDataset.create(scope, selectorInputDataset, dataInputDatasets, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset containing elements of first component of {@code input_dataset} having true in the last component.
+   *
+   * @param inputDataset the inputDataset value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of FilterByLastComponentDataset
+   */
+  public FilterByLastComponentDataset filterByLastComponentDataset(
+      Operand<? extends TType> inputDataset, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return FilterByLastComponentDataset.create(scope, inputDataset, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset containing elements of {@code input_dataset} matching {@code predicate}.
+   *  The {@code predicate} function must return a scalar boolean and accept the
+   *  following arguments:
+   *  <ul>
+   *  <li>One tensor for each component of an element of {@code input_dataset}.</li>
+   *  <li>One tensor for each value in {@code other_arguments}.</li>
+   *  </ul>
+   *
+   * @param inputDataset the inputDataset value
+   * @param otherArguments A list of tensors, typically values that were captured when
+   *  building a closure for {@code predicate}.
+   * @param predicate A function returning a scalar boolean.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of FilterDataset
+   */
+  public FilterDataset filterDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction predicate,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return FilterDataset.create(scope, inputDataset, otherArguments, predicate, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset by applying {@code tf.data.Options} to {@code input_dataset}.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of FinalizeDataset
+   */
+  public FinalizeDataset finalizeDataset(Operand<? extends TType> inputDataset,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      FinalizeDataset.Options... options) {
+    return FinalizeDataset.create(scope, inputDataset, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * The FixedLengthRecordDatasetV2 operation
+   *
+   * @param filenames the filenames value
+   * @param headerBytes the headerBytes value
+   * @param recordBytes the recordBytes value
+   * @param footerBytes the footerBytes value
+   * @param bufferSize the bufferSize value
+   * @param compressionType the compressionType value
+   * @return a new instance of FixedLengthRecordDataset
+   */
+  public FixedLengthRecordDataset fixedLengthRecordDataset(Operand<TString> filenames,
+      Operand<TInt64> headerBytes, Operand<TInt64> recordBytes, Operand<TInt64> footerBytes,
+      Operand<TInt64> bufferSize, Operand<TString> compressionType) {
+    return FixedLengthRecordDataset.create(scope, filenames, headerBytes, recordBytes, footerBytes, bufferSize, compressionType);
+  }
+
+  /**
+   * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+   *  Unlike MapDataset, the {@code f} in FlatMapDataset is expected to return a
+   *  Dataset variant, and FlatMapDataset will flatten successive results
+   *  into a single Dataset.
+   *
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   *  {@code other_arguments}, to a Dataset variant that contains elements matching
+   *  {@code output_types} and {@code output_shapes}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of FlatMapDataset
+   */
+  public FlatMapDataset flatMapDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return FlatMapDataset.create(scope, inputDataset, otherArguments, f, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that invokes a function to generate elements.
+   *
+   * @param initFuncOtherArgs the initFuncOtherArgs value
+   * @param nextFuncOtherArgs the nextFuncOtherArgs value
+   * @param finalizeFuncOtherArgs the finalizeFuncOtherArgs value
+   * @param initFunc the value of the initFunc property
+   * @param nextFunc the value of the nextFunc property
+   * @param finalizeFunc the value of the finalizeFunc property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of GeneratorDataset
+   */
+  public GeneratorDataset generatorDataset(Iterable<Operand<?>> initFuncOtherArgs,
+      Iterable<Operand<?>> nextFuncOtherArgs, Iterable<Operand<?>> finalizeFuncOtherArgs,
+      ConcreteFunction initFunc, ConcreteFunction nextFunc, ConcreteFunction finalizeFunc,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return GeneratorDataset.create(scope, initFuncOtherArgs, nextFuncOtherArgs, finalizeFuncOtherArgs, initFunc, nextFunc, finalizeFunc, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that computes a group-by on {@code input_dataset}.
+   *  Creates a dataset that computes a group-by on {@code input_dataset}.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param keyFuncOtherArguments A list of tensors, typically values that were captured when
+   *  building a closure for {@code key_func}.
+   * @param initFuncOtherArguments A list of tensors, typically values that were captured when
+   *  building a closure for {@code init_func}.
+   * @param reduceFuncOtherArguments A list of tensors, typically values that were captured when
+   *  building a closure for {@code reduce_func}.
+   * @param finalizeFuncOtherArguments A list of tensors, typically values that were captured when
+   *  building a closure for {@code finalize_func}.
+   * @param keyFunc A function mapping an element of {@code input_dataset}, concatenated
+   *  with {@code key_func_other_arguments} to a scalar value of type DT_INT64.
+   * @param initFunc A function mapping a key of type DT_INT64, concatenated with
+   *  {@code init_func_other_arguments} to the initial reducer state.
+   * @param reduceFunc A function mapping the current reducer state and an element of {@code input_dataset},
+   *  concatenated with {@code reduce_func_other_arguments} to a new reducer state.
+   * @param finalizeFunc A function mapping the final reducer state to an output element.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of GroupByReducerDataset
+   */
+  public GroupByReducerDataset groupByReducerDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> keyFuncOtherArguments, Iterable<Operand<?>> initFuncOtherArguments,
+      Iterable<Operand<?>> reduceFuncOtherArguments,
+      Iterable<Operand<?>> finalizeFuncOtherArguments, ConcreteFunction keyFunc,
+      ConcreteFunction initFunc, ConcreteFunction reduceFunc, ConcreteFunction finalizeFunc,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return GroupByReducerDataset.create(scope, inputDataset, keyFuncOtherArguments, initFuncOtherArguments, reduceFuncOtherArguments, finalizeFuncOtherArguments, keyFunc, initFunc, reduceFunc, finalizeFunc, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that computes a windowed group-by on {@code input_dataset}.
+   *  // TODO(mrry): Support non-int64 keys.
+   *
+   * @param inputDataset the inputDataset value
+   * @param keyFuncOtherArguments the keyFuncOtherArguments value
+   * @param reduceFuncOtherArguments the reduceFuncOtherArguments value
+   * @param windowSizeFuncOtherArguments the windowSizeFuncOtherArguments value
+   * @param keyFunc A function mapping an element of {@code input_dataset}, concatenated
+   *  with {@code key_func_other_arguments} to a scalar value of type DT_INT64.
+   * @param reduceFunc the value of the reduceFunc property
+   * @param windowSizeFunc the value of the windowSizeFunc property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of GroupByWindowDataset
+   */
+  public GroupByWindowDataset groupByWindowDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> keyFuncOtherArguments, Iterable<Operand<?>> reduceFuncOtherArguments,
+      Iterable<Operand<?>> windowSizeFuncOtherArguments, ConcreteFunction keyFunc,
+      ConcreteFunction reduceFunc, ConcreteFunction windowSizeFunc,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return GroupByWindowDataset.create(scope, inputDataset, keyFuncOtherArguments, reduceFuncOtherArguments, windowSizeFuncOtherArguments, keyFunc, reduceFunc, windowSizeFunc, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that contains the elements of {@code input_dataset} ignoring errors.
+   *
+   * @param inputDataset the inputDataset value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of IgnoreErrorsDataset
+   */
+  public IgnoreErrorsDataset ignoreErrorsDataset(Operand<? extends TType> inputDataset,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      IgnoreErrorsDataset.Options... options) {
+    return IgnoreErrorsDataset.create(scope, inputDataset, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * The InitializeTableFromDataset operation
+   *
+   * @param tableHandle the tableHandle value
+   * @param dataset the dataset value
+   * @return a new instance of InitializeTableFromDataset
+   */
+  public InitializeTableFromDataset initializeTableFromDataset(Operand<? extends TType> tableHandle,
+      Operand<? extends TType> dataset) {
+    return InitializeTableFromDataset.create(scope, tableHandle, dataset);
+  }
+
+  /**
+   * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+   *  Unlike MapDataset, the {@code f} in InterleaveDataset is expected to return
+   *  a Dataset variant, and InterleaveDataset will flatten successive
+   *  results into a single Dataset. Unlike FlatMapDataset,
+   *  InterleaveDataset will interleave sequences of up to {@code block_length}
+   *  consecutive elements from {@code cycle_length} input elements.
+   *
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param cycleLength the cycleLength value
+   * @param blockLength the blockLength value
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   *  {@code other_arguments}, to a Dataset variant that contains elements matching
+   *  {@code output_types} and {@code output_shapes}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of InterleaveDataset
+   */
+  public InterleaveDataset interleaveDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> cycleLength, Operand<TInt64> blockLength,
+      ConcreteFunction f, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return InterleaveDataset.create(scope, inputDataset, otherArguments, cycleLength, blockLength, f, outputTypes, outputShapes);
+  }
+
+  /**
+   * The IteratorV2 operation
+   *
+   * @param sharedName the value of the sharedName property
+   * @param container the value of the container property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of Iterator
+   */
+  public Iterator iterator(String sharedName, String container,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return Iterator.create(scope, sharedName, container, outputTypes, outputShapes);
+  }
+
+  /**
+   * Gets the next output from the given iterator .
+   *
+   * @param iterator the iterator value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of IteratorGetNext
+   */
+  public IteratorGetNext iteratorGetNext(Operand<? extends TType> iterator,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return IteratorGetNext.create(scope, iterator, outputTypes, outputShapes);
+  }
+
+  /**
+   * Gets the next output from the given iterator as an Optional variant.
+   *
+   * @param iterator the iterator value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of IteratorGetNextAsOptional
+   */
+  public IteratorGetNextAsOptional iteratorGetNextAsOptional(Operand<? extends TType> iterator,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return IteratorGetNextAsOptional.create(scope, iterator, outputTypes, outputShapes);
+  }
+
+  /**
+   * Gets the next output from the given iterator.
+   *  This operation is a synchronous version IteratorGetNext. It should only be used
+   *  in situations where the iterator does not block the calling thread, or where
+   *  the calling thread is not a member of the thread pool used to execute parallel
+   *  operations (e.g. in eager mode).
+   *
+   * @param iterator the iterator value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of IteratorGetNextSync
+   */
+  public IteratorGetNextSync iteratorGetNextSync(Operand<? extends TType> iterator,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return IteratorGetNextSync.create(scope, iterator, outputTypes, outputShapes);
+  }
+
+  /**
+   * Converts the given {@code resource_handle} representing an iterator to a string.
+   *
+   * @param resourceHandle A handle to an iterator resource.
+   * @return a new instance of IteratorToStringHandle
+   */
+  public IteratorToStringHandle iteratorToStringHandle(Operand<? extends TType> resourceHandle) {
+    return IteratorToStringHandle.create(scope, resourceHandle);
+  }
+
+  /**
+   * Creates a dataset that emits the key-value pairs in one or more LMDB files.
+   *  The Lightning Memory-Mapped Database Manager, or LMDB, is an embedded binary
+   *  key-value database. This dataset can read the contents of LMDB database files,
+   *  the names of which generally have the {@code .mdb} suffix.
+   *  <p>Each output element consists of a key-value pair represented as a pair of
+   *  scalar string {@code Tensor}s, where the first {@code Tensor} contains the key and the
+   *  second {@code Tensor} contains the value.
+   *  <p>LMDB uses different file formats on big- and little-endian machines.
+   *  {@code data.LMDBDataset} can only read files in the format of the host machine.
+   *
+   * @param filenames A scalar or a vector containing the name(s) of the binary file(s) to be
+   *  read.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of LMDBDataset
+   */
+  public LMDBDataset lMDBDataset(Operand<TString> filenames,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return LMDBDataset.create(scope, filenames, outputTypes, outputShapes);
+  }
+
+  /**
+   * Records the latency of producing {@code input_dataset} elements in a StatsAggregator.
+   *
+   * @param inputDataset the inputDataset value
+   * @param tag the tag value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of LatencyStatsDataset
+   */
+  public LatencyStatsDataset latencyStatsDataset(Operand<? extends TType> inputDataset,
+      Operand<TString> tag, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return LatencyStatsDataset.create(scope, inputDataset, tag, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+   *  The resulting dataset is similar to the {@code InterleaveDataset}, with the exception
+   *  that if retrieving the next value from a dataset would cause the requester to
+   *  block, it will skip that input dataset. This dataset is especially useful
+   *  when loading data from a variable-latency datastores (e.g. HDFS, GCS), as it
+   *  allows the training step to proceed so long as some data is available.
+   *  <p>!! WARNING !! This dataset is not deterministic!
+   *
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param cycleLength the cycleLength value
+   * @param blockLength the blockLength value
+   * @param bufferOutputElements the bufferOutputElements value
+   * @param prefetchInputElements the prefetchInputElements value
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   *  {@code other_arguments}, to a Dataset variant that contains elements matching
+   *  {@code output_types} and {@code output_shapes}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of LegacyParallelInterleaveDataset
+   */
+  public LegacyParallelInterleaveDataset legacyParallelInterleaveDataset(
+      Operand<? extends TType> inputDataset, Iterable<Operand<?>> otherArguments,
+      Operand<TInt64> cycleLength, Operand<TInt64> blockLength,
+      Operand<TInt64> bufferOutputElements, Operand<TInt64> prefetchInputElements,
+      ConcreteFunction f, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      LegacyParallelInterleaveDataset.Options... options) {
+    return LegacyParallelInterleaveDataset.create(scope, inputDataset, otherArguments, cycleLength, blockLength, bufferOutputElements, prefetchInputElements, f, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * The LoadDataset operation
+   *
+   * @param path the path value
+   * @param readerFuncOtherArgs the readerFuncOtherArgs value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param readerFunc the value of the readerFunc property
+   * @param options carries optional attribute values
+   * @return a new instance of LoadDataset
+   */
+  public LoadDataset loadDataset(Operand<TString> path, Iterable<Operand<?>> readerFuncOtherArgs,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ConcreteFunction readerFunc, LoadDataset.Options... options) {
+    return LoadDataset.create(scope, path, readerFuncOtherArgs, outputTypes, outputShapes, readerFunc, options);
+  }
+
+  /**
+   * Makes a new iterator from the given {@code dataset} and stores it in {@code iterator}.
+   *  This operation may be executed multiple times. Each execution will reset the
+   *  iterator in {@code iterator} to the first element of {@code dataset}.
+   *
+   * @param dataset the dataset value
+   * @param iterator the iterator value
+   * @return a new instance of MakeIterator
+   */
+  public MakeIterator makeIterator(Operand<? extends TType> dataset,
+      Operand<? extends TType> iterator) {
+    return MakeIterator.create(scope, dataset, iterator);
+  }
+
+  /**
+   * Creates a dataset that fuses mapping with batching.
+   *  Creates a dataset that applies {@code f} to the outputs of {@code input_dataset} and then
+   *  batches {@code batch_size} of them.
+   *  <p>Unlike a &quot;MapDataset&quot;, which applies {@code f} sequentially, this dataset invokes up
+   *  to {@code batch_size * num_parallel_batches} copies of {@code f} in parallel.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param otherArguments A list of tensors, typically values that were captured when building a closure
+   *  for {@code f}.
+   * @param batchSize A scalar representing the number of elements to accumulate in a
+   *  batch. It determines the number of concurrent invocations of {@code f} that process
+   *  elements from {@code input_dataset} in parallel.
+   * @param numParallelCalls A scalar representing the maximum number of parallel invocations of the {@code map_fn}
+   *  function. Applying the {@code map_fn} on consecutive input elements in parallel has
+   *  the potential to improve input pipeline throughput.
+   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
+   *  is smaller than desired.
+   * @param f A function to apply to the outputs of {@code input_dataset}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of MapAndBatchDataset
+   */
+  public MapAndBatchDataset mapAndBatchDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> batchSize,
+      Operand<TInt64> numParallelCalls, Operand<TBool> dropRemainder, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      MapAndBatchDataset.Options... options) {
+    return MapAndBatchDataset.create(scope, inputDataset, otherArguments, batchSize, numParallelCalls, dropRemainder, f, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+   *
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param f the value of the f property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of MapDataset
+   */
+  public MapDataset mapDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      MapDataset.Options... options) {
+    return MapDataset.create(scope, inputDataset, otherArguments, f, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * The MatchingFilesDataset operation
+   *
+   * @param patterns the patterns value
+   * @return a new instance of MatchingFilesDataset
+   */
+  public MatchingFilesDataset matchingFilesDataset(Operand<TString> patterns) {
+    return MatchingFilesDataset.create(scope, patterns);
+  }
+
+  /**
+   * Creates a dataset that overrides the maximum intra-op parallelism.
+   *
+   * @param inputDataset the inputDataset value
+   * @param maxIntraOpParallelism Identifies the maximum intra-op parallelism to use.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of MaxIntraOpParallelismDataset
+   */
+  public MaxIntraOpParallelismDataset maxIntraOpParallelismDataset(
+      Operand<? extends TType> inputDataset, Operand<TInt64> maxIntraOpParallelism,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return MaxIntraOpParallelismDataset.create(scope, inputDataset, maxIntraOpParallelism, outputTypes, outputShapes);
+  }
+
+  /**
+   * Identity transformation that models performance.
+   *  Identity transformation that models performance.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ModelDataset
+   */
+  public ModelDataset modelDataset(Operand<? extends TType> inputDataset,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ModelDataset.Options... options) {
+    return ModelDataset.create(scope, inputDataset, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * The NonSerializableDataset operation
+   *
+   * @param inputDataset the inputDataset value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of NonSerializableDataset
+   */
+  public NonSerializableDataset nonSerializableDataset(Operand<? extends TType> inputDataset,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return NonSerializableDataset.create(scope, inputDataset, outputTypes, outputShapes);
+  }
+
+  /**
+   * Makes a &quot;one-shot&quot; iterator that can be iterated only once.
+   *  A one-shot iterator bundles the logic for defining the dataset and
+   *  the state of the iterator in a single op, which allows simple input
+   *  pipelines to be defined without an additional initialization
+   *  (&quot;MakeIterator&quot;) step.
+   *  <p>One-shot iterators have the following limitations:
+   *  <ul>
+   *  <li>They do not support parameterization: all logic for creating the underlying
+   *  dataset must be bundled in the {@code dataset_factory} function.</li>
+   *  <li>They are not resettable. Once a one-shot iterator reaches the end of its
+   *  underlying dataset, subsequent &quot;IteratorGetNext&quot; operations on that
+   *  iterator will always produce an {@code OutOfRange} error.</li>
+   *  </ul>
+   *  <p>For greater flexibility, use &quot;Iterator&quot; and &quot;MakeIterator&quot; to define
+   *  an iterator using an arbitrary subgraph, which may capture tensors
+   *  (including fed values) as parameters, and which may be reset multiple
+   *  times by rerunning &quot;MakeIterator&quot;.
+   *
+   * @param datasetFactory A function of type {@code () -> DT_VARIANT}, where the returned
+   *  DT_VARIANT is a dataset.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of OneShotIterator
+   */
+  public OneShotIterator oneShotIterator(ConcreteFunction datasetFactory,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      OneShotIterator.Options... options) {
+    return OneShotIterator.create(scope, datasetFactory, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * Creates a dataset by applying related optimizations to {@code input_dataset}.
+   *  Creates a dataset by applying related optimizations to {@code input_dataset}.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param optimizationsEnabled A {@code tf.string} vector {@code tf.Tensor} identifying user enabled optimizations.
+   * @param optimizationsDisabled A {@code tf.string} vector {@code tf.Tensor} identifying user disabled optimizations.
+   * @param optimizationsDefault A {@code tf.string} vector {@code tf.Tensor} identifying optimizations by default.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of OptimizeDataset
+   */
+  public OptimizeDataset optimizeDataset(Operand<? extends TType> inputDataset,
+      Operand<TString> optimizationsEnabled, Operand<TString> optimizationsDisabled,
+      Operand<TString> optimizationsDefault, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, OptimizeDataset.Options... options) {
+    return OptimizeDataset.create(scope, inputDataset, optimizationsEnabled, optimizationsDisabled, optimizationsDefault, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * Constructs an Optional variant from a tuple of tensors.
+   *
+   * @param components the components value
+   * @return a new instance of OptionalFromValue
+   */
+  public OptionalFromValue optionalFromValue(Iterable<Operand<?>> components) {
+    return OptionalFromValue.create(scope, components);
+  }
+
+  /**
+   * Returns the value stored in an Optional variant or raises an error if none exists.
+   *
+   * @param optional the optional value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of OptionalGetValue
+   */
+  public OptionalGetValue optionalGetValue(Operand<? extends TType> optional,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return OptionalGetValue.create(scope, optional, outputTypes, outputShapes);
+  }
+
+  /**
+   * Returns true if and only if the given Optional variant has a value.
+   *
+   * @param optional the optional value
+   * @return a new instance of OptionalHasValue
+   */
+  public OptionalHasValue optionalHasValue(Operand<? extends TType> optional) {
+    return OptionalHasValue.create(scope, optional);
+  }
+
+  /**
+   * Creates an Optional variant with no value.
+   *
+   * @return a new instance of OptionalNone
+   */
+  public OptionalNone optionalNone() {
+    return OptionalNone.create(scope);
+  }
+
+  /**
+   * Creates a dataset by attaching tf.data.Options to {@code input_dataset}.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param serializedOptions A {@code tf.string} scalar {@code tf.Tensor} of serialized {@code tf.data.Options} protocol buffer.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of OptionsDataset
+   */
+  public OptionsDataset optionsDataset(Operand<? extends TType> inputDataset,
+      String serializedOptions, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return OptionsDataset.create(scope, inputDataset, serializedOptions, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that batches and pads {@code batch_size} elements from the input.
+   *
+   * @param inputDataset the inputDataset value
+   * @param batchSize A scalar representing the number of elements to accumulate in a
+   *  batch.
+   * @param paddedShapes A list of int64 tensors representing the desired padded shapes
+   *  of the corresponding output components. These shapes may be partially
+   *  specified, using {@code -1} to indicate that a particular dimension should be
+   *  padded to the maximum size of all batch elements.
+   * @param paddingValues A list of scalars containing the padding value to use for
+   *  each of the outputs.
+   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
+   *  is smaller than desired.
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of PaddedBatchDataset
+   */
+  public PaddedBatchDataset paddedBatchDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> batchSize, Iterable<Operand<TInt64>> paddedShapes,
+      Iterable<Operand<?>> paddingValues, Operand<TBool> dropRemainder, List<Shape> outputShapes,
+      PaddedBatchDataset.Options... options) {
+    return PaddedBatchDataset.create(scope, inputDataset, batchSize, paddedShapes, paddingValues, dropRemainder, outputShapes, options);
+  }
+
+  /**
+   * The ParallelBatchDataset operation
+   *
+   * @param inputDataset the inputDataset value
+   * @param batchSize the batchSize value
+   * @param numParallelCalls the numParallelCalls value
+   * @param dropRemainder the dropRemainder value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ParallelBatchDataset
+   */
+  public ParallelBatchDataset parallelBatchDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> batchSize, Operand<TInt64> numParallelCalls, Operand<TBool> dropRemainder,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ParallelBatchDataset.Options... options) {
+    return ParallelBatchDataset.create(scope, inputDataset, batchSize, numParallelCalls, dropRemainder, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+   *  The resulting dataset is similar to the {@code InterleaveDataset}, except that the
+   *  dataset will fetch records from the interleaved datasets in parallel.
+   *  <p>The {@code tf.data} Python API creates instances of this op from
+   *  {@code Dataset.interleave()} when the {@code num_parallel_calls} parameter of that method
+   *  is set to any value other than {@code None}.
+   *  <p>By default, the output of this dataset will be deterministic, which may result
+   *  in the dataset blocking if the next data item to be returned isn't available.
+   *  In order to avoid head-of-line blocking, one can either set the {@code deterministic}
+   *  attribute to &quot;false&quot;, or leave it as &quot;default&quot; and set the
+   *  {@code experimental_deterministic} parameter of {@code tf.data.Options} to {@code False}.
+   *  This can improve performance at the expense of non-determinism.
+   *
+   * @param inputDataset Dataset that produces a stream of arguments for the function {@code f}.
+   * @param otherArguments Additional arguments to pass to {@code f} beyond those produced by {@code input_dataset}.
+   *  Evaluated once when the dataset is instantiated.
+   * @param cycleLength Number of datasets (each created by applying {@code f} to the elements of
+   *  {@code input_dataset}) among which the {@code ParallelInterleaveDatasetV2} will cycle in a
+   *  round-robin fashion.
+   * @param blockLength Number of elements at a time to produce from each interleaved invocation of a
+   *  dataset returned by {@code f}.
+   * @param bufferOutputElements The number of elements each iterator being interleaved should buffer (similar
+   *  to the {@code .prefetch()} transformation for each interleaved iterator).
+   * @param prefetchInputElements Determines the number of iterators to prefetch, allowing buffers to warm up and
+   *  data to be pre-fetched without blocking the main thread.
+   * @param numParallelCalls Determines the number of threads that should be used for fetching data from
+   *  input datasets in parallel. The Python API {@code tf.data.experimental.AUTOTUNE}
+   *  constant can be used to indicate that the level of parallelism should be autotuned.
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   *  {@code other_arguments}, to a Dataset variant that contains elements matching
+   *  {@code output_types} and {@code output_shapes}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ParallelInterleaveDataset
+   */
+  public ParallelInterleaveDataset parallelInterleaveDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> cycleLength, Operand<TInt64> blockLength,
+      Operand<TInt64> bufferOutputElements, Operand<TInt64> prefetchInputElements,
+      Operand<TInt64> numParallelCalls, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ParallelInterleaveDataset.Options... options) {
+    return ParallelInterleaveDataset.create(scope, inputDataset, otherArguments, cycleLength, blockLength, bufferOutputElements, prefetchInputElements, numParallelCalls, f, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+   *  Unlike a &quot;MapDataset&quot;, which applies {@code f} sequentially, this dataset invokes up
+   *  to {@code num_parallel_calls} copies of {@code f} in parallel.
+   *
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param numParallelCalls The number of concurrent invocations of {@code f} that process
+   *  elements from {@code input_dataset} in parallel.
+   * @param f the value of the f property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ParallelMapDataset
+   */
+  public ParallelMapDataset parallelMapDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> numParallelCalls, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ParallelMapDataset.Options... options) {
+    return ParallelMapDataset.create(scope, inputDataset, otherArguments, numParallelCalls, f, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * Transforms {@code input_dataset} containing {@code Example} protos as vectors of DT_STRING into a dataset of {@code Tensor} or {@code SparseTensor} objects representing the parsed features.
+   *
+   * @param inputDataset the inputDataset value
+   * @param numParallelCalls the numParallelCalls value
+   * @param denseDefaults A dict mapping string keys to {@code Tensor}s.
+   *  The keys of the dict must match the dense_keys of the feature.
+   * @param sparseKeys A list of string keys in the examples features.
+   *  The results for these keys will be returned as {@code SparseTensor} objects.
+   * @param denseKeys A list of Ndense string Tensors (scalars).
+   *  The keys expected in the Examples features associated with dense values.
+   * @param sparseTypes A list of {@code DTypes} of the same length as {@code sparse_keys}.
+   *  Only {@code tf.float32} ({@code FloatList}), {@code tf.int64} ({@code Int64List}),
+   *  and {@code tf.string} ({@code BytesList}) are supported.
+   * @param denseShapes List of tuples with the same length as {@code dense_keys}.
+   *  The shape of the data for each dense feature referenced by {@code dense_keys}.
+   *  Required for any input tensors identified by {@code dense_keys}.  Must be
+   *  either fully defined, or may contain an unknown first dimension.
+   *  An unknown first dimension means the feature is treated as having
+   *  a variable number of blocks, and the output shape along this dimension
+   *  is considered unknown at graph build time.  Padding is applied for
+   *  minibatch elements smaller than the maximum number of blocks for the
+   *  given feature along this dimension.
+   * @param outputTypes The type list for the return values.
+   * @param outputShapes The list of shapes being produced.
+   * @param raggedValueTypes the value of the raggedValueTypes property
+   * @param raggedSplitTypes the value of the raggedSplitTypes property
+   * @param options carries optional attribute values
+   * @return a new instance of ParseExampleDataset
+   */
+  public ParseExampleDataset parseExampleDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> numParallelCalls, Iterable<Operand<?>> denseDefaults, List<String> sparseKeys,
+      List<String> denseKeys, List<Class<? extends TType>> sparseTypes, List<Shape> denseShapes,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      List<Class<? extends TType>> raggedValueTypes,
+      List<Class<? extends TNumber>> raggedSplitTypes, ParseExampleDataset.Options... options) {
+    return ParseExampleDataset.create(scope, inputDataset, numParallelCalls, denseDefaults, sparseKeys, denseKeys, sparseTypes, denseShapes, outputTypes, outputShapes, raggedValueTypes, raggedSplitTypes, options);
+  }
+
+  /**
+   * Creates a dataset that asynchronously prefetches elements from {@code input_dataset}.
+   *
+   * @param inputDataset the inputDataset value
+   * @param bufferSize The maximum number of elements to buffer in an iterator over
+   *  this dataset.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of PrefetchDataset
+   */
+  public PrefetchDataset prefetchDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> bufferSize, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, PrefetchDataset.Options... options) {
+    return PrefetchDataset.create(scope, inputDataset, bufferSize, outputTypes, outputShapes, options);
+  }
+
+  /**
+   * Creates a dataset that uses a custom thread pool to compute {@code input_dataset}.
    *
    * @param inputDataset the inputDataset value
-   * @param anotherDataset the anotherDataset value
+   * @param numThreads Identifies the number of threads to use for the private threadpool.
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of ConcatenateDataset
+   * @return a new instance of PrivateThreadPoolDataset
    */
-  public ConcatenateDataset concatenateDataset(Operand<? extends TType> inputDataset,
-      Operand<? extends TType> anotherDataset, List<Class<? extends TType>> outputTypes,
+  public PrivateThreadPoolDataset privateThreadPoolDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> numThreads, List<Class<? extends TType>> outputTypes,
       List<Shape> outputShapes) {
-    return ConcatenateDataset.create(scope, inputDataset, anotherDataset, outputTypes, outputShapes);
+    return PrivateThreadPoolDataset.create(scope, inputDataset, numThreads, outputTypes, outputShapes);
   }
 
   /**
-   * A container for an iterator resource.
+   * Creates a Dataset that returns pseudorandom numbers.
+   *  Creates a Dataset that returns a stream of uniformly distributed
+   *  pseudorandom 64-bit signed integers.
+   *  <p>In the TensorFlow Python API, you can instantiate this dataset via the
+   *  class {@code tf.data.experimental.RandomDataset}.
+   *  <p>Instances of this dataset are also created as a result of the
+   *  {@code hoist_random_uniform} static optimization. Whether this optimization is
+   *  performed is determined by the {@code experimental_optimization.hoist_random_uniform}
+   *  option of {@code tf.data.Options}.
    *
-   * @param handle A handle to the iterator to delete.
-   * @param deleter A variant deleter.
-   * @return a new instance of DeleteIterator
+   * @param seed A scalar seed for the random number generator. If either seed or
+   *  seed2 is set to be non-zero, the random number generator is seeded
+   *  by the given seed.  Otherwise, a random seed is used.
+   * @param seed2 A second scalar seed to avoid seed collision.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of RandomDataset
    */
-  public DeleteIterator deleteIterator(Operand<? extends TType> handle,
-      Operand<? extends TType> deleter) {
-    return DeleteIterator.create(scope, handle, deleter);
+  public RandomDataset randomDataset(Operand<TInt64> seed, Operand<TInt64> seed2,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return RandomDataset.create(scope, seed, seed2, outputTypes, outputShapes);
   }
 
   /**
-   * Converts the given variant tensor to an iterator and stores it in the given resource.
+   * Creates a dataset with a range of values. Corresponds to python's xrange.
    *
-   * @param resourceHandle A handle to an iterator resource.
-   * @param serialized A variant tensor storing the state of the iterator contained in the
-   *  resource.
-   * @return a new instance of DeserializeIterator
+   * @param start corresponds to start in python's xrange().
+   * @param stop corresponds to stop in python's xrange().
+   * @param step corresponds to step in python's xrange().
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of RangeDataset
    */
-  public DeserializeIterator deserializeIterator(Operand<? extends TType> resourceHandle,
-      Operand<? extends TType> serialized) {
-    return DeserializeIterator.create(scope, resourceHandle, serialized);
+  public RangeDataset rangeDataset(Operand<TInt64> start, Operand<TInt64> stop,
+      Operand<TInt64> step, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return RangeDataset.create(scope, start, stop, step, outputTypes, outputShapes);
   }
 
   /**
-   * The IteratorV2 operation
+   * Creates a dataset that changes the batch size.
+   *  Creates a dataset that rebatches elements from {@code input_dataset} into new batch
+   *  sizes.
    *
-   * @param sharedName the value of the sharedName property
-   * @param container the value of the container property
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param batchSizes A vector of integers representing the size of batches to produce. These values
+   *  are cycled through in order.
+   * @param dropRemainder the dropRemainder value
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of Iterator
+   * @return a new instance of RebatchDatasetV2
    */
-  public Iterator iterator(String sharedName, String container,
+  public RebatchDatasetV2 rebatchDatasetV2(Operand<? extends TType> inputDataset,
+      Operand<TInt64> batchSizes, Operand<TBool> dropRemainder,
       List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    return Iterator.create(scope, sharedName, container, outputTypes, outputShapes);
+    return RebatchDatasetV2.create(scope, inputDataset, batchSizes, dropRemainder, outputTypes, outputShapes);
   }
 
   /**
-   * Gets the next output from the given iterator .
+   * Reduces the input dataset to a singleton using a reduce function.
    *
-   * @param iterator the iterator value
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param initialState A nested structure of tensors, representing the initial state of the
+   *  transformation.
+   * @param otherArguments the otherArguments value
+   * @param f A function that maps {@code (old_state, input_element)} to {@code new_state}. It must take
+   *  two arguments and return a nested structures of tensors. The structure of
+   *  {@code new_state} must match the structure of {@code initial_state}.
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of IteratorGetNext
+   * @param options carries optional attribute values
+   * @return a new instance of ReduceDataset
    */
-  public IteratorGetNext iteratorGetNext(Operand<? extends TType> iterator,
-      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    return IteratorGetNext.create(scope, iterator, outputTypes, outputShapes);
+  public ReduceDataset reduceDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> initialState, Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ReduceDataset.Options... options) {
+    return ReduceDataset.create(scope, inputDataset, initialState, otherArguments, f, outputTypes, outputShapes, options);
   }
 
   /**
-   * Gets the next output from the given iterator as an Optional variant.
+   * Registers a dataset with the tf.data service.
    *
-   * @param iterator the iterator value
+   * @param dataset the dataset value
+   * @param address the address value
+   * @param protocol the protocol value
+   * @param externalStatePolicy the value of the externalStatePolicy property
+   * @return a new instance of RegisterDataset
+   */
+  public RegisterDataset registerDataset(Operand<? extends TType> dataset, Operand<TString> address,
+      Operand<TString> protocol, Long externalStatePolicy) {
+    return RegisterDataset.create(scope, dataset, address, protocol, externalStatePolicy);
+  }
+
+  /**
+   * Creates a dataset that emits the outputs of {@code input_dataset} {@code count} times.
+   *
+   * @param inputDataset the inputDataset value
+   * @param count A scalar representing the number of times that {@code input_dataset} should
+   *  be repeated. A value of {@code -1} indicates that it should be repeated infinitely.
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of IteratorGetNextAsOptional
+   * @return a new instance of RepeatDataset
    */
-  public IteratorGetNextAsOptional iteratorGetNextAsOptional(Operand<? extends TType> iterator,
+  public RepeatDataset repeatDataset(Operand<? extends TType> inputDataset, Operand<TInt64> count,
       List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    return IteratorGetNextAsOptional.create(scope, iterator, outputTypes, outputShapes);
+    return RepeatDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
   }
 
   /**
-   * Gets the next output from the given iterator.
-   *  This operation is a synchronous version IteratorGetNext. It should only be used
-   *  in situations where the iterator does not block the calling thread, or where
-   *  the calling thread is not a member of the thread pool used to execute parallel
-   *  operations (e.g. in eager mode).
+   * Creates a dataset that takes a Bernoulli sample of the contents of another dataset.
+   *  There is no transformation in the {@code tf.data} Python API for creating this dataset.
+   *  Instead, it is created as a result of the {@code filter_with_random_uniform_fusion}
+   *  static optimization. Whether this optimization is performed is determined by the
+   *  {@code experimental_optimization.filter_with_random_uniform_fusion} option of
+   *  {@code tf.data.Options}.
    *
-   * @param iterator the iterator value
+   * @param inputDataset the inputDataset value
+   * @param rate A scalar representing the sample rate. Each element of {@code input_dataset} is
+   *  retained with this probability, independent of all other elements.
+   * @param seed A scalar representing seed of random number generator.
+   * @param seed2 A scalar representing seed2 of random number generator.
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of IteratorGetNextSync
+   * @return a new instance of SamplingDataset
    */
-  public IteratorGetNextSync iteratorGetNextSync(Operand<? extends TType> iterator,
+  public SamplingDataset samplingDataset(Operand<? extends TType> inputDataset,
+      Operand<TFloat32> rate, Operand<TInt64> seed, Operand<TInt64> seed2,
       List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    return IteratorGetNextSync.create(scope, iterator, outputTypes, outputShapes);
+    return SamplingDataset.create(scope, inputDataset, rate, seed, seed2, outputTypes, outputShapes);
   }
 
   /**
-   * Converts the given {@code resource_handle} representing an iterator to a string.
+   * The SaveDataset operation
    *
-   * @param resourceHandle A handle to an iterator resource.
-   * @return a new instance of IteratorToStringHandle
+   * @param inputDataset the inputDataset value
+   * @param path the path value
+   * @param shardFuncOtherArgs the shardFuncOtherArgs value
+   * @param shardFunc the value of the shardFunc property
+   * @param options carries optional attribute values
+   * @return a new instance of SaveDataset
    */
-  public IteratorToStringHandle iteratorToStringHandle(Operand<? extends TType> resourceHandle) {
-    return IteratorToStringHandle.create(scope, resourceHandle);
+  public SaveDataset saveDataset(Operand<? extends TType> inputDataset, Operand<TString> path,
+      Iterable<Operand<?>> shardFuncOtherArgs, ConcreteFunction shardFunc,
+      SaveDataset.Options... options) {
+    return SaveDataset.create(scope, inputDataset, path, shardFuncOtherArgs, shardFunc, options);
   }
 
   /**
-   * Makes a new iterator from the given {@code dataset} and stores it in {@code iterator}.
-   *  This operation may be executed multiple times. Each execution will reset the
-   *  iterator in {@code iterator} to the first element of {@code dataset}.
+   * Creates a dataset successively reduces {@code f} over the elements of {@code input_dataset}.
    *
-   * @param dataset the dataset value
-   * @param iterator the iterator value
-   * @return a new instance of MakeIterator
+   * @param inputDataset the inputDataset value
+   * @param initialState the initialState value
+   * @param otherArguments the otherArguments value
+   * @param f the value of the f property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ScanDataset
    */
-  public MakeIterator makeIterator(Operand<? extends TType> dataset,
-      Operand<? extends TType> iterator) {
-    return MakeIterator.create(scope, dataset, iterator);
+  public ScanDataset scanDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> initialState, Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ScanDataset.Options... options) {
+    return ScanDataset.create(scope, inputDataset, initialState, otherArguments, f, outputTypes, outputShapes, options);
   }
 
   /**
-   * Constructs an Optional variant from a tuple of tensors.
+   * Converts the given {@code resource_handle} representing an iterator to a variant tensor.
    *
-   * @param components the components value
-   * @return a new instance of OptionalFromValue
+   * @param resourceHandle A handle to an iterator resource.
+   * @param options carries optional attribute values
+   * @return a new instance of SerializeIterator
    */
-  public OptionalFromValue optionalFromValue(Iterable<Operand<?>> components) {
-    return OptionalFromValue.create(scope, components);
+  public SerializeIterator serializeIterator(Operand<? extends TType> resourceHandle,
+      SerializeIterator.Options... options) {
+    return SerializeIterator.create(scope, resourceHandle, options);
   }
 
   /**
-   * Returns the value stored in an Optional variant or raises an error if none exists.
+   * The SetStatsAggregatorDataset operation
    *
-   * @param optional the optional value
+   * @param inputDataset the inputDataset value
+   * @param statsAggregator the statsAggregator value
+   * @param tag the tag value
+   * @param counterPrefix the counterPrefix value
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of OptionalGetValue
+   * @return a new instance of SetStatsAggregatorDataset
    */
-  public OptionalGetValue optionalGetValue(Operand<? extends TType> optional,
-      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    return OptionalGetValue.create(scope, optional, outputTypes, outputShapes);
+  public SetStatsAggregatorDataset setStatsAggregatorDataset(Operand<? extends TType> inputDataset,
+      Operand<? extends TType> statsAggregator, Operand<TString> tag,
+      Operand<TString> counterPrefix, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return SetStatsAggregatorDataset.create(scope, inputDataset, statsAggregator, tag, counterPrefix, outputTypes, outputShapes);
   }
 
   /**
-   * Returns true if and only if the given Optional variant has a value.
+   * Creates a {@code Dataset} that includes only 1/{@code num_shards} of this dataset.
    *
-   * @param optional the optional value
-   * @return a new instance of OptionalHasValue
+   * @param inputDataset the inputDataset value
+   * @param numShards An integer representing the number of shards operating in parallel.
+   * @param index An integer representing the current worker index.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ShardDataset
    */
-  public OptionalHasValue optionalHasValue(Operand<? extends TType> optional) {
-    return OptionalHasValue.create(scope, optional);
+  public ShardDataset shardDataset(Operand<? extends TType> inputDataset, Operand<TInt64> numShards,
+      Operand<TInt64> index, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ShardDataset.Options... options) {
+    return ShardDataset.create(scope, inputDataset, numShards, index, outputTypes, outputShapes, options);
   }
 
   /**
-   * Creates an Optional variant with no value.
+   * The ShuffleAndRepeatDatasetV2 operation
    *
-   * @return a new instance of OptionalNone
+   * @param inputDataset the inputDataset value
+   * @param bufferSize the bufferSize value
+   * @param seed the seed value
+   * @param seed2 the seed2 value
+   * @param count the count value
+   * @param seedGenerator the seedGenerator value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ShuffleAndRepeatDataset
    */
-  public OptionalNone optionalNone() {
-    return OptionalNone.create(scope);
+  public ShuffleAndRepeatDataset shuffleAndRepeatDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2,
+      Operand<TInt64> count, Operand<? extends TType> seedGenerator,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      ShuffleAndRepeatDataset.Options... options) {
+    return ShuffleAndRepeatDataset.create(scope, inputDataset, bufferSize, seed, seed2, count, seedGenerator, outputTypes, outputShapes, options);
   }
 
   /**
-   * Creates a dataset with a range of values. Corresponds to python's xrange.
+   * The ShuffleDatasetV3 operation
    *
-   * @param start corresponds to start in python's xrange().
-   * @param stop corresponds to stop in python's xrange().
-   * @param step corresponds to step in python's xrange().
+   * @param inputDataset the inputDataset value
+   * @param bufferSize the bufferSize value
+   * @param seed the seed value
+   * @param seed2 the seed2 value
+   * @param seedGenerator the seedGenerator value
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of RangeDataset
+   * @param options carries optional attribute values
+   * @return a new instance of ShuffleDataset
    */
-  public RangeDataset rangeDataset(Operand<TInt64> start, Operand<TInt64> stop,
-      Operand<TInt64> step, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    return RangeDataset.create(scope, start, stop, step, outputTypes, outputShapes);
+  public ShuffleDataset shuffleDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2,
+      Operand<? extends TType> seedGenerator, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, ShuffleDataset.Options... options) {
+    return ShuffleDataset.create(scope, inputDataset, bufferSize, seed, seed2, seedGenerator, outputTypes, outputShapes, options);
   }
 
   /**
-   * Creates a dataset that emits the outputs of {@code input_dataset} {@code count} times.
+   * Creates a dataset that skips {@code count} elements from the {@code input_dataset}.
    *
    * @param inputDataset the inputDataset value
-   * @param count A scalar representing the number of times that {@code input_dataset} should
-   *  be repeated. A value of {@code -1} indicates that it should be repeated infinitely.
+   * @param count A scalar representing the number of elements from the {@code input_dataset}
+   *  that should be skipped.  If count is -1, skips everything.
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of RepeatDataset
+   * @return a new instance of SkipDataset
    */
-  public RepeatDataset repeatDataset(Operand<? extends TType> inputDataset, Operand<TInt64> count,
+  public SkipDataset skipDataset(Operand<? extends TType> inputDataset, Operand<TInt64> count,
       List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    return RepeatDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
+    return SkipDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
   }
 
   /**
-   * Converts the given {@code resource_handle} representing an iterator to a variant tensor.
+   * The SleepDataset operation
    *
-   * @param resourceHandle A handle to an iterator resource.
-   * @param options carries optional attribute values
-   * @return a new instance of SerializeIterator
+   * @param inputDataset the inputDataset value
+   * @param sleepMicroseconds the sleepMicroseconds value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of SleepDataset
    */
-  public SerializeIterator serializeIterator(Operand<? extends TType> resourceHandle,
-      SerializeIterator.Options... options) {
-    return SerializeIterator.create(scope, resourceHandle, options);
+  public SleepDataset sleepDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> sleepMicroseconds, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return SleepDataset.create(scope, inputDataset, sleepMicroseconds, outputTypes, outputShapes);
   }
 
   /**
-   * Creates a dataset that skips {@code count} elements from the {@code input_dataset}.
+   * Creates a dataset that passes a sliding window over {@code input_dataset}.
    *
    * @param inputDataset the inputDataset value
-   * @param count A scalar representing the number of elements from the {@code input_dataset}
-   *  that should be skipped.  If count is -1, skips everything.
+   * @param windowSize A scalar representing the number of elements in the
+   *  sliding window.
+   * @param windowShift A scalar representing the steps moving the sliding window
+   *  forward in one iteration. It must be positive.
+   * @param windowStride A scalar representing the stride of the input elements of the sliding window.
+   *  It must be positive.
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of SkipDataset
+   * @return a new instance of SlidingWindowDataset
    */
-  public SkipDataset skipDataset(Operand<? extends TType> inputDataset, Operand<TInt64> count,
+  public SlidingWindowDataset slidingWindowDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride,
       List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    return SkipDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
+    return SlidingWindowDataset.create(scope, inputDataset, windowSize, windowShift, windowStride, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that will write to / read from a snapshot.
+   *  This dataset attempts to determine whether a valid snapshot exists at the
+   *  {@code snapshot_path}, and reads from the snapshot in lieu of using {@code input_dataset}.
+   *  If not, it will run the preprocessing pipeline as usual, and write out a
+   *  snapshot of the data processed for future use.
+   *
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param path The path we should write snapshots to / read snapshots from.
+   * @param readerFuncOtherArgs the readerFuncOtherArgs value
+   * @param shardFuncOtherArgs the shardFuncOtherArgs value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param readerFunc Optional. A function to control how to read data from snapshot shards.
+   * @param shardFunc Optional. A function to control how to shard data when writing a snapshot.
+   * @param options carries optional attribute values
+   * @return a new instance of SnapshotDataset
+   */
+  public SnapshotDataset snapshotDataset(Operand<? extends TType> inputDataset,
+      Operand<TString> path, Iterable<Operand<?>> readerFuncOtherArgs,
+      Iterable<Operand<?>> shardFuncOtherArgs, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, ConcreteFunction readerFunc, ConcreteFunction shardFunc,
+      SnapshotDataset.Options... options) {
+    return SnapshotDataset.create(scope, inputDataset, path, readerFuncOtherArgs, shardFuncOtherArgs, outputTypes, outputShapes, readerFunc, shardFunc, options);
+  }
+
+  /**
+   * Creates a dataset that splits a SparseTensor into elements row-wise.
+   *
+   * @param indices the indices value
+   * @param values the values value
+   * @param denseShape the denseShape value
+   * @return a new instance of SparseTensorSliceDataset
+   */
+  public SparseTensorSliceDataset sparseTensorSliceDataset(Operand<TInt64> indices,
+      Operand<? extends TType> values, Operand<TInt64> denseShape) {
+    return SparseTensorSliceDataset.create(scope, indices, values, denseShape);
+  }
+
+  /**
+   * Creates a dataset that executes a SQL query and emits rows of the result set.
+   *
+   * @param driverName The database type. Currently, the only supported type is 'sqlite'.
+   * @param dataSourceName A connection string to connect to the database.
+   * @param query A SQL query to execute.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of SqlDataset
+   */
+  public SqlDataset sqlDataset(Operand<TString> driverName, Operand<TString> dataSourceName,
+      Operand<TString> query, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return SqlDataset.create(scope, driverName, dataSourceName, query, outputTypes, outputShapes);
   }
 
   /**
@@ -330,6 +1641,40 @@ public TakeDataset takeDataset(Operand<? extends TType> inputDataset, Operand<TI
     return TakeDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
   }
 
+  /**
+   * Creates a dataset that stops iteration when predicate` is false.
+   *  The {@code predicate} function must return a scalar boolean and accept the
+   *  following arguments:
+   *  <ul>
+   *  <li>One tensor for each component of an element of {@code input_dataset}.</li>
+   *  <li>One tensor for each value in {@code other_arguments}.</li>
+   *  </ul>
+   *
+   * @param inputDataset the inputDataset value
+   * @param otherArguments A list of tensors, typically values that were captured when
+   *  building a closure for {@code predicate}.
+   * @param predicate A function returning a scalar boolean.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of TakeWhileDataset
+   */
+  public TakeWhileDataset takeWhileDataset(Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction predicate,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return TakeWhileDataset.create(scope, inputDataset, otherArguments, predicate, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that emits {@code components} as a tuple of tensors once.
+   *
+   * @param components the components value
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of TensorDataset
+   */
+  public TensorDataset tensorDataset(Iterable<Operand<?>> components, List<Shape> outputShapes) {
+    return TensorDataset.create(scope, components, outputShapes);
+  }
+
   /**
    * Creates a dataset that emits each dim-0 slice of {@code components} once.
    *
@@ -373,6 +1718,127 @@ public TfRecordDataset tfRecordDataset(Operand<TString> filenames,
     return TfRecordDataset.create(scope, filenames, compressionType, bufferSize);
   }
 
+  /**
+   * Creates a dataset that uses a custom thread pool to compute {@code input_dataset}.
+   *
+   * @param inputDataset the inputDataset value
+   * @param threadPool A resource produced by the ThreadPoolHandle op.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of ThreadPoolDataset
+   */
+  public ThreadPoolDataset threadPoolDataset(Operand<? extends TType> inputDataset,
+      Operand<? extends TType> threadPool, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return ThreadPoolDataset.create(scope, inputDataset, threadPool, outputTypes, outputShapes);
+  }
+
+  /**
+   * A dataset that splits the elements of its input into multiple elements.
+   *
+   * @param inputDataset the inputDataset value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of UnbatchDataset
+   */
+  public UnbatchDataset unbatchDataset(Operand<? extends TType> inputDataset,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return UnbatchDataset.create(scope, inputDataset, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that contains the unique elements of {@code input_dataset}.
+   *
+   * @param inputDataset the inputDataset value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of UniqueDataset
+   */
+  public UniqueDataset uniqueDataset(Operand<? extends TType> inputDataset,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return UniqueDataset.create(scope, inputDataset, outputTypes, outputShapes);
+  }
+
+  /**
+   * The UnwrapDatasetVariant operation
+   *
+   * @param inputHandle the inputHandle value
+   * @return a new instance of UnwrapDatasetVariant
+   */
+  public UnwrapDatasetVariant unwrapDatasetVariant(Operand<? extends TType> inputHandle) {
+    return UnwrapDatasetVariant.create(scope, inputHandle);
+  }
+
+  /**
+   * Combines (nests of) input elements into a dataset of (nests of) windows.
+   *  <p>A &quot;window&quot; is a finite dataset of flat elements of size {@code size} (or possibly
+   *  fewer if there are not enough input elements to fill the window and
+   *  {@code drop_remainder} evaluates to false).
+   *  <p>The {@code shift} argument determines the number of input elements by which
+   *  the window moves on each iteration.  The first element in the {@code k}th window
+   *  will be element
+   *  <pre>
+   *  1 + (k-1) * shift
+   *  </pre>
+   *  <p>of the input dataset. In particular, the first element of the first window
+   *  will always be the first element of the input dataset.
+   *  <p>If the {@code stride} parameter is greater than 1, then each window will skip
+   *  {@code (stride - 1)} input elements between each element that appears in the
+   *  window. Output windows will still contain {@code size} elements regardless of
+   *  the value of {@code stride}.
+   *  <p>The {@code stride} argument determines the stride of the input elements, and the
+   *  {@code shift} argument determines the shift of the window.
+   *  <p>For example, letting {@code {...}} to represent a Dataset:
+   *  <ul>
+   *  <li>{@code tf.data.Dataset.range(7).window(2)} produces
+   *  {@code {{0, 1}, {2, 3}, {4, 5}, {6}}}</li>
+   *  <li>{@code tf.data.Dataset.range(7).window(3, 2, 1, True)} produces
+   *  {@code {{0, 1, 2}, {2, 3, 4}, {4, 5, 6}}}</li>
+   *  <li>{@code tf.data.Dataset.range(7).window(3, 1, 2, True)} produces
+   *  {@code {{0, 2, 4}, {1, 3, 5}, {2, 4, 6}}}</li>
+   *  </ul>
+   *  <p>Note that when the {@code window} transformation is applied to a dataset of
+   *  nested elements, it produces a dataset of nested windows.
+   *  <p>For example:
+   *  <ul>
+   *  <li>{@code tf.data.Dataset.from_tensor_slices((range(4), range(4))).window(2)}
+   *  produces {@code {({0, 1}, {0, 1}), ({2, 3}, {2, 3})}}</li>
+   *  <li>{@code tf.data.Dataset.from_tensor_slices({"a": range(4)}).window(2)}
+   *  produces {@code {{"a": {0, 1}}, {"a": {2, 3}}}}</li>
+   *  </ul>
+   *
+   * @param inputDataset the inputDataset value
+   * @param sizeOutput An integer scalar, representing the number of elements
+   *  of the input dataset to combine into a window. Must be positive.
+   * @param shift An integer scalar, representing the number of input elements
+   *  by which the window moves in each iteration.  Defaults to {@code size}.
+   *  Must be positive.
+   * @param stride An integer scalar, representing the stride of the input elements
+   *  in the sliding window. Must be positive. The default value of 1 means
+   *  &quot;retain every input element&quot;.
+   * @param dropRemainder A Boolean scalar, representing whether the last window should be
+   *  dropped if its size is smaller than {@code window_size}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of WindowDataset
+   */
+  public WindowDataset windowDataset(Operand<? extends TType> inputDataset,
+      Operand<TInt64> sizeOutput, Operand<TInt64> shift, Operand<TInt64> stride,
+      Operand<TBool> dropRemainder, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return WindowDataset.create(scope, inputDataset, sizeOutput, shift, stride, dropRemainder, outputTypes, outputShapes);
+  }
+
+  /**
+   * The WrapDatasetVariant operation
+   *
+   * @param inputHandle the inputHandle value
+   * @return a new instance of WrapDatasetVariant
+   */
+  public WrapDatasetVariant wrapDatasetVariant(Operand<? extends TType> inputHandle) {
+    return WrapDatasetVariant.create(scope, inputHandle);
+  }
+
   /**
    * Creates a dataset that zips together {@code input_datasets}.
    *  The elements of the resulting dataset are created by zipping corresponding
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
index a4a7f5d6dbc..7abb451be2d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
@@ -58,6 +58,7 @@
 import org.tensorflow.op.core.BarrierReadySize;
 import org.tensorflow.op.core.BarrierTakeMany;
 import org.tensorflow.op.core.Batch;
+import org.tensorflow.op.core.BatchFunction;
 import org.tensorflow.op.core.BatchToSpace;
 import org.tensorflow.op.core.BatchToSpaceNd;
 import org.tensorflow.op.core.Bitcast;
@@ -66,6 +67,7 @@
 import org.tensorflow.op.core.BroadcastDynamicShape;
 import org.tensorflow.op.core.BroadcastTo;
 import org.tensorflow.op.core.Bucketize;
+import org.tensorflow.op.core.Case;
 import org.tensorflow.op.core.ClipByValue;
 import org.tensorflow.op.core.Concat;
 import org.tensorflow.op.core.Constant;
@@ -89,6 +91,7 @@
 import org.tensorflow.op.core.ExtractVolumePatches;
 import org.tensorflow.op.core.Fill;
 import org.tensorflow.op.core.Fingerprint;
+import org.tensorflow.op.core.For;
 import org.tensorflow.op.core.Function;
 import org.tensorflow.op.core.Gather;
 import org.tensorflow.op.core.GatherNd;
@@ -101,6 +104,7 @@
 import org.tensorflow.op.core.HistogramFixedWidth;
 import org.tensorflow.op.core.Identity;
 import org.tensorflow.op.core.IdentityN;
+import org.tensorflow.op.core.If;
 import org.tensorflow.op.core.ImmutableConst;
 import org.tensorflow.op.core.Init;
 import org.tensorflow.op.core.InitializeTable;
@@ -149,6 +153,7 @@
 import org.tensorflow.op.core.Pad;
 import org.tensorflow.op.core.ParallelConcat;
 import org.tensorflow.op.core.ParallelDynamicStitch;
+import org.tensorflow.op.core.PartitionedCall;
 import org.tensorflow.op.core.Placeholder;
 import org.tensorflow.op.core.PlaceholderWithDefault;
 import org.tensorflow.op.core.Print;
@@ -166,6 +171,7 @@
 import org.tensorflow.op.core.RefNextIteration;
 import org.tensorflow.op.core.RefSelect;
 import org.tensorflow.op.core.RefSwitch;
+import org.tensorflow.op.core.RemoteCall;
 import org.tensorflow.op.core.Reshape;
 import org.tensorflow.op.core.ResourceCountUpTo;
 import org.tensorflow.op.core.ResourceGather;
@@ -215,6 +221,13 @@
 import org.tensorflow.op.core.StageClear;
 import org.tensorflow.op.core.StagePeek;
 import org.tensorflow.op.core.StageSize;
+import org.tensorflow.op.core.StatefulCase;
+import org.tensorflow.op.core.StatefulIf;
+import org.tensorflow.op.core.StatefulPartitionedCall;
+import org.tensorflow.op.core.StatefulWhile;
+import org.tensorflow.op.core.StatelessIf;
+import org.tensorflow.op.core.StatelessPartitionedCall;
+import org.tensorflow.op.core.StatelessWhile;
 import org.tensorflow.op.core.StopGradient;
 import org.tensorflow.op.core.StridedSlice;
 import org.tensorflow.op.core.StridedSliceAssign;
@@ -281,11 +294,7 @@
 import org.tensorflow.op.core.Variable;
 import org.tensorflow.op.core.VariableShape;
 import org.tensorflow.op.core.Where;
-import org.tensorflow.op.core.XlaConvV2;
-import org.tensorflow.op.core.XlaDotV2;
-import org.tensorflow.op.core.XlaSetDynamicDimensionSize;
-import org.tensorflow.op.core.XlaSpmdFullToShardShape;
-import org.tensorflow.op.core.XlaSpmdShardToFullShape;
+import org.tensorflow.op.core.While;
 import org.tensorflow.op.core.Zeros;
 import org.tensorflow.op.core.ZerosLike;
 import org.tensorflow.types.TBool;
@@ -357,20 +366,20 @@ public final class Ops {
 
   public final SparseOps sparse;
 
-  public final BitwiseOps bitwise;
-
   public final TpuOps tpu;
 
+  public final BitwiseOps bitwise;
+
   public final MathOps math;
 
   public final AudioOps audio;
 
   public final SignalOps signal;
 
-  public final TrainOps train;
-
   public final QuantizationOps quantization;
 
+  public final TrainOps train;
+
   private final Scope scope;
 
   private Ops(Scope scope) {
@@ -388,13 +397,13 @@ private Ops(Scope scope) {
     random = new RandomOps(this);
     strings = new StringsOps(this);
     sparse = new SparseOps(this);
-    bitwise = new BitwiseOps(this);
     tpu = new TpuOps(this);
+    bitwise = new BitwiseOps(this);
     math = new MathOps(this);
     audio = new AudioOps(this);
     signal = new SignalOps(this);
-    train = new TrainOps(this);
     quantization = new QuantizationOps(this);
+    train = new TrainOps(this);
   }
 
   /**
@@ -776,6 +785,61 @@ public Batch batch(Iterable<Operand<?>> inTensors, Long numBatchThreads, Long ma
     return Batch.create(scope, inTensors, numBatchThreads, maxBatchSize, batchTimeoutMicros, gradTimeoutMicros, options);
   }
 
+  /**
+   * Batches all the inputs tensors to the computation done by the function.
+   *  So, for example, in the following code
+   *  <pre>
+   *
+   *  # This input will be captured.
+   *  y = tf.placeholder_with_default(1.0, shape=[])
+   *
+   *  {@literal @}tf.Defun(tf.float32)
+   *  def computation(a):
+   *    return tf.matmul(a, a) + y
+   *
+   *  b = gen_batch_ops.batch_function(
+   *          f=computation
+   *          in_tensors=[a],
+   *          captured_tensors=computation.captured_inputs,
+   *          Tout=[o.type for o in computation.definition.signature.output_arg],
+   *          num_batch_threads=1,
+   *          max_batch_size=10,
+   *          batch_timeout_micros=100000,  # 100ms
+   *          allowed_batch_sizes=[3, 10],
+   *          batching_queue=&quot;&quot;)
+   *  </pre>
+   *  <p>If more than one session.run call is simultaneously trying to compute {@code b}
+   *  the values of {@code a} will be gathered, non-deterministically concatenated
+   *  along the first axis, and only one thread will run the computation.
+   *  <p>Assumes that all arguments of the function are Tensors which will be batched
+   *  along their first dimension.
+   *  <p>Arguments that are captured, are not batched. The session.run call which does
+   *  the concatenation, will use the values of the captured tensors available to it.
+   *  Therefore, typical uses of captured tensors should involve values which remain
+   *  unchanged across session.run calls. Inference is a good example of this.
+   *  <p>SparseTensor is not supported. The return value of the decorated function
+   *  must be a Tensor or a list/tuple of Tensors.
+   *
+   * @param inTensors The tensors to be batched.
+   * @param capturedTensors The tensors which are captured in the function, and don't need
+   *  to be batched.
+   * @param f the value of the f property
+   * @param numBatchThreads Number of scheduling threads for processing batches of work.
+   *  Determines the number of batches processed in parallel.
+   * @param maxBatchSize Batch sizes will never be bigger than this.
+   * @param batchTimeoutMicros Maximum number of microseconds to wait before outputting
+   *  an incomplete batch.
+   * @param Tout the types of the output tensors.
+   * @param options carries optional attribute values
+   * @return a new instance of BatchFunction
+   */
+  public BatchFunction batchFunction(Iterable<Operand<?>> inTensors,
+      Iterable<Operand<?>> capturedTensors, ConcreteFunction f, Long numBatchThreads,
+      Long maxBatchSize, Long batchTimeoutMicros, List<Class<? extends TType>> Tout,
+      BatchFunction.Options... options) {
+    return BatchFunction.create(scope, inTensors, capturedTensors, f, numBatchThreads, maxBatchSize, batchTimeoutMicros, Tout, options);
+  }
+
   /**
    * BatchToSpace for 4-D tensors of type T.
    *  This is a legacy version of the more general BatchToSpaceND.
@@ -1144,6 +1208,44 @@ public Map<String, Operand<?>> call(ConcreteFunction function,
     return Function.call(scope, function, arguments);
   }
 
+  /**
+   * An n-way switch statement which calls a single branch function.
+   *  <pre>
+   *  An n-way switch statement, implementing the following:
+   *  ```
+   *  switch (branch_index) {
+   *    case 0:
+   *      output = branches[0](input);
+   *      break;
+   *    case 1:
+   *      output = branches[1](input);
+   *      break;
+   *    ...
+   *    case [[nbranches-1]]:
+   *    default:
+   *      output = branches[nbranches-1](input);
+   *      break;
+   *  }
+   *  ```
+   *  </pre>
+   *
+   *  <p>Selects between {@link StatefulCase} and {@link StatelessCase} based on the statefulness of the function arguments.
+   *
+   * @param branchIndex The branch selector, an int32 Tensor.
+   * @param input A list of input tensors passed to the branch function.
+   * @param Tout A list of output types.
+   * @param branches <pre>
+   *    A list of functions each of which takes 'inputs' and returns a list of
+   *    tensors, whose types are the same as what every other branch returns.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of Case
+   */
+  public Case caseOp(Operand<TInt32> branchIndex, Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, List<ConcreteFunction> branches, Case.Options... options) {
+    return Case.create(scope, branchIndex, input, Tout, branches, options);
+  }
+
   /**
    * Clips tensor values to a specified min and max.
    *  Given a tensor {@code t}, this operation returns a tensor of the same type and
@@ -2475,6 +2577,28 @@ public Fingerprint fingerprint(Operand<? extends TType> data, Operand<TString> m
     return Fingerprint.create(scope, data, method);
   }
 
+  /**
+   * <pre>
+   *   output = input;
+   *   for i in range(start, limit, delta)
+   *     output = body(i, output);
+   *  </pre>
+   *
+   * @param start The lower bound. An int32
+   * @param limit The upper bound. An int32
+   * @param delta The increment. An int32
+   * @param input A list of input tensors whose types are T.
+   * @param body <pre>
+   *  A function that takes a list of tensors (int32, T) and returns another
+   *  list of tensors (T).
+   *  </pre>
+   * @return a new instance of For
+   */
+  public For forOp(Operand<TInt32> start, Operand<TInt32> limit, Operand<TInt32> delta,
+      Iterable<Operand<?>> input, ConcreteFunction body) {
+    return For.create(scope, start, limit, delta, input, body);
+  }
+
   /**
    * Gather slices from {@code params} axis {@code axis} according to {@code indices}.
    *  {@code indices} must be an integer tensor of any dimension (usually 0-D or 1-D).
@@ -2820,6 +2944,38 @@ public IdentityN identityN(Iterable<Operand<?>> input) {
     return IdentityN.create(scope, input);
   }
 
+  /**
+   * output = cond ? then_branch(input) : else_branch(input)
+   *
+   *  <p>Selects between {@link StatefulIf} and {@link StatelessIf} based on the statefulness of the function arguments.
+   *
+   * @param cond <pre>
+   *    A Tensor. If the tensor is a scalar of non-boolean type, the
+   *    scalar is converted to a boolean according to the
+   *    following rule: if the scalar is a numerical value, non-zero means
+   *    `True` and zero means False; if the scalar is a string, non-empty
+   *    means `True` and empty means `False`. If the tensor is not a scalar,
+   *    being empty means False and being non-empty means True.
+   *  </pre>
+   * @param input A list of input tensors.
+   * @param Tout A list of output types.
+   * @param thenBranch <pre>
+   *    A function that takes 'inputs' and returns a list of tensors, whose
+   *    types are the same as what else_branch returns.
+   *  </pre>
+   * @param elseBranch <pre>
+   *  A function that takes 'inputs' and returns a list of tensors, whose
+   *  types are the same as what then_branch returns.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of If
+   */
+  public If ifOp(Operand<? extends TType> cond, Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, ConcreteFunction thenBranch, ConcreteFunction elseBranch,
+      If.Options... options) {
+    return If.create(scope, cond, input, Tout, thenBranch, elseBranch, options);
+  }
+
   /**
    * Returns immutable tensor from memory region.
    *  The current implementation memmaps the tensor from a file.
@@ -3859,6 +4015,28 @@ public <T extends TType> ParallelDynamicStitch<T> parallelDynamicStitch(
     return ParallelDynamicStitch.create(scope, indices, data);
   }
 
+  /**
+   * returns {@code f(inputs)}, where {@code f}'s body is placed and partitioned.
+   *
+   *  <p>Selects between {@link StatefulPartitionedCall} and {@link StatelessPartitionedCall} based on the statefulness of the function arguments.
+   *
+   * @param args A list of input tensors.
+   * @param Tout A list of output types.
+   * @param f <pre>
+   *    A function that takes 'args', a list of tensors, and returns 'output',
+   *    another list of tensors. Input and output types are specified by 'Tin'
+   *    and 'Tout'. The function body of f will be placed and partitioned across
+   *    devices, setting this op apart from the regular Call op. This op is
+   *    stateful.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of PartitionedCall
+   */
+  public PartitionedCall partitionedCall(Iterable<Operand<?>> args,
+      List<Class<? extends TType>> Tout, ConcreteFunction f, PartitionedCall.Options... options) {
+    return PartitionedCall.create(scope, args, Tout, f, options);
+  }
+
   /**
    * A placeholder op for a value that will be fed into the computation.
    *  N.B. This operation will fail with an error if it is executed. It is
@@ -4158,6 +4336,20 @@ public <T extends TType> RefSwitch<T> refSwitch(Operand<T> data, Operand<TBool>
     return RefSwitch.create(scope, data, pred);
   }
 
+  /**
+   * Runs function {@code f} on a remote device indicated by {@code target}.
+   *
+   * @param target A fully specified device name where we want to run the function.
+   * @param args A list of arguments for the function.
+   * @param Tout The type list for the return values.
+   * @param f The function to run remotely.
+   * @return a new instance of RemoteCall
+   */
+  public RemoteCall remoteCall(Operand<TString> target, Iterable<Operand<?>> args,
+      List<Class<? extends TType>> Tout, ConcreteFunction f) {
+    return RemoteCall.create(scope, target, args, Tout, f);
+  }
+
   /**
    * Reshapes a tensor.
    *  Given {@code tensor}, this operation returns a tensor that has the same values
@@ -5840,6 +6032,199 @@ public StageSize stageSize(List<Class<? extends TType>> dtypes, StageSize.Option
     return StageSize.create(scope, dtypes, options);
   }
 
+  /**
+   * An n-way switch statement which calls a single branch function.
+   *  <pre>
+   *  An n-way switch statement, implementing the following:
+   *  ```
+   *  switch (branch_index) {
+   *    case 0:
+   *      output = branches[0](input);
+   *      break;
+   *    case 1:
+   *      output = branches[1](input);
+   *      break;
+   *    ...
+   *    case [[nbranches-1]]:
+   *    default:
+   *      output = branches[nbranches-1](input);
+   *      break;
+   *  }
+   *  ```
+   *  </pre>
+   *
+   * @param branchIndex The branch selector, an int32 Tensor.
+   * @param input A list of input tensors passed to the branch function.
+   * @param Tout A list of output types.
+   * @param branches <pre>
+   *    A list of functions each of which takes 'inputs' and returns a list of
+   *    tensors, whose types are the same as what every other branch returns.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatefulCase
+   */
+  public StatefulCase statefulCase(Operand<TInt32> branchIndex, Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, List<ConcreteFunction> branches, Case.Options... options) {
+    return StatefulCase.create(scope, branchIndex, input, Tout, branches, options);
+  }
+
+  /**
+   * output = cond ? then_branch(input) : else_branch(input)
+   *
+   * @param cond <pre>
+   *    A Tensor. If the tensor is a scalar of non-boolean type, the
+   *    scalar is converted to a boolean according to the
+   *    following rule: if the scalar is a numerical value, non-zero means
+   *    `True` and zero means False; if the scalar is a string, non-empty
+   *    means `True` and empty means `False`. If the tensor is not a scalar,
+   *    being empty means False and being non-empty means True.
+   *  </pre>
+   * @param input A list of input tensors.
+   * @param Tout A list of output types.
+   * @param thenBranch <pre>
+   *    A function that takes 'inputs' and returns a list of tensors, whose
+   *    types are the same as what else_branch returns.
+   *  </pre>
+   * @param elseBranch <pre>
+   *  A function that takes 'inputs' and returns a list of tensors, whose
+   *  types are the same as what then_branch returns.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatefulIf
+   */
+  public StatefulIf statefulIf(Operand<? extends TType> cond, Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, ConcreteFunction thenBranch, ConcreteFunction elseBranch,
+      If.Options... options) {
+    return StatefulIf.create(scope, cond, input, Tout, thenBranch, elseBranch, options);
+  }
+
+  /**
+   * returns {@code f(inputs)}, where {@code f}'s body is placed and partitioned.
+   *
+   * @param args A list of input tensors.
+   * @param Tout A list of output types.
+   * @param f <pre>
+   *    A function that takes 'args', a list of tensors, and returns 'output',
+   *    another list of tensors. Input and output types are specified by 'Tin'
+   *    and 'Tout'. The function body of f will be placed and partitioned across
+   *    devices, setting this op apart from the regular Call op. This op is
+   *    stateful.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatefulPartitionedCall
+   */
+  public StatefulPartitionedCall statefulPartitionedCall(Iterable<Operand<?>> args,
+      List<Class<? extends TType>> Tout, ConcreteFunction f, PartitionedCall.Options... options) {
+    return StatefulPartitionedCall.create(scope, args, Tout, f, options);
+  }
+
+  /**
+   * output = input; While (Cond(output)) { output = Body(output) }
+   *
+   * @param input A list of input tensors whose types are T.
+   * @param cond <pre>
+   *    A function takes 'input' and returns a tensor.  If the tensor is
+   *    a scalar of non-boolean, the scalar is converted to a boolean
+   *    according to the following rule: if the scalar is a numerical
+   *    value, non-zero means True and zero means False; if the scalar is
+   *    a string, non-empty means True and empty means False. If the
+   *    tensor is not a scalar, non-emptiness means True and False
+   *    otherwise.
+   *  </pre>
+   * @param body <pre>
+   *    A function that takes a list of tensors and returns another
+   *    list of tensors. Both lists have the same types as specified
+   *    by T.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatefulWhile
+   */
+  public StatefulWhile statefulWhile(Iterable<Operand<?>> input, ConcreteFunction cond,
+      ConcreteFunction body, While.Options... options) {
+    return StatefulWhile.create(scope, input, cond, body, options);
+  }
+
+  /**
+   * output = cond ? then_branch(input) : else_branch(input)
+   *
+   * @param cond <pre>
+   *    A Tensor. If the tensor is a scalar of non-boolean type, the
+   *    scalar is converted to a boolean according to the
+   *    following rule: if the scalar is a numerical value, non-zero means
+   *    `True` and zero means False; if the scalar is a string, non-empty
+   *    means `True` and empty means `False`. If the tensor is not a scalar,
+   *    being empty means False and being non-empty means True.
+   *
+   *    This should only be used when the if then/else body functions do not
+   *    have stateful ops.
+   *  </pre>
+   * @param input A list of input tensors.
+   * @param Tout A list of output types.
+   * @param thenBranch <pre>
+   *    A function that takes 'inputs' and returns a list of tensors, whose
+   *    types are the same as what else_branch returns.
+   *  </pre>
+   * @param elseBranch <pre>
+   *  A function that takes 'inputs' and returns a list of tensors, whose
+   *  types are the same as what then_branch returns.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatelessIf
+   */
+  public StatelessIf statelessIf(Operand<? extends TType> cond, Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, ConcreteFunction thenBranch, ConcreteFunction elseBranch,
+      If.Options... options) {
+    return StatelessIf.create(scope, cond, input, Tout, thenBranch, elseBranch, options);
+  }
+
+  /**
+   * returns {@code f(inputs)}, where {@code f}'s body is placed and partitioned.
+   *
+   * @param args A list of input tensors.
+   * @param Tout A list of output types.
+   * @param f <pre>
+   *    A function that takes 'args', a list of tensors, and returns 'output',
+   *    another list of tensors. Input and output types are specified by 'Tin'
+   *    and 'Tout'. The function body of f will be placed and partitioned across
+   *    devices, setting this op apart from the regular Call op.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatelessPartitionedCall
+   */
+  public StatelessPartitionedCall statelessPartitionedCall(Iterable<Operand<?>> args,
+      List<Class<? extends TType>> Tout, ConcreteFunction f, PartitionedCall.Options... options) {
+    return StatelessPartitionedCall.create(scope, args, Tout, f, options);
+  }
+
+  /**
+   * output = input; While (Cond(output)) { output = Body(output) }
+   *
+   * @param input A list of input tensors whose types are T.
+   * @param cond <pre>
+   *    A function takes 'input' and returns a tensor.  If the tensor is
+   *    a scalar of non-boolean, the scalar is converted to a boolean
+   *    according to the following rule: if the scalar is a numerical
+   *    value, non-zero means True and zero means False; if the scalar is
+   *    a string, non-empty means True and empty means False. If the
+   *    tensor is not a scalar, non-emptiness means True and False
+   *    otherwise.
+   *
+   *    This should only be used when the while condition and body functions
+   *    do not have stateful ops.
+   *  </pre>
+   * @param body <pre>
+   *    A function that takes a list of tensors and returns another
+   *    list of tensors. Both lists have the same types as specified
+   *    by T.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatelessWhile
+   */
+  public StatelessWhile statelessWhile(Iterable<Operand<?>> input, ConcreteFunction cond,
+      ConcreteFunction body, While.Options... options) {
+    return StatelessWhile.create(scope, input, cond, body, options);
+  }
+
   /**
    * Stops gradient computation.
    *  When executed in a graph, this op outputs its input tensor as-is.
@@ -7698,105 +8083,31 @@ public Where where(Operand<? extends TType> condition) {
   }
 
   /**
-   * Wraps the XLA ConvGeneralDilated operator, documented at
-   *  https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
-   *  .
+   * output = input; While (Cond(output)) { output = Body(output) }
    *
-   * @param <W> data type for {@code output} output
-   * @param lhs the input tensor
-   * @param rhs the kernel tensor
-   * @param windowStrides the inter-window strides
-   * @param padding the padding to apply at the start and end of each input dimensions
-   * @param lhsDilation dilation to apply between input elements
-   * @param rhsDilation dilation to apply between kernel elements
-   * @param featureGroupCount number of feature groups for grouped convolution.
-   * @param dimensionNumbers a serialized xla::ConvolutionDimensionNumbers proto.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @param preferredElementType The type of the tensor.
-   * @param <W> data type for {@code XlaConvV2} output and operands
-   * @param <V> data type for {@code XlaConvV2} output and operands
-   * @return a new instance of XlaConvV2
-   */
-  public <W extends TType, V extends TNumber> XlaConvV2<W> xlaConvV2(Operand<? extends TType> lhs,
-      Operand<? extends TType> rhs, Operand<V> windowStrides, Operand<V> padding,
-      Operand<V> lhsDilation, Operand<V> rhsDilation, Operand<V> featureGroupCount,
-      String dimensionNumbers, String precisionConfig, Class<W> preferredElementType) {
-    return XlaConvV2.create(scope, lhs, rhs, windowStrides, padding, lhsDilation, rhsDilation, featureGroupCount, dimensionNumbers, precisionConfig, preferredElementType);
-  }
-
-  /**
-   * Wraps the XLA DotGeneral operator, documented at
-   *  https://www.tensorflow.org/performance/xla/operation_semantics#dotgeneral
-   *  .
+   *  <p>Selects between {@link StatefulWhile} and {@link StatelessWhile} based on the statefulness of the function arguments.
    *
-   * @param <V> data type for {@code output} output
-   * @param lhs the LHS tensor
-   * @param rhs the RHS tensor
-   * @param dimensionNumbers a serialized xla::DotDimensionNumbers proto.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @param preferredElementType The type of the tensor.
-   * @param <V> data type for {@code XlaDotV2} output and operands
-   * @return a new instance of XlaDotV2
-   */
-  public <V extends TType> XlaDotV2<V> xlaDotV2(Operand<? extends TType> lhs,
-      Operand<? extends TType> rhs, String dimensionNumbers, String precisionConfig,
-      Class<V> preferredElementType) {
-    return XlaDotV2.create(scope, lhs, rhs, dimensionNumbers, precisionConfig, preferredElementType);
-  }
-
-  /**
-   * Make a static dimension into a xla bounded dynamic dimension.
-   *  <pre>
-   *      The current static dimension size will become the bound and the second
-   *      operand becomes the dynamic size of the dimension.
+   * @param input A list of input tensors whose types are T.
+   * @param cond <pre>
+   *    A function takes 'input' and returns a tensor.  If the tensor is
+   *    a scalar of non-boolean, the scalar is converted to a boolean
+   *    according to the following rule: if the scalar is a numerical
+   *    value, non-zero means True and zero means False; if the scalar is
+   *    a string, non-empty means True and empty means False. If the
+   *    tensor is not a scalar, non-emptiness means True and False
+   *    otherwise.
    *  </pre>
-   *
-   * @param <T> data type for {@code output} output
-   * @param input the input value
-   * @param dimIndex the dimIndex value
-   * @param sizeOutput the sizeOutput value
-   * @param <T> data type for {@code XlaSetDynamicDimensionSize} output and operands
-   * @return a new instance of XlaSetDynamicDimensionSize
-   */
-  public <T extends TType> XlaSetDynamicDimensionSize<T> xlaSetDynamicDimensionSize(
-      Operand<T> input, Operand<TInt32> dimIndex, Operand<TInt32> sizeOutput) {
-    return XlaSetDynamicDimensionSize.create(scope, input, dimIndex, sizeOutput);
-  }
-
-  /**
-   * An op used by XLA SPMD partitioner to switch from automatic partitioning to
-   *  manual partitioning. It annotates the input (full-shape, to be automatically
-   *  partitioned) with the same sharding used by manual partitioning, and outputs a
-   *  shard-shaped tensor to be consumed by later manually-partitioned ops. If the
-   *  shape is not evenly partitionable, the padding region will be masked with 0s.
-   *
-   * @param <T> data type for {@code output} output
-   * @param input the input value
-   * @param manualSharding the value of the manualSharding property
-   * @param <T> data type for {@code XlaSpmdFullToShardShape} output and operands
-   * @return a new instance of XlaSpmdFullToShardShape
+   * @param body <pre>
+   *    A function that takes a list of tensors and returns another
+   *    list of tensors. Both lists have the same types as specified
+   *    by T.
+   *  </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of While
    */
-  public <T extends TType> XlaSpmdFullToShardShape<T> xlaSpmdFullToShardShape(Operand<T> input,
-      String manualSharding) {
-    return XlaSpmdFullToShardShape.create(scope, input, manualSharding);
-  }
-
-  /**
-   * An op used by XLA SPMD partitioner to switch from manual partitioning to
-   *  automatic partitioning. It converts the shard-shaped, manually partitioned input
-   *  into full-shaped tensor to be partitioned automatically with the same sharding
-   *  used by manual partitioning.
-   *
-   * @param <T> data type for {@code output} output
-   * @param input the input value
-   * @param manualSharding the value of the manualSharding property
-   * @param fullShape the value of the fullShape property
-   * @param <T> data type for {@code XlaSpmdShardToFullShape} output and operands
-   * @return a new instance of XlaSpmdShardToFullShape
-   */
-  public <T extends TType> XlaSpmdShardToFullShape<T> xlaSpmdShardToFullShape(Operand<T> input,
-      String manualSharding, Shape fullShape) {
-    return XlaSpmdShardToFullShape.create(scope, input, manualSharding, fullShape);
+  public While whileOp(Iterable<Operand<?>> input, ConcreteFunction cond, ConcreteFunction body,
+      While.Options... options) {
+    return While.create(scope, input, cond, body, options);
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TpuOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TpuOps.java
index 1278494020f..6cd4a872851 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TpuOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TpuOps.java
@@ -18,12 +18,15 @@
 package org.tensorflow.op;
 
 import java.util.List;
+import org.tensorflow.ConcreteFunction;
 import org.tensorflow.Operand;
+import org.tensorflow.op.tpu.Compile;
 import org.tensorflow.op.tpu.CompileSucceededAssert;
 import org.tensorflow.op.tpu.Execute;
 import org.tensorflow.op.tpu.ExecuteAndUpdateVariables;
 import org.tensorflow.op.tpu.PartitionedInput;
 import org.tensorflow.op.tpu.PartitionedOutput;
+import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
@@ -42,6 +45,36 @@ public final class TpuOps {
     this.ops = ops;
   }
 
+  /**
+   * Compiles a computations for execution on one or more TPU devices.
+   *  For the internal use of the distributed TPU compiler.
+   *  <p>'num_computations' is the number of computations to be compiled.
+   *  'function' is a function containing the computation to compile.
+   *  'dynamic_shapes' contains dynamic shapes of arguments whose shapes were not
+   *  known statically at TPUReplication rewrite time.
+   *  'guaranteed_constants' is a list of tensors which have been guaranteed to not
+   *  change their values during the session lifetime. These contain tensors marked as
+   *  constant using the GuaranteeConstOp.
+   *  'metadata' is a serialized TPUCompileMetadataProto describing
+   *  the shapes and types of the inputs to the computation, as well as a mapping onto
+   *  the TPU pod topology.
+   *  Each 'program' output is a string key that is passed to the _TPUExecute op and
+   *  used to look up the program in the compilation cache.
+   *  'may_modify_variables' indicates whether variables may be modified.
+   *
+   * @param dynamicShapes the dynamicShapes value
+   * @param guaranteedConstants the guaranteedConstants value
+   * @param numComputations the value of the numComputations property
+   * @param function the value of the function property
+   * @param metadata the value of the metadata property
+   * @return a new instance of Compile
+   */
+  public Compile compile(Iterable<Operand<TInt64>> dynamicShapes,
+      Iterable<Operand<?>> guaranteedConstants, Long numComputations, ConcreteFunction function,
+      String metadata) {
+    return Compile.create(scope, dynamicShapes, guaranteedConstants, numComputations, function, metadata);
+  }
+
   /**
    * Asserts that compilation succeeded. This op produces no output and closes the
    *  device during failure to ensure all pending device interactions fail.
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
index 305b973c139..2661165e101 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
@@ -18,6 +18,7 @@
 package org.tensorflow.op;
 
 import java.util.List;
+import org.tensorflow.ConcreteFunction;
 import org.tensorflow.Operand;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.train.AccumulatorApplyGradient;
@@ -81,6 +82,7 @@
 import org.tensorflow.op.train.SparseApplyProximalAdagrad;
 import org.tensorflow.op.train.SparseApplyProximalGradientDescent;
 import org.tensorflow.op.train.SparseApplyRmsProp;
+import org.tensorflow.op.train.SymbolicGradient;
 import org.tensorflow.op.train.TileGrad;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
@@ -1610,6 +1612,32 @@ public <T extends TType> SparseApplyRmsProp<T> sparseApplyRmsProp(Operand<T> var
     return SparseApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
   }
 
+  /**
+   * Computes the gradient function for function f via backpropagation.
+   *
+   * @param input a list of input tensors of size N + M;
+   * @param Tout the type list for the input list.
+   * @param f The function we want to compute the gradient for.
+   *  <p>The function 'f' must be a numerical function which takes N inputs and
+   *  produces M outputs. Its gradient function 'g', which is computed by
+   *  this SymbolicGradient op is a function taking N + M inputs and
+   *  produces N outputs.
+   *  <p>I.e. if we have
+   *  (y1, y2, ..., y_M) = f(x1, x2, ..., x_N),
+   *  then, g is
+   *  (dL/dx1, dL/dx2, ..., dL/dx_N) = g(x1, x2, ..., x_N,
+   *  dL/dy1, dL/dy2, ..., dL/dy_M),
+   *  <p>where L is a scalar-value function of (x1, x2, ..., xN) (e.g., the
+   *  loss function). dL/dx_i is the partial derivative of L with respect
+   *  to x_i.
+   *  <p>(Needs some math expert to say the comment above better.)
+   * @return a new instance of SymbolicGradient
+   */
+  public SymbolicGradient symbolicGradient(Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, ConcreteFunction f) {
+    return SymbolicGradient.create(scope, input, Tout, f);
+  }
+
   /**
    * Returns the gradient of {@code Tile}.
    *  Since {@code Tile} takes an input and repeats the input {@code multiples} times
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
index 99caae1fdc2..07dd99967d1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
@@ -17,6 +17,8 @@
 //
 package org.tensorflow.op;
 
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
 import org.tensorflow.Operand;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.xla.BroadcastHelper;
@@ -28,18 +30,31 @@
 import org.tensorflow.op.xla.DynamicUpdateSlice;
 import org.tensorflow.op.xla.Einsum;
 import org.tensorflow.op.xla.Gather;
+import org.tensorflow.op.xla.If;
 import org.tensorflow.op.xla.KeyValueSort;
 import org.tensorflow.op.xla.Pad;
 import org.tensorflow.op.xla.Recv;
+import org.tensorflow.op.xla.Reduce;
+import org.tensorflow.op.xla.ReduceWindow;
 import org.tensorflow.op.xla.ReplicaId;
+import org.tensorflow.op.xla.Scatter;
+import org.tensorflow.op.xla.SelectAndScatter;
 import org.tensorflow.op.xla.SelfAdjointEig;
 import org.tensorflow.op.xla.Send;
+import org.tensorflow.op.xla.SetDynamicDimensionSize;
 import org.tensorflow.op.xla.Sharding;
 import org.tensorflow.op.xla.Sort;
+import org.tensorflow.op.xla.SpmdFullToShardShape;
+import org.tensorflow.op.xla.SpmdShardToFullShape;
 import org.tensorflow.op.xla.Svd;
+import org.tensorflow.op.xla.While;
+import org.tensorflow.op.xla.XlaHostCompute;
+import org.tensorflow.op.xla.XlaLaunch;
 import org.tensorflow.op.xla.XlaRecvFromHost;
 import org.tensorflow.op.xla.XlaSendToHost;
 import org.tensorflow.op.xla.XlaSetBound;
+import org.tensorflow.op.xla.XlaVariadicReduce;
+import org.tensorflow.op.xla.XlaVariadicSort;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
@@ -94,7 +109,7 @@ public <T extends TType> ClusterOutput<T> clusterOutput(Operand<T> input) {
    *  https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
    *  .
    *
-   * @param <T> data type for {@code output} output
+   * @param <W> data type for {@code output} output
    * @param lhs the input tensor
    * @param rhs the kernel tensor
    * @param windowStrides the inter-window strides
@@ -104,14 +119,16 @@ public <T extends TType> ClusterOutput<T> clusterOutput(Operand<T> input) {
    * @param featureGroupCount number of feature groups for grouped convolution.
    * @param dimensionNumbers a serialized xla::ConvolutionDimensionNumbers proto.
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @param <T> data type for {@code XlaConv} output and operands
-   * @param <U> data type for {@code XlaConv} output and operands
+   * @param preferredElementType The type of the tensor.
+   * @param <W> data type for {@code XlaConvV2} output and operands
+   * @param <V> data type for {@code XlaConvV2} output and operands
    * @return a new instance of Conv
    */
-  public <T extends TType, U extends TNumber> Conv<T> conv(Operand<T> lhs, Operand<T> rhs,
-      Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation,
-      Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
-    return Conv.create(scope, lhs, rhs, windowStrides, padding, lhsDilation, rhsDilation, featureGroupCount, dimensionNumbers, precisionConfig);
+  public <W extends TType, V extends TNumber> Conv<W> conv(Operand<? extends TType> lhs,
+      Operand<? extends TType> rhs, Operand<V> windowStrides, Operand<V> padding,
+      Operand<V> lhsDilation, Operand<V> rhsDilation, Operand<V> featureGroupCount,
+      String dimensionNumbers, String precisionConfig, Class<W> preferredElementType) {
+    return Conv.create(scope, lhs, rhs, windowStrides, padding, lhsDilation, rhsDilation, featureGroupCount, dimensionNumbers, precisionConfig, preferredElementType);
   }
 
   /**
@@ -136,17 +153,18 @@ public Dequantize dequantize(Operand<? extends TType> input, Float minRange, Flo
    *  https://www.tensorflow.org/performance/xla/operation_semantics#dotgeneral
    *  .
    *
-   * @param <T> data type for {@code output} output
+   * @param <V> data type for {@code output} output
    * @param lhs the LHS tensor
    * @param rhs the RHS tensor
    * @param dimensionNumbers a serialized xla::DotDimensionNumbers proto.
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @param <T> data type for {@code XlaDot} output and operands
+   * @param preferredElementType The type of the tensor.
+   * @param <V> data type for {@code XlaDotV2} output and operands
    * @return a new instance of Dot
    */
-  public <T extends TType> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dimensionNumbers,
-      String precisionConfig) {
-    return Dot.create(scope, lhs, rhs, dimensionNumbers, precisionConfig);
+  public <V extends TType> Dot<V> dot(Operand<? extends TType> lhs, Operand<? extends TType> rhs,
+      String dimensionNumbers, String precisionConfig, Class<V> preferredElementType) {
+    return Dot.create(scope, lhs, rhs, dimensionNumbers, precisionConfig, preferredElementType);
   }
 
   /**
@@ -234,6 +252,23 @@ public <T extends TType, U extends TNumber> Gather<T> gather(Operand<T> operand,
     return Gather.create(scope, operand, startIndices, sliceSizes, dimensionNumbers, indicesAreSorted);
   }
 
+  /**
+   * output = cond ? then_branch(inputs) : else_branch(inputs).
+   *
+   * @param cond A boolean scalar.
+   * @param inputs A list of input tensors.
+   * @param thenBranch A function takes 'inputs' and returns a list of tensors,
+   *  whose types are the same as what else_branch returns.
+   * @param elseBranch A function takes 'inputs' and returns a list of tensors.
+   *  whose types are the same as what then_branch returns.
+   * @param Tout the value of the Tout property
+   * @return a new instance of If
+   */
+  public If ifOp(Operand<? extends TType> cond, Iterable<Operand<?>> inputs,
+      ConcreteFunction thenBranch, ConcreteFunction elseBranch, List<Class<? extends TType>> Tout) {
+    return If.create(scope, cond, inputs, thenBranch, elseBranch, Tout);
+  }
+
   /**
    * Wraps the XLA Sort operator, documented at
    *  https://www.tensorflow.org/performance/xla/operation_semantics#sort
@@ -293,6 +328,47 @@ public <T extends TType> Recv<T> recv(Class<T> dtype, String tensorName, Shape s
     return Recv.create(scope, dtype, tensorName, shape);
   }
 
+  /**
+   * Wraps the XLA Reduce operator, documented at
+   *  https://www.tensorflow.org/performance/xla/operation_semantics#reduce .
+   *
+   * @param <T> data type for {@code output} output
+   * @param input the input tensor
+   * @param initValue a scalar representing the initial value for the reduction
+   * @param dimensionsToReduce dimension numbers over which to reduce
+   * @param reducer a reducer function to apply
+   * @param <T> data type for {@code XlaReduce} output and operands
+   * @return a new instance of Reduce
+   */
+  public <T extends TType> Reduce<T> reduce(Operand<T> input, Operand<T> initValue,
+      List<Long> dimensionsToReduce, ConcreteFunction reducer) {
+    return Reduce.create(scope, input, initValue, dimensionsToReduce, reducer);
+  }
+
+  /**
+   * Wraps the XLA ReduceWindow operator, documented at
+   *  https://www.tensorflow.org/performance/xla/operation_semantics#reducewindow .
+   *
+   * @param <T> data type for {@code output} output
+   * @param input the input tensor
+   * @param initValue a scalar representing the initial value for the reduction
+   * @param windowDimensions the shape of the window
+   * @param windowStrides the inter-window strides
+   * @param baseDilations the baseDilations value
+   * @param windowDilations the windowDilations value
+   * @param padding the padding to apply at the start and end of each input dimensions
+   * @param computation a reducer function to apply
+   * @param <T> data type for {@code XlaReduceWindow} output and operands
+   * @param <U> data type for {@code XlaReduceWindow} output and operands
+   * @return a new instance of ReduceWindow
+   */
+  public <T extends TType, U extends TNumber> ReduceWindow<T> reduceWindow(Operand<T> input,
+      Operand<T> initValue, Operand<U> windowDimensions, Operand<U> windowStrides,
+      Operand<U> baseDilations, Operand<U> windowDilations, Operand<U> padding,
+      ConcreteFunction computation) {
+    return ReduceWindow.create(scope, input, initValue, windowDimensions, windowStrides, baseDilations, windowDilations, padding, computation);
+  }
+
   /**
    * Replica ID.
    *
@@ -302,6 +378,52 @@ public ReplicaId replicaId() {
     return ReplicaId.create(scope);
   }
 
+  /**
+   * Wraps the XLA Scatter operator documented at
+   *  https://www.tensorflow.org/xla/operation_semantics#scatter.
+   *
+   * @param <T> data type for {@code output} output
+   * @param operand Array to be scattered into.
+   * @param scatterIndices Array containing the starting indices of the slices that must
+   *  be scattered to.
+   * @param updates Array containing the values that must be used for scattering.
+   * @param updateComputation Computation to be used for combining the existing values in
+   *  the input array and the updates during scatter.
+   * @param dimensionNumbers A serialized xla::ScatterDimensionNumbers proto.
+   * @param indicesAreSorted Boolean indicating if the indices are sorted.
+   * @param <T> data type for {@code XlaScatter} output and operands
+   * @return a new instance of Scatter
+   */
+  public <T extends TType> Scatter<T> scatter(Operand<T> operand,
+      Operand<? extends TNumber> scatterIndices, Operand<T> updates,
+      ConcreteFunction updateComputation, String dimensionNumbers, Boolean indicesAreSorted) {
+    return Scatter.create(scope, operand, scatterIndices, updates, updateComputation, dimensionNumbers, indicesAreSorted);
+  }
+
+  /**
+   * Wraps the XLA SelectAndScatter operator, documented at
+   *  https://www.tensorflow.org/performance/xla/operation_semantics#selectandscatter
+   *  .
+   *
+   * @param <T> data type for {@code output} output
+   * @param operand the input tensor
+   * @param windowDimensions the shape of the window
+   * @param windowStrides the inter-window strides
+   * @param padding the padding to apply at the start and end of each input dimensions
+   * @param source a tensor of values to scatter
+   * @param initValue a scalar representing the initial value for the output tensor
+   * @param select a selection function to apply
+   * @param scatter a scatter function to apply
+   * @param <T> data type for {@code XlaSelectAndScatter} output and operands
+   * @param <U> data type for {@code XlaSelectAndScatter} output and operands
+   * @return a new instance of SelectAndScatter
+   */
+  public <T extends TType, U extends TNumber> SelectAndScatter<T> selectAndScatter(
+      Operand<T> operand, Operand<U> windowDimensions, Operand<U> windowStrides, Operand<U> padding,
+      Operand<T> source, Operand<T> initValue, ConcreteFunction select, ConcreteFunction scatter) {
+    return SelectAndScatter.create(scope, operand, windowDimensions, windowStrides, padding, source, initValue, select, scatter);
+  }
+
   /**
    * Computes the eigen decomposition of a batch of self-adjoint matrices
    *  (Note: Only real inputs are supported).
@@ -339,6 +461,25 @@ public Send send(Operand<? extends TType> tensor, String tensorName) {
     return Send.create(scope, tensor, tensorName);
   }
 
+  /**
+   * Make a static dimension into a xla bounded dynamic dimension.
+   *  <pre>
+   *      The current static dimension size will become the bound and the second
+   *      operand becomes the dynamic size of the dimension.
+   *  </pre>
+   *
+   * @param <T> data type for {@code output} output
+   * @param input the input value
+   * @param dimIndex the dimIndex value
+   * @param sizeOutput the sizeOutput value
+   * @param <T> data type for {@code XlaSetDynamicDimensionSize} output and operands
+   * @return a new instance of SetDynamicDimensionSize
+   */
+  public <T extends TType> SetDynamicDimensionSize<T> setDynamicDimensionSize(Operand<T> input,
+      Operand<TInt32> dimIndex, Operand<TInt32> sizeOutput) {
+    return SetDynamicDimensionSize.create(scope, input, dimIndex, sizeOutput);
+  }
+
   /**
    * An op which shards the input based on the given sharding attribute.
    *
@@ -367,6 +508,42 @@ public <T extends TType> Sort<T> sort(Operand<T> input) {
     return Sort.create(scope, input);
   }
 
+  /**
+   * An op used by XLA SPMD partitioner to switch from automatic partitioning to
+   *  manual partitioning. It annotates the input (full-shape, to be automatically
+   *  partitioned) with the same sharding used by manual partitioning, and outputs a
+   *  shard-shaped tensor to be consumed by later manually-partitioned ops. If the
+   *  shape is not evenly partitionable, the padding region will be masked with 0s.
+   *
+   * @param <T> data type for {@code output} output
+   * @param input the input value
+   * @param manualSharding the value of the manualSharding property
+   * @param <T> data type for {@code XlaSpmdFullToShardShape} output and operands
+   * @return a new instance of SpmdFullToShardShape
+   */
+  public <T extends TType> SpmdFullToShardShape<T> spmdFullToShardShape(Operand<T> input,
+      String manualSharding) {
+    return SpmdFullToShardShape.create(scope, input, manualSharding);
+  }
+
+  /**
+   * An op used by XLA SPMD partitioner to switch from manual partitioning to
+   *  automatic partitioning. It converts the shard-shaped, manually partitioned input
+   *  into full-shaped tensor to be partitioned automatically with the same sharding
+   *  used by manual partitioning.
+   *
+   * @param <T> data type for {@code output} output
+   * @param input the input value
+   * @param manualSharding the value of the manualSharding property
+   * @param fullShape the value of the fullShape property
+   * @param <T> data type for {@code XlaSpmdShardToFullShape} output and operands
+   * @return a new instance of SpmdShardToFullShape
+   */
+  public <T extends TType> SpmdShardToFullShape<T> spmdShardToFullShape(Operand<T> input,
+      String manualSharding, Shape fullShape) {
+    return SpmdShardToFullShape.create(scope, input, manualSharding, fullShape);
+  }
+
   /**
    * Computes the eigen decomposition of a batch of self-adjoint matrices
    *  (Note: Only real inputs are supported).
@@ -389,6 +566,61 @@ public <T extends TType> Svd<T> svd(Operand<T> a, Long maxIter, Float epsilon,
     return Svd.create(scope, a, maxIter, epsilon, precisionConfig);
   }
 
+  /**
+   * output = input; While (Cond(output)) { output = Body(output) }
+   *
+   * @param input A list of input tensors whose types are T.
+   * @param cond A function takes 'input' and returns a tensor.  If the tensor is
+   *  a scalar of non-boolean, the scalar is converted to a boolean
+   *  according to the following rule: if the scalar is a numerical
+   *  value, non-zero means True and zero means False; if the scalar is
+   *  a string, non-empty means True and empty means False. If the
+   *  tensor is not a scalar, non-emptiness means True and False
+   *  otherwise.
+   * @param body A function that takes a list of tensors and returns another
+   *  list of tensors. Both lists have the same types as specified by T.
+   * @return a new instance of While
+   */
+  public While whileOp(Iterable<Operand<?>> input, ConcreteFunction cond, ConcreteFunction body) {
+    return While.create(scope, input, cond, body);
+  }
+
+  /**
+   * A pseudo-op to represent host-side computation in an XLA program.
+   *
+   * @param inputs A list of tensors that will be sent to the host.
+   * @param Toutputs The element types of each element in {@code outputs}.
+   * @param ancestors A list of names of HostCompute computations that must be
+   *  sequenced before this computation.
+   * @param shapes If shape_inference_graph is empty, a list of the shapes of {@code outputs}.
+   * @param shapeInferenceGraph If non-empty, a serialized GraphDef representing a graph
+   *  that must be analyzed at compile time to determine the shapes of the outputs.
+   * @param key A unique identifier for this region used to match up host transfers.
+   * @param options carries optional attribute values
+   * @return a new instance of XlaHostCompute
+   */
+  public XlaHostCompute xlaHostCompute(Iterable<Operand<?>> inputs,
+      List<Class<? extends TType>> Toutputs, List<String> ancestors, List<Shape> shapes,
+      ConcreteFunction shapeInferenceGraph, String key, XlaHostCompute.Options... options) {
+    return XlaHostCompute.create(scope, inputs, Toutputs, ancestors, shapes, shapeInferenceGraph, key, options);
+  }
+
+  /**
+   * XLA Launch Op. For use by the XLA JIT only.
+   *
+   * @param constants the constants value
+   * @param args the args value
+   * @param resources the resources value
+   * @param Tresults the value of the Tresults property
+   * @param function the value of the function property
+   * @return a new instance of XlaLaunch
+   */
+  public XlaLaunch xlaLaunch(Iterable<Operand<?>> constants, Iterable<Operand<?>> args,
+      Iterable<Operand<? extends TType>> resources, List<Class<? extends TType>> Tresults,
+      ConcreteFunction function) {
+    return XlaLaunch.create(scope, constants, args, resources, Tresults, function);
+  }
+
   /**
    * An op to receive a tensor from the host.
    *  output: the tensor that will be received from the host.
@@ -436,6 +668,44 @@ public XlaSetBound xlaSetBound(Operand<TInt32> input, Operand<TInt32> bound) {
     return XlaSetBound.create(scope, input, bound);
   }
 
+  /**
+   * Wraps the variadic XLA Reduce operator.
+   *  Semantics are documented at
+   *  https://www.tensorflow.org/performance/xla/operation_semantics#variadic_reduce.
+   *
+   * @param <T> data type for {@code output} output
+   * @param input the input tensor(s)
+   * @param initValue scalar initial value(s) for the reduction
+   * @param dimensionsToReduce dimension numbers over which to reduce
+   * @param reducer a reducer function to apply
+   * @param <T> data type for {@code XlaVariadicReduce} output and operands
+   * @return a new instance of XlaVariadicReduce
+   */
+  public <T extends TType> XlaVariadicReduce<T> xlaVariadicReduce(Iterable<Operand<T>> input,
+      Iterable<Operand<T>> initValue, List<Long> dimensionsToReduce, ConcreteFunction reducer) {
+    return XlaVariadicReduce.create(scope, input, initValue, dimensionsToReduce, reducer);
+  }
+
+  /**
+   * Wraps the XLA Sort operator, documented at
+   *  https://www.tensorflow.org/performance/xla/operation_semantics#sort
+   *  .
+   *  <p>Sorts one or more tensors, with support for custom comparator, dimension, and
+   *  is_stable attributes.
+   *
+   * @param inputs A list of {@code Tensor} of identical shape but possibly different types.
+   * @param dimension The dimension along which to sort. Must be a compile-time constant.
+   * @param comparator A comparator function to apply to 2*N scalars and returning a
+   *  boolean. N is the number of sort inputs. If you want to sort in ascending
+   *  order then the comparator should perform a less-than comparison.
+   * @param isStable Whether to use stable sort.
+   * @return a new instance of XlaVariadicSort
+   */
+  public XlaVariadicSort xlaVariadicSort(Iterable<Operand<?>> inputs, Operand<TInt32> dimension,
+      ConcreteFunction comparator, Boolean isStable) {
+    return XlaVariadicSort.create(scope, inputs, dimension, comparator, isStable);
+  }
+
   /**
    * Get the parent {@link Ops} object.
    */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
index 2441bc1af65..b345ab4dad2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
@@ -3150,7 +3150,7 @@ public static native void TF_RegisterFilesystemPlugin(
 // TF_InitKernel to do op/kernel registration.
 // Plugin should implement TF_InitKernel to register kernels. This function
 // should register all kernels in a plugin.
-public static native void TF_InitKernel();
+
 // Targeting ../Create_func_TF_OpKernelConstruction.java
 
 
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchFunction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchFunction.java
new file mode 100644
index 00000000000..db64e937645
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchFunction.java
@@ -0,0 +1,353 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Batches all the inputs tensors to the computation done by the function.
+ * So, for example, in the following code
+ * <pre>
+ *
+ * # This input will be captured.
+ * y = tf.placeholder_with_default(1.0, shape=[])
+ *
+ * {@literal @}tf.Defun(tf.float32)
+ * def computation(a):
+ *   return tf.matmul(a, a) + y
+ *
+ * b = gen_batch_ops.batch_function(
+ *         f=computation
+ *         in_tensors=[a],
+ *         captured_tensors=computation.captured_inputs,
+ *         Tout=[o.type for o in computation.definition.signature.output_arg],
+ *         num_batch_threads=1,
+ *         max_batch_size=10,
+ *         batch_timeout_micros=100000,  # 100ms
+ *         allowed_batch_sizes=[3, 10],
+ *         batching_queue=&quot;&quot;)
+ * </pre>
+ * <p>If more than one session.run call is simultaneously trying to compute {@code b}
+ * the values of {@code a} will be gathered, non-deterministically concatenated
+ * along the first axis, and only one thread will run the computation.
+ * <p>Assumes that all arguments of the function are Tensors which will be batched
+ * along their first dimension.
+ * <p>Arguments that are captured, are not batched. The session.run call which does
+ * the concatenation, will use the values of the captured tensors available to it.
+ * Therefore, typical uses of captured tensors should involve values which remain
+ * unchanged across session.run calls. Inference is a good example of this.
+ * <p>SparseTensor is not supported. The return value of the decorated function
+ * must be a Tensor or a list/tuple of Tensors.
+ */
+@Operator
+public final class BatchFunction extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "BatchFunction";
+
+  private List<Output<?>> outTensors;
+
+  @SuppressWarnings("unchecked")
+  private BatchFunction(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outTensorsLength = operation.outputListLength("out_tensors");
+    outTensors = Arrays.asList(operation.outputList(outputIdx, outTensorsLength));
+    outputIdx += outTensorsLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new BatchFunction operation.
+   *
+   * @param scope current scope
+   * @param inTensors The tensors to be batched.
+   * @param capturedTensors The tensors which are captured in the function, and don't need
+   * to be batched.
+   * @param f the value of the f property
+   * @param numBatchThreads Number of scheduling threads for processing batches of work.
+   * Determines the number of batches processed in parallel.
+   * @param maxBatchSize Batch sizes will never be bigger than this.
+   * @param batchTimeoutMicros Maximum number of microseconds to wait before outputting
+   * an incomplete batch.
+   * @param Tout the types of the output tensors.
+   * @param options carries optional attribute values
+   * @return a new instance of BatchFunction
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static BatchFunction create(Scope scope, Iterable<Operand<?>> inTensors,
+      Iterable<Operand<?>> capturedTensors, ConcreteFunction f, Long numBatchThreads,
+      Long maxBatchSize, Long batchTimeoutMicros, List<Class<? extends TType>> Tout,
+      Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("BatchFunction"));
+    opBuilder.addInputList(Operands.asOutputs(inTensors));
+    opBuilder.addInputList(Operands.asOutputs(capturedTensors));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("num_batch_threads", numBatchThreads);
+    opBuilder.setAttr("max_batch_size", maxBatchSize);
+    opBuilder.setAttr("batch_timeout_micros", batchTimeoutMicros);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxEnqueuedBatches != null) {
+          opBuilder.setAttr("max_enqueued_batches", opts.maxEnqueuedBatches);
+        }
+        if (opts.allowedBatchSizes != null) {
+          long[] allowedBatchSizesArray = new long[opts.allowedBatchSizes.size()];
+          for (int i = 0 ; i < allowedBatchSizesArray.length ; i++) {
+            allowedBatchSizesArray[i] = opts.allowedBatchSizes.get(i);
+          }
+          opBuilder.setAttr("allowed_batch_sizes", allowedBatchSizesArray);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.batchingQueue != null) {
+          opBuilder.setAttr("batching_queue", opts.batchingQueue);
+        }
+        if (opts.enableLargeBatchSplitting != null) {
+          opBuilder.setAttr("enable_large_batch_splitting", opts.enableLargeBatchSplitting);
+        }
+      }
+    }
+    return new BatchFunction(opBuilder.build());
+  }
+
+  /**
+   * Sets the maxEnqueuedBatches option.
+   *
+   * @param maxEnqueuedBatches Maximum number of batches enqueued. Default: 10.
+   * @return this Options instance.
+   */
+  public static Options maxEnqueuedBatches(Long maxEnqueuedBatches) {
+    return new Options().maxEnqueuedBatches(maxEnqueuedBatches);
+  }
+
+  /**
+   * Sets the allowedBatchSizes option.
+   *
+   * @param allowedBatchSizes Optional list of allowed batch sizes. If left empty, does
+   * nothing. Otherwise, supplies a list of batch sizes, causing the op to pad
+   * batches up to one of those sizes. The entries must increase monotonically.
+   * If enable_large_batch_splitting is false (i.e., large-input-split is not
+   * enabled) the final entry must equal max_batch_size.
+   * @return this Options instance.
+   */
+  public static Options allowedBatchSizes(List<Long> allowedBatchSizes) {
+    return new Options().allowedBatchSizes(allowedBatchSizes);
+  }
+
+  /**
+   * Sets the allowedBatchSizes option.
+   *
+   * @param allowedBatchSizes Optional list of allowed batch sizes. If left empty, does
+   * nothing. Otherwise, supplies a list of batch sizes, causing the op to pad
+   * batches up to one of those sizes. The entries must increase monotonically.
+   * If enable_large_batch_splitting is false (i.e., large-input-split is not
+   * enabled) the final entry must equal max_batch_size.
+   * @return this Options instance.
+   */
+  public static Options allowedBatchSizes(Long[] allowedBatchSizes) {
+    return new Options().allowedBatchSizes(allowedBatchSizes);
+  }
+
+  /**
+   * Sets the container option.
+   *
+   * @param container Controls the scope of sharing of this batch.
+   * @return this Options instance.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+
+  /**
+   * Sets the sharedName option.
+   *
+   * @param sharedName Concurrently running instances of batch in the same device with the
+   * same container and shared_name will batch their elements together. If left
+   * empty, the op name will be used as the shared name.
+   * @return this Options instance.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+
+  /**
+   * Sets the batchingQueue option.
+   *
+   * @param batchingQueue the batchingQueue option
+   * @return this Options instance.
+   */
+  public static Options batchingQueue(String batchingQueue) {
+    return new Options().batchingQueue(batchingQueue);
+  }
+
+  /**
+   * Sets the enableLargeBatchSplitting option.
+   *
+   * @param enableLargeBatchSplitting input with a large size (i.e., larger than the largest value of
+   * {@code allowed_batch_sizes}) will be splitted into multiple batches with batch size.
+   * @return this Options instance.
+   */
+  public static Options enableLargeBatchSplitting(Boolean enableLargeBatchSplitting) {
+    return new Options().enableLargeBatchSplitting(enableLargeBatchSplitting);
+  }
+
+  /**
+   * Gets outTensors.
+   * The output tensors.
+   * @return outTensors.
+   */
+  public List<Output<?>> outTensors() {
+    return outTensors;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) outTensors.iterator();
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.BatchFunction}
+   */
+  public static class Options {
+    private Long maxEnqueuedBatches;
+
+    private List<Long> allowedBatchSizes;
+
+    private String container;
+
+    private String sharedName;
+
+    private String batchingQueue;
+
+    private Boolean enableLargeBatchSplitting;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the maxEnqueuedBatches option.
+     *
+     * @param maxEnqueuedBatches Maximum number of batches enqueued. Default: 10.
+     * @return this Options instance.
+     */
+    public Options maxEnqueuedBatches(Long maxEnqueuedBatches) {
+      this.maxEnqueuedBatches = maxEnqueuedBatches;
+      return this;
+    }
+
+    /**
+     * Sets the allowedBatchSizes option.
+     *
+     * @param allowedBatchSizes Optional list of allowed batch sizes. If left empty, does
+     * nothing. Otherwise, supplies a list of batch sizes, causing the op to pad
+     * batches up to one of those sizes. The entries must increase monotonically.
+     * If enable_large_batch_splitting is false (i.e., large-input-split is not
+     * enabled) the final entry must equal max_batch_size.
+     * @return this Options instance.
+     */
+    public Options allowedBatchSizes(List<Long> allowedBatchSizes) {
+      this.allowedBatchSizes = allowedBatchSizes;
+      return this;
+    }
+
+    /**
+     * Sets the allowedBatchSizes option.
+     *
+     * @param allowedBatchSizes Optional list of allowed batch sizes. If left empty, does
+     * nothing. Otherwise, supplies a list of batch sizes, causing the op to pad
+     * batches up to one of those sizes. The entries must increase monotonically.
+     * If enable_large_batch_splitting is false (i.e., large-input-split is not
+     * enabled) the final entry must equal max_batch_size.
+     * @return this Options instance.
+     */
+    public Options allowedBatchSizes(Long... allowedBatchSizes) {
+      this.allowedBatchSizes = Arrays.asList(allowedBatchSizes);
+      return this;
+    }
+
+    /**
+     * Sets the container option.
+     *
+     * @param container Controls the scope of sharing of this batch.
+     * @return this Options instance.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+
+    /**
+     * Sets the sharedName option.
+     *
+     * @param sharedName Concurrently running instances of batch in the same device with the
+     * same container and shared_name will batch their elements together. If left
+     * empty, the op name will be used as the shared name.
+     * @return this Options instance.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+
+    /**
+     * Sets the batchingQueue option.
+     *
+     * @param batchingQueue the batchingQueue option
+     * @return this Options instance.
+     */
+    public Options batchingQueue(String batchingQueue) {
+      this.batchingQueue = batchingQueue;
+      return this;
+    }
+
+    /**
+     * Sets the enableLargeBatchSplitting option.
+     *
+     * @param enableLargeBatchSplitting input with a large size (i.e., larger than the largest value of
+     * {@code allowed_batch_sizes}) will be splitted into multiple batches with batch size.
+     * @return this Options instance.
+     */
+    public Options enableLargeBatchSplitting(Boolean enableLargeBatchSplitting) {
+      this.enableLargeBatchSplitting = enableLargeBatchSplitting;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Case.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Case.java
new file mode 100644
index 00000000000..b137e28f283
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Case.java
@@ -0,0 +1,151 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An n-way switch statement which calls a single branch function.
+ * <pre>
+ * An n-way switch statement, implementing the following:
+ * ```
+ * switch (branch_index) {
+ *   case 0:
+ *     output = branches[0](input);
+ *     break;
+ *   case 1:
+ *     output = branches[1](input);
+ *     break;
+ *   ...
+ *   case [[nbranches-1]]:
+ *   default:
+ *     output = branches[nbranches-1](input);
+ *     break;
+ * }
+ * ```
+ * </pre>
+ *
+ * <p>Selects between {@link StatefulCase} and {@link StatelessCase} based on the statefulness of the function arguments.
+ */
+@Operator
+public interface Case extends Iterable<Operand<TType>> {
+  /**
+   * Factory method to create a class wrapping a new Case operation.
+   *
+   * @param scope current scope
+   * @param branchIndex The branch selector, an int32 Tensor.
+   * @param input A list of input tensors passed to the branch function.
+   * @param Tout A list of output types.
+   * @param branches <pre>
+   *   A list of functions each of which takes 'inputs' and returns a list of
+   *   tensors, whose types are the same as what every other branch returns.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of Case
+   */
+  @Endpoint(
+      describeByClass = true,
+      name = "caseOp"
+  )
+  static Case create(Scope scope, Operand<TInt32> branchIndex, Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, List<ConcreteFunction> branches, Options... options) {
+    boolean isStateful = false;
+    if (branches.stream().anyMatch(x -> x.isStateful())) {
+      isStateful = true;
+    }
+    if (isStateful) {
+      return StatefulCase.create(scope, branchIndex, input, Tout, branches, options);
+    } else {
+      return StatelessCase.create(scope, branchIndex, input, Tout, branches, options);
+    }
+  }
+
+  /**
+   * Sets the outputShapes option.
+   *
+   * @param outputShapes the outputShapes option
+   * @return this Options instance.
+   */
+  static Options outputShapes(List<Shape> outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+
+  /**
+   * Sets the outputShapes option.
+   *
+   * @param outputShapes the outputShapes option
+   * @return this Options instance.
+   */
+  static Options outputShapes(Shape[] outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  List<Output<?>> output();
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  Iterator<Operand<TType>> iterator();
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Case}
+   */
+  class Options {
+    List<Shape> outputShapes;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the outputShapes option.
+     *
+     * @param outputShapes the outputShapes option
+     * @return this Options instance.
+     */
+    public Options outputShapes(List<Shape> outputShapes) {
+      this.outputShapes = outputShapes;
+      return this;
+    }
+
+    /**
+     * Sets the outputShapes option.
+     *
+     * @param outputShapes the outputShapes option
+     * @return this Options instance.
+     */
+    public Options outputShapes(Shape... outputShapes) {
+      this.outputShapes = Arrays.asList(outputShapes);
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/For.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/For.java
new file mode 100644
index 00000000000..96392a1a3cc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/For.java
@@ -0,0 +1,105 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * <pre>
+ *  output = input;
+ *  for i in range(start, limit, delta)
+ *    output = body(i, output);
+ * </pre>
+ */
+@Operator
+public final class For extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "For";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private For(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new For operation.
+   *
+   * @param scope current scope
+   * @param start The lower bound. An int32
+   * @param limit The upper bound. An int32
+   * @param delta The increment. An int32
+   * @param input A list of input tensors whose types are T.
+   * @param body <pre>
+   * A function that takes a list of tensors (int32, T) and returns another
+   * list of tensors (T).
+   * </pre>
+   * @return a new instance of For
+   */
+  @Endpoint(
+      describeByClass = true,
+      name = "forOp"
+  )
+  public static For create(Scope scope, Operand<TInt32> start, Operand<TInt32> limit,
+      Operand<TInt32> delta, Iterable<Operand<?>> input, ConcreteFunction body) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("For"));
+    opBuilder.addInput(start.asOutput());
+    opBuilder.addInput(limit.asOutput());
+    opBuilder.addInput(delta.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("body", body);
+    return new For(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of output tensors whose types are T.
+   * @return output.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/If.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/If.java
new file mode 100644
index 00000000000..2ba19c25f25
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/If.java
@@ -0,0 +1,147 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * output = cond ? then_branch(input) : else_branch(input)
+ *
+ * <p>Selects between {@link StatefulIf} and {@link StatelessIf} based on the statefulness of the function arguments.
+ */
+@Operator
+public interface If extends Iterable<Operand<TType>> {
+  /**
+   * Factory method to create a class wrapping a new If operation.
+   *
+   * @param scope current scope
+   * @param cond <pre>
+   *   A Tensor. If the tensor is a scalar of non-boolean type, the
+   *   scalar is converted to a boolean according to the
+   *   following rule: if the scalar is a numerical value, non-zero means
+   *   `True` and zero means False; if the scalar is a string, non-empty
+   *   means `True` and empty means `False`. If the tensor is not a scalar,
+   *   being empty means False and being non-empty means True.
+   * </pre>
+   * @param input A list of input tensors.
+   * @param Tout A list of output types.
+   * @param thenBranch <pre>
+   *   A function that takes 'inputs' and returns a list of tensors, whose
+   *   types are the same as what else_branch returns.
+   * </pre>
+   * @param elseBranch <pre>
+   * A function that takes 'inputs' and returns a list of tensors, whose
+   * types are the same as what then_branch returns.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of If
+   */
+  @Endpoint(
+      describeByClass = true,
+      name = "ifOp"
+  )
+  static If create(Scope scope, Operand<? extends TType> cond, Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, ConcreteFunction thenBranch, ConcreteFunction elseBranch,
+      Options... options) {
+    boolean isStateful = false;
+    if (thenBranch.isStateful()) {
+      isStateful = true;
+    }
+    if (elseBranch.isStateful()) {
+      isStateful = true;
+    }
+    if (isStateful) {
+      return StatefulIf.create(scope, cond, input, Tout, thenBranch, elseBranch, options);
+    } else {
+      return StatelessIf.create(scope, cond, input, Tout, thenBranch, elseBranch, options);
+    }
+  }
+
+  /**
+   * Sets the outputShapes option.
+   *
+   * @param outputShapes the outputShapes option
+   * @return this Options instance.
+   */
+  static Options outputShapes(List<Shape> outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+
+  /**
+   * Sets the outputShapes option.
+   *
+   * @param outputShapes the outputShapes option
+   * @return this Options instance.
+   */
+  static Options outputShapes(Shape[] outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  List<Output<?>> output();
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  Iterator<Operand<TType>> iterator();
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.If}
+   */
+  class Options {
+    List<Shape> outputShapes;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the outputShapes option.
+     *
+     * @param outputShapes the outputShapes option
+     * @return this Options instance.
+     */
+    public Options outputShapes(List<Shape> outputShapes) {
+      this.outputShapes = outputShapes;
+      return this;
+    }
+
+    /**
+     * Sets the outputShapes option.
+     *
+     * @param outputShapes the outputShapes option
+     * @return this Options instance.
+     */
+    public Options outputShapes(Shape... outputShapes) {
+      this.outputShapes = Arrays.asList(outputShapes);
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapDefun.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapDefun.java
new file mode 100644
index 00000000000..60fb426abd6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapDefun.java
@@ -0,0 +1,158 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Maps a function on the list of tensors unpacked from arguments on dimension 0.
+ * The function given by {@code f} is assumed to be stateless, and is executed
+ * concurrently on all the slices; up to batch_size (i.e. the size of the 0th
+ * dimension of each argument) functions will be scheduled at once.
+ * <p>The {@code max_intra_op_parallelism} attr, which defaults to 1, can be used to
+ * limit the intra op parallelism. To limit inter-op parallelism, a user can
+ * set a private threadpool on the dataset using {@code tf.data.Options}'s
+ * {@code ThreadingOptions}.
+ * <p>Note that this op is not exposed to users directly, but is invoked in tf.data
+ * rewrites.
+ */
+public final class MapDefun extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "MapDefun";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private MapDefun(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new MapDefun operation.
+   *
+   * @param scope current scope
+   * @param arguments <pre>
+   * A list of tensors whose types are `Targuments`, corresponding to the inputs
+   * the function should be mapped over.
+   * </pre>
+   * @param capturedInputs <pre>
+   * A list of tensors whose types are `Tcaptured`, corresponding to the captured
+   * inputs of the defun.
+   * </pre>
+   * @param outputTypes A list of types.
+   * @param outputShapes A list of shapes.
+   * @param f the value of the f property
+   * @param options carries optional attribute values
+   * @return a new instance of MapDefun
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static MapDefun create(Scope scope, Iterable<Operand<?>> arguments,
+      Iterable<Operand<?>> capturedInputs, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, ConcreteFunction f, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("MapDefun"));
+    opBuilder.addInputList(Operands.asOutputs(arguments));
+    opBuilder.addInputList(Operands.asOutputs(capturedInputs));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    opBuilder.setAttr("f", f);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxIntraOpParallelism != null) {
+          opBuilder.setAttr("max_intra_op_parallelism", opts.maxIntraOpParallelism);
+        }
+      }
+    }
+    return new MapDefun(opBuilder.build());
+  }
+
+  /**
+   * Sets the maxIntraOpParallelism option.
+   *
+   * @param maxIntraOpParallelism the maxIntraOpParallelism option
+   * @return this Options instance.
+   */
+  public static Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
+    return new Options().maxIntraOpParallelism(maxIntraOpParallelism);
+  }
+
+  /**
+   * Gets output.
+   * <pre>
+   * A list of output tensors whose types are `output_types` and whose dimensions
+   * 0 are the same as the dimensions 0 of the tensors in `arguments`, and whose
+   * remaining dimensions correspond to those in `output_shapes`.
+   * </pre>
+   * @return output.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MapDefun}
+   */
+  public static class Options {
+    private Long maxIntraOpParallelism;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the maxIntraOpParallelism option.
+     *
+     * @param maxIntraOpParallelism the maxIntraOpParallelism option
+     * @return this Options instance.
+     */
+    public Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
+      this.maxIntraOpParallelism = maxIntraOpParallelism;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PartitionedCall.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PartitionedCall.java
new file mode 100644
index 00000000000..7deb23131c8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PartitionedCall.java
@@ -0,0 +1,156 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Output;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * returns {@code f(inputs)}, where {@code f}'s body is placed and partitioned.
+ *
+ * <p>Selects between {@link StatefulPartitionedCall} and {@link StatelessPartitionedCall} based on the statefulness of the function arguments.
+ */
+@Operator
+public interface PartitionedCall extends Iterable<Operand<TType>> {
+  /**
+   * Factory method to create a class wrapping a new StatefulPartitionedCall operation.
+   *
+   * @param scope current scope
+   * @param args A list of input tensors.
+   * @param Tout A list of output types.
+   * @param f <pre>
+   *   A function that takes 'args', a list of tensors, and returns 'output',
+   *   another list of tensors. Input and output types are specified by 'Tin'
+   *   and 'Tout'. The function body of f will be placed and partitioned across
+   *   devices, setting this op apart from the regular Call op. This op is
+   *   stateful.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of PartitionedCall
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  static PartitionedCall create(Scope scope, Iterable<Operand<?>> args,
+      List<Class<? extends TType>> Tout, ConcreteFunction f, Options... options) {
+    boolean isStateful = false;
+    if (f.isStateful()) {
+      isStateful = true;
+    }
+    if (isStateful) {
+      return StatefulPartitionedCall.create(scope, args, Tout, f, options);
+    } else {
+      return StatelessPartitionedCall.create(scope, args, Tout, f, options);
+    }
+  }
+
+  /**
+   * Sets the config option.
+   *
+   * @param config the config option
+   * @return this Options instance.
+   */
+  static Options config(String config) {
+    return new Options().config(config);
+  }
+
+  /**
+   * Sets the configProto option.
+   *
+   * @param configProto the configProto option
+   * @return this Options instance.
+   */
+  static Options configProto(String configProto) {
+    return new Options().configProto(configProto);
+  }
+
+  /**
+   * Sets the executorType option.
+   *
+   * @param executorType the executorType option
+   * @return this Options instance.
+   */
+  static Options executorType(String executorType) {
+    return new Options().executorType(executorType);
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  List<Output<?>> output();
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  Iterator<Operand<TType>> iterator();
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.PartitionedCall}
+   */
+  class Options {
+    String config;
+
+    String configProto;
+
+    String executorType;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the config option.
+     *
+     * @param config the config option
+     * @return this Options instance.
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+
+    /**
+     * Sets the configProto option.
+     *
+     * @param configProto the configProto option
+     * @return this Options instance.
+     */
+    public Options configProto(String configProto) {
+      this.configProto = configProto;
+      return this;
+    }
+
+    /**
+     * Sets the executorType option.
+     *
+     * @param executorType the executorType option
+     * @return this Options instance.
+     */
+    public Options executorType(String executorType) {
+      this.executorType = executorType;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteCall.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteCall.java
new file mode 100644
index 00000000000..3848a6ff0e7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteCall.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Runs function {@code f} on a remote device indicated by {@code target}.
+ */
+@Operator
+public final class RemoteCall extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "RemoteCall";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private RemoteCall(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new RemoteCall operation.
+   *
+   * @param scope current scope
+   * @param target A fully specified device name where we want to run the function.
+   * @param args A list of arguments for the function.
+   * @param Tout The type list for the return values.
+   * @param f The function to run remotely.
+   * @return a new instance of RemoteCall
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static RemoteCall create(Scope scope, Operand<TString> target, Iterable<Operand<?>> args,
+      List<Class<? extends TType>> Tout, ConcreteFunction f) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("RemoteCall"));
+    opBuilder.addInput(target.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(args));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    opBuilder.setAttr("f", f);
+    return new RemoteCall(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulCase.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulCase.java
new file mode 100644
index 00000000000..817539cb187
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulCase.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An n-way switch statement which calls a single branch function.
+ * <pre>
+ * An n-way switch statement, implementing the following:
+ * ```
+ * switch (branch_index) {
+ *   case 0:
+ *     output = branches[0](input);
+ *     break;
+ *   case 1:
+ *     output = branches[1](input);
+ *     break;
+ *   ...
+ *   case [[nbranches-1]]:
+ *   default:
+ *     output = branches[nbranches-1](input);
+ *     break;
+ * }
+ * ```
+ * </pre>
+ */
+@Operator
+public final class StatefulCase extends RawOp implements Case {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "Case";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private StatefulCase(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new Case operation.
+   *
+   * @param scope current scope
+   * @param branchIndex The branch selector, an int32 Tensor.
+   * @param input A list of input tensors passed to the branch function.
+   * @param Tout A list of output types.
+   * @param branches <pre>
+   *   A list of functions each of which takes 'inputs' and returns a list of
+   *   tensors, whose types are the same as what every other branch returns.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatefulCase
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static StatefulCase create(Scope scope, Operand<TInt32> branchIndex,
+      Iterable<Operand<?>> input, List<Class<? extends TType>> Tout,
+      List<ConcreteFunction> branches, Case.Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("StatefulCase"));
+    opBuilder.addInput(branchIndex.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    ConcreteFunction[] branchesArray = new ConcreteFunction[branches.size()];
+    for (int i = 0 ; i < branchesArray.length ; i++) {
+      branchesArray[i] = branches.get(i);
+    }
+    opBuilder.setAttr("branches", branchesArray);
+    if (options != null) {
+      for (Case.Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0 ; i < outputShapesArray.length ; i++) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+      }
+    }
+    return new StatefulCase(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  @Override
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulIf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulIf.java
new file mode 100644
index 00000000000..e1f13750f3e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulIf.java
@@ -0,0 +1,124 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * output = cond ? then_branch(input) : else_branch(input)
+ */
+@Operator
+public final class StatefulIf extends RawOp implements If {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "If";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private StatefulIf(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new If operation.
+   *
+   * @param scope current scope
+   * @param cond <pre>
+   *   A Tensor. If the tensor is a scalar of non-boolean type, the
+   *   scalar is converted to a boolean according to the
+   *   following rule: if the scalar is a numerical value, non-zero means
+   *   `True` and zero means False; if the scalar is a string, non-empty
+   *   means `True` and empty means `False`. If the tensor is not a scalar,
+   *   being empty means False and being non-empty means True.
+   * </pre>
+   * @param input A list of input tensors.
+   * @param Tout A list of output types.
+   * @param thenBranch <pre>
+   *   A function that takes 'inputs' and returns a list of tensors, whose
+   *   types are the same as what else_branch returns.
+   * </pre>
+   * @param elseBranch <pre>
+   * A function that takes 'inputs' and returns a list of tensors, whose
+   * types are the same as what then_branch returns.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatefulIf
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static StatefulIf create(Scope scope, Operand<? extends TType> cond,
+      Iterable<Operand<?>> input, List<Class<? extends TType>> Tout, ConcreteFunction thenBranch,
+      ConcreteFunction elseBranch, If.Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("StatefulIf"));
+    opBuilder.addInput(cond.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    opBuilder.setAttr("then_branch", thenBranch);
+    opBuilder.setAttr("else_branch", elseBranch);
+    if (options != null) {
+      for (If.Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0 ; i < outputShapesArray.length ; i++) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+      }
+    }
+    return new StatefulIf(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  @Override
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulPartitionedCall.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulPartitionedCall.java
new file mode 100644
index 00000000000..ce789a235f9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulPartitionedCall.java
@@ -0,0 +1,113 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * returns {@code f(inputs)}, where {@code f}'s body is placed and partitioned.
+ */
+@Operator
+public final class StatefulPartitionedCall extends RawOp implements PartitionedCall {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "StatefulPartitionedCall";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private StatefulPartitionedCall(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new StatefulPartitionedCall operation.
+   *
+   * @param scope current scope
+   * @param args A list of input tensors.
+   * @param Tout A list of output types.
+   * @param f <pre>
+   *   A function that takes 'args', a list of tensors, and returns 'output',
+   *   another list of tensors. Input and output types are specified by 'Tin'
+   *   and 'Tout'. The function body of f will be placed and partitioned across
+   *   devices, setting this op apart from the regular Call op. This op is
+   *   stateful.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatefulPartitionedCall
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static StatefulPartitionedCall create(Scope scope, Iterable<Operand<?>> args,
+      List<Class<? extends TType>> Tout, ConcreteFunction f, PartitionedCall.Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("StatefulPartitionedCall"));
+    opBuilder.addInputList(Operands.asOutputs(args));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    opBuilder.setAttr("f", f);
+    if (options != null) {
+      for (PartitionedCall.Options opts : options) {
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+        if (opts.configProto != null) {
+          opBuilder.setAttr("config_proto", opts.configProto);
+        }
+        if (opts.executorType != null) {
+          opBuilder.setAttr("executor_type", opts.executorType);
+        }
+      }
+    }
+    return new StatefulPartitionedCall(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  @Override
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulWhile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulWhile.java
new file mode 100644
index 00000000000..5d8fdcff241
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatefulWhile.java
@@ -0,0 +1,121 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * output = input; While (Cond(output)) { output = Body(output) }
+ */
+@Operator
+public final class StatefulWhile extends RawOp implements While {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "While";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private StatefulWhile(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new While operation.
+   *
+   * @param scope current scope
+   * @param input A list of input tensors whose types are T.
+   * @param cond <pre>
+   *   A function takes 'input' and returns a tensor.  If the tensor is
+   *   a scalar of non-boolean, the scalar is converted to a boolean
+   *   according to the following rule: if the scalar is a numerical
+   *   value, non-zero means True and zero means False; if the scalar is
+   *   a string, non-empty means True and empty means False. If the
+   *   tensor is not a scalar, non-emptiness means True and False
+   *   otherwise.
+   * </pre>
+   * @param body <pre>
+   *   A function that takes a list of tensors and returns another
+   *   list of tensors. Both lists have the same types as specified
+   *   by T.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatefulWhile
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static StatefulWhile create(Scope scope, Iterable<Operand<?>> input, ConcreteFunction cond,
+      ConcreteFunction body, While.Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("StatefulWhile"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("cond", cond);
+    opBuilder.setAttr("body", body);
+    if (options != null) {
+      for (While.Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0 ; i < outputShapesArray.length ; i++) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+        if (opts.parallelIterations != null) {
+          opBuilder.setAttr("parallel_iterations", opts.parallelIterations);
+        }
+      }
+    }
+    return new StatefulWhile(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of output tensors whose types are T.
+   * @return output.
+   */
+  @Override
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessCase.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessCase.java
new file mode 100644
index 00000000000..31326021300
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessCase.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An n-way switch statement which calls a single branch function.
+ * <pre>
+ * An n-way switch statement, implementing the following:
+ * ```
+ * switch (branch_index) {
+ *   case 0:
+ *     output = branches[0](input);
+ *     break;
+ *   case 1:
+ *     output = branches[1](input);
+ *     break;
+ *   ...
+ *   case [[nbranches-1]]:
+ *   default:
+ *     output = branches[nbranches-1](input);
+ *     break;
+ * }
+ * ```
+ *
+ * This should only be used when the none of branches has stateful ops.
+ * </pre>
+ */
+public final class StatelessCase extends RawOp implements Case {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "StatelessCase";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private StatelessCase(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new StatelessCase operation.
+   *
+   * @param scope current scope
+   * @param branchIndex The branch selector, an int32 Tensor.
+   * @param input A list of input tensors passed to the branch function.
+   * @param Tout A list of output types.
+   * @param branches <pre>
+   *   A list of functions each of which takes 'inputs' and returns a list of
+   *   tensors, whose types are the same as what every other branch returns.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatelessCase
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static StatelessCase create(Scope scope, Operand<TInt32> branchIndex,
+      Iterable<Operand<?>> input, List<Class<? extends TType>> Tout,
+      List<ConcreteFunction> branches, Case.Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("StatelessCase"));
+    opBuilder.addInput(branchIndex.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    ConcreteFunction[] branchesArray = new ConcreteFunction[branches.size()];
+    for (int i = 0 ; i < branchesArray.length ; i++) {
+      branchesArray[i] = branches.get(i);
+    }
+    opBuilder.setAttr("branches", branchesArray);
+    if (options != null) {
+      for (Case.Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0 ; i < outputShapesArray.length ; i++) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+      }
+    }
+    return new StatelessCase(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  @Override
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessIf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessIf.java
new file mode 100644
index 00000000000..83589e73a34
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessIf.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * output = cond ? then_branch(input) : else_branch(input)
+ */
+@Operator
+public final class StatelessIf extends RawOp implements If {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "StatelessIf";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private StatelessIf(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new StatelessIf operation.
+   *
+   * @param scope current scope
+   * @param cond <pre>
+   *   A Tensor. If the tensor is a scalar of non-boolean type, the
+   *   scalar is converted to a boolean according to the
+   *   following rule: if the scalar is a numerical value, non-zero means
+   *   `True` and zero means False; if the scalar is a string, non-empty
+   *   means `True` and empty means `False`. If the tensor is not a scalar,
+   *   being empty means False and being non-empty means True.
+   *
+   *   This should only be used when the if then/else body functions do not
+   *   have stateful ops.
+   * </pre>
+   * @param input A list of input tensors.
+   * @param Tout A list of output types.
+   * @param thenBranch <pre>
+   *   A function that takes 'inputs' and returns a list of tensors, whose
+   *   types are the same as what else_branch returns.
+   * </pre>
+   * @param elseBranch <pre>
+   * A function that takes 'inputs' and returns a list of tensors, whose
+   * types are the same as what then_branch returns.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatelessIf
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static StatelessIf create(Scope scope, Operand<? extends TType> cond,
+      Iterable<Operand<?>> input, List<Class<? extends TType>> Tout, ConcreteFunction thenBranch,
+      ConcreteFunction elseBranch, If.Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("StatelessIf"));
+    opBuilder.addInput(cond.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    opBuilder.setAttr("then_branch", thenBranch);
+    opBuilder.setAttr("else_branch", elseBranch);
+    if (options != null) {
+      for (If.Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0 ; i < outputShapesArray.length ; i++) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+      }
+    }
+    return new StatelessIf(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  @Override
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessPartitionedCall.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessPartitionedCall.java
new file mode 100644
index 00000000000..f497454692b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessPartitionedCall.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * returns {@code f(inputs)}, where {@code f}'s body is placed and partitioned.
+ */
+@Operator
+public final class StatelessPartitionedCall extends RawOp implements PartitionedCall {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "PartitionedCall";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private StatelessPartitionedCall(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new PartitionedCall operation.
+   *
+   * @param scope current scope
+   * @param args A list of input tensors.
+   * @param Tout A list of output types.
+   * @param f <pre>
+   *   A function that takes 'args', a list of tensors, and returns 'output',
+   *   another list of tensors. Input and output types are specified by 'Tin'
+   *   and 'Tout'. The function body of f will be placed and partitioned across
+   *   devices, setting this op apart from the regular Call op.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatelessPartitionedCall
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static StatelessPartitionedCall create(Scope scope, Iterable<Operand<?>> args,
+      List<Class<? extends TType>> Tout, ConcreteFunction f, PartitionedCall.Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("StatelessPartitionedCall"));
+    opBuilder.addInputList(Operands.asOutputs(args));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    opBuilder.setAttr("f", f);
+    if (options != null) {
+      for (PartitionedCall.Options opts : options) {
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+        if (opts.configProto != null) {
+          opBuilder.setAttr("config_proto", opts.configProto);
+        }
+        if (opts.executorType != null) {
+          opBuilder.setAttr("executor_type", opts.executorType);
+        }
+      }
+    }
+    return new StatelessPartitionedCall(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of return values.
+   * @return output.
+   */
+  @Override
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessWhile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessWhile.java
new file mode 100644
index 00000000000..77c067dfa42
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StatelessWhile.java
@@ -0,0 +1,124 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * output = input; While (Cond(output)) { output = Body(output) }
+ */
+@Operator
+public final class StatelessWhile extends RawOp implements While {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "StatelessWhile";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private StatelessWhile(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new StatelessWhile operation.
+   *
+   * @param scope current scope
+   * @param input A list of input tensors whose types are T.
+   * @param cond <pre>
+   *   A function takes 'input' and returns a tensor.  If the tensor is
+   *   a scalar of non-boolean, the scalar is converted to a boolean
+   *   according to the following rule: if the scalar is a numerical
+   *   value, non-zero means True and zero means False; if the scalar is
+   *   a string, non-empty means True and empty means False. If the
+   *   tensor is not a scalar, non-emptiness means True and False
+   *   otherwise.
+   *
+   *   This should only be used when the while condition and body functions
+   *   do not have stateful ops.
+   * </pre>
+   * @param body <pre>
+   *   A function that takes a list of tensors and returns another
+   *   list of tensors. Both lists have the same types as specified
+   *   by T.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of StatelessWhile
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static StatelessWhile create(Scope scope, Iterable<Operand<?>> input,
+      ConcreteFunction cond, ConcreteFunction body, While.Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("StatelessWhile"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("cond", cond);
+    opBuilder.setAttr("body", body);
+    if (options != null) {
+      for (While.Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0 ; i < outputShapesArray.length ; i++) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+        if (opts.parallelIterations != null) {
+          opBuilder.setAttr("parallel_iterations", opts.parallelIterations);
+        }
+      }
+    }
+    return new StatelessWhile(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of output tensors whose types are T.
+   * @return output.
+   */
+  @Override
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/While.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/While.java
new file mode 100644
index 00000000000..714929f426c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/While.java
@@ -0,0 +1,166 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * output = input; While (Cond(output)) { output = Body(output) }
+ *
+ * <p>Selects between {@link StatefulWhile} and {@link StatelessWhile} based on the statefulness of the function arguments.
+ */
+@Operator
+public interface While extends Iterable<Operand<TType>> {
+  /**
+   * Factory method to create a class wrapping a new While operation.
+   *
+   * @param scope current scope
+   * @param input A list of input tensors whose types are T.
+   * @param cond <pre>
+   *   A function takes 'input' and returns a tensor.  If the tensor is
+   *   a scalar of non-boolean, the scalar is converted to a boolean
+   *   according to the following rule: if the scalar is a numerical
+   *   value, non-zero means True and zero means False; if the scalar is
+   *   a string, non-empty means True and empty means False. If the
+   *   tensor is not a scalar, non-emptiness means True and False
+   *   otherwise.
+   * </pre>
+   * @param body <pre>
+   *   A function that takes a list of tensors and returns another
+   *   list of tensors. Both lists have the same types as specified
+   *   by T.
+   * </pre>
+   * @param options carries optional attribute values
+   * @return a new instance of While
+   */
+  @Endpoint(
+      describeByClass = true,
+      name = "whileOp"
+  )
+  static While create(Scope scope, Iterable<Operand<?>> input, ConcreteFunction cond,
+      ConcreteFunction body, Options... options) {
+    boolean isStateful = false;
+    if (cond.isStateful()) {
+      isStateful = true;
+    }
+    if (body.isStateful()) {
+      isStateful = true;
+    }
+    if (isStateful) {
+      return StatefulWhile.create(scope, input, cond, body, options);
+    } else {
+      return StatelessWhile.create(scope, input, cond, body, options);
+    }
+  }
+
+  /**
+   * Sets the outputShapes option.
+   *
+   * @param outputShapes the outputShapes option
+   * @return this Options instance.
+   */
+  static Options outputShapes(List<Shape> outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+
+  /**
+   * Sets the outputShapes option.
+   *
+   * @param outputShapes the outputShapes option
+   * @return this Options instance.
+   */
+  static Options outputShapes(Shape[] outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+
+  /**
+   * Sets the parallelIterations option.
+   *
+   * @param parallelIterations the parallelIterations option
+   * @return this Options instance.
+   */
+  static Options parallelIterations(Long parallelIterations) {
+    return new Options().parallelIterations(parallelIterations);
+  }
+
+  /**
+   * Gets output.
+   * A list of output tensors whose types are T.
+   * @return output.
+   */
+  List<Output<?>> output();
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  Iterator<Operand<TType>> iterator();
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.While}
+   */
+  class Options {
+    List<Shape> outputShapes;
+
+    Long parallelIterations;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the outputShapes option.
+     *
+     * @param outputShapes the outputShapes option
+     * @return this Options instance.
+     */
+    public Options outputShapes(List<Shape> outputShapes) {
+      this.outputShapes = outputShapes;
+      return this;
+    }
+
+    /**
+     * Sets the outputShapes option.
+     *
+     * @param outputShapes the outputShapes option
+     * @return this Options instance.
+     */
+    public Options outputShapes(Shape... outputShapes) {
+      this.outputShapes = Arrays.asList(outputShapes);
+      return this;
+    }
+
+    /**
+     * Sets the parallelIterations option.
+     *
+     * @param parallelIterations the parallelIterations option
+     * @return this Options instance.
+     */
+    public Options parallelIterations(Long parallelIterations) {
+      this.parallelIterations = parallelIterations;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaConvV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaConvV2.java
deleted file mode 100644
index fe95ace9832..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaConvV2.java
+++ /dev/null
@@ -1,108 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. 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.
-You may obtain a copy of the License at
-
-    http://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.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA ConvGeneralDilated operator, documented at
- * https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
- * .
- *
- * @param <W> data type for {@code output} output
- */
-@Operator
-public final class XlaConvV2<W extends TType> extends RawOp implements Operand<W> {
-  /**
-   * The name of this op, as known by TensorFlow core engine
-   */
-  public static final String OP_NAME = "XlaConvV2";
-
-  private Output<W> output;
-
-  private XlaConvV2(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-
-  /**
-   * Factory method to create a class wrapping a new XlaConvV2 operation.
-   *
-   * @param scope current scope
-   * @param lhs the input tensor
-   * @param rhs the kernel tensor
-   * @param windowStrides the inter-window strides
-   * @param padding the padding to apply at the start and end of each input dimensions
-   * @param lhsDilation dilation to apply between input elements
-   * @param rhsDilation dilation to apply between kernel elements
-   * @param featureGroupCount number of feature groups for grouped convolution.
-   * @param dimensionNumbers a serialized xla::ConvolutionDimensionNumbers proto.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @param preferredElementType The type of the tensor.
-   * @param <W> data type for {@code XlaConvV2} output and operands
-   * @param <V> data type for {@code XlaConvV2} output and operands
-   * @return a new instance of XlaConvV2
-   */
-  @Endpoint(
-      describeByClass = true
-  )
-  public static <W extends TType, V extends TNumber> XlaConvV2<W> create(Scope scope,
-      Operand<? extends TType> lhs, Operand<? extends TType> rhs, Operand<V> windowStrides,
-      Operand<V> padding, Operand<V> lhsDilation, Operand<V> rhsDilation,
-      Operand<V> featureGroupCount, String dimensionNumbers, String precisionConfig,
-      Class<W> preferredElementType) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaConvV2"));
-    opBuilder.addInput(lhs.asOutput());
-    opBuilder.addInput(rhs.asOutput());
-    opBuilder.addInput(windowStrides.asOutput());
-    opBuilder.addInput(padding.asOutput());
-    opBuilder.addInput(lhsDilation.asOutput());
-    opBuilder.addInput(rhsDilation.asOutput());
-    opBuilder.addInput(featureGroupCount.asOutput());
-    opBuilder = scope.apply(opBuilder);
-    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
-    opBuilder.setAttr("precision_config", precisionConfig);
-    opBuilder.setAttr("preferred_element_type", Operands.toDataType(preferredElementType));
-    return new XlaConvV2<>(opBuilder.build());
-  }
-
-  /**
-   * Gets output.
-   *
-   * @return output.
-   */
-  public Output<W> output() {
-    return output;
-  }
-
-  @Override
-  public Output<W> asOutput() {
-    return output;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertCardinalityDataset.java
similarity index 96%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertCardinalityDataset.java
index 90ff976d9e5..f23c263c973 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertCardinalityDataset.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.data.experimental;
+package org.tensorflow.op.data;
 
 import java.util.List;
 import org.tensorflow.Operand;
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * The AssertCardinalityDataset operation
  */
+@Operator(
+    group = "data"
+)
 public final class AssertCardinalityDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
index 55d04a508aa..193a9c7e21f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
@@ -40,6 +41,9 @@
  * means that the check happens <em>after</em> any static optimizations are applied
  * to the dataset graph.
  */
+@Operator(
+    group = "data"
+)
 public final class AssertNextDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
index 901aa906739..fd68bffe522 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
@@ -39,6 +40,9 @@
  * <p>This dataset will throw a NotFound error if we cannot shard the dataset
  * automatically.
  */
+@Operator(
+    group = "data"
+)
 public final class AutoShardDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
index d2965b028d7..2bea64b5ee9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
  * Records the bytes size of each element of {@code input_dataset} in a StatsAggregator.
  */
+@Operator(
+    group = "data"
+)
 public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
index 6eaaf8f621c..321cfb95669 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
@@ -27,19 +27,23 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
- * The CSVDataset operation
+ * The CSVDatasetV2 operation
  */
+@Operator(
+    group = "data"
+)
 public final class CSVDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "CSVDataset";
+  public static final String OP_NAME = "CSVDatasetV2";
 
   private Output<? extends TType> handle;
 
@@ -51,7 +55,7 @@ private CSVDataset(Operation operation) {
   }
 
   /**
-   * Factory method to create a class wrapping a new CSVDataset operation.
+   * Factory method to create a class wrapping a new CSVDatasetV2 operation.
    *
    * @param scope current scope
    * @param filenames the filenames value
@@ -63,6 +67,7 @@ private CSVDataset(Operation operation) {
    * @param naValue the naValue value
    * @param selectCols the selectCols value
    * @param recordDefaults the recordDefaults value
+   * @param excludeCols the excludeCols value
    * @param outputShapes the value of the outputShapes property
    * @return a new instance of CSVDataset
    */
@@ -72,7 +77,8 @@ private CSVDataset(Operation operation) {
   public static CSVDataset create(Scope scope, Operand<TString> filenames,
       Operand<TString> compressionType, Operand<TInt64> bufferSize, Operand<TBool> header,
       Operand<TString> fieldDelim, Operand<TBool> useQuoteDelim, Operand<TString> naValue,
-      Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
+      Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, Operand<TInt64> excludeCols,
+      List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("CSVDataset"));
     opBuilder.addInput(filenames.asOutput());
     opBuilder.addInput(compressionType.asOutput());
@@ -83,6 +89,7 @@ public static CSVDataset create(Scope scope, Operand<TString> filenames,
     opBuilder.addInput(naValue.asOutput());
     opBuilder.addInput(selectCols.asOutput());
     opBuilder.addInputList(Operands.asOutputs(recordDefaults));
+    opBuilder.addInput(excludeCols.asOutput());
     opBuilder = scope.apply(opBuilder);
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
     for (int i = 0 ; i < outputShapesArray.length ; i++) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
index b610b7af19a..b9903b62d21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
@@ -27,21 +27,21 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
- * Creates a dataset that caches elements from {@code input_dataset}.
- * A CacheDataset will iterate over the input_dataset, and store tensors. If the
- * cache already exists, the cache will be used. If the cache is inappropriate
- * (e.g. cannot be opened, contains tensors of the wrong shape / size), an error
- * will the returned when used.
+ * The CacheDatasetV2 operation
  */
+@Operator(
+    group = "data"
+)
 public final class CacheDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "CacheDataset";
+  public static final String OP_NAME = "CacheDatasetV2";
 
   private Output<? extends TType> handle;
 
@@ -53,12 +53,12 @@ private CacheDataset(Operation operation) {
   }
 
   /**
-   * Factory method to create a class wrapping a new CacheDataset operation.
+   * Factory method to create a class wrapping a new CacheDatasetV2 operation.
    *
    * @param scope current scope
    * @param inputDataset the inputDataset value
-   * @param filename A path on the filesystem where we should cache the dataset. Note: this
-   * will be a directory.
+   * @param filename the filename value
+   * @param cache the cache value
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
    * @return a new instance of CacheDataset
@@ -67,11 +67,12 @@ private CacheDataset(Operation operation) {
       describeByClass = true
   )
   public static CacheDataset create(Scope scope, Operand<? extends TType> inputDataset,
-      Operand<TString> filename, List<Class<? extends TType>> outputTypes,
-      List<Shape> outputShapes) {
+      Operand<TString> filename, Operand<? extends TType> cache,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("CacheDataset"));
     opBuilder.addInput(inputDataset.asOutput());
     opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(cache.asOutput());
     opBuilder = scope.apply(opBuilder);
     opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestBranchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestBranchDataset.java
new file mode 100644
index 00000000000..5623bcf21b5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestBranchDataset.java
@@ -0,0 +1,119 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * The ChooseFastestBranchDataset operation
+ */
+@Operator(
+    group = "data"
+)
+public final class ChooseFastestBranchDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ChooseFastestBranchDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private ChooseFastestBranchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ChooseFastestBranchDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param ratioNumerator the ratioNumerator value
+   * @param ratioDenominator the ratioDenominator value
+   * @param otherArguments the otherArguments value
+   * @param numElementsPerBranch the value of the numElementsPerBranch property
+   * @param branches the value of the branches property
+   * @param otherArgumentsLengths the value of the otherArgumentsLengths property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of ChooseFastestBranchDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static ChooseFastestBranchDataset create(Scope scope,
+      Operand<? extends TType> inputDataset, Operand<TInt64> ratioNumerator,
+      Operand<TInt64> ratioDenominator, Iterable<Operand<?>> otherArguments,
+      Long numElementsPerBranch, List<ConcreteFunction> branches, List<Long> otherArgumentsLengths,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ChooseFastestBranchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(ratioNumerator.asOutput());
+    opBuilder.addInput(ratioDenominator.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("num_elements_per_branch", numElementsPerBranch);
+    ConcreteFunction[] branchesArray = new ConcreteFunction[branches.size()];
+    for (int i = 0 ; i < branchesArray.length ; i++) {
+      branchesArray[i] = branches.get(i);
+    }
+    opBuilder.setAttr("branches", branchesArray);
+    long[] otherArgumentsLengthsArray = new long[otherArgumentsLengths.size()];
+    for (int i = 0 ; i < otherArgumentsLengthsArray.length ; i++) {
+      otherArgumentsLengthsArray[i] = otherArgumentsLengths.get(i);
+    }
+    opBuilder.setAttr("other_arguments_lengths", otherArgumentsLengthsArray);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new ChooseFastestBranchDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
index d91ed316fe5..e5c26edcba8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * The ChooseFastestDataset operation
  */
+@Operator(
+    group = "data"
+)
 public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CompressElement.java
similarity index 98%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CompressElement.java
index 30caf1434f3..10be8060ede 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CompressElement.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.data.experimental;
+package org.tensorflow.op.data;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/DataServiceDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DataServiceDatasetV2.java
similarity index 96%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/DataServiceDatasetV2.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DataServiceDatasetV2.java
index 96059a264e4..69c97855be0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/DataServiceDatasetV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DataServiceDatasetV2.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.rawops;
+package org.tensorflow.op.data;
 
 import java.util.List;
 import org.tensorflow.Operand;
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
@@ -34,6 +35,9 @@
 /**
  * Creates a dataset that reads data from the tf.data service.
  */
+@Operator(
+    group = "data"
+)
 public final class DataServiceDatasetV2 extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
@@ -141,7 +145,7 @@ public Output<TType> asOutput() {
   }
 
   /**
-   * Optional attributes for {@link org.tensorflow.op.rawops.DataServiceDatasetV2}
+   * Optional attributes for {@link org.tensorflow.op.data.DataServiceDatasetV2}
    */
   public static class Options {
     private Long taskRefreshIntervalHintMs;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
index 2d1b8a24ba5..8ec90a071c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
@@ -24,6 +24,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
@@ -31,6 +32,9 @@
  * Returns the cardinality of {@code input_dataset}.
  * Returns the cardinality of {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class DatasetCardinality extends RawOp implements Operand<TInt64> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
index 0064b4a1efa..d54a9ce08de 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
@@ -24,6 +24,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
@@ -31,6 +32,9 @@
  * Creates a dataset from the given {@code graph_def}.
  * Creates a dataset from the provided {@code graph_def}.
  */
+@Operator(
+    group = "data"
+)
 public final class DatasetFromGraph extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
index 1f550abaf7d..d799766128b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
@@ -24,6 +24,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
@@ -31,6 +32,9 @@
  * Returns a serialized GraphDef representing {@code input_dataset}.
  * Returns a graph representation for {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class DatasetToGraph extends RawOp implements Operand<TString> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
index dffd7bebf02..9b302486b3b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
@@ -29,11 +29,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Outputs the single element from the given dataset.
  */
+@Operator(
+    group = "data"
+)
 public final class DatasetToSingleElement extends RawOp implements Iterable<Operand<TType>> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
index 4ccdc21bafa..9e8b386200f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
@@ -23,12 +23,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
  * Writes the given dataset to the given file using the TFRecord format.
  */
+@Operator(
+    group = "data"
+)
 public final class DatasetToTfRecord extends RawOp {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
index 07d490b1008..ecea765507d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches input elements into a SparseTensor.
  */
+@Operator(
+    group = "data"
+)
 public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
index 5f4f262a18c..58637eb9f7f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * A substitute for {@code InterleaveDataset} on a fixed list of {@code N} datasets.
  */
+@Operator(
+    group = "data"
+)
 public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DummyIterationCounter.java
similarity index 98%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DummyIterationCounter.java
index d7e6e4ce05e..3b6fcd763b7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DummyIterationCounter.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.data.experimental;
+package org.tensorflow.op.data;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
index 22ef6836c25..0dfdd3eb57d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset containing elements of first component of {@code input_dataset} having true in the last component.
  */
+@Operator(
+    group = "data"
+)
 public final class FilterByLastComponentDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterDataset.java
similarity index 53%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDatasetV2.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterDataset.java
index 47db26f6dd6..75f9b98aa8a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDatasetV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterDataset.java
@@ -18,6 +18,7 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
+import org.tensorflow.ConcreteFunction;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -27,72 +28,66 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
- * The CSVDatasetV2 operation
+ * Creates a dataset containing elements of {@code input_dataset} matching {@code predicate}.
+ * The {@code predicate} function must return a scalar boolean and accept the
+ * following arguments:
+ * <ul>
+ * <li>One tensor for each component of an element of {@code input_dataset}.</li>
+ * <li>One tensor for each value in {@code other_arguments}.</li>
+ * </ul>
  */
-public final class CSVDatasetV2 extends RawOp implements Operand<TType> {
+@Operator(
+    group = "data"
+)
+public final class FilterDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "CSVDatasetV2";
+  public static final String OP_NAME = "FilterDataset";
 
   private Output<? extends TType> handle;
 
   @SuppressWarnings("unchecked")
-  private CSVDatasetV2(Operation operation) {
+  private FilterDataset(Operation operation) {
     super(operation);
     int outputIdx = 0;
     handle = operation.output(outputIdx++);
   }
 
   /**
-   * Factory method to create a class wrapping a new CSVDatasetV2 operation.
+   * Factory method to create a class wrapping a new FilterDataset operation.
    *
    * @param scope current scope
-   * @param filenames the filenames value
-   * @param compressionType the compressionType value
-   * @param bufferSize the bufferSize value
-   * @param header the header value
-   * @param fieldDelim the fieldDelim value
-   * @param useQuoteDelim the useQuoteDelim value
-   * @param naValue the naValue value
-   * @param selectCols the selectCols value
-   * @param recordDefaults the recordDefaults value
-   * @param excludeCols the excludeCols value
+   * @param inputDataset the inputDataset value
+   * @param otherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code predicate}.
+   * @param predicate A function returning a scalar boolean.
+   * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of CSVDatasetV2
+   * @return a new instance of FilterDataset
    */
   @Endpoint(
       describeByClass = true
   )
-  public static CSVDatasetV2 create(Scope scope, Operand<TString> filenames,
-      Operand<TString> compressionType, Operand<TInt64> bufferSize, Operand<TBool> header,
-      Operand<TString> fieldDelim, Operand<TBool> useQuoteDelim, Operand<TString> naValue,
-      Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, Operand<TInt64> excludeCols,
-      List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("CSVDatasetV2"));
-    opBuilder.addInput(filenames.asOutput());
-    opBuilder.addInput(compressionType.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
-    opBuilder.addInput(header.asOutput());
-    opBuilder.addInput(fieldDelim.asOutput());
-    opBuilder.addInput(useQuoteDelim.asOutput());
-    opBuilder.addInput(naValue.asOutput());
-    opBuilder.addInput(selectCols.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(recordDefaults));
-    opBuilder.addInput(excludeCols.asOutput());
+  public static FilterDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction predicate,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("FilterDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
     opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("predicate", predicate);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
     for (int i = 0 ; i < outputShapesArray.length ; i++) {
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new CSVDatasetV2(opBuilder.build());
+    return new FilterDataset(opBuilder.build());
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/FinalizeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FinalizeDataset.java
similarity index 95%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/FinalizeDataset.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FinalizeDataset.java
index 6b37412d717..a56b7d175bf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/FinalizeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FinalizeDataset.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.rawops;
+package org.tensorflow.op.data;
 
 import java.util.List;
 import org.tensorflow.Operand;
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset by applying {@code tf.data.Options} to {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class FinalizeDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
@@ -107,7 +111,7 @@ public Output<TType> asOutput() {
   }
 
   /**
-   * Optional attributes for {@link org.tensorflow.op.rawops.FinalizeDataset}
+   * Optional attributes for {@link org.tensorflow.op.data.FinalizeDataset}
    */
   public static class Options {
     private Boolean hasCapturedRef;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
index 6c5f4ec9fec..adf6f2a31d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
@@ -24,6 +24,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
@@ -31,6 +32,9 @@
 /**
  * The FixedLengthRecordDatasetV2 operation
  */
+@Operator(
+    group = "data"
+)
 public final class FixedLengthRecordDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FlatMapDataset.java
similarity index 66%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FlatMapDataset.java
index 60b371e0f4c..6e4a48642ce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FlatMapDataset.java
@@ -18,6 +18,7 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
+import org.tensorflow.ConcreteFunction;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -27,56 +28,64 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.types.TString;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
- * The CacheDatasetV2 operation
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+ * Unlike MapDataset, the {@code f} in FlatMapDataset is expected to return a
+ * Dataset variant, and FlatMapDataset will flatten successive results
+ * into a single Dataset.
  */
-public final class CacheDatasetV2 extends RawOp implements Operand<TType> {
+@Operator(
+    group = "data"
+)
+public final class FlatMapDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "CacheDatasetV2";
+  public static final String OP_NAME = "FlatMapDataset";
 
   private Output<? extends TType> handle;
 
   @SuppressWarnings("unchecked")
-  private CacheDatasetV2(Operation operation) {
+  private FlatMapDataset(Operation operation) {
     super(operation);
     int outputIdx = 0;
     handle = operation.output(outputIdx++);
   }
 
   /**
-   * Factory method to create a class wrapping a new CacheDatasetV2 operation.
+   * Factory method to create a class wrapping a new FlatMapDataset operation.
    *
    * @param scope current scope
    * @param inputDataset the inputDataset value
-   * @param filename the filename value
-   * @param cache the cache value
+   * @param otherArguments the otherArguments value
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   * {@code other_arguments}, to a Dataset variant that contains elements matching
+   * {@code output_types} and {@code output_shapes}.
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
-   * @return a new instance of CacheDatasetV2
+   * @return a new instance of FlatMapDataset
    */
   @Endpoint(
       describeByClass = true
   )
-  public static CacheDatasetV2 create(Scope scope, Operand<? extends TType> inputDataset,
-      Operand<TString> filename, Operand<? extends TType> cache,
+  public static FlatMapDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction f,
       List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("CacheDatasetV2"));
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("FlatMapDataset"));
     opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(filename.asOutput());
-    opBuilder.addInput(cache.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
     opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
     opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
     for (int i = 0 ; i < outputShapesArray.length ; i++) {
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new CacheDatasetV2(opBuilder.build());
+    return new FlatMapDataset(opBuilder.build());
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GeneratorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GeneratorDataset.java
new file mode 100644
index 00000000000..0887f089adb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GeneratorDataset.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that invokes a function to generate elements.
+ */
+@Operator(
+    group = "data"
+)
+public final class GeneratorDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "GeneratorDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private GeneratorDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new GeneratorDataset operation.
+   *
+   * @param scope current scope
+   * @param initFuncOtherArgs the initFuncOtherArgs value
+   * @param nextFuncOtherArgs the nextFuncOtherArgs value
+   * @param finalizeFuncOtherArgs the finalizeFuncOtherArgs value
+   * @param initFunc the value of the initFunc property
+   * @param nextFunc the value of the nextFunc property
+   * @param finalizeFunc the value of the finalizeFunc property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of GeneratorDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static GeneratorDataset create(Scope scope, Iterable<Operand<?>> initFuncOtherArgs,
+      Iterable<Operand<?>> nextFuncOtherArgs, Iterable<Operand<?>> finalizeFuncOtherArgs,
+      ConcreteFunction initFunc, ConcreteFunction nextFunc, ConcreteFunction finalizeFunc,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("GeneratorDataset"));
+    opBuilder.addInputList(Operands.asOutputs(initFuncOtherArgs));
+    opBuilder.addInputList(Operands.asOutputs(nextFuncOtherArgs));
+    opBuilder.addInputList(Operands.asOutputs(finalizeFuncOtherArgs));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("init_func", initFunc);
+    opBuilder.setAttr("next_func", nextFunc);
+    opBuilder.setAttr("finalize_func", finalizeFunc);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new GeneratorDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GroupByReducerDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GroupByReducerDataset.java
new file mode 100644
index 00000000000..e9c2828179e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GroupByReducerDataset.java
@@ -0,0 +1,123 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that computes a group-by on {@code input_dataset}.
+ * Creates a dataset that computes a group-by on {@code input_dataset}.
+ */
+@Operator(
+    group = "data"
+)
+public final class GroupByReducerDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "GroupByReducerDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private GroupByReducerDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new GroupByReducerDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param keyFuncOtherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code key_func}.
+   * @param initFuncOtherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code init_func}.
+   * @param reduceFuncOtherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code reduce_func}.
+   * @param finalizeFuncOtherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code finalize_func}.
+   * @param keyFunc A function mapping an element of {@code input_dataset}, concatenated
+   * with {@code key_func_other_arguments} to a scalar value of type DT_INT64.
+   * @param initFunc A function mapping a key of type DT_INT64, concatenated with
+   * {@code init_func_other_arguments} to the initial reducer state.
+   * @param reduceFunc A function mapping the current reducer state and an element of {@code input_dataset},
+   * concatenated with {@code reduce_func_other_arguments} to a new reducer state.
+   * @param finalizeFunc A function mapping the final reducer state to an output element.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of GroupByReducerDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static GroupByReducerDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> keyFuncOtherArguments, Iterable<Operand<?>> initFuncOtherArguments,
+      Iterable<Operand<?>> reduceFuncOtherArguments,
+      Iterable<Operand<?>> finalizeFuncOtherArguments, ConcreteFunction keyFunc,
+      ConcreteFunction initFunc, ConcreteFunction reduceFunc, ConcreteFunction finalizeFunc,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("GroupByReducerDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(keyFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(initFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(reduceFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(finalizeFuncOtherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("key_func", keyFunc);
+    opBuilder.setAttr("init_func", initFunc);
+    opBuilder.setAttr("reduce_func", reduceFunc);
+    opBuilder.setAttr("finalize_func", finalizeFunc);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new GroupByReducerDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GroupByWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GroupByWindowDataset.java
new file mode 100644
index 00000000000..f84ec83d20c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/GroupByWindowDataset.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that computes a windowed group-by on {@code input_dataset}.
+ * // TODO(mrry): Support non-int64 keys.
+ */
+@Operator(
+    group = "data"
+)
+public final class GroupByWindowDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "GroupByWindowDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private GroupByWindowDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new GroupByWindowDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param keyFuncOtherArguments the keyFuncOtherArguments value
+   * @param reduceFuncOtherArguments the reduceFuncOtherArguments value
+   * @param windowSizeFuncOtherArguments the windowSizeFuncOtherArguments value
+   * @param keyFunc A function mapping an element of {@code input_dataset}, concatenated
+   * with {@code key_func_other_arguments} to a scalar value of type DT_INT64.
+   * @param reduceFunc the value of the reduceFunc property
+   * @param windowSizeFunc the value of the windowSizeFunc property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of GroupByWindowDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static GroupByWindowDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> keyFuncOtherArguments, Iterable<Operand<?>> reduceFuncOtherArguments,
+      Iterable<Operand<?>> windowSizeFuncOtherArguments, ConcreteFunction keyFunc,
+      ConcreteFunction reduceFunc, ConcreteFunction windowSizeFunc,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("GroupByWindowDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(keyFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(reduceFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(windowSizeFuncOtherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("key_func", keyFunc);
+    opBuilder.setAttr("reduce_func", reduceFunc);
+    opBuilder.setAttr("window_size_func", windowSizeFunc);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new GroupByWindowDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
index 152d5696a91..5eb18474f4f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the elements of {@code input_dataset} ignoring errors.
  */
+@Operator(
+    group = "data"
+)
 public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
index b57533279cd..7afd248da3d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
@@ -23,11 +23,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * The InitializeTableFromDataset operation
  */
+@Operator(
+    group = "data"
+)
 public final class InitializeTableFromDataset extends RawOp {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InterleaveDataset.java
new file mode 100644
index 00000000000..64f177ec539
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InterleaveDataset.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+ * Unlike MapDataset, the {@code f} in InterleaveDataset is expected to return
+ * a Dataset variant, and InterleaveDataset will flatten successive
+ * results into a single Dataset. Unlike FlatMapDataset,
+ * InterleaveDataset will interleave sequences of up to {@code block_length}
+ * consecutive elements from {@code cycle_length} input elements.
+ */
+@Operator(
+    group = "data"
+)
+public final class InterleaveDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "InterleaveDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private InterleaveDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new InterleaveDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param cycleLength the cycleLength value
+   * @param blockLength the blockLength value
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   * {@code other_arguments}, to a Dataset variant that contains elements matching
+   * {@code output_types} and {@code output_shapes}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of InterleaveDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static InterleaveDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> cycleLength, Operand<TInt64> blockLength,
+      ConcreteFunction f, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("InterleaveDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder.addInput(cycleLength.asOutput());
+    opBuilder.addInput(blockLength.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new InterleaveDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
index 77d37448de1..9aef6cf746b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
@@ -41,6 +42,9 @@
  * <p>LMDB uses different file formats on big- and little-endian machines.
  * {@code data.LMDBDataset} can only read files in the format of the host machine.
  */
+@Operator(
+    group = "data"
+)
 public final class LMDBDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
index 29256f43c80..2d30dae30b2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
  * Records the latency of producing {@code input_dataset} elements in a StatsAggregator.
  */
+@Operator(
+    group = "data"
+)
 public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LegacyParallelInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LegacyParallelInterleaveDataset.java
new file mode 100644
index 00000000000..6f769b45b97
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LegacyParallelInterleaveDataset.java
@@ -0,0 +1,159 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+ * The resulting dataset is similar to the {@code InterleaveDataset}, with the exception
+ * that if retrieving the next value from a dataset would cause the requester to
+ * block, it will skip that input dataset. This dataset is especially useful
+ * when loading data from a variable-latency datastores (e.g. HDFS, GCS), as it
+ * allows the training step to proceed so long as some data is available.
+ * <p>!! WARNING !! This dataset is not deterministic!
+ */
+@Operator(
+    group = "data"
+)
+public final class LegacyParallelInterleaveDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "LegacyParallelInterleaveDatasetV2";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private LegacyParallelInterleaveDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new LegacyParallelInterleaveDatasetV2 operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param cycleLength the cycleLength value
+   * @param blockLength the blockLength value
+   * @param bufferOutputElements the bufferOutputElements value
+   * @param prefetchInputElements the prefetchInputElements value
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   * {@code other_arguments}, to a Dataset variant that contains elements matching
+   * {@code output_types} and {@code output_shapes}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of LegacyParallelInterleaveDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static LegacyParallelInterleaveDataset create(Scope scope,
+      Operand<? extends TType> inputDataset, Iterable<Operand<?>> otherArguments,
+      Operand<TInt64> cycleLength, Operand<TInt64> blockLength,
+      Operand<TInt64> bufferOutputElements, Operand<TInt64> prefetchInputElements,
+      ConcreteFunction f, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("LegacyParallelInterleaveDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder.addInput(cycleLength.asOutput());
+    opBuilder.addInput(blockLength.asOutput());
+    opBuilder.addInput(bufferOutputElements.asOutput());
+    opBuilder.addInput(prefetchInputElements.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deterministic != null) {
+          opBuilder.setAttr("deterministic", opts.deterministic);
+        }
+      }
+    }
+    return new LegacyParallelInterleaveDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the deterministic option.
+   *
+   * @param deterministic the deterministic option
+   * @return this Options instance.
+   */
+  public static Options deterministic(String deterministic) {
+    return new Options().deterministic(deterministic);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.LegacyParallelInterleaveDataset}
+   */
+  public static class Options {
+    private String deterministic;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the deterministic option.
+     *
+     * @param deterministic the deterministic option
+     * @return this Options instance.
+     */
+    public Options deterministic(String deterministic) {
+      this.deterministic = deterministic;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LoadDataset.java
similarity index 59%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LoadDataset.java
index 874587c68e7..5897e41b71e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LoadDataset.java
@@ -18,6 +18,7 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
+import org.tensorflow.ConcreteFunction;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -27,51 +28,52 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.types.TInt64;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
- * Creates a dataset that changes the batch size.
- * Creates a dataset that changes the batch size of the dataset to current batch
- * size // num_workers.
+ * The LoadDataset operation
  */
-public final class RebatchDataset extends RawOp implements Operand<TType> {
+@Operator(
+    group = "data"
+)
+public final class LoadDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "RebatchDataset";
+  public static final String OP_NAME = "LoadDataset";
 
   private Output<? extends TType> handle;
 
   @SuppressWarnings("unchecked")
-  private RebatchDataset(Operation operation) {
+  private LoadDataset(Operation operation) {
     super(operation);
     int outputIdx = 0;
     handle = operation.output(outputIdx++);
   }
 
   /**
-   * Factory method to create a class wrapping a new RebatchDataset operation.
+   * Factory method to create a class wrapping a new LoadDataset operation.
    *
    * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param numReplicas A scalar representing the number of replicas to distribute this batch across. As
-   * a result of this transformation the current batch size would end up being
-   * divided  by this parameter.
+   * @param path the path value
+   * @param readerFuncOtherArgs the readerFuncOtherArgs value
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
+   * @param readerFunc the value of the readerFunc property
    * @param options carries optional attribute values
-   * @return a new instance of RebatchDataset
+   * @return a new instance of LoadDataset
    */
   @Endpoint(
       describeByClass = true
   )
-  public static RebatchDataset create(Scope scope, Operand<? extends TType> inputDataset,
-      Operand<TInt64> numReplicas, List<Class<? extends TType>> outputTypes,
-      List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("RebatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numReplicas.asOutput());
+  public static LoadDataset create(Scope scope, Operand<TString> path,
+      Iterable<Operand<?>> readerFuncOtherArgs, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, ConcreteFunction readerFunc, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("LoadDataset"));
+    opBuilder.addInput(path.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(readerFuncOtherArgs));
     opBuilder = scope.apply(opBuilder);
     opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
@@ -79,24 +81,25 @@ public static RebatchDataset create(Scope scope, Operand<? extends TType> inputD
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
+    opBuilder.setAttr("reader_func", readerFunc);
     if (options != null) {
       for (Options opts : options) {
-        if (opts.useFallback != null) {
-          opBuilder.setAttr("use_fallback", opts.useFallback);
+        if (opts.compression != null) {
+          opBuilder.setAttr("compression", opts.compression);
         }
       }
     }
-    return new RebatchDataset(opBuilder.build());
+    return new LoadDataset(opBuilder.build());
   }
 
   /**
-   * Sets the useFallback option.
+   * Sets the compression option.
    *
-   * @param useFallback the useFallback option
+   * @param compression the compression option
    * @return this Options instance.
    */
-  public static Options useFallback(Boolean useFallback) {
-    return new Options().useFallback(useFallback);
+  public static Options compression(String compression) {
+    return new Options().compression(compression);
   }
 
   /**
@@ -115,22 +118,22 @@ public Output<TType> asOutput() {
   }
 
   /**
-   * Optional attributes for {@link org.tensorflow.op.data.RebatchDataset}
+   * Optional attributes for {@link org.tensorflow.op.data.LoadDataset}
    */
   public static class Options {
-    private Boolean useFallback;
+    private String compression;
 
     private Options() {
     }
 
     /**
-     * Sets the useFallback option.
+     * Sets the compression option.
      *
-     * @param useFallback the useFallback option
+     * @param compression the compression option
      * @return this Options instance.
      */
-    public Options useFallback(Boolean useFallback) {
-      this.useFallback = useFallback;
+    public Options compression(String compression) {
+      this.compression = compression;
       return this;
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MapAndBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MapAndBatchDataset.java
new file mode 100644
index 00000000000..3526f689b30
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MapAndBatchDataset.java
@@ -0,0 +1,158 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that fuses mapping with batching.
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset} and then
+ * batches {@code batch_size} of them.
+ * <p>Unlike a &quot;MapDataset&quot;, which applies {@code f} sequentially, this dataset invokes up
+ * to {@code batch_size * num_parallel_batches} copies of {@code f} in parallel.
+ */
+@Operator(
+    group = "data"
+)
+public final class MapAndBatchDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "MapAndBatchDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private MapAndBatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new MapAndBatchDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param otherArguments A list of tensors, typically values that were captured when building a closure
+   * for {@code f}.
+   * @param batchSize A scalar representing the number of elements to accumulate in a
+   * batch. It determines the number of concurrent invocations of {@code f} that process
+   * elements from {@code input_dataset} in parallel.
+   * @param numParallelCalls A scalar representing the maximum number of parallel invocations of the {@code map_fn}
+   * function. Applying the {@code map_fn} on consecutive input elements in parallel has
+   * the potential to improve input pipeline throughput.
+   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
+   * is smaller than desired.
+   * @param f A function to apply to the outputs of {@code input_dataset}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of MapAndBatchDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static MapAndBatchDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> batchSize,
+      Operand<TInt64> numParallelCalls, Operand<TBool> dropRemainder, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("MapAndBatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder.addInput(batchSize.asOutput());
+    opBuilder.addInput(numParallelCalls.asOutput());
+    opBuilder.addInput(dropRemainder.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.preserveCardinality != null) {
+          opBuilder.setAttr("preserve_cardinality", opts.preserveCardinality);
+        }
+      }
+    }
+    return new MapAndBatchDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the preserveCardinality option.
+   *
+   * @param preserveCardinality the preserveCardinality option
+   * @return this Options instance.
+   */
+  public static Options preserveCardinality(Boolean preserveCardinality) {
+    return new Options().preserveCardinality(preserveCardinality);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.MapAndBatchDataset}
+   */
+  public static class Options {
+    private Boolean preserveCardinality;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the preserveCardinality option.
+     *
+     * @param preserveCardinality the preserveCardinality option
+     * @return this Options instance.
+     */
+    public Options preserveCardinality(Boolean preserveCardinality) {
+      this.preserveCardinality = preserveCardinality;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MapDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MapDataset.java
new file mode 100644
index 00000000000..5232a58cd1d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MapDataset.java
@@ -0,0 +1,165 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+ */
+@Operator(
+    group = "data"
+)
+public final class MapDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "MapDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private MapDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new MapDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param f the value of the f property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of MapDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static MapDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("MapDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useInterOpParallelism != null) {
+          opBuilder.setAttr("use_inter_op_parallelism", opts.useInterOpParallelism);
+        }
+        if (opts.preserveCardinality != null) {
+          opBuilder.setAttr("preserve_cardinality", opts.preserveCardinality);
+        }
+      }
+    }
+    return new MapDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the useInterOpParallelism option.
+   *
+   * @param useInterOpParallelism the useInterOpParallelism option
+   * @return this Options instance.
+   */
+  public static Options useInterOpParallelism(Boolean useInterOpParallelism) {
+    return new Options().useInterOpParallelism(useInterOpParallelism);
+  }
+
+  /**
+   * Sets the preserveCardinality option.
+   *
+   * @param preserveCardinality the preserveCardinality option
+   * @return this Options instance.
+   */
+  public static Options preserveCardinality(Boolean preserveCardinality) {
+    return new Options().preserveCardinality(preserveCardinality);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.MapDataset}
+   */
+  public static class Options {
+    private Boolean useInterOpParallelism;
+
+    private Boolean preserveCardinality;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the useInterOpParallelism option.
+     *
+     * @param useInterOpParallelism the useInterOpParallelism option
+     * @return this Options instance.
+     */
+    public Options useInterOpParallelism(Boolean useInterOpParallelism) {
+      this.useInterOpParallelism = useInterOpParallelism;
+      return this;
+    }
+
+    /**
+     * Sets the preserveCardinality option.
+     *
+     * @param preserveCardinality the preserveCardinality option
+     * @return this Options instance.
+     */
+    public Options preserveCardinality(Boolean preserveCardinality) {
+      this.preserveCardinality = preserveCardinality;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
index a4fefc83572..d90d48a6011 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
@@ -24,12 +24,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
  * The MatchingFilesDataset operation
  */
+@Operator(
+    group = "data"
+)
 public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
index b63d4d577e3..406c90a988c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that overrides the maximum intra-op parallelism.
  */
+@Operator(
+    group = "data"
+)
 public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
index 4133f57bddf..a88c103708f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Identity transformation that models performance.
  * Identity transformation that models performance.
  */
+@Operator(
+    group = "data"
+)
 public final class ModelDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
index 8acb8a1a9c7..7db69b45f8d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * The NonSerializableDataset operation
  */
+@Operator(
+    group = "data"
+)
 public final class NonSerializableDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OneShotIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OneShotIterator.java
new file mode 100644
index 00000000000..59421497931
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OneShotIterator.java
@@ -0,0 +1,178 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Makes a &quot;one-shot&quot; iterator that can be iterated only once.
+ * A one-shot iterator bundles the logic for defining the dataset and
+ * the state of the iterator in a single op, which allows simple input
+ * pipelines to be defined without an additional initialization
+ * (&quot;MakeIterator&quot;) step.
+ * <p>One-shot iterators have the following limitations:
+ * <ul>
+ * <li>They do not support parameterization: all logic for creating the underlying
+ * dataset must be bundled in the {@code dataset_factory} function.</li>
+ * <li>They are not resettable. Once a one-shot iterator reaches the end of its
+ * underlying dataset, subsequent &quot;IteratorGetNext&quot; operations on that
+ * iterator will always produce an {@code OutOfRange} error.</li>
+ * </ul>
+ * <p>For greater flexibility, use &quot;Iterator&quot; and &quot;MakeIterator&quot; to define
+ * an iterator using an arbitrary subgraph, which may capture tensors
+ * (including fed values) as parameters, and which may be reset multiple
+ * times by rerunning &quot;MakeIterator&quot;.
+ */
+@Operator(
+    group = "data"
+)
+public final class OneShotIterator extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "OneShotIterator";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private OneShotIterator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new OneShotIterator operation.
+   *
+   * @param scope current scope
+   * @param datasetFactory A function of type {@code () -> DT_VARIANT}, where the returned
+   * DT_VARIANT is a dataset.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of OneShotIterator
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static OneShotIterator create(Scope scope, ConcreteFunction datasetFactory,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("OneShotIterator"));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("dataset_factory", datasetFactory);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new OneShotIterator(opBuilder.build());
+  }
+
+  /**
+   * Sets the container option.
+   *
+   * @param container the container option
+   * @return this Options instance.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+
+  /**
+   * Sets the sharedName option.
+   *
+   * @param sharedName the sharedName option
+   * @return this Options instance.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+
+  /**
+   * Gets handle.
+   * A handle to the iterator that can be passed to an &quot;IteratorGetNext&quot;
+   * op.
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.OneShotIterator}
+   */
+  public static class Options {
+    private String container;
+
+    private String sharedName;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the container option.
+     *
+     * @param container the container option
+     * @return this Options instance.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+
+    /**
+     * Sets the sharedName option.
+     *
+     * @param sharedName the sharedName option
+     * @return this Options instance.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
index e4c68e1585b..7c1352a9e00 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
@@ -28,18 +28,22 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
- * Creates a dataset by applying optimizations to {@code input_dataset}.
- * Creates a dataset by applying optimizations to {@code input_dataset}.
+ * Creates a dataset by applying related optimizations to {@code input_dataset}.
+ * Creates a dataset by applying related optimizations to {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class OptimizeDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "OptimizeDataset";
+  public static final String OP_NAME = "OptimizeDatasetV2";
 
   private Output<? extends TType> handle;
 
@@ -51,11 +55,13 @@ private OptimizeDataset(Operation operation) {
   }
 
   /**
-   * Factory method to create a class wrapping a new OptimizeDataset operation.
+   * Factory method to create a class wrapping a new OptimizeDatasetV2 operation.
    *
    * @param scope current scope
    * @param inputDataset A variant tensor representing the input dataset.
-   * @param optimizations A {@code tf.string} vector {@code tf.Tensor} identifying optimizations to use.
+   * @param optimizationsEnabled A {@code tf.string} vector {@code tf.Tensor} identifying user enabled optimizations.
+   * @param optimizationsDisabled A {@code tf.string} vector {@code tf.Tensor} identifying user disabled optimizations.
+   * @param optimizationsDefault A {@code tf.string} vector {@code tf.Tensor} identifying optimizations by default.
    * @param outputTypes the value of the outputTypes property
    * @param outputShapes the value of the outputShapes property
    * @param options carries optional attribute values
@@ -65,11 +71,14 @@ private OptimizeDataset(Operation operation) {
       describeByClass = true
   )
   public static OptimizeDataset create(Scope scope, Operand<? extends TType> inputDataset,
-      Operand<TString> optimizations, List<Class<? extends TType>> outputTypes,
+      Operand<TString> optimizationsEnabled, Operand<TString> optimizationsDisabled,
+      Operand<TString> optimizationsDefault, List<Class<? extends TType>> outputTypes,
       List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("OptimizeDataset"));
     opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(optimizations.asOutput());
+    opBuilder.addInput(optimizationsEnabled.asOutput());
+    opBuilder.addInput(optimizationsDisabled.asOutput());
+    opBuilder.addInput(optimizationsDefault.asOutput());
     opBuilder = scope.apply(opBuilder);
     opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDatasetV2.java
deleted file mode 100644
index f2a7659cd76..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDatasetV2.java
+++ /dev/null
@@ -1,165 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. 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.
-You may obtain a copy of the License at
-
-    http://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.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset by applying related optimizations to {@code input_dataset}.
- * Creates a dataset by applying related optimizations to {@code input_dataset}.
- */
-public final class OptimizeDatasetV2 extends RawOp implements Operand<TType> {
-  /**
-   * The name of this op, as known by TensorFlow core engine
-   */
-  public static final String OP_NAME = "OptimizeDatasetV2";
-
-  private Output<? extends TType> handle;
-
-  @SuppressWarnings("unchecked")
-  private OptimizeDatasetV2(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-
-  /**
-   * Factory method to create a class wrapping a new OptimizeDatasetV2 operation.
-   *
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param optimizationsEnabled A {@code tf.string} vector {@code tf.Tensor} identifying user enabled optimizations.
-   * @param optimizationsDisabled A {@code tf.string} vector {@code tf.Tensor} identifying user disabled optimizations.
-   * @param optimizationsDefault A {@code tf.string} vector {@code tf.Tensor} identifying optimizations by default.
-   * @param outputTypes the value of the outputTypes property
-   * @param outputShapes the value of the outputShapes property
-   * @param options carries optional attribute values
-   * @return a new instance of OptimizeDatasetV2
-   */
-  @Endpoint(
-      describeByClass = true
-  )
-  public static OptimizeDatasetV2 create(Scope scope, Operand<? extends TType> inputDataset,
-      Operand<TString> optimizationsEnabled, Operand<TString> optimizationsDisabled,
-      Operand<TString> optimizationsDefault, List<Class<? extends TType>> outputTypes,
-      List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("OptimizeDatasetV2"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(optimizationsEnabled.asOutput());
-    opBuilder.addInput(optimizationsDisabled.asOutput());
-    opBuilder.addInput(optimizationsDefault.asOutput());
-    opBuilder = scope.apply(opBuilder);
-    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0 ; i < outputShapesArray.length ; i++) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.optimizationConfigs != null) {
-          String[] optimizationConfigsArray = new String[opts.optimizationConfigs.size()];
-          for (int i = 0 ; i < optimizationConfigsArray.length ; i++) {
-            optimizationConfigsArray[i] = opts.optimizationConfigs.get(i);
-          }
-          opBuilder.setAttr("optimization_configs", optimizationConfigsArray);
-        }
-      }
-    }
-    return new OptimizeDatasetV2(opBuilder.build());
-  }
-
-  /**
-   * Sets the optimizationConfigs option.
-   *
-   * @param optimizationConfigs the optimizationConfigs option
-   * @return this Options instance.
-   */
-  public static Options optimizationConfigs(List<String> optimizationConfigs) {
-    return new Options().optimizationConfigs(optimizationConfigs);
-  }
-
-  /**
-   * Sets the optimizationConfigs option.
-   *
-   * @param optimizationConfigs the optimizationConfigs option
-   * @return this Options instance.
-   */
-  public static Options optimizationConfigs(String[] optimizationConfigs) {
-    return new Options().optimizationConfigs(optimizationConfigs);
-  }
-
-  /**
-   * Gets handle.
-   *
-   * @return handle.
-   */
-  public Output<? extends TType> handle() {
-    return handle;
-  }
-
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
-  }
-
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.OptimizeDatasetV2}
-   */
-  public static class Options {
-    private List<String> optimizationConfigs;
-
-    private Options() {
-    }
-
-    /**
-     * Sets the optimizationConfigs option.
-     *
-     * @param optimizationConfigs the optimizationConfigs option
-     * @return this Options instance.
-     */
-    public Options optimizationConfigs(List<String> optimizationConfigs) {
-      this.optimizationConfigs = optimizationConfigs;
-      return this;
-    }
-
-    /**
-     * Sets the optimizationConfigs option.
-     *
-     * @param optimizationConfigs the optimizationConfigs option
-     * @return this Options instance.
-     */
-    public Options optimizationConfigs(String... optimizationConfigs) {
-      this.optimizationConfigs = Arrays.asList(optimizationConfigs);
-      return this;
-    }
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/OptionsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionsDataset.java
similarity index 96%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/OptionsDataset.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionsDataset.java
index ec3bf2560c2..2ff6923fc71 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/OptionsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionsDataset.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.rawops;
+package org.tensorflow.op.data;
 
 import java.util.List;
 import org.tensorflow.Operand;
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset by attaching tf.data.Options to {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class OptionsDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
index 5e74c59c2dc..058015155ea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
@@ -34,6 +35,9 @@
 /**
  * Creates a dataset that batches and pads {@code batch_size} elements from the input.
  */
+@Operator(
+    group = "data"
+)
 public final class PaddedBatchDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/ParallelBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelBatchDataset.java
similarity index 95%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/ParallelBatchDataset.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelBatchDataset.java
index 23fe867b5ae..f313afbf0b8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/rawops/ParallelBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelBatchDataset.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.rawops;
+package org.tensorflow.op.data;
 
 import java.util.List;
 import org.tensorflow.Operand;
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
@@ -34,6 +35,9 @@
 /**
  * The ParallelBatchDataset operation
  */
+@Operator(
+    group = "data"
+)
 public final class ParallelBatchDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
@@ -116,7 +120,7 @@ public Output<TType> asOutput() {
   }
 
   /**
-   * Optional attributes for {@link org.tensorflow.op.rawops.ParallelBatchDataset}
+   * Optional attributes for {@link org.tensorflow.op.data.ParallelBatchDataset}
    */
   public static class Options {
     private String deterministic;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelInterleaveDataset.java
new file mode 100644
index 00000000000..bf87079b34c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelInterleaveDataset.java
@@ -0,0 +1,181 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+ * The resulting dataset is similar to the {@code InterleaveDataset}, except that the
+ * dataset will fetch records from the interleaved datasets in parallel.
+ * <p>The {@code tf.data} Python API creates instances of this op from
+ * {@code Dataset.interleave()} when the {@code num_parallel_calls} parameter of that method
+ * is set to any value other than {@code None}.
+ * <p>By default, the output of this dataset will be deterministic, which may result
+ * in the dataset blocking if the next data item to be returned isn't available.
+ * In order to avoid head-of-line blocking, one can either set the {@code deterministic}
+ * attribute to &quot;false&quot;, or leave it as &quot;default&quot; and set the
+ * {@code experimental_deterministic} parameter of {@code tf.data.Options} to {@code False}.
+ * This can improve performance at the expense of non-determinism.
+ */
+@Operator(
+    group = "data"
+)
+public final class ParallelInterleaveDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ParallelInterleaveDatasetV4";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private ParallelInterleaveDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ParallelInterleaveDatasetV4 operation.
+   *
+   * @param scope current scope
+   * @param inputDataset Dataset that produces a stream of arguments for the function {@code f}.
+   * @param otherArguments Additional arguments to pass to {@code f} beyond those produced by {@code input_dataset}.
+   * Evaluated once when the dataset is instantiated.
+   * @param cycleLength Number of datasets (each created by applying {@code f} to the elements of
+   * {@code input_dataset}) among which the {@code ParallelInterleaveDatasetV2} will cycle in a
+   * round-robin fashion.
+   * @param blockLength Number of elements at a time to produce from each interleaved invocation of a
+   * dataset returned by {@code f}.
+   * @param bufferOutputElements The number of elements each iterator being interleaved should buffer (similar
+   * to the {@code .prefetch()} transformation for each interleaved iterator).
+   * @param prefetchInputElements Determines the number of iterators to prefetch, allowing buffers to warm up and
+   * data to be pre-fetched without blocking the main thread.
+   * @param numParallelCalls Determines the number of threads that should be used for fetching data from
+   * input datasets in parallel. The Python API {@code tf.data.experimental.AUTOTUNE}
+   * constant can be used to indicate that the level of parallelism should be autotuned.
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   * {@code other_arguments}, to a Dataset variant that contains elements matching
+   * {@code output_types} and {@code output_shapes}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ParallelInterleaveDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static ParallelInterleaveDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> cycleLength, Operand<TInt64> blockLength,
+      Operand<TInt64> bufferOutputElements, Operand<TInt64> prefetchInputElements,
+      Operand<TInt64> numParallelCalls, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ParallelInterleaveDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder.addInput(cycleLength.asOutput());
+    opBuilder.addInput(blockLength.asOutput());
+    opBuilder.addInput(bufferOutputElements.asOutput());
+    opBuilder.addInput(prefetchInputElements.asOutput());
+    opBuilder.addInput(numParallelCalls.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deterministic != null) {
+          opBuilder.setAttr("deterministic", opts.deterministic);
+        }
+      }
+    }
+    return new ParallelInterleaveDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the deterministic option.
+   *
+   * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
+   * whether the interleave is allowed to return elements out of order if the next
+   * element to be returned isn't available, but a later element is. Options are
+   * &quot;true&quot;, &quot;false&quot;, and &quot;default&quot;. &quot;default&quot; indicates that determinism should be
+   * decided by the {@code experimental_deterministic} parameter of {@code tf.data.Options}.
+   * @return this Options instance.
+   */
+  public static Options deterministic(String deterministic) {
+    return new Options().deterministic(deterministic);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ParallelInterleaveDataset}
+   */
+  public static class Options {
+    private String deterministic;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the deterministic option.
+     *
+     * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
+     * whether the interleave is allowed to return elements out of order if the next
+     * element to be returned isn't available, but a later element is. Options are
+     * &quot;true&quot;, &quot;false&quot;, and &quot;default&quot;. &quot;default&quot; indicates that determinism should be
+     * decided by the {@code experimental_deterministic} parameter of {@code tf.data.Options}.
+     * @return this Options instance.
+     */
+    public Options deterministic(String deterministic) {
+      this.deterministic = deterministic;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelMapDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelMapDataset.java
new file mode 100644
index 00000000000..8d3c1686b5f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParallelMapDataset.java
@@ -0,0 +1,197 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+ * Unlike a &quot;MapDataset&quot;, which applies {@code f} sequentially, this dataset invokes up
+ * to {@code num_parallel_calls} copies of {@code f} in parallel.
+ */
+@Operator(
+    group = "data"
+)
+public final class ParallelMapDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ParallelMapDatasetV2";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private ParallelMapDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ParallelMapDatasetV2 operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param numParallelCalls The number of concurrent invocations of {@code f} that process
+   * elements from {@code input_dataset} in parallel.
+   * @param f the value of the f property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ParallelMapDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static ParallelMapDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> numParallelCalls, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ParallelMapDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder.addInput(numParallelCalls.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useInterOpParallelism != null) {
+          opBuilder.setAttr("use_inter_op_parallelism", opts.useInterOpParallelism);
+        }
+        if (opts.deterministic != null) {
+          opBuilder.setAttr("deterministic", opts.deterministic);
+        }
+        if (opts.preserveCardinality != null) {
+          opBuilder.setAttr("preserve_cardinality", opts.preserveCardinality);
+        }
+      }
+    }
+    return new ParallelMapDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the useInterOpParallelism option.
+   *
+   * @param useInterOpParallelism the useInterOpParallelism option
+   * @return this Options instance.
+   */
+  public static Options useInterOpParallelism(Boolean useInterOpParallelism) {
+    return new Options().useInterOpParallelism(useInterOpParallelism);
+  }
+
+  /**
+   * Sets the deterministic option.
+   *
+   * @param deterministic the deterministic option
+   * @return this Options instance.
+   */
+  public static Options deterministic(String deterministic) {
+    return new Options().deterministic(deterministic);
+  }
+
+  /**
+   * Sets the preserveCardinality option.
+   *
+   * @param preserveCardinality the preserveCardinality option
+   * @return this Options instance.
+   */
+  public static Options preserveCardinality(Boolean preserveCardinality) {
+    return new Options().preserveCardinality(preserveCardinality);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ParallelMapDataset}
+   */
+  public static class Options {
+    private Boolean useInterOpParallelism;
+
+    private String deterministic;
+
+    private Boolean preserveCardinality;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the useInterOpParallelism option.
+     *
+     * @param useInterOpParallelism the useInterOpParallelism option
+     * @return this Options instance.
+     */
+    public Options useInterOpParallelism(Boolean useInterOpParallelism) {
+      this.useInterOpParallelism = useInterOpParallelism;
+      return this;
+    }
+
+    /**
+     * Sets the deterministic option.
+     *
+     * @param deterministic the deterministic option
+     * @return this Options instance.
+     */
+    public Options deterministic(String deterministic) {
+      this.deterministic = deterministic;
+      return this;
+    }
+
+    /**
+     * Sets the preserveCardinality option.
+     *
+     * @param preserveCardinality the preserveCardinality option
+     * @return this Options instance.
+     */
+    public Options preserveCardinality(Boolean preserveCardinality) {
+      this.preserveCardinality = preserveCardinality;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParseExampleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParseExampleDataset.java
new file mode 100644
index 00000000000..6389e91a952
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ParseExampleDataset.java
@@ -0,0 +1,240 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transforms {@code input_dataset} containing {@code Example} protos as vectors of DT_STRING into a dataset of {@code Tensor} or {@code SparseTensor} objects representing the parsed features.
+ */
+@Operator(
+    group = "data"
+)
+public final class ParseExampleDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ParseExampleDatasetV2";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private ParseExampleDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ParseExampleDatasetV2 operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param numParallelCalls the numParallelCalls value
+   * @param denseDefaults A dict mapping string keys to {@code Tensor}s.
+   * The keys of the dict must match the dense_keys of the feature.
+   * @param sparseKeys A list of string keys in the examples features.
+   * The results for these keys will be returned as {@code SparseTensor} objects.
+   * @param denseKeys A list of Ndense string Tensors (scalars).
+   * The keys expected in the Examples features associated with dense values.
+   * @param sparseTypes A list of {@code DTypes} of the same length as {@code sparse_keys}.
+   * Only {@code tf.float32} ({@code FloatList}), {@code tf.int64} ({@code Int64List}),
+   * and {@code tf.string} ({@code BytesList}) are supported.
+   * @param denseShapes List of tuples with the same length as {@code dense_keys}.
+   * The shape of the data for each dense feature referenced by {@code dense_keys}.
+   * Required for any input tensors identified by {@code dense_keys}.  Must be
+   * either fully defined, or may contain an unknown first dimension.
+   * An unknown first dimension means the feature is treated as having
+   * a variable number of blocks, and the output shape along this dimension
+   * is considered unknown at graph build time.  Padding is applied for
+   * minibatch elements smaller than the maximum number of blocks for the
+   * given feature along this dimension.
+   * @param outputTypes The type list for the return values.
+   * @param outputShapes The list of shapes being produced.
+   * @param raggedValueTypes the value of the raggedValueTypes property
+   * @param raggedSplitTypes the value of the raggedSplitTypes property
+   * @param options carries optional attribute values
+   * @return a new instance of ParseExampleDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static ParseExampleDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Operand<TInt64> numParallelCalls, Iterable<Operand<?>> denseDefaults, List<String> sparseKeys,
+      List<String> denseKeys, List<Class<? extends TType>> sparseTypes, List<Shape> denseShapes,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      List<Class<? extends TType>> raggedValueTypes,
+      List<Class<? extends TNumber>> raggedSplitTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ParseExampleDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numParallelCalls.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(denseDefaults));
+    opBuilder = scope.apply(opBuilder);
+    String[] sparseKeysArray = new String[sparseKeys.size()];
+    for (int i = 0 ; i < sparseKeysArray.length ; i++) {
+      sparseKeysArray[i] = sparseKeys.get(i);
+    }
+    opBuilder.setAttr("sparse_keys", sparseKeysArray);
+    String[] denseKeysArray = new String[denseKeys.size()];
+    for (int i = 0 ; i < denseKeysArray.length ; i++) {
+      denseKeysArray[i] = denseKeys.get(i);
+    }
+    opBuilder.setAttr("dense_keys", denseKeysArray);
+    opBuilder.setAttr("sparse_types", Operands.toDataTypes(sparseTypes));
+    Shape[] denseShapesArray = new Shape[denseShapes.size()];
+    for (int i = 0 ; i < denseShapesArray.length ; i++) {
+      denseShapesArray[i] = denseShapes.get(i);
+    }
+    opBuilder.setAttr("dense_shapes", denseShapesArray);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    opBuilder.setAttr("ragged_value_types", Operands.toDataTypes(raggedValueTypes));
+    opBuilder.setAttr("ragged_split_types", Operands.toDataTypes(raggedSplitTypes));
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deterministic != null) {
+          opBuilder.setAttr("deterministic", opts.deterministic);
+        }
+        if (opts.raggedKeys != null) {
+          String[] raggedKeysArray = new String[opts.raggedKeys.size()];
+          for (int i = 0 ; i < raggedKeysArray.length ; i++) {
+            raggedKeysArray[i] = opts.raggedKeys.get(i);
+          }
+          opBuilder.setAttr("ragged_keys", raggedKeysArray);
+        }
+      }
+    }
+    return new ParseExampleDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the deterministic option.
+   *
+   * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
+   * whether the dataset is allowed to return elements out of order if the next
+   * element to be returned isn't available, but a later element is. Options are
+   * &quot;true&quot;, &quot;false&quot;, and &quot;default&quot;. &quot;default&quot; indicates that determinism should be
+   * decided by the {@code experimental_deterministic} parameter of {@code tf.data.Options}.
+   * @return this Options instance.
+   */
+  public static Options deterministic(String deterministic) {
+    return new Options().deterministic(deterministic);
+  }
+
+  /**
+   * Sets the raggedKeys option.
+   *
+   * @param raggedKeys the raggedKeys option
+   * @return this Options instance.
+   */
+  public static Options raggedKeys(List<String> raggedKeys) {
+    return new Options().raggedKeys(raggedKeys);
+  }
+
+  /**
+   * Sets the raggedKeys option.
+   *
+   * @param raggedKeys the raggedKeys option
+   * @return this Options instance.
+   */
+  public static Options raggedKeys(String[] raggedKeys) {
+    return new Options().raggedKeys(raggedKeys);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ParseExampleDataset}
+   */
+  public static class Options {
+    private String deterministic;
+
+    private List<String> raggedKeys;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the deterministic option.
+     *
+     * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
+     * whether the dataset is allowed to return elements out of order if the next
+     * element to be returned isn't available, but a later element is. Options are
+     * &quot;true&quot;, &quot;false&quot;, and &quot;default&quot;. &quot;default&quot; indicates that determinism should be
+     * decided by the {@code experimental_deterministic} parameter of {@code tf.data.Options}.
+     * @return this Options instance.
+     */
+    public Options deterministic(String deterministic) {
+      this.deterministic = deterministic;
+      return this;
+    }
+
+    /**
+     * Sets the raggedKeys option.
+     *
+     * @param raggedKeys the raggedKeys option
+     * @return this Options instance.
+     */
+    public Options raggedKeys(List<String> raggedKeys) {
+      this.raggedKeys = raggedKeys;
+      return this;
+    }
+
+    /**
+     * Sets the raggedKeys option.
+     *
+     * @param raggedKeys the raggedKeys option
+     * @return this Options instance.
+     */
+    public Options raggedKeys(String... raggedKeys) {
+      this.raggedKeys = Arrays.asList(raggedKeys);
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
index 8ef26922652..92d74b4742a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that asynchronously prefetches elements from {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class PrefetchDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
index 3290840da75..e5e387da993 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
index 788ff579222..ef1b7a72431 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
@@ -41,6 +42,9 @@
  * performed is determined by the {@code experimental_optimization.hoist_random_uniform}
  * option of {@code tf.data.Options}.
  */
+@Operator(
+    group = "data"
+)
 public final class RandomDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDatasetV2.java
index b98013d55bc..2c50b6cf519 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDatasetV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDatasetV2.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
@@ -36,6 +37,9 @@
  * Creates a dataset that rebatches elements from {@code input_dataset} into new batch
  * sizes.
  */
+@Operator(
+    group = "data"
+)
 public final class RebatchDatasetV2 extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ReduceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ReduceDataset.java
new file mode 100644
index 00000000000..143ec183fee
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ReduceDataset.java
@@ -0,0 +1,148 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reduces the input dataset to a singleton using a reduce function.
+ */
+@Operator(
+    group = "data"
+)
+public final class ReduceDataset extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ReduceDataset";
+
+  private List<Output<?>> components;
+
+  @SuppressWarnings("unchecked")
+  private ReduceDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ReduceDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param initialState A nested structure of tensors, representing the initial state of the
+   * transformation.
+   * @param otherArguments the otherArguments value
+   * @param f A function that maps {@code (old_state, input_element)} to {@code new_state}. It must take
+   * two arguments and return a nested structures of tensors. The structure of
+   * {@code new_state} must match the structure of {@code initial_state}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ReduceDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static ReduceDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> initialState, Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ReduceDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(initialState));
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useInterOpParallelism != null) {
+          opBuilder.setAttr("use_inter_op_parallelism", opts.useInterOpParallelism);
+        }
+      }
+    }
+    return new ReduceDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the useInterOpParallelism option.
+   *
+   * @param useInterOpParallelism the useInterOpParallelism option
+   * @return this Options instance.
+   */
+  public static Options useInterOpParallelism(Boolean useInterOpParallelism) {
+    return new Options().useInterOpParallelism(useInterOpParallelism);
+  }
+
+  /**
+   * Gets components.
+   *
+   * @return components.
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ReduceDataset}
+   */
+  public static class Options {
+    private Boolean useInterOpParallelism;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the useInterOpParallelism option.
+     *
+     * @param useInterOpParallelism the useInterOpParallelism option
+     * @return this Options instance.
+     */
+    public Options useInterOpParallelism(Boolean useInterOpParallelism) {
+      this.useInterOpParallelism = useInterOpParallelism;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
index 5809a919ebd..e389557f99f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
@@ -24,6 +24,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
@@ -31,6 +32,9 @@
 /**
  * Registers a dataset with the tf.data service.
  */
+@Operator(
+    group = "data"
+)
 public final class RegisterDataset extends RawOp implements Operand<TInt64> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
index 27778a1dd49..550df76210e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
@@ -39,6 +40,9 @@
  * {@code experimental_optimization.filter_with_random_uniform_fusion} option of
  * {@code tf.data.Options}.
  */
+@Operator(
+    group = "data"
+)
 public final class SamplingDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SaveDataset.java
new file mode 100644
index 00000000000..74a20665508
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SaveDataset.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * The SaveDataset operation
+ */
+@Operator(
+    group = "data"
+)
+public final class SaveDataset extends RawOp {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "SaveDataset";
+
+  private SaveDataset(Operation operation) {
+    super(operation);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new SaveDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param path the path value
+   * @param shardFuncOtherArgs the shardFuncOtherArgs value
+   * @param shardFunc the value of the shardFunc property
+   * @param options carries optional attribute values
+   * @return a new instance of SaveDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static SaveDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Operand<TString> path, Iterable<Operand<?>> shardFuncOtherArgs, ConcreteFunction shardFunc,
+      Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("SaveDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(path.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(shardFuncOtherArgs));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("shard_func", shardFunc);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.compression != null) {
+          opBuilder.setAttr("compression", opts.compression);
+        }
+        if (opts.useShardFunc != null) {
+          opBuilder.setAttr("use_shard_func", opts.useShardFunc);
+        }
+      }
+    }
+    return new SaveDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the compression option.
+   *
+   * @param compression the compression option
+   * @return this Options instance.
+   */
+  public static Options compression(String compression) {
+    return new Options().compression(compression);
+  }
+
+  /**
+   * Sets the useShardFunc option.
+   *
+   * @param useShardFunc the useShardFunc option
+   * @return this Options instance.
+   */
+  public static Options useShardFunc(Boolean useShardFunc) {
+    return new Options().useShardFunc(useShardFunc);
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.SaveDataset}
+   */
+  public static class Options {
+    private String compression;
+
+    private Boolean useShardFunc;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the compression option.
+     *
+     * @param compression the compression option
+     * @return this Options instance.
+     */
+    public Options compression(String compression) {
+      this.compression = compression;
+      return this;
+    }
+
+    /**
+     * Sets the useShardFunc option.
+     *
+     * @param useShardFunc the useShardFunc option
+     * @return this Options instance.
+     */
+    public Options useShardFunc(Boolean useShardFunc) {
+      this.useShardFunc = useShardFunc;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ScanDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ScanDataset.java
new file mode 100644
index 00000000000..ab71a2a7da3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ScanDataset.java
@@ -0,0 +1,167 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset successively reduces {@code f} over the elements of {@code input_dataset}.
+ */
+@Operator(
+    group = "data"
+)
+public final class ScanDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ScanDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private ScanDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ScanDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param initialState the initialState value
+   * @param otherArguments the otherArguments value
+   * @param f the value of the f property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ScanDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static ScanDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> initialState, Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ScanDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(initialState));
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.preserveCardinality != null) {
+          opBuilder.setAttr("preserve_cardinality", opts.preserveCardinality);
+        }
+        if (opts.useDefaultDevice != null) {
+          opBuilder.setAttr("use_default_device", opts.useDefaultDevice);
+        }
+      }
+    }
+    return new ScanDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the preserveCardinality option.
+   *
+   * @param preserveCardinality the preserveCardinality option
+   * @return this Options instance.
+   */
+  public static Options preserveCardinality(Boolean preserveCardinality) {
+    return new Options().preserveCardinality(preserveCardinality);
+  }
+
+  /**
+   * Sets the useDefaultDevice option.
+   *
+   * @param useDefaultDevice the useDefaultDevice option
+   * @return this Options instance.
+   */
+  public static Options useDefaultDevice(Boolean useDefaultDevice) {
+    return new Options().useDefaultDevice(useDefaultDevice);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ScanDataset}
+   */
+  public static class Options {
+    private Boolean preserveCardinality;
+
+    private Boolean useDefaultDevice;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the preserveCardinality option.
+     *
+     * @param preserveCardinality the preserveCardinality option
+     * @return this Options instance.
+     */
+    public Options preserveCardinality(Boolean preserveCardinality) {
+      this.preserveCardinality = preserveCardinality;
+      return this;
+    }
+
+    /**
+     * Sets the useDefaultDevice option.
+     *
+     * @param useDefaultDevice the useDefaultDevice option
+     * @return this Options instance.
+     */
+    public Options useDefaultDevice(Boolean useDefaultDevice) {
+      this.useDefaultDevice = useDefaultDevice;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
index 63bce56359c..fd935d5862f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
  * The SetStatsAggregatorDataset operation
  */
+@Operator(
+    group = "data"
+)
 public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
index ea9b7059eea..1609721dbde 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a {@code Dataset} that includes only 1/{@code num_shards} of this dataset.
  */
+@Operator(
+    group = "data"
+)
 public final class ShardDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
index 1a7d3667632..c57edd1bb10 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * The ShuffleAndRepeatDatasetV2 operation
  */
+@Operator(
+    group = "data"
+)
 public final class ShuffleAndRepeatDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
index 4ae9f5a57cf..4bdbd0abc2c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * The ShuffleDatasetV3 operation
  */
+@Operator(
+    group = "data"
+)
 public final class ShuffleDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
index bc3b16e01a2..74992fd71fa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * The SleepDataset operation
  */
+@Operator(
+    group = "data"
+)
 public final class SleepDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
index 123a5780271..4a5d1629816 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that passes a sliding window over {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
new file mode 100644
index 00000000000..a1cbcd27635
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
@@ -0,0 +1,256 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that will write to / read from a snapshot.
+ * This dataset attempts to determine whether a valid snapshot exists at the
+ * {@code snapshot_path}, and reads from the snapshot in lieu of using {@code input_dataset}.
+ * If not, it will run the preprocessing pipeline as usual, and write out a
+ * snapshot of the data processed for future use.
+ */
+@Operator(
+    group = "data"
+)
+public final class SnapshotDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "SnapshotDatasetV2";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private SnapshotDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new SnapshotDatasetV2 operation.
+   *
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param path The path we should write snapshots to / read snapshots from.
+   * @param readerFuncOtherArgs the readerFuncOtherArgs value
+   * @param shardFuncOtherArgs the shardFuncOtherArgs value
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param readerFunc Optional. A function to control how to read data from snapshot shards.
+   * @param shardFunc Optional. A function to control how to shard data when writing a snapshot.
+   * @param options carries optional attribute values
+   * @return a new instance of SnapshotDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static SnapshotDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Operand<TString> path, Iterable<Operand<?>> readerFuncOtherArgs,
+      Iterable<Operand<?>> shardFuncOtherArgs, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, ConcreteFunction readerFunc, ConcreteFunction shardFunc,
+      Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("SnapshotDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(path.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(readerFuncOtherArgs));
+    opBuilder.addInputList(Operands.asOutputs(shardFuncOtherArgs));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    opBuilder.setAttr("reader_func", readerFunc);
+    opBuilder.setAttr("shard_func", shardFunc);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.compression != null) {
+          opBuilder.setAttr("compression", opts.compression);
+        }
+        if (opts.readerPrefix != null) {
+          opBuilder.setAttr("reader_prefix", opts.readerPrefix);
+        }
+        if (opts.writerPrefix != null) {
+          opBuilder.setAttr("writer_prefix", opts.writerPrefix);
+        }
+        if (opts.hashValid != null) {
+          opBuilder.setAttr("hash_valid", opts.hashValid);
+        }
+        if (opts.hash != null) {
+          opBuilder.setAttr("hash", opts.hash);
+        }
+      }
+    }
+    return new SnapshotDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the compression option.
+   *
+   * @param compression The type of compression to be applied to the saved snapshot files.
+   * @return this Options instance.
+   */
+  public static Options compression(String compression) {
+    return new Options().compression(compression);
+  }
+
+  /**
+   * Sets the readerPrefix option.
+   *
+   * @param readerPrefix the readerPrefix option
+   * @return this Options instance.
+   */
+  public static Options readerPrefix(String readerPrefix) {
+    return new Options().readerPrefix(readerPrefix);
+  }
+
+  /**
+   * Sets the writerPrefix option.
+   *
+   * @param writerPrefix the writerPrefix option
+   * @return this Options instance.
+   */
+  public static Options writerPrefix(String writerPrefix) {
+    return new Options().writerPrefix(writerPrefix);
+  }
+
+  /**
+   * Sets the hashValid option.
+   *
+   * @param hashValid the hashValid option
+   * @return this Options instance.
+   */
+  public static Options hashValid(Boolean hashValid) {
+    return new Options().hashValid(hashValid);
+  }
+
+  /**
+   * Sets the hash option.
+   *
+   * @param hash the hash option
+   * @return this Options instance.
+   */
+  public static Options hash(Long hash) {
+    return new Options().hash(hash);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.SnapshotDataset}
+   */
+  public static class Options {
+    private String compression;
+
+    private String readerPrefix;
+
+    private String writerPrefix;
+
+    private Boolean hashValid;
+
+    private Long hash;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the compression option.
+     *
+     * @param compression The type of compression to be applied to the saved snapshot files.
+     * @return this Options instance.
+     */
+    public Options compression(String compression) {
+      this.compression = compression;
+      return this;
+    }
+
+    /**
+     * Sets the readerPrefix option.
+     *
+     * @param readerPrefix the readerPrefix option
+     * @return this Options instance.
+     */
+    public Options readerPrefix(String readerPrefix) {
+      this.readerPrefix = readerPrefix;
+      return this;
+    }
+
+    /**
+     * Sets the writerPrefix option.
+     *
+     * @param writerPrefix the writerPrefix option
+     * @return this Options instance.
+     */
+    public Options writerPrefix(String writerPrefix) {
+      this.writerPrefix = writerPrefix;
+      return this;
+    }
+
+    /**
+     * Sets the hashValid option.
+     *
+     * @param hashValid the hashValid option
+     * @return this Options instance.
+     */
+    public Options hashValid(Boolean hashValid) {
+      this.hashValid = hashValid;
+      return this;
+    }
+
+    /**
+     * Sets the hash option.
+     *
+     * @param hash the hash option
+     * @return this Options instance.
+     */
+    public Options hash(Long hash) {
+      this.hash = hash;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
index 188bf6f6a8b..6de2eb40539 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
@@ -24,12 +24,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that splits a SparseTensor into elements row-wise.
  */
+@Operator(
+    group = "data"
+)
 public final class SparseTensorSliceDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
index 7240a21d3b4..5fd6a3d5784 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
@@ -27,12 +27,16 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that executes a SQL query and emits rows of the result set.
  */
+@Operator(
+    group = "data"
+)
 public final class SqlDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
index a4cd180c217..99c227869fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
@@ -27,13 +27,13 @@
 import org.tensorflow.types.family.TType;
 
 /**
- * Creates a statistics manager resource.
+ * The StatsAggregatorHandleV2 operation
  */
 public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "StatsAggregatorHandle";
+  public static final String OP_NAME = "StatsAggregatorHandleV2";
 
   private Output<? extends TType> handle;
 
@@ -45,7 +45,7 @@ private StatsAggregatorHandle(Operation operation) {
   }
 
   /**
-   * Factory method to create a class wrapping a new StatsAggregatorHandle operation.
+   * Factory method to create a class wrapping a new StatsAggregatorHandleV2 operation.
    *
    * @param scope current scope
    * @param options carries optional attribute values
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorSetSummaryWriter.java
similarity index 97%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorSetSummaryWriter.java
index a0630991efe..aa3e56da61b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorSetSummaryWriter.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.data.experimental;
+package org.tensorflow.op.data;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeWhileDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeWhileDataset.java
new file mode 100644
index 00000000000..5f916997113
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeWhileDataset.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that stops iteration when predicate` is false.
+ * The {@code predicate} function must return a scalar boolean and accept the
+ * following arguments:
+ * <ul>
+ * <li>One tensor for each component of an element of {@code input_dataset}.</li>
+ * <li>One tensor for each value in {@code other_arguments}.</li>
+ * </ul>
+ */
+@Operator(
+    group = "data"
+)
+public final class TakeWhileDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "TakeWhileDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private TakeWhileDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new TakeWhileDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param otherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code predicate}.
+   * @param predicate A function returning a scalar boolean.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of TakeWhileDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static TakeWhileDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction predicate,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("TakeWhileDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("predicate", predicate);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new TakeWhileDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
index 1c93d86e475..a089e8e1be5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits {@code components} as a tuple of tensors once.
  */
+@Operator(
+    group = "data"
+)
 public final class TensorDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
index 9f67fdf0a66..17a7c0c9065 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
index a827341e997..2fd64618e8e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * A dataset that splits the elements of its input into multiple elements.
  */
+@Operator(
+    group = "data"
+)
 public final class UnbatchDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UncompressElement.java
similarity index 98%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UncompressElement.java
index 0f498626609..344915555c9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UncompressElement.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.data.experimental;
+package org.tensorflow.op.data;
 
 import java.util.Arrays;
 import java.util.Iterator;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
index 79e21583277..6938c786469 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
@@ -27,11 +27,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the unique elements of {@code input_dataset}.
  */
+@Operator(
+    group = "data"
+)
 public final class UniqueDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
index df30e1749d9..2bbc602b5fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
@@ -24,11 +24,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * The UnwrapDatasetVariant operation
  */
+@Operator(
+    group = "data"
+)
 public final class UnwrapDatasetVariant extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
index 3e2069a2a93..3988ddcd99d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
@@ -69,6 +70,9 @@
  * produces {@code {{"a": {0, 1}}, {"a": {2, 3}}}}</li>
  * </ul>
  */
+@Operator(
+    group = "data"
+)
 public final class WindowDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
index 8f6de33fda7..aec61c37b21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
@@ -24,11 +24,15 @@
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TType;
 
 /**
  * The WrapDatasetVariant operation
  */
+@Operator(
+    group = "data"
+)
 public final class WrapDatasetVariant extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
deleted file mode 100644
index dd4a3e3f6d9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. 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.
-You may obtain a copy of the License at
-
-    http://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.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that reads data from the tf.data service.
- */
-public final class DataServiceDataset extends RawOp implements Operand<TType> {
-  /**
-   * The name of this op, as known by TensorFlow core engine
-   */
-  public static final String OP_NAME = "DataServiceDataset";
-
-  private Output<? extends TType> handle;
-
-  @SuppressWarnings("unchecked")
-  private DataServiceDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-
-  /**
-   * Factory method to create a class wrapping a new DataServiceDataset operation.
-   *
-   * @param scope current scope
-   * @param datasetId the datasetId value
-   * @param processingMode the processingMode value
-   * @param address the address value
-   * @param protocol the protocol value
-   * @param jobName the jobName value
-   * @param maxOutstandingRequests the maxOutstandingRequests value
-   * @param iterationCounter the iterationCounter value
-   * @param outputTypes the value of the outputTypes property
-   * @param outputShapes the value of the outputShapes property
-   * @param options carries optional attribute values
-   * @return a new instance of DataServiceDataset
-   */
-  @Endpoint(
-      describeByClass = true
-  )
-  public static DataServiceDataset create(Scope scope, Operand<TInt64> datasetId,
-      Operand<TString> processingMode, Operand<TString> address, Operand<TString> protocol,
-      Operand<TString> jobName, Operand<TInt64> maxOutstandingRequests,
-      Operand<? extends TType> iterationCounter, List<Class<? extends TType>> outputTypes,
-      List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("DataServiceDataset"));
-    opBuilder.addInput(datasetId.asOutput());
-    opBuilder.addInput(processingMode.asOutput());
-    opBuilder.addInput(address.asOutput());
-    opBuilder.addInput(protocol.asOutput());
-    opBuilder.addInput(jobName.asOutput());
-    opBuilder.addInput(maxOutstandingRequests.asOutput());
-    opBuilder.addInput(iterationCounter.asOutput());
-    opBuilder = scope.apply(opBuilder);
-    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0 ; i < outputShapesArray.length ; i++) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.taskRefreshIntervalHintMs != null) {
-          opBuilder.setAttr("task_refresh_interval_hint_ms", opts.taskRefreshIntervalHintMs);
-        }
-        if (opts.dataTransferProtocol != null) {
-          opBuilder.setAttr("data_transfer_protocol", opts.dataTransferProtocol);
-        }
-      }
-    }
-    return new DataServiceDataset(opBuilder.build());
-  }
-
-  /**
-   * Sets the taskRefreshIntervalHintMs option.
-   *
-   * @param taskRefreshIntervalHintMs the taskRefreshIntervalHintMs option
-   * @return this Options instance.
-   */
-  public static Options taskRefreshIntervalHintMs(Long taskRefreshIntervalHintMs) {
-    return new Options().taskRefreshIntervalHintMs(taskRefreshIntervalHintMs);
-  }
-
-  /**
-   * Sets the dataTransferProtocol option.
-   *
-   * @param dataTransferProtocol the dataTransferProtocol option
-   * @return this Options instance.
-   */
-  public static Options dataTransferProtocol(String dataTransferProtocol) {
-    return new Options().dataTransferProtocol(dataTransferProtocol);
-  }
-
-  /**
-   * Gets handle.
-   *
-   * @return handle.
-   */
-  public Output<? extends TType> handle() {
-    return handle;
-  }
-
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
-  }
-
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.experimental.DataServiceDataset}
-   */
-  public static class Options {
-    private Long taskRefreshIntervalHintMs;
-
-    private String dataTransferProtocol;
-
-    private Options() {
-    }
-
-    /**
-     * Sets the taskRefreshIntervalHintMs option.
-     *
-     * @param taskRefreshIntervalHintMs the taskRefreshIntervalHintMs option
-     * @return this Options instance.
-     */
-    public Options taskRefreshIntervalHintMs(Long taskRefreshIntervalHintMs) {
-      this.taskRefreshIntervalHintMs = taskRefreshIntervalHintMs;
-      return this;
-    }
-
-    /**
-     * Sets the dataTransferProtocol option.
-     *
-     * @param dataTransferProtocol the dataTransferProtocol option
-     * @return this Options instance.
-     */
-    public Options dataTransferProtocol(String dataTransferProtocol) {
-      this.dataTransferProtocol = dataTransferProtocol;
-      return this;
-    }
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/GroupByReducerDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/GroupByReducerDataset.java
new file mode 100644
index 00000000000..1b2c46e2de1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/GroupByReducerDataset.java
@@ -0,0 +1,119 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that computes a group-by on {@code input_dataset}.
+ * Creates a dataset that computes a group-by on {@code input_dataset}.
+ */
+public final class GroupByReducerDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ExperimentalGroupByReducerDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private GroupByReducerDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ExperimentalGroupByReducerDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param keyFuncOtherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code key_func}.
+   * @param initFuncOtherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code init_func}.
+   * @param reduceFuncOtherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code reduce_func}.
+   * @param finalizeFuncOtherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code finalize_func}.
+   * @param keyFunc A function mapping an element of {@code input_dataset}, concatenated
+   * with {@code key_func_other_arguments} to a scalar value of type DT_INT64.
+   * @param initFunc A function mapping a key of type DT_INT64, concatenated with
+   * {@code init_func_other_arguments} to the initial reducer state.
+   * @param reduceFunc A function mapping the current reducer state and an element of {@code input_dataset},
+   * concatenated with {@code reduce_func_other_arguments} to a new reducer state.
+   * @param finalizeFunc A function mapping the final reducer state to an output element.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of GroupByReducerDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static GroupByReducerDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> keyFuncOtherArguments, Iterable<Operand<?>> initFuncOtherArguments,
+      Iterable<Operand<?>> reduceFuncOtherArguments,
+      Iterable<Operand<?>> finalizeFuncOtherArguments, ConcreteFunction keyFunc,
+      ConcreteFunction initFunc, ConcreteFunction reduceFunc, ConcreteFunction finalizeFunc,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("GroupByReducerDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(keyFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(initFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(reduceFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(finalizeFuncOtherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("key_func", keyFunc);
+    opBuilder.setAttr("init_func", initFunc);
+    opBuilder.setAttr("reduce_func", reduceFunc);
+    opBuilder.setAttr("finalize_func", finalizeFunc);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new GroupByReducerDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/GroupByWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/GroupByWindowDataset.java
new file mode 100644
index 00000000000..043b87f399b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/GroupByWindowDataset.java
@@ -0,0 +1,108 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that computes a windowed group-by on {@code input_dataset}.
+ * // TODO(mrry): Support non-int64 keys.
+ */
+public final class GroupByWindowDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ExperimentalGroupByWindowDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private GroupByWindowDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ExperimentalGroupByWindowDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param keyFuncOtherArguments the keyFuncOtherArguments value
+   * @param reduceFuncOtherArguments the reduceFuncOtherArguments value
+   * @param windowSizeFuncOtherArguments the windowSizeFuncOtherArguments value
+   * @param keyFunc A function mapping an element of {@code input_dataset}, concatenated
+   * with {@code key_func_other_arguments} to a scalar value of type DT_INT64.
+   * @param reduceFunc the value of the reduceFunc property
+   * @param windowSizeFunc the value of the windowSizeFunc property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of GroupByWindowDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static GroupByWindowDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> keyFuncOtherArguments, Iterable<Operand<?>> reduceFuncOtherArguments,
+      Iterable<Operand<?>> windowSizeFuncOtherArguments, ConcreteFunction keyFunc,
+      ConcreteFunction reduceFunc, ConcreteFunction windowSizeFunc,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("GroupByWindowDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(keyFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(reduceFuncOtherArguments));
+    opBuilder.addInputList(Operands.asOutputs(windowSizeFuncOtherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("key_func", keyFunc);
+    opBuilder.setAttr("reduce_func", reduceFunc);
+    opBuilder.setAttr("window_size_func", windowSizeFunc);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new GroupByWindowDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MapAndBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MapAndBatchDataset.java
new file mode 100644
index 00000000000..3a86ebd3ed3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MapAndBatchDataset.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that fuses mapping with batching.
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset} and then
+ * batches {@code batch_size} of them.
+ * <p>Unlike a &quot;MapDataset&quot;, which applies {@code f} sequentially, this dataset invokes up
+ * to {@code batch_size * num_parallel_batches} copies of {@code f} in parallel.
+ */
+public final class MapAndBatchDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ExperimentalMapAndBatchDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private MapAndBatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ExperimentalMapAndBatchDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param otherArguments A list of tensors, typically values that were captured when building a closure
+   * for {@code f}.
+   * @param batchSize A scalar representing the number of elements to accumulate in a
+   * batch. It determines the number of concurrent invocations of {@code f} that process
+   * elements from {@code input_dataset} in parallel.
+   * @param numParallelCalls A scalar representing the maximum number of parallel invocations of the {@code map_fn}
+   * function. Applying the {@code map_fn} on consecutive input elements in parallel has
+   * the potential to improve input pipeline throughput.
+   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
+   * is smaller than desired.
+   * @param f A function to apply to the outputs of {@code input_dataset}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of MapAndBatchDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static MapAndBatchDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> batchSize,
+      Operand<TInt64> numParallelCalls, Operand<TBool> dropRemainder, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("MapAndBatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder.addInput(batchSize.asOutput());
+    opBuilder.addInput(numParallelCalls.asOutput());
+    opBuilder.addInput(dropRemainder.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.preserveCardinality != null) {
+          opBuilder.setAttr("preserve_cardinality", opts.preserveCardinality);
+        }
+      }
+    }
+    return new MapAndBatchDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the preserveCardinality option.
+   *
+   * @param preserveCardinality the preserveCardinality option
+   * @return this Options instance.
+   */
+  public static Options preserveCardinality(Boolean preserveCardinality) {
+    return new Options().preserveCardinality(preserveCardinality);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.MapAndBatchDataset}
+   */
+  public static class Options {
+    private Boolean preserveCardinality;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the preserveCardinality option.
+     *
+     * @param preserveCardinality the preserveCardinality option
+     * @return this Options instance.
+     */
+    public Options preserveCardinality(Boolean preserveCardinality) {
+      this.preserveCardinality = preserveCardinality;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MapDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MapDataset.java
new file mode 100644
index 00000000000..efab3bb9ef3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MapDataset.java
@@ -0,0 +1,161 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+ */
+public final class MapDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ExperimentalMapDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private MapDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ExperimentalMapDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param f the value of the f property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of MapDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static MapDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("MapDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useInterOpParallelism != null) {
+          opBuilder.setAttr("use_inter_op_parallelism", opts.useInterOpParallelism);
+        }
+        if (opts.preserveCardinality != null) {
+          opBuilder.setAttr("preserve_cardinality", opts.preserveCardinality);
+        }
+      }
+    }
+    return new MapDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the useInterOpParallelism option.
+   *
+   * @param useInterOpParallelism the useInterOpParallelism option
+   * @return this Options instance.
+   */
+  public static Options useInterOpParallelism(Boolean useInterOpParallelism) {
+    return new Options().useInterOpParallelism(useInterOpParallelism);
+  }
+
+  /**
+   * Sets the preserveCardinality option.
+   *
+   * @param preserveCardinality the preserveCardinality option
+   * @return this Options instance.
+   */
+  public static Options preserveCardinality(Boolean preserveCardinality) {
+    return new Options().preserveCardinality(preserveCardinality);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.MapDataset}
+   */
+  public static class Options {
+    private Boolean useInterOpParallelism;
+
+    private Boolean preserveCardinality;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the useInterOpParallelism option.
+     *
+     * @param useInterOpParallelism the useInterOpParallelism option
+     * @return this Options instance.
+     */
+    public Options useInterOpParallelism(Boolean useInterOpParallelism) {
+      this.useInterOpParallelism = useInterOpParallelism;
+      return this;
+    }
+
+    /**
+     * Sets the preserveCardinality option.
+     *
+     * @param preserveCardinality the preserveCardinality option
+     * @return this Options instance.
+     */
+    public Options preserveCardinality(Boolean preserveCardinality) {
+      this.preserveCardinality = preserveCardinality;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParallelInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParallelInterleaveDataset.java
new file mode 100644
index 00000000000..0aa546f2f9c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParallelInterleaveDataset.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that applies {@code f} to the outputs of {@code input_dataset}.
+ * The resulting dataset is similar to the {@code InterleaveDataset}, with the exception
+ * that if retrieving the next value from a dataset would cause the requester to
+ * block, it will skip that input dataset. This dataset is especially useful
+ * when loading data from a variable-latency datastores (e.g. HDFS, GCS), as it
+ * allows the training step to proceed so long as some data is available.
+ * <p>!! WARNING !! This dataset is not deterministic!
+ */
+public final class ParallelInterleaveDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ExperimentalParallelInterleaveDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private ParallelInterleaveDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ExperimentalParallelInterleaveDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param otherArguments the otherArguments value
+   * @param cycleLength the cycleLength value
+   * @param blockLength the blockLength value
+   * @param sloppy the sloppy value
+   * @param bufferOutputElements the bufferOutputElements value
+   * @param prefetchInputElements the prefetchInputElements value
+   * @param f A function mapping elements of {@code input_dataset}, concatenated with
+   * {@code other_arguments}, to a Dataset variant that contains elements matching
+   * {@code output_types} and {@code output_shapes}.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of ParallelInterleaveDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static ParallelInterleaveDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, Operand<TInt64> cycleLength, Operand<TInt64> blockLength,
+      Operand<TBool> sloppy, Operand<TInt64> bufferOutputElements,
+      Operand<TInt64> prefetchInputElements, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ParallelInterleaveDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder.addInput(cycleLength.asOutput());
+    opBuilder.addInput(blockLength.asOutput());
+    opBuilder.addInput(sloppy.asOutput());
+    opBuilder.addInput(bufferOutputElements.asOutput());
+    opBuilder.addInput(prefetchInputElements.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new ParallelInterleaveDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
index c3e179c29db..6a569cb0c68 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.data.experimental;
 
-import java.util.Arrays;
 import java.util.List;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
@@ -29,7 +28,6 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -39,7 +37,7 @@ public final class ParseExampleDataset extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "ParseExampleDatasetV2";
+  public static final String OP_NAME = "ExperimentalParseExampleDataset";
 
   private Output<? extends TType> handle;
 
@@ -51,7 +49,7 @@ private ParseExampleDataset(Operation operation) {
   }
 
   /**
-   * Factory method to create a class wrapping a new ParseExampleDatasetV2 operation.
+   * Factory method to create a class wrapping a new ExperimentalParseExampleDataset operation.
    *
    * @param scope current scope
    * @param inputDataset the inputDataset value
@@ -76,8 +74,6 @@ private ParseExampleDataset(Operation operation) {
    * given feature along this dimension.
    * @param outputTypes The type list for the return values.
    * @param outputShapes The list of shapes being produced.
-   * @param raggedValueTypes the value of the raggedValueTypes property
-   * @param raggedSplitTypes the value of the raggedSplitTypes property
    * @param options carries optional attribute values
    * @return a new instance of ParseExampleDataset
    */
@@ -87,9 +83,7 @@ private ParseExampleDataset(Operation operation) {
   public static ParseExampleDataset create(Scope scope, Operand<? extends TType> inputDataset,
       Operand<TInt64> numParallelCalls, Iterable<Operand<?>> denseDefaults, List<String> sparseKeys,
       List<String> denseKeys, List<Class<? extends TType>> sparseTypes, List<Shape> denseShapes,
-      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
-      List<Class<? extends TType>> raggedValueTypes,
-      List<Class<? extends TNumber>> raggedSplitTypes, Options... options) {
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ParseExampleDataset"));
     opBuilder.addInput(inputDataset.asOutput());
     opBuilder.addInput(numParallelCalls.asOutput());
@@ -117,19 +111,10 @@ public static ParseExampleDataset create(Scope scope, Operand<? extends TType> i
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
-    opBuilder.setAttr("ragged_value_types", Operands.toDataTypes(raggedValueTypes));
-    opBuilder.setAttr("ragged_split_types", Operands.toDataTypes(raggedSplitTypes));
     if (options != null) {
       for (Options opts : options) {
-        if (opts.deterministic != null) {
-          opBuilder.setAttr("deterministic", opts.deterministic);
-        }
-        if (opts.raggedKeys != null) {
-          String[] raggedKeysArray = new String[opts.raggedKeys.size()];
-          for (int i = 0 ; i < raggedKeysArray.length ; i++) {
-            raggedKeysArray[i] = opts.raggedKeys.get(i);
-          }
-          opBuilder.setAttr("ragged_keys", raggedKeysArray);
+        if (opts.sloppy != null) {
+          opBuilder.setAttr("sloppy", opts.sloppy);
         }
       }
     }
@@ -137,37 +122,13 @@ public static ParseExampleDataset create(Scope scope, Operand<? extends TType> i
   }
 
   /**
-   * Sets the deterministic option.
-   *
-   * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
-   * whether the dataset is allowed to return elements out of order if the next
-   * element to be returned isn't available, but a later element is. Options are
-   * &quot;true&quot;, &quot;false&quot;, and &quot;default&quot;. &quot;default&quot; indicates that determinism should be
-   * decided by the {@code experimental_deterministic} parameter of {@code tf.data.Options}.
-   * @return this Options instance.
-   */
-  public static Options deterministic(String deterministic) {
-    return new Options().deterministic(deterministic);
-  }
-
-  /**
-   * Sets the raggedKeys option.
+   * Sets the sloppy option.
    *
-   * @param raggedKeys the raggedKeys option
+   * @param sloppy the sloppy option
    * @return this Options instance.
    */
-  public static Options raggedKeys(List<String> raggedKeys) {
-    return new Options().raggedKeys(raggedKeys);
-  }
-
-  /**
-   * Sets the raggedKeys option.
-   *
-   * @param raggedKeys the raggedKeys option
-   * @return this Options instance.
-   */
-  public static Options raggedKeys(String[] raggedKeys) {
-    return new Options().raggedKeys(raggedKeys);
+  public static Options sloppy(Boolean sloppy) {
+    return new Options().sloppy(sloppy);
   }
 
   /**
@@ -189,47 +150,19 @@ public Output<TType> asOutput() {
    * Optional attributes for {@link org.tensorflow.op.data.experimental.ParseExampleDataset}
    */
   public static class Options {
-    private String deterministic;
-
-    private List<String> raggedKeys;
+    private Boolean sloppy;
 
     private Options() {
     }
 
     /**
-     * Sets the deterministic option.
-     *
-     * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
-     * whether the dataset is allowed to return elements out of order if the next
-     * element to be returned isn't available, but a later element is. Options are
-     * &quot;true&quot;, &quot;false&quot;, and &quot;default&quot;. &quot;default&quot; indicates that determinism should be
-     * decided by the {@code experimental_deterministic} parameter of {@code tf.data.Options}.
-     * @return this Options instance.
-     */
-    public Options deterministic(String deterministic) {
-      this.deterministic = deterministic;
-      return this;
-    }
-
-    /**
-     * Sets the raggedKeys option.
-     *
-     * @param raggedKeys the raggedKeys option
-     * @return this Options instance.
-     */
-    public Options raggedKeys(List<String> raggedKeys) {
-      this.raggedKeys = raggedKeys;
-      return this;
-    }
-
-    /**
-     * Sets the raggedKeys option.
+     * Sets the sloppy option.
      *
-     * @param raggedKeys the raggedKeys option
+     * @param sloppy the sloppy option
      * @return this Options instance.
      */
-    public Options raggedKeys(String... raggedKeys) {
-      this.raggedKeys = Arrays.asList(raggedKeys);
+    public Options sloppy(Boolean sloppy) {
+      this.sloppy = sloppy;
       return this;
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ScanDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ScanDataset.java
new file mode 100644
index 00000000000..e60d6024026
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ScanDataset.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset successively reduces {@code f} over the elements of {@code input_dataset}.
+ */
+public final class ScanDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ExperimentalScanDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private ScanDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ExperimentalScanDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param initialState the initialState value
+   * @param otherArguments the otherArguments value
+   * @param f the value of the f property
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @param options carries optional attribute values
+   * @return a new instance of ScanDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static ScanDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> initialState, Iterable<Operand<?>> otherArguments, ConcreteFunction f,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ScanDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(initialState));
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("f", f);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.preserveCardinality != null) {
+          opBuilder.setAttr("preserve_cardinality", opts.preserveCardinality);
+        }
+      }
+    }
+    return new ScanDataset(opBuilder.build());
+  }
+
+  /**
+   * Sets the preserveCardinality option.
+   *
+   * @param preserveCardinality the preserveCardinality option
+   * @return this Options instance.
+   */
+  public static Options preserveCardinality(Boolean preserveCardinality) {
+    return new Options().preserveCardinality(preserveCardinality);
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.ScanDataset}
+   */
+  public static class Options {
+    private Boolean preserveCardinality;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the preserveCardinality option.
+     *
+     * @param preserveCardinality the preserveCardinality option
+     * @return this Options instance.
+     */
+    public Options preserveCardinality(Boolean preserveCardinality) {
+      this.preserveCardinality = preserveCardinality;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
index e4183305b0f..4f1ea92e22d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
@@ -27,13 +27,13 @@
 import org.tensorflow.types.family.TType;
 
 /**
- * The StatsAggregatorHandleV2 operation
+ * Creates a statistics manager resource.
  */
 public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "StatsAggregatorHandleV2";
+  public static final String OP_NAME = "ExperimentalStatsAggregatorHandle";
 
   private Output<? extends TType> handle;
 
@@ -45,7 +45,7 @@ private StatsAggregatorHandle(Operation operation) {
   }
 
   /**
-   * Factory method to create a class wrapping a new StatsAggregatorHandleV2 operation.
+   * Factory method to create a class wrapping a new ExperimentalStatsAggregatorHandle operation.
    *
    * @param scope current scope
    * @param options carries optional attribute values
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/TakeWhileDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/TakeWhileDataset.java
new file mode 100644
index 00000000000..9f283a7eb1f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/TakeWhileDataset.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that stops iteration when predicate` is false.
+ * The {@code predicate} function must return a scalar boolean and accept the
+ * following arguments:
+ * <ul>
+ * <li>One tensor for each component of an element of {@code input_dataset}.</li>
+ * <li>One tensor for each value in {@code other_arguments}.</li>
+ * </ul>
+ */
+public final class TakeWhileDataset extends RawOp implements Operand<TType> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "ExperimentalTakeWhileDataset";
+
+  private Output<? extends TType> handle;
+
+  @SuppressWarnings("unchecked")
+  private TakeWhileDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new ExperimentalTakeWhileDataset operation.
+   *
+   * @param scope current scope
+   * @param inputDataset the inputDataset value
+   * @param otherArguments A list of tensors, typically values that were captured when
+   * building a closure for {@code predicate}.
+   * @param predicate A function returning a scalar boolean.
+   * @param outputTypes the value of the outputTypes property
+   * @param outputShapes the value of the outputShapes property
+   * @return a new instance of TakeWhileDataset
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static TakeWhileDataset create(Scope scope, Operand<? extends TType> inputDataset,
+      Iterable<Operand<?>> otherArguments, ConcreteFunction predicate,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("TakeWhileDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(otherArguments));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("predicate", predicate);
+    opBuilder.setAttr("output_types", Operands.toDataTypes(outputTypes));
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0 ; i < outputShapesArray.length ; i++) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new TakeWhileDataset(opBuilder.build());
+  }
+
+  /**
+   * Gets handle.
+   *
+   * @return handle.
+   */
+  public Output<? extends TType> handle() {
+    return handle;
+  }
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DummySeedGenerator.java
similarity index 97%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DummySeedGenerator.java
index b5f8baccd92..068f85a76df 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DummySeedGenerator.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.random.experimental;
+package org.tensorflow.op.random;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/risc/RiscCondition.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/risc/RiscCondition.java
new file mode 100644
index 00000000000..f6c37680a15
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/risc/RiscCondition.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.risc;
+
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * The RiscCondition operation
+ *
+ * @param <U> data type for {@code output} output
+ */
+public final class RiscCondition<U extends TNumber> extends RawOp implements Operand<U> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "RiscCondition";
+
+  private Output<U> output;
+
+  private RiscCondition(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new RiscCondition operation.
+   *
+   * @param scope current scope
+   * @param pred the pred value
+   * @param inputTrue the inputTrue value
+   * @param inputFalse the inputFalse value
+   * @param funcTrue the value of the funcTrue property
+   * @param funcFalse the value of the funcFalse property
+   * @param DstT the value of the DstT property
+   * @param <U> data type for {@code RiscCondition} output and operands
+   * @param <T> data type for {@code RiscCondition} output and operands
+   * @return a new instance of RiscCondition
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static <U extends TNumber, T extends TNumber> RiscCondition<U> create(Scope scope,
+      Operand<TBool> pred, Operand<T> inputTrue, Operand<T> inputFalse, ConcreteFunction funcTrue,
+      ConcreteFunction funcFalse, Class<U> DstT) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("RiscCondition"));
+    opBuilder.addInput(pred.asOutput());
+    opBuilder.addInput(inputTrue.asOutput());
+    opBuilder.addInput(inputFalse.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("func_true", funcTrue);
+    opBuilder.setAttr("func_false", funcFalse);
+    opBuilder.setAttr("DstT", Operands.toDataType(DstT));
+    return new RiscCondition<>(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   *
+   * @return output.
+   */
+  public Output<U> output() {
+    return output;
+  }
+
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/risc/RiscWhile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/risc/RiscWhile.java
new file mode 100644
index 00000000000..328f098c70d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/risc/RiscWhile.java
@@ -0,0 +1,181 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.risc;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.family.TType;
+
+/**
+ * The RiscWhile operation
+ */
+public final class RiscWhile extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "RiscWhile";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private RiscWhile(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new RiscWhile operation.
+   *
+   * @param scope current scope
+   * @param input the input value
+   * @param cond the value of the cond property
+   * @param body the value of the body property
+   * @param options carries optional attribute values
+   * @return a new instance of RiscWhile
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static RiscWhile create(Scope scope, Iterable<Operand<?>> input, ConcreteFunction cond,
+      ConcreteFunction body, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("RiscWhile"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("cond", cond);
+    opBuilder.setAttr("body", body);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0 ; i < outputShapesArray.length ; i++) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+        if (opts.parallelIterations != null) {
+          opBuilder.setAttr("parallel_iterations", opts.parallelIterations);
+        }
+      }
+    }
+    return new RiscWhile(opBuilder.build());
+  }
+
+  /**
+   * Sets the outputShapes option.
+   *
+   * @param outputShapes the outputShapes option
+   * @return this Options instance.
+   */
+  public static Options outputShapes(List<Shape> outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+
+  /**
+   * Sets the outputShapes option.
+   *
+   * @param outputShapes the outputShapes option
+   * @return this Options instance.
+   */
+  public static Options outputShapes(Shape[] outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+
+  /**
+   * Sets the parallelIterations option.
+   *
+   * @param parallelIterations the parallelIterations option
+   * @return this Options instance.
+   */
+  public static Options parallelIterations(Long parallelIterations) {
+    return new Options().parallelIterations(parallelIterations);
+  }
+
+  /**
+   * Gets output.
+   *
+   * @return output.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.risc.RiscWhile}
+   */
+  public static class Options {
+    private List<Shape> outputShapes;
+
+    private Long parallelIterations;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the outputShapes option.
+     *
+     * @param outputShapes the outputShapes option
+     * @return this Options instance.
+     */
+    public Options outputShapes(List<Shape> outputShapes) {
+      this.outputShapes = outputShapes;
+      return this;
+    }
+
+    /**
+     * Sets the outputShapes option.
+     *
+     * @param outputShapes the outputShapes option
+     * @return this Options instance.
+     */
+    public Options outputShapes(Shape... outputShapes) {
+      this.outputShapes = Arrays.asList(outputShapes);
+      return this;
+    }
+
+    /**
+     * Sets the parallelIterations option.
+     *
+     * @param parallelIterations the parallelIterations option
+     * @return this Options instance.
+     */
+    public Options parallelIterations(Long parallelIterations) {
+      this.parallelIterations = parallelIterations;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Compile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Compile.java
new file mode 100644
index 00000000000..43725ba9b0b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Compile.java
@@ -0,0 +1,134 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Compiles a computations for execution on one or more TPU devices.
+ * For the internal use of the distributed TPU compiler.
+ * <p>'num_computations' is the number of computations to be compiled.
+ * 'function' is a function containing the computation to compile.
+ * 'dynamic_shapes' contains dynamic shapes of arguments whose shapes were not
+ * known statically at TPUReplication rewrite time.
+ * 'guaranteed_constants' is a list of tensors which have been guaranteed to not
+ * change their values during the session lifetime. These contain tensors marked as
+ * constant using the GuaranteeConstOp.
+ * 'metadata' is a serialized TPUCompileMetadataProto describing
+ * the shapes and types of the inputs to the computation, as well as a mapping onto
+ * the TPU pod topology.
+ * Each 'program' output is a string key that is passed to the _TPUExecute op and
+ * used to look up the program in the compilation cache.
+ * 'may_modify_variables' indicates whether variables may be modified.
+ */
+@Operator(
+    group = "tpu"
+)
+public final class Compile extends RawOp {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "TPUCompile";
+
+  private Output<TString> compilationStatus;
+
+  private List<Output<TString>> program;
+
+  private List<Output<TBool>> mayModifyVariables;
+
+  @SuppressWarnings("unchecked")
+  private Compile(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    compilationStatus = operation.output(outputIdx++);
+    int programLength = operation.outputListLength("program");
+    program = Arrays.asList((Output<TString>[]) operation.outputList(outputIdx, programLength));
+    outputIdx += programLength;
+    int mayModifyVariablesLength = operation.outputListLength("may_modify_variables");
+    mayModifyVariables = Arrays.asList((Output<TBool>[]) operation.outputList(outputIdx, mayModifyVariablesLength));
+    outputIdx += mayModifyVariablesLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new TPUCompile operation.
+   *
+   * @param scope current scope
+   * @param dynamicShapes the dynamicShapes value
+   * @param guaranteedConstants the guaranteedConstants value
+   * @param numComputations the value of the numComputations property
+   * @param function the value of the function property
+   * @param metadata the value of the metadata property
+   * @return a new instance of Compile
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static Compile create(Scope scope, Iterable<Operand<TInt64>> dynamicShapes,
+      Iterable<Operand<?>> guaranteedConstants, Long numComputations, ConcreteFunction function,
+      String metadata) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("Compile"));
+    opBuilder.addInputList(Operands.asOutputs(dynamicShapes));
+    opBuilder.addInputList(Operands.asOutputs(guaranteedConstants));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("num_computations", numComputations);
+    opBuilder.setAttr("function", function);
+    opBuilder.setAttr("metadata", metadata);
+    return new Compile(opBuilder.build());
+  }
+
+  /**
+   * Gets compilationStatus.
+   *
+   * @return compilationStatus.
+   */
+  public Output<TString> compilationStatus() {
+    return compilationStatus;
+  }
+
+  /**
+   * Gets program.
+   *
+   * @return program.
+   */
+  public List<Output<TString>> program() {
+    return program;
+  }
+
+  /**
+   * Gets mayModifyVariables.
+   *
+   * @return mayModifyVariables.
+   */
+  public List<Output<TBool>> mayModifyVariables() {
+    return mayModifyVariables;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PartitionedCall.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PartitionedCall.java
new file mode 100644
index 00000000000..a4086db1728
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PartitionedCall.java
@@ -0,0 +1,133 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Calls a function placed on a specified TPU device.
+ */
+public final class PartitionedCall extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "TPUPartitionedCall";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private PartitionedCall(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new TPUPartitionedCall operation.
+   *
+   * @param scope current scope
+   * @param args The arguments to the function.
+   * @param deviceOrdinal The TPU device ordinal to run the function on.
+   * @param Tout The types of the outputs of the function.
+   * @param f The function to call.
+   * @param options carries optional attribute values
+   * @return a new instance of PartitionedCall
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static PartitionedCall create(Scope scope, Iterable<Operand<?>> args,
+      Operand<TInt32> deviceOrdinal, List<Class<? extends TType>> Tout, ConcreteFunction f,
+      Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("PartitionedCall"));
+    opBuilder.addInputList(Operands.asOutputs(args));
+    opBuilder.addInput(deviceOrdinal.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    opBuilder.setAttr("f", f);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.autotunerThresh != null) {
+          opBuilder.setAttr("autotuner_thresh", opts.autotunerThresh);
+        }
+      }
+    }
+    return new PartitionedCall(opBuilder.build());
+  }
+
+  /**
+   * Sets the autotunerThresh option.
+   *
+   * @param autotunerThresh the autotunerThresh option
+   * @return this Options instance.
+   */
+  public static Options autotunerThresh(Long autotunerThresh) {
+    return new Options().autotunerThresh(autotunerThresh);
+  }
+
+  /**
+   * Gets output.
+   * The output of the function call.
+   * @return output.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.PartitionedCall}
+   */
+  public static class Options {
+    private Long autotunerThresh;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the autotunerThresh option.
+     *
+     * @param autotunerThresh the autotunerThresh option
+     * @return this Options instance.
+     */
+    public Options autotunerThresh(Long autotunerThresh) {
+      this.autotunerThresh = autotunerThresh;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SymbolicGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SymbolicGradient.java
new file mode 100644
index 00000000000..e58ce0b2afb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SymbolicGradient.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the gradient function for function f via backpropagation.
+ */
+@Operator(
+    group = "train"
+)
+public final class SymbolicGradient extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "SymbolicGradient";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private SymbolicGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new SymbolicGradient operation.
+   *
+   * @param scope current scope
+   * @param input a list of input tensors of size N + M;
+   * @param Tout the type list for the input list.
+   * @param f The function we want to compute the gradient for.
+   * <p>The function 'f' must be a numerical function which takes N inputs and
+   * produces M outputs. Its gradient function 'g', which is computed by
+   * this SymbolicGradient op is a function taking N + M inputs and
+   * produces N outputs.
+   * <p>I.e. if we have
+   * (y1, y2, ..., y_M) = f(x1, x2, ..., x_N),
+   * then, g is
+   * (dL/dx1, dL/dx2, ..., dL/dx_N) = g(x1, x2, ..., x_N,
+   * dL/dy1, dL/dy2, ..., dL/dy_M),
+   * <p>where L is a scalar-value function of (x1, x2, ..., xN) (e.g., the
+   * loss function). dL/dx_i is the partial derivative of L with respect
+   * to x_i.
+   * <p>(Needs some math expert to say the comment above better.)
+   * @return a new instance of SymbolicGradient
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static SymbolicGradient create(Scope scope, Iterable<Operand<?>> input,
+      List<Class<? extends TType>> Tout, ConcreteFunction f) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("SymbolicGradient"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    opBuilder.setAttr("f", f);
+    return new SymbolicGradient(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * a list of output tensors of size N;
+   * @return output.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
index 1fe493c4223..fc238e7ebc5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,18 +34,18 @@
  * https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
  * .
  *
- * @param <T> data type for {@code output} output
+ * @param <W> data type for {@code output} output
  */
 @Operator(
     group = "xla"
 )
-public final class Conv<T extends TType> extends RawOp implements Operand<T> {
+public final class Conv<W extends TType> extends RawOp implements Operand<W> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "XlaConv";
+  public static final String OP_NAME = "XlaConvV2";
 
-  private Output<T> output;
+  private Output<W> output;
 
   private Conv(Operation operation) {
     super(operation);
@@ -53,7 +54,7 @@ private Conv(Operation operation) {
   }
 
   /**
-   * Factory method to create a class wrapping a new XlaConv operation.
+   * Factory method to create a class wrapping a new XlaConvV2 operation.
    *
    * @param scope current scope
    * @param lhs the input tensor
@@ -65,17 +66,19 @@ private Conv(Operation operation) {
    * @param featureGroupCount number of feature groups for grouped convolution.
    * @param dimensionNumbers a serialized xla::ConvolutionDimensionNumbers proto.
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @param <T> data type for {@code XlaConv} output and operands
-   * @param <U> data type for {@code XlaConv} output and operands
+   * @param preferredElementType The type of the tensor.
+   * @param <W> data type for {@code XlaConvV2} output and operands
+   * @param <V> data type for {@code XlaConvV2} output and operands
    * @return a new instance of Conv
    */
   @Endpoint(
       describeByClass = true
   )
-  public static <T extends TType, U extends TNumber> Conv<T> create(Scope scope, Operand<T> lhs,
-      Operand<T> rhs, Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation,
-      Operand<U> rhsDilation, Operand<U> featureGroupCount, String dimensionNumbers,
-      String precisionConfig) {
+  public static <W extends TType, V extends TNumber> Conv<W> create(Scope scope,
+      Operand<? extends TType> lhs, Operand<? extends TType> rhs, Operand<V> windowStrides,
+      Operand<V> padding, Operand<V> lhsDilation, Operand<V> rhsDilation,
+      Operand<V> featureGroupCount, String dimensionNumbers, String precisionConfig,
+      Class<W> preferredElementType) {
     OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("Conv"));
     opBuilder.addInput(lhs.asOutput());
     opBuilder.addInput(rhs.asOutput());
@@ -87,6 +90,7 @@ public static <T extends TType, U extends TNumber> Conv<T> create(Scope scope, O
     opBuilder = scope.apply(opBuilder);
     opBuilder.setAttr("dimension_numbers", dimensionNumbers);
     opBuilder.setAttr("precision_config", precisionConfig);
+    opBuilder.setAttr("preferred_element_type", Operands.toDataType(preferredElementType));
     return new Conv<>(opBuilder.build());
   }
 
@@ -95,12 +99,12 @@ public static <T extends TType, U extends TNumber> Conv<T> create(Scope scope, O
    *
    * @return output.
    */
-  public Output<T> output() {
+  public Output<W> output() {
     return output;
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<W> asOutput() {
     return output;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
index 3de0491e718..420ad933a2b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,18 +33,18 @@
  * https://www.tensorflow.org/performance/xla/operation_semantics#dotgeneral
  * .
  *
- * @param <T> data type for {@code output} output
+ * @param <V> data type for {@code output} output
  */
 @Operator(
     group = "xla"
 )
-public final class Dot<T extends TType> extends RawOp implements Operand<T> {
+public final class Dot<V extends TType> extends RawOp implements Operand<V> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "XlaDot";
+  public static final String OP_NAME = "XlaDotV2";
 
-  private Output<T> output;
+  private Output<V> output;
 
   private Dot(Operation operation) {
     super(operation);
@@ -52,27 +53,30 @@ private Dot(Operation operation) {
   }
 
   /**
-   * Factory method to create a class wrapping a new XlaDot operation.
+   * Factory method to create a class wrapping a new XlaDotV2 operation.
    *
    * @param scope current scope
    * @param lhs the LHS tensor
    * @param rhs the RHS tensor
    * @param dimensionNumbers a serialized xla::DotDimensionNumbers proto.
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @param <T> data type for {@code XlaDot} output and operands
+   * @param preferredElementType The type of the tensor.
+   * @param <V> data type for {@code XlaDotV2} output and operands
    * @return a new instance of Dot
    */
   @Endpoint(
       describeByClass = true
   )
-  public static <T extends TType> Dot<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs,
-      String dimensionNumbers, String precisionConfig) {
+  public static <V extends TType> Dot<V> create(Scope scope, Operand<? extends TType> lhs,
+      Operand<? extends TType> rhs, String dimensionNumbers, String precisionConfig,
+      Class<V> preferredElementType) {
     OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("Dot"));
     opBuilder.addInput(lhs.asOutput());
     opBuilder.addInput(rhs.asOutput());
     opBuilder = scope.apply(opBuilder);
     opBuilder.setAttr("dimension_numbers", dimensionNumbers);
     opBuilder.setAttr("precision_config", precisionConfig);
+    opBuilder.setAttr("preferred_element_type", Operands.toDataType(preferredElementType));
     return new Dot<>(opBuilder.build());
   }
 
@@ -81,12 +85,12 @@ public static <T extends TType> Dot<T> create(Scope scope, Operand<T> lhs, Opera
    *
    * @return output.
    */
-  public Output<T> output() {
+  public Output<V> output() {
     return output;
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<V> asOutput() {
     return output;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/If.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/If.java
new file mode 100644
index 00000000000..c0bc7d5001b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/If.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * output = cond ? then_branch(inputs) : else_branch(inputs).
+ */
+@Operator(
+    group = "xla"
+)
+public final class If extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaIf";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private If(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaIf operation.
+   *
+   * @param scope current scope
+   * @param cond A boolean scalar.
+   * @param inputs A list of input tensors.
+   * @param thenBranch A function takes 'inputs' and returns a list of tensors,
+   * whose types are the same as what else_branch returns.
+   * @param elseBranch A function takes 'inputs' and returns a list of tensors.
+   * whose types are the same as what then_branch returns.
+   * @param Tout the value of the Tout property
+   * @return a new instance of If
+   */
+  @Endpoint(
+      describeByClass = true,
+      name = "ifOp"
+  )
+  public static If create(Scope scope, Operand<? extends TType> cond, Iterable<Operand<?>> inputs,
+      ConcreteFunction thenBranch, ConcreteFunction elseBranch, List<Class<? extends TType>> Tout) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("If"));
+    opBuilder.addInput(cond.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("then_branch", thenBranch);
+    opBuilder.setAttr("else_branch", elseBranch);
+    opBuilder.setAttr("Tout", Operands.toDataTypes(Tout));
+    return new If(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of tensors returned by either then_branch(inputs) or
+   * else_branch(inputs). The input shapes of the then_branch and
+   * else_branch must match.
+   * @return output.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaDotV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Reduce.java
similarity index 50%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaDotV2.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Reduce.java
index 6d94739df68..ec39d1def2c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaDotV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Reduce.java
@@ -15,13 +15,14 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.core;
+package org.tensorflow.op.xla;
 
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,53 +30,55 @@
 import org.tensorflow.types.family.TType;
 
 /**
- * Wraps the XLA DotGeneral operator, documented at
- * https://www.tensorflow.org/performance/xla/operation_semantics#dotgeneral
- * .
+ * Wraps the XLA Reduce operator, documented at
+ * https://www.tensorflow.org/performance/xla/operation_semantics#reduce .
  *
- * @param <V> data type for {@code output} output
+ * @param <T> data type for {@code output} output
  */
-@Operator
-public final class XlaDotV2<V extends TType> extends RawOp implements Operand<V> {
+@Operator(
+    group = "xla"
+)
+public final class Reduce<T extends TType> extends RawOp implements Operand<T> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
-  public static final String OP_NAME = "XlaDotV2";
+  public static final String OP_NAME = "XlaReduce";
 
-  private Output<V> output;
+  private Output<T> output;
 
-  private XlaDotV2(Operation operation) {
+  private Reduce(Operation operation) {
     super(operation);
     int outputIdx = 0;
     output = operation.output(outputIdx++);
   }
 
   /**
-   * Factory method to create a class wrapping a new XlaDotV2 operation.
+   * Factory method to create a class wrapping a new XlaReduce operation.
    *
    * @param scope current scope
-   * @param lhs the LHS tensor
-   * @param rhs the RHS tensor
-   * @param dimensionNumbers a serialized xla::DotDimensionNumbers proto.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @param preferredElementType The type of the tensor.
-   * @param <V> data type for {@code XlaDotV2} output and operands
-   * @return a new instance of XlaDotV2
+   * @param input the input tensor
+   * @param initValue a scalar representing the initial value for the reduction
+   * @param dimensionsToReduce dimension numbers over which to reduce
+   * @param reducer a reducer function to apply
+   * @param <T> data type for {@code XlaReduce} output and operands
+   * @return a new instance of Reduce
    */
   @Endpoint(
       describeByClass = true
   )
-  public static <V extends TType> XlaDotV2<V> create(Scope scope, Operand<? extends TType> lhs,
-      Operand<? extends TType> rhs, String dimensionNumbers, String precisionConfig,
-      Class<V> preferredElementType) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaDotV2"));
-    opBuilder.addInput(lhs.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+  public static <T extends TType> Reduce<T> create(Scope scope, Operand<T> input,
+      Operand<T> initValue, List<Long> dimensionsToReduce, ConcreteFunction reducer) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("Reduce"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(initValue.asOutput());
     opBuilder = scope.apply(opBuilder);
-    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
-    opBuilder.setAttr("precision_config", precisionConfig);
-    opBuilder.setAttr("preferred_element_type", Operands.toDataType(preferredElementType));
-    return new XlaDotV2<>(opBuilder.build());
+    long[] dimensionsToReduceArray = new long[dimensionsToReduce.size()];
+    for (int i = 0 ; i < dimensionsToReduceArray.length ; i++) {
+      dimensionsToReduceArray[i] = dimensionsToReduce.get(i);
+    }
+    opBuilder.setAttr("dimensions_to_reduce", dimensionsToReduceArray);
+    opBuilder.setAttr("reducer", reducer);
+    return new Reduce<>(opBuilder.build());
   }
 
   /**
@@ -83,12 +86,12 @@ public static <V extends TType> XlaDotV2<V> create(Scope scope, Operand<? extend
    *
    * @return output.
    */
-  public Output<V> output() {
+  public Output<T> output() {
     return output;
   }
 
   @Override
-  public Output<V> asOutput() {
+  public Output<T> asOutput() {
     return output;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReduceWindow.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReduceWindow.java
new file mode 100644
index 00000000000..7e2a4eaf100
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReduceWindow.java
@@ -0,0 +1,104 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA ReduceWindow operator, documented at
+ * https://www.tensorflow.org/performance/xla/operation_semantics#reducewindow .
+ *
+ * @param <T> data type for {@code output} output
+ */
+@Operator(
+    group = "xla"
+)
+public final class ReduceWindow<T extends TType> extends RawOp implements Operand<T> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaReduceWindow";
+
+  private Output<T> output;
+
+  private ReduceWindow(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaReduceWindow operation.
+   *
+   * @param scope current scope
+   * @param input the input tensor
+   * @param initValue a scalar representing the initial value for the reduction
+   * @param windowDimensions the shape of the window
+   * @param windowStrides the inter-window strides
+   * @param baseDilations the baseDilations value
+   * @param windowDilations the windowDilations value
+   * @param padding the padding to apply at the start and end of each input dimensions
+   * @param computation a reducer function to apply
+   * @param <T> data type for {@code XlaReduceWindow} output and operands
+   * @param <U> data type for {@code XlaReduceWindow} output and operands
+   * @return a new instance of ReduceWindow
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static <T extends TType, U extends TNumber> ReduceWindow<T> create(Scope scope,
+      Operand<T> input, Operand<T> initValue, Operand<U> windowDimensions, Operand<U> windowStrides,
+      Operand<U> baseDilations, Operand<U> windowDilations, Operand<U> padding,
+      ConcreteFunction computation) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("ReduceWindow"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(initValue.asOutput());
+    opBuilder.addInput(windowDimensions.asOutput());
+    opBuilder.addInput(windowStrides.asOutput());
+    opBuilder.addInput(baseDilations.asOutput());
+    opBuilder.addInput(windowDilations.asOutput());
+    opBuilder.addInput(padding.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("computation", computation);
+    return new ReduceWindow<>(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   *
+   * @return output.
+   */
+  public Output<T> output() {
+    return output;
+  }
+
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Scatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Scatter.java
new file mode 100644
index 00000000000..9a249b0e862
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Scatter.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA Scatter operator documented at
+ * https://www.tensorflow.org/xla/operation_semantics#scatter.
+ *
+ * @param <T> data type for {@code output} output
+ */
+@Operator(
+    group = "xla"
+)
+public final class Scatter<T extends TType> extends RawOp implements Operand<T> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaScatter";
+
+  private Output<T> output;
+
+  private Scatter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaScatter operation.
+   *
+   * @param scope current scope
+   * @param operand Array to be scattered into.
+   * @param scatterIndices Array containing the starting indices of the slices that must
+   * be scattered to.
+   * @param updates Array containing the values that must be used for scattering.
+   * @param updateComputation Computation to be used for combining the existing values in
+   * the input array and the updates during scatter.
+   * @param dimensionNumbers A serialized xla::ScatterDimensionNumbers proto.
+   * @param indicesAreSorted Boolean indicating if the indices are sorted.
+   * @param <T> data type for {@code XlaScatter} output and operands
+   * @return a new instance of Scatter
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static <T extends TType> Scatter<T> create(Scope scope, Operand<T> operand,
+      Operand<? extends TNumber> scatterIndices, Operand<T> updates,
+      ConcreteFunction updateComputation, String dimensionNumbers, Boolean indicesAreSorted) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("Scatter"));
+    opBuilder.addInput(operand.asOutput());
+    opBuilder.addInput(scatterIndices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("update_computation", updateComputation);
+    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
+    opBuilder.setAttr("indices_are_sorted", indicesAreSorted);
+    return new Scatter<>(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   *
+   * @return output.
+   */
+  public Output<T> output() {
+    return output;
+  }
+
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelectAndScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelectAndScatter.java
new file mode 100644
index 00000000000..4e3b9753f96
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelectAndScatter.java
@@ -0,0 +1,104 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA SelectAndScatter operator, documented at
+ * https://www.tensorflow.org/performance/xla/operation_semantics#selectandscatter
+ * .
+ *
+ * @param <T> data type for {@code output} output
+ */
+@Operator(
+    group = "xla"
+)
+public final class SelectAndScatter<T extends TType> extends RawOp implements Operand<T> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaSelectAndScatter";
+
+  private Output<T> output;
+
+  private SelectAndScatter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaSelectAndScatter operation.
+   *
+   * @param scope current scope
+   * @param operand the input tensor
+   * @param windowDimensions the shape of the window
+   * @param windowStrides the inter-window strides
+   * @param padding the padding to apply at the start and end of each input dimensions
+   * @param source a tensor of values to scatter
+   * @param initValue a scalar representing the initial value for the output tensor
+   * @param select a selection function to apply
+   * @param scatter a scatter function to apply
+   * @param <T> data type for {@code XlaSelectAndScatter} output and operands
+   * @param <U> data type for {@code XlaSelectAndScatter} output and operands
+   * @return a new instance of SelectAndScatter
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static <T extends TType, U extends TNumber> SelectAndScatter<T> create(Scope scope,
+      Operand<T> operand, Operand<U> windowDimensions, Operand<U> windowStrides, Operand<U> padding,
+      Operand<T> source, Operand<T> initValue, ConcreteFunction select, ConcreteFunction scatter) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("SelectAndScatter"));
+    opBuilder.addInput(operand.asOutput());
+    opBuilder.addInput(windowDimensions.asOutput());
+    opBuilder.addInput(windowStrides.asOutput());
+    opBuilder.addInput(padding.asOutput());
+    opBuilder.addInput(source.asOutput());
+    opBuilder.addInput(initValue.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("select", select);
+    opBuilder.setAttr("scatter", scatter);
+    return new SelectAndScatter<>(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   *
+   * @return output.
+   */
+  public Output<T> output() {
+    return output;
+  }
+
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSetDynamicDimensionSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SetDynamicDimensionSize.java
similarity index 81%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSetDynamicDimensionSize.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SetDynamicDimensionSize.java
index 2ab7f6605ab..7388d3d9048 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSetDynamicDimensionSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SetDynamicDimensionSize.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.core;
+package org.tensorflow.op.xla;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
@@ -37,8 +37,10 @@
  *
  * @param <T> data type for {@code output} output
  */
-@Operator
-public final class XlaSetDynamicDimensionSize<T extends TType> extends RawOp implements Operand<T> {
+@Operator(
+    group = "xla"
+)
+public final class SetDynamicDimensionSize<T extends TType> extends RawOp implements Operand<T> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
@@ -46,7 +48,7 @@ public final class XlaSetDynamicDimensionSize<T extends TType> extends RawOp imp
 
   private Output<T> output;
 
-  private XlaSetDynamicDimensionSize(Operation operation) {
+  private SetDynamicDimensionSize(Operation operation) {
     super(operation);
     int outputIdx = 0;
     output = operation.output(outputIdx++);
@@ -60,19 +62,19 @@ private XlaSetDynamicDimensionSize(Operation operation) {
    * @param dimIndex the dimIndex value
    * @param sizeOutput the sizeOutput value
    * @param <T> data type for {@code XlaSetDynamicDimensionSize} output and operands
-   * @return a new instance of XlaSetDynamicDimensionSize
+   * @return a new instance of SetDynamicDimensionSize
    */
   @Endpoint(
       describeByClass = true
   )
-  public static <T extends TType> XlaSetDynamicDimensionSize<T> create(Scope scope,
-      Operand<T> input, Operand<TInt32> dimIndex, Operand<TInt32> sizeOutput) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaSetDynamicDimensionSize"));
+  public static <T extends TType> SetDynamicDimensionSize<T> create(Scope scope, Operand<T> input,
+      Operand<TInt32> dimIndex, Operand<TInt32> sizeOutput) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("SetDynamicDimensionSize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(dimIndex.asOutput());
     opBuilder.addInput(sizeOutput.asOutput());
     opBuilder = scope.apply(opBuilder);
-    return new XlaSetDynamicDimensionSize<>(opBuilder.build());
+    return new SetDynamicDimensionSize<>(opBuilder.build());
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SpmdFullToShardShape.java
similarity index 84%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SpmdFullToShardShape.java
index cedee394b63..2cb2377382a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SpmdFullToShardShape.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.core;
+package org.tensorflow.op.xla;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
@@ -36,8 +36,10 @@
  *
  * @param <T> data type for {@code output} output
  */
-@Operator
-public final class XlaSpmdFullToShardShape<T extends TType> extends RawOp implements Operand<T> {
+@Operator(
+    group = "xla"
+)
+public final class SpmdFullToShardShape<T extends TType> extends RawOp implements Operand<T> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
@@ -45,7 +47,7 @@ public final class XlaSpmdFullToShardShape<T extends TType> extends RawOp implem
 
   private Output<T> output;
 
-  private XlaSpmdFullToShardShape(Operation operation) {
+  private SpmdFullToShardShape(Operation operation) {
     super(operation);
     int outputIdx = 0;
     output = operation.output(outputIdx++);
@@ -58,18 +60,18 @@ private XlaSpmdFullToShardShape(Operation operation) {
    * @param input the input value
    * @param manualSharding the value of the manualSharding property
    * @param <T> data type for {@code XlaSpmdFullToShardShape} output and operands
-   * @return a new instance of XlaSpmdFullToShardShape
+   * @return a new instance of SpmdFullToShardShape
    */
   @Endpoint(
       describeByClass = true
   )
-  public static <T extends TType> XlaSpmdFullToShardShape<T> create(Scope scope, Operand<T> input,
+  public static <T extends TType> SpmdFullToShardShape<T> create(Scope scope, Operand<T> input,
       String manualSharding) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaSpmdFullToShardShape"));
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("SpmdFullToShardShape"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.apply(opBuilder);
     opBuilder.setAttr("manual_sharding", manualSharding);
-    return new XlaSpmdFullToShardShape<>(opBuilder.build());
+    return new SpmdFullToShardShape<>(opBuilder.build());
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SpmdShardToFullShape.java
similarity index 84%
rename from tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
rename to tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SpmdShardToFullShape.java
index dc5e3de5834..c4e49461a22 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SpmdShardToFullShape.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.core;
+package org.tensorflow.op.xla;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
@@ -36,8 +36,10 @@
  *
  * @param <T> data type for {@code output} output
  */
-@Operator
-public final class XlaSpmdShardToFullShape<T extends TType> extends RawOp implements Operand<T> {
+@Operator(
+    group = "xla"
+)
+public final class SpmdShardToFullShape<T extends TType> extends RawOp implements Operand<T> {
   /**
    * The name of this op, as known by TensorFlow core engine
    */
@@ -45,7 +47,7 @@ public final class XlaSpmdShardToFullShape<T extends TType> extends RawOp implem
 
   private Output<T> output;
 
-  private XlaSpmdShardToFullShape(Operation operation) {
+  private SpmdShardToFullShape(Operation operation) {
     super(operation);
     int outputIdx = 0;
     output = operation.output(outputIdx++);
@@ -59,19 +61,19 @@ private XlaSpmdShardToFullShape(Operation operation) {
    * @param manualSharding the value of the manualSharding property
    * @param fullShape the value of the fullShape property
    * @param <T> data type for {@code XlaSpmdShardToFullShape} output and operands
-   * @return a new instance of XlaSpmdShardToFullShape
+   * @return a new instance of SpmdShardToFullShape
    */
   @Endpoint(
       describeByClass = true
   )
-  public static <T extends TType> XlaSpmdShardToFullShape<T> create(Scope scope, Operand<T> input,
+  public static <T extends TType> SpmdShardToFullShape<T> create(Scope scope, Operand<T> input,
       String manualSharding, Shape fullShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaSpmdShardToFullShape"));
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("SpmdShardToFullShape"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.apply(opBuilder);
     opBuilder.setAttr("manual_sharding", manualSharding);
     opBuilder.setAttr("full_shape", fullShape);
-    return new XlaSpmdShardToFullShape<>(opBuilder.build());
+    return new SpmdShardToFullShape<>(opBuilder.build());
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/While.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/While.java
new file mode 100644
index 00000000000..61e1e3a1ec6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/While.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * output = input; While (Cond(output)) { output = Body(output) }
+ */
+@Operator(
+    group = "xla"
+)
+public final class While extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaWhile";
+
+  private List<Output<?>> output;
+
+  @SuppressWarnings("unchecked")
+  private While(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaWhile operation.
+   *
+   * @param scope current scope
+   * @param input A list of input tensors whose types are T.
+   * @param cond A function takes 'input' and returns a tensor.  If the tensor is
+   * a scalar of non-boolean, the scalar is converted to a boolean
+   * according to the following rule: if the scalar is a numerical
+   * value, non-zero means True and zero means False; if the scalar is
+   * a string, non-empty means True and empty means False. If the
+   * tensor is not a scalar, non-emptiness means True and False
+   * otherwise.
+   * @param body A function that takes a list of tensors and returns another
+   * list of tensors. Both lists have the same types as specified by T.
+   * @return a new instance of While
+   */
+  @Endpoint(
+      describeByClass = true,
+      name = "whileOp"
+  )
+  public static While create(Scope scope, Iterable<Operand<?>> input, ConcreteFunction cond,
+      ConcreteFunction body) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("While"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("cond", cond);
+    opBuilder.setAttr("body", body);
+    return new While(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   * A list of output tensors whose types are T.
+   * @return output.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaHostCompute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaHostCompute.java
new file mode 100644
index 00000000000..ff628c183c5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaHostCompute.java
@@ -0,0 +1,229 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A pseudo-op to represent host-side computation in an XLA program.
+ */
+@Operator(
+    group = "xla"
+)
+public final class XlaHostCompute extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaHostCompute";
+
+  private List<Output<?>> outputs;
+
+  @SuppressWarnings("unchecked")
+  private XlaHostCompute(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaHostCompute operation.
+   *
+   * @param scope current scope
+   * @param inputs A list of tensors that will be sent to the host.
+   * @param Toutputs The element types of each element in {@code outputs}.
+   * @param ancestors A list of names of HostCompute computations that must be
+   * sequenced before this computation.
+   * @param shapes If shape_inference_graph is empty, a list of the shapes of {@code outputs}.
+   * @param shapeInferenceGraph If non-empty, a serialized GraphDef representing a graph
+   * that must be analyzed at compile time to determine the shapes of the outputs.
+   * @param key A unique identifier for this region used to match up host transfers.
+   * @param options carries optional attribute values
+   * @return a new instance of XlaHostCompute
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static XlaHostCompute create(Scope scope, Iterable<Operand<?>> inputs,
+      List<Class<? extends TType>> Toutputs, List<String> ancestors, List<Shape> shapes,
+      ConcreteFunction shapeInferenceGraph, String key, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaHostCompute"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Toutputs", Operands.toDataTypes(Toutputs));
+    String[] ancestorsArray = new String[ancestors.size()];
+    for (int i = 0 ; i < ancestorsArray.length ; i++) {
+      ancestorsArray[i] = ancestors.get(i);
+    }
+    opBuilder.setAttr("ancestors", ancestorsArray);
+    Shape[] shapesArray = new Shape[shapes.size()];
+    for (int i = 0 ; i < shapesArray.length ; i++) {
+      shapesArray[i] = shapes.get(i);
+    }
+    opBuilder.setAttr("shapes", shapesArray);
+    opBuilder.setAttr("shape_inference_graph", shapeInferenceGraph);
+    opBuilder.setAttr("key", key);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.sendKey != null) {
+          opBuilder.setAttr("send_key", opts.sendKey);
+        }
+        if (opts.recvKey != null) {
+          opBuilder.setAttr("recv_key", opts.recvKey);
+        }
+        if (opts.costEstimateNs != null) {
+          opBuilder.setAttr("cost_estimate_ns", opts.costEstimateNs);
+        }
+        if (opts.tpuCore != null) {
+          opBuilder.setAttr("tpu_core", opts.tpuCore);
+        }
+      }
+    }
+    return new XlaHostCompute(opBuilder.build());
+  }
+
+  /**
+   * Sets the sendKey option.
+   *
+   * @param sendKey the sendKey option
+   * @return this Options instance.
+   */
+  public static Options sendKey(String sendKey) {
+    return new Options().sendKey(sendKey);
+  }
+
+  /**
+   * Sets the recvKey option.
+   *
+   * @param recvKey the recvKey option
+   * @return this Options instance.
+   */
+  public static Options recvKey(String recvKey) {
+    return new Options().recvKey(recvKey);
+  }
+
+  /**
+   * Sets the costEstimateNs option.
+   *
+   * @param costEstimateNs Estimated duration of the host computation in nanoseconds.
+   * @return this Options instance.
+   */
+  public static Options costEstimateNs(Long costEstimateNs) {
+    return new Options().costEstimateNs(costEstimateNs);
+  }
+
+  /**
+   * Sets the tpuCore option.
+   *
+   * @param tpuCore Default core to use for host to device transfers.
+   * @return this Options instance.
+   */
+  public static Options tpuCore(Long tpuCore) {
+    return new Options().tpuCore(tpuCore);
+  }
+
+  /**
+   * Gets outputs.
+   * A list of tensors that will be returned to the device.
+   * @return outputs.
+   */
+  public List<Output<?>> outputs() {
+    return outputs;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+
+  /**
+   * Optional attributes for {@link org.tensorflow.op.xla.XlaHostCompute}
+   */
+  public static class Options {
+    private String sendKey;
+
+    private String recvKey;
+
+    private Long costEstimateNs;
+
+    private Long tpuCore;
+
+    private Options() {
+    }
+
+    /**
+     * Sets the sendKey option.
+     *
+     * @param sendKey the sendKey option
+     * @return this Options instance.
+     */
+    public Options sendKey(String sendKey) {
+      this.sendKey = sendKey;
+      return this;
+    }
+
+    /**
+     * Sets the recvKey option.
+     *
+     * @param recvKey the recvKey option
+     * @return this Options instance.
+     */
+    public Options recvKey(String recvKey) {
+      this.recvKey = recvKey;
+      return this;
+    }
+
+    /**
+     * Sets the costEstimateNs option.
+     *
+     * @param costEstimateNs Estimated duration of the host computation in nanoseconds.
+     * @return this Options instance.
+     */
+    public Options costEstimateNs(Long costEstimateNs) {
+      this.costEstimateNs = costEstimateNs;
+      return this;
+    }
+
+    /**
+     * Sets the tpuCore option.
+     *
+     * @param tpuCore Default core to use for host to device transfers.
+     * @return this Options instance.
+     */
+    public Options tpuCore(Long tpuCore) {
+      this.tpuCore = tpuCore;
+      return this;
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaLaunch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaLaunch.java
new file mode 100644
index 00000000000..110b723def4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaLaunch.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * XLA Launch Op. For use by the XLA JIT only.
+ */
+@Operator(
+    group = "xla"
+)
+public final class XlaLaunch extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaLaunch";
+
+  private List<Output<?>> results;
+
+  @SuppressWarnings("unchecked")
+  private XlaLaunch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int resultsLength = operation.outputListLength("results");
+    results = Arrays.asList(operation.outputList(outputIdx, resultsLength));
+    outputIdx += resultsLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaLaunch operation.
+   *
+   * @param scope current scope
+   * @param constants the constants value
+   * @param args the args value
+   * @param resources the resources value
+   * @param Tresults the value of the Tresults property
+   * @param function the value of the function property
+   * @return a new instance of XlaLaunch
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static XlaLaunch create(Scope scope, Iterable<Operand<?>> constants,
+      Iterable<Operand<?>> args, Iterable<Operand<? extends TType>> resources,
+      List<Class<? extends TType>> Tresults, ConcreteFunction function) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaLaunch"));
+    opBuilder.addInputList(Operands.asOutputs(constants));
+    opBuilder.addInputList(Operands.asOutputs(args));
+    opBuilder.addInputList(Operands.asOutputs(resources));
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("Tresults", Operands.toDataTypes(Tresults));
+    opBuilder.setAttr("function", function);
+    return new XlaLaunch(opBuilder.build());
+  }
+
+  /**
+   * Gets results.
+   *
+   * @return results.
+   */
+  public List<Output<?>> results() {
+    return results;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) results.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaVariadicReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaVariadicReduce.java
new file mode 100644
index 00000000000..317df94f72e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaVariadicReduce.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the variadic XLA Reduce operator.
+ * Semantics are documented at
+ * https://www.tensorflow.org/performance/xla/operation_semantics#variadic_reduce.
+ *
+ * @param <T> data type for {@code output} output
+ */
+@Operator(
+    group = "xla"
+)
+public final class XlaVariadicReduce<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaVariadicReduce";
+
+  private List<Output<T>> output;
+
+  @SuppressWarnings("unchecked")
+  private XlaVariadicReduce(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList((Output<T>[]) operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaVariadicReduce operation.
+   *
+   * @param scope current scope
+   * @param input the input tensor(s)
+   * @param initValue scalar initial value(s) for the reduction
+   * @param dimensionsToReduce dimension numbers over which to reduce
+   * @param reducer a reducer function to apply
+   * @param <T> data type for {@code XlaVariadicReduce} output and operands
+   * @return a new instance of XlaVariadicReduce
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static <T extends TType> XlaVariadicReduce<T> create(Scope scope,
+      Iterable<Operand<T>> input, Iterable<Operand<T>> initValue, List<Long> dimensionsToReduce,
+      ConcreteFunction reducer) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaVariadicReduce"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder.addInputList(Operands.asOutputs(initValue));
+    opBuilder = scope.apply(opBuilder);
+    long[] dimensionsToReduceArray = new long[dimensionsToReduce.size()];
+    for (int i = 0 ; i < dimensionsToReduceArray.length ; i++) {
+      dimensionsToReduceArray[i] = dimensionsToReduce.get(i);
+    }
+    opBuilder.setAttr("dimensions_to_reduce", dimensionsToReduceArray);
+    opBuilder.setAttr("reducer", reducer);
+    return new XlaVariadicReduce<>(opBuilder.build());
+  }
+
+  /**
+   * Gets output.
+   *
+   * @return output.
+   */
+  public List<Output<T>> output() {
+    return output;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<T>> iterator() {
+    return (Iterator) output.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaVariadicSort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaVariadicSort.java
new file mode 100644
index 00000000000..fdb7d680cc6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/XlaVariadicSort.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA Sort operator, documented at
+ * https://www.tensorflow.org/performance/xla/operation_semantics#sort
+ * .
+ * <p>Sorts one or more tensors, with support for custom comparator, dimension, and
+ * is_stable attributes.
+ */
+@Operator(
+    group = "xla"
+)
+public final class XlaVariadicSort extends RawOp implements Iterable<Operand<TType>> {
+  /**
+   * The name of this op, as known by TensorFlow core engine
+   */
+  public static final String OP_NAME = "XlaVariadicSort";
+
+  private List<Output<?>> outputs;
+
+  @SuppressWarnings("unchecked")
+  private XlaVariadicSort(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+
+  /**
+   * Factory method to create a class wrapping a new XlaVariadicSort operation.
+   *
+   * @param scope current scope
+   * @param inputs A list of {@code Tensor} of identical shape but possibly different types.
+   * @param dimension The dimension along which to sort. Must be a compile-time constant.
+   * @param comparator A comparator function to apply to 2*N scalars and returning a
+   * boolean. N is the number of sort inputs. If you want to sort in ascending
+   * order then the comparator should perform a less-than comparison.
+   * @param isStable Whether to use stable sort.
+   * @return a new instance of XlaVariadicSort
+   */
+  @Endpoint(
+      describeByClass = true
+  )
+  public static XlaVariadicSort create(Scope scope, Iterable<Operand<?>> inputs,
+      Operand<TInt32> dimension, ConcreteFunction comparator, Boolean isStable) {
+    OperationBuilder opBuilder = scope.env().opBuilder(OP_NAME, scope.makeOpName("XlaVariadicSort"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInput(dimension.asOutput());
+    opBuilder = scope.apply(opBuilder);
+    opBuilder.setAttr("comparator", comparator);
+    opBuilder.setAttr("is_stable", isStable);
+    return new XlaVariadicSort(opBuilder.build());
+  }
+
+  /**
+   * Gets outputs.
+   * A list of {@code Tensor} of same shape and types as the {@code input}.
+   * @return outputs.
+   */
+  public List<Output<?>> outputs() {
+    return outputs;
+  }
+
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/resources/ops.pb b/tensorflow-core/tensorflow-core-api/src/gen/resources/ops.pb
index 4c3e6bef038..40322dc8780 100644
Binary files a/tensorflow-core/tensorflow-core-api/src/gen/resources/ops.pb and b/tensorflow-core/tensorflow-core-api/src/gen/resources/ops.pb differ
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
index 3e264e0e25d..f5dcd7c2ce3 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
@@ -1,18 +1,18 @@
 /* Copyright 2020-2021 The TensorFlow Authors. 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.
- You may obtain a copy of the License at
-
-     http://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.
- =======================================================================
- */
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================
+*/
 package org.tensorflow;
 
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_FunctionSetAttrValueProto;
@@ -38,8 +38,10 @@
 import org.tensorflow.internal.c_api.TF_Operation;
 import org.tensorflow.internal.c_api.TF_Output;
 import org.tensorflow.internal.c_api.TF_Status;
+import org.tensorflow.internal.types.registry.TensorTypeRegistry;
 import org.tensorflow.op.Ops;
 import org.tensorflow.op.Scope;
+import org.tensorflow.op.core.PartitionedCall;
 import org.tensorflow.op.core.Placeholder;
 import org.tensorflow.op.core.PlaceholderWithDefault;
 import org.tensorflow.proto.framework.AttrValue;
@@ -207,11 +209,6 @@ public String toString() {
     return signature.toString();
   }
 
-  // TODO migrate to the actual ops once they are generated
-  public static final String CALL_OP = "PartitionedCall";
-  // TODO migrate to the actual ops once they are generated
-  public static final String STATEFUL_CALL_OP = "StatefulPartitionedCall";
-
   /**
    * Calls the function in an execution environment, adding its graph as a function if it isn't
    * already present. The inputs and outputs are keyed by the names set in the {@code Signature}.
@@ -221,11 +218,8 @@ public String toString() {
    * @return the outputs of the function
    */
   public Map<String, Operand<?>> call(Scope scope, Map<String, Operand<?>> arguments) {
-    List<Operand<?>> inputList = new ArrayList<>();
-
-    Output<?>[] inputs = new Output<?>[signature().inputNames().size()];
+    List<Operand<?>> inputList = new ArrayList<>(signature.inputNames().size());
 
-    int i = 0;
     for (String inputName : signature().inputNames()) {
       if (!arguments.containsKey(inputName)) {
         throw new IllegalArgumentException(
@@ -243,40 +237,23 @@ public Map<String, Operand<?>> call(Scope scope, Map<String, Operand<?>> argumen
                 + inputName
                 + "\" was null.");
       }
-      inputs[i] = input.asOutput();
-      i++;
+      inputList.add(input);
     }
 
-    scope.env().attachFunction(this);
-    String name = getDefinedName();
-
-    String displayName = Scope.isValidOpName(name) ? name : "FunctionCall";
-
-    OperationBuilder opBuilder =
-        scope
-            .env()
-            .opBuilder(isStateful() ? STATEFUL_CALL_OP : CALL_OP, scope.makeOpName(displayName));
-
-    opBuilder.addInputList(inputs);
-
-    opBuilder.setAttr("f", this);
-    opBuilder.setAttr("Tin", inputDtypes);
-    opBuilder.setAttr("Tout", outputDtypes);
-
-    opBuilder = scope.apply(opBuilder);
-    Operation op = opBuilder.build();
-
-    int numOutputs1 = op.numOutputs();
-    List<Operand<?>> outputList = new ArrayList<>(signature().outputNames().size());
-
-    for (i = 0; i < numOutputs1; i++) {
-      outputList.add(op.output(i));
-    }
+    List<Output<?>> outputList =
+        PartitionedCall.create(
+                scope,
+                inputList,
+                Arrays.stream(inputDtypes)
+                    .map(x -> TensorTypeRegistry.find(x).type())
+                    .collect(Collectors.toList()),
+                this)
+            .output();
 
     Map<String, Operand<?>> namedOutputs = new LinkedHashMap<>(signature().outputNames().size());
 
     List<String> outputNames = new ArrayList<>(signature().outputNames());
-    for (i = 0; i < outputNames.size(); i++) {
+    for (int i = 0; i < outputNames.size(); i++) {
       String outputName = outputNames.get(i);
 
       if (i > outputList.size()) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/NativeFunction.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/NativeFunction.java
index faab6dbca7b..05e89a39722 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/NativeFunction.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/NativeFunction.java
@@ -1,18 +1,18 @@
 /* Copyright 2021 The TensorFlow Authors. 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.
- You may obtain a copy of the License at
-
-     http://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.
- =======================================================================
- */
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================
+*/
 package org.tensorflow;
 
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_FunctionName;
@@ -76,10 +76,21 @@ public synchronized List<String> getDependencies() {
       Set<String> deps = new LinkedHashSet<>();
 
       for (NodeDef node : getFunctionDef().getNodeDefList()) {
-        if (node.getOp().equals(ConcreteFunction.CALL_OP)
-            || node.getOp().equals(ConcreteFunction.STATEFUL_CALL_OP)) {
-          deps.add(node.getAttrMap().get("f").getFunc().getName());
-        }
+        node.getAttrMap()
+            .values()
+            .forEach(
+                (attr) -> {
+                  if (attr.hasFunc()) {
+                    deps.add(attr.getFunc().getName());
+                  } else if (attr.hasList()) {
+                    attr.getList()
+                        .getFuncList()
+                        .forEach(
+                            funcs -> {
+                              deps.add(funcs.getName());
+                            });
+                  }
+                });
       }
       dependencies = Collections.unmodifiableList(new ArrayList<>(deps));
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/IfTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/IfTest.java
new file mode 100644
index 00000000000..57bc0bc9ffb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/IfTest.java
@@ -0,0 +1,173 @@
+/* Copyright 2021 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================
+*/
+package org.tensorflow.op.core;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+
+import java.util.Arrays;
+import java.util.Collections;
+import org.junit.jupiter.api.Test;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.EagerSession;
+import org.tensorflow.Graph;
+import org.tensorflow.Operand;
+import org.tensorflow.Session;
+import org.tensorflow.Signature;
+import org.tensorflow.op.Ops;
+import org.tensorflow.types.TInt32;
+
+public class IfTest {
+
+  private static Operand<TInt32> basicIf(Ops tf, Operand<TInt32> a, Operand<TInt32> b) {
+    ConcreteFunction thenBranch =
+        ConcreteFunction.create(
+            (ops) -> {
+              Operand<TInt32> a1 = ops.placeholder(TInt32.class);
+              Operand<TInt32> b1 = ops.placeholder(TInt32.class);
+              return Signature.builder().input("a", a1).input("b", b1).output("y", a1).build();
+            });
+
+    ConcreteFunction elseBranch =
+        ConcreteFunction.create(
+            (ops) -> {
+              Operand<TInt32> a1 = ops.placeholder(TInt32.class);
+              Operand<TInt32> b1 = ops.placeholder(TInt32.class);
+              Operand<TInt32> y = ops.math.neg(b1);
+              return Signature.builder().input("a", a1).input("b", b1).output("y", y).build();
+            });
+
+    return (Operand<TInt32>)
+        tf.ifOp(
+                tf.math.greater(a, b),
+                Arrays.asList(a, b),
+                Arrays.asList(TInt32.class),
+                thenBranch,
+                elseBranch)
+            .output()
+            .get(0);
+  }
+
+  @Test
+  public void testGraph() {
+    try (Graph g = new Graph();
+        Session s = new Session(g)) {
+      Ops tf = Ops.create(g);
+      Operand<TInt32> a = tf.placeholder(TInt32.class);
+      Operand<TInt32> b = tf.placeholder(TInt32.class);
+      Operand<TInt32> c = basicIf(tf, a, b);
+
+      assertEquals(StatelessIf.OP_NAME, c.op().type());
+
+      try (TInt32 out =
+          (TInt32)
+              s.runner()
+                  .feed(a, TInt32.scalarOf(2))
+                  .feed(b, TInt32.scalarOf(1))
+                  .fetch(c)
+                  .run()
+                  .get(0)) {
+        assertEquals(2, out.getInt());
+      }
+
+      try (TInt32 out =
+          (TInt32)
+              s.runner()
+                  .feed(a, TInt32.scalarOf(2))
+                  .feed(b, TInt32.scalarOf(3))
+                  .fetch(c)
+                  .run()
+                  .get(0)) {
+        assertEquals(-3, out.getInt());
+      }
+    }
+  }
+
+  @Test
+  public void testStatefullness() {
+    try (Graph g = new Graph()) {
+      Ops tf = Ops.create(g);
+      Operand<TInt32> a = tf.placeholder(TInt32.class);
+      Operand<TInt32> b = tf.placeholder(TInt32.class);
+
+      ConcreteFunction thenBranch =
+          ConcreteFunction.create(
+              (ops) -> {
+                Operand<TInt32> a1 = ops.placeholder(TInt32.class);
+                Operand<TInt32> b1 = ops.placeholder(TInt32.class);
+                Operand<TInt32> result =
+                    (Operand<TInt32>)
+                        ops.statefulIf(
+                                ops.constant(false),
+                                Collections.emptyList(),
+                                Arrays.asList(TInt32.class),
+                                ConcreteFunction.create(
+                                    (ops1) ->
+                                        Signature.builder().output("y", ops1.constant(1)).build()),
+                                ConcreteFunction.create(
+                                    (ops1) ->
+                                        Signature.builder().output("y", ops1.constant(1)).build()))
+                            .output()
+                            .get(0);
+                return Signature.builder()
+                    .input("a", a1)
+                    .input("b", b1)
+                    .output("y", result)
+                    .build();
+              });
+
+      ConcreteFunction elseBranch =
+          ConcreteFunction.create(
+              (ops) -> {
+                Operand<TInt32> a1 = ops.placeholder(TInt32.class);
+                Operand<TInt32> b1 = ops.placeholder(TInt32.class);
+                Operand<TInt32> y = ops.math.neg(b1);
+                return Signature.builder().input("a", a1).input("b", b1).output("y", y).build();
+              });
+
+      Operand<TInt32> output =
+          (Operand<TInt32>)
+              tf.ifOp(
+                      tf.math.greater(a, b),
+                      Arrays.asList(a, b),
+                      Arrays.asList(TInt32.class),
+                      thenBranch,
+                      elseBranch)
+                  .output()
+                  .get(0);
+
+      assertEquals(StatefulIf.OP_NAME, output.op().type());
+    }
+  }
+
+  @Test
+  public void testEager() {
+    try (EagerSession e = EagerSession.create()) {
+      Ops tf = Ops.create(e);
+
+      Operand<TInt32> out1 = basicIf(tf, tf.constant(2), tf.constant(1));
+
+      assertEquals(StatelessIf.OP_NAME, out1.op().type());
+
+      try (TInt32 out = out1.asTensor()) {
+        assertEquals(2, out.getInt());
+      }
+
+      try (TInt32 out = basicIf(tf, tf.constant(2), tf.constant(3)).asTensor()) {
+        assertEquals(-3, out.getInt());
+      }
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/PartitionedCallTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/PartitionedCallTest.java
new file mode 100644
index 00000000000..f075934d2cc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/PartitionedCallTest.java
@@ -0,0 +1,68 @@
+/* Copyright 2021 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================
+*/
+package org.tensorflow.op.core;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+
+import java.util.Arrays;
+import org.junit.jupiter.api.Test;
+import org.tensorflow.ConcreteFunction;
+import org.tensorflow.EagerSession;
+import org.tensorflow.Graph;
+import org.tensorflow.Operand;
+import org.tensorflow.Session;
+import org.tensorflow.Signature;
+import org.tensorflow.op.Ops;
+import org.tensorflow.types.TInt32;
+
+public class PartitionedCallTest {
+
+  public static Signature plusTwo(Ops tf) {
+    Operand<TInt32> x = tf.placeholder(TInt32.class);
+    Operand<TInt32> y = tf.math.add(x, tf.constant(2));
+    return Signature.builder().input("x", x).output("y", y).build();
+  }
+
+  @Test
+  public void testEager() {
+    try (EagerSession e = EagerSession.create();
+        ConcreteFunction f = ConcreteFunction.create(PartitionedCallTest::plusTwo)) {
+      Ops tf = Ops.create(e);
+      Operand<TInt32> x = tf.constant(3);
+      Operand<TInt32> y =
+          (Operand<TInt32>)
+              tf.partitionedCall(Arrays.asList(x), Arrays.asList(TInt32.class), f).output().get(0);
+      assertEquals(5, y.asTensor().getInt());
+    }
+  }
+
+  @Test
+  public void testGraph() {
+    try (Graph g = new Graph();
+        ConcreteFunction f = ConcreteFunction.create(PartitionedCallTest::plusTwo)) {
+      Ops tf = Ops.create(g);
+      Operand<TInt32> x = tf.placeholder(TInt32.class);
+      Operand<TInt32> y =
+          (Operand<TInt32>)
+              tf.partitionedCall(Arrays.asList(x), Arrays.asList(TInt32.class), f).output().get(0);
+
+      try (Session s = new Session(g);
+          TInt32 out = (TInt32) s.runner().feed(x, TInt32.scalarOf(3)).fetch(y).run().get(0)) {
+        assertEquals(5, out.getInt());
+      }
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/ClassGenerator.java b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/ClassGenerator.java
index be6d0a32392..d5d617158c7 100644
--- a/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/ClassGenerator.java
+++ b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/ClassGenerator.java
@@ -41,6 +41,7 @@
 import java.util.List;
 import java.util.Map;
 import java.util.Set;
+import java.util.StringJoiner;
 import javax.lang.model.element.Modifier;
 import org.tensorflow.Names;
 import org.tensorflow.proto.framework.ApiDef;
@@ -57,7 +58,6 @@ final class ClassGenerator {
   /** Return true if we can generate the operation class for {@code op}. */
   static boolean canGenerateOp(OpDef op, ApiDef apiDef) {
     return apiDef.getVisibility() != Visibility.SKIP
-        && !op.getAttrList().stream().anyMatch(x -> x.getType().contains("func"))
         && !op.getName()
             .startsWith("_"); // TODO do I want this?  Some interesting ops like _XlaCompile
   }
@@ -70,6 +70,8 @@ enum RenderMode {
     OPERAND;
   }
 
+  private static final String OP_NAME_FIELD = "OP_NAME";
+
   /** The in-progress class builder for the top level op class. */
   private final TypeSpec.Builder builder;
 
@@ -97,6 +99,10 @@ enum RenderMode {
   /** The endpoint being generated in this class. */
   private final Endpoint endpoint;
 
+  private final StatefulPair statefulPair;
+  private final boolean isStateSelector;
+  private final boolean isStateSubclass;
+
   /**
    * The generated options class, or null if it doesn't have one or {@link #buildOptionsClass()} has
    * not been ran.
@@ -129,7 +135,8 @@ enum RenderMode {
       String fullPackage,
       String group,
       String className,
-      Endpoint endpoint) {
+      Endpoint endpoint,
+      StatefulPair statefulPair) {
 
     this.builder = builder;
     this.op = op;
@@ -140,6 +147,9 @@ enum RenderMode {
     this.group = group;
     this.className = className;
     this.endpoint = endpoint;
+    this.statefulPair = statefulPair;
+    this.isStateSelector = statefulPair != null && className.equals(statefulPair.selectorClassName);
+    this.isStateSubclass = statefulPair != null && !isStateSelector;
 
     op.getAttrList()
         .forEach(
@@ -207,8 +217,15 @@ private String fullClassName() {
 
   /** Build the class. */
   void buildClass() {
-    builder.addModifiers(Modifier.PUBLIC, Modifier.FINAL);
-    builder.superclass(Names.RawOp);
+    builder.addModifiers(Modifier.PUBLIC);
+    if (!isStateSelector) {
+      builder.addModifiers(Modifier.FINAL);
+      builder.superclass(Names.RawOp);
+    }
+
+    if (isStateSubclass) {
+      builder.addSuperinterface(ClassName.get(fullPackage, statefulPair.selectorClassName));
+    }
 
     // add class javadocs
     String summary = parseDocumentation(apiDef.getSummary());
@@ -222,6 +239,16 @@ void buildClass() {
       builder.addJavadoc("$L", desc + "\n");
     }
 
+    if (isStateSelector) {
+      builder.addJavadoc("\n<p>");
+      builder.addJavadoc(
+          "Selects between {@link "
+              + statefulPair.statefulClassName
+              + "} and {@link "
+              + statefulPair.statelessClassName
+              + "} based on the statefulness of the function arguments.");
+    }
+
     // add superinterface and set mode
     if (op.getOutputArgCount() == 1) {
       ArgDef output = op.getOutputArg(0);
@@ -233,11 +260,15 @@ void buildClass() {
 
       if (iterable) {
         mode = RenderMode.LIST_OPERAND;
-        builder.addSuperinterface(
-            ParameterizedTypeName.get(ClassName.get(Iterable.class), operandType));
+        if (!isStateSubclass) {
+          builder.addSuperinterface(
+              ParameterizedTypeName.get(ClassName.get(Iterable.class), operandType));
+        }
       } else {
         mode = RenderMode.OPERAND;
-        builder.addSuperinterface(operandType);
+        if (!isStateSubclass) {
+          builder.addSuperinterface(operandType);
+        }
       }
     }
 
@@ -277,7 +308,7 @@ void buildClass() {
       builder.addAnnotation(annotation.build());
     }
 
-    if (!optionalAttributes.isEmpty()) {
+    if (!optionalAttributes.isEmpty() && !isStateSubclass) {
       buildOptionsClass();
     }
 
@@ -287,29 +318,31 @@ void buildClass() {
       buildInterfaceImpl();
     }
 
-    // add op name field
-    builder.addField(
-        FieldSpec.builder(
-                TypeResolver.STRING,
-                OP_NAME_FIELD_NAME,
-                Modifier.PUBLIC,
-                Modifier.STATIC,
-                Modifier.FINAL)
-            .addJavadoc("$L", "The name of this op, as known by TensorFlow core engine")
-            .initializer("$S", op.getName())
-            .build());
+    if (!isStateSelector) {
+      // add op name field
+      builder.addField(
+          FieldSpec.builder(
+                  TypeResolver.STRING,
+                  OP_NAME_FIELD,
+                  Modifier.PUBLIC,
+                  Modifier.STATIC,
+                  Modifier.FINAL)
+              .addJavadoc("$L", "The name of this op, as known by TensorFlow core engine")
+              .initializer("$S", op.getName())
+              .build());
 
-    // add output fields
-    if (op.getOutputArgCount() > 0) {
-      for (ArgDef output : op.getOutputArgList()) {
-        builder.addField(
-            resolver.typeOf(output).listIfIterable().javaType,
-            getJavaName(output),
-            Modifier.PRIVATE);
+      // add output fields
+      if (op.getOutputArgCount() > 0) {
+        for (ArgDef output : op.getOutputArgList()) {
+          builder.addField(
+              resolver.typeOf(output).listIfIterable().javaType,
+              getJavaName(output),
+              Modifier.PRIVATE);
+        }
       }
-    }
 
-    buildConstructor();
+      buildConstructor();
+    }
   }
 
   /** Add a nested class for Options */
@@ -376,9 +409,14 @@ private void buildOptionsClass() {
                 .build());
       }
 
+      FieldSpec.Builder field =
+          FieldSpec.builder(type.classIfGeneric().listIfIterable().javaType, name);
+      if (!isStateSelector) {
+        field.addModifiers(Modifier.PRIVATE);
+      }
+
       // add the field
-      optionsBuilder.addField(
-          type.classIfGeneric().listIfIterable().javaType, name, Modifier.PRIVATE);
+      optionsBuilder.addField(field.build());
     }
 
     // add a private constructor
@@ -444,8 +482,16 @@ private void buildFactoryMethods() {
     }
     factoryBuilder.returns(returnType);
 
-    factoryBuilder.addAnnotation(
-        AnnotationSpec.builder(Names.Endpoint).addMember("describeByClass", "true").build());
+    AnnotationSpec.Builder endpointAnnotation =
+        AnnotationSpec.builder(Names.Endpoint).addMember("describeByClass", "true");
+
+    String methodName = GeneratorUtils.getOpMethodName(className);
+
+    if (methodName != null) {
+      endpointAnnotation.addMember("name", "$S", methodName);
+    }
+
+    factoryBuilder.addAnnotation(endpointAnnotation.build());
 
     factoryBuilder.addJavadoc(
         "Factory method to create a class wrapping a new $L operation.\n", op.getName());
@@ -463,15 +509,26 @@ private void buildFactoryMethods() {
     body.addStatement(
         "$T opBuilder = scope.env().opBuilder($L, scope.makeOpName($S))",
         Names.OperationBuilder,
-        OP_NAME_FIELD_NAME,
+        OP_NAME_FIELD,
         className);
 
+    List<String> functionArgs = new ArrayList<>();
+    List<String> iterableFunctionArgs = new ArrayList<>();
+
     // add the inputs as parameters, and add them to the op builder
     for (ArgDef input : op.getInputArgList()) {
       ApiDef.Arg argDef = argApis.get(input);
       ResolvedType type = resolver.typeOf(input);
       String name = getJavaName(input);
 
+      if (type.javaType.equals(Names.ConcreteFunction)) {
+        if (type.iterable) {
+          iterableFunctionArgs.add(name);
+        } else {
+          functionArgs.add(name);
+        }
+      }
+
       ParameterSpec.Builder param = ParameterSpec.builder(type.iterableIfIterable().javaType, name);
       String description =
           argDef.getDescription().isEmpty()
@@ -502,11 +559,19 @@ private void buildFactoryMethods() {
 
       ResolvedType type = resolver.typeOf(attr);
       ApiDef.Attr apiAttr = attrApis.get(attr);
+      String javaName = getJavaName(attr);
+
+      if (type.javaType.equals(Names.ConcreteFunction)) {
+        if (type.iterable) {
+          iterableFunctionArgs.add(javaName);
+        } else {
+          functionArgs.add(javaName);
+        }
+      }
 
       ParameterSpec.Builder builder =
           ParameterSpec.builder(type.classIfGeneric().listIfIterable().javaType, getJavaName(attr));
 
-      String javaName = getJavaName(attr);
       String description =
           apiAttr.getDescription().isEmpty()
               ? String.format("the value of the %s property", javaName)
@@ -529,18 +594,26 @@ private void buildFactoryMethods() {
       writeSetAttr(body, attr, type, false);
     }
 
+    // TODO optional function attrs (there currently aren't any)
+
     // add optional attributes
-    if (optionsClass != null) {
+    if (optionsClass != null || (isStateSubclass && statefulPair.hasOptionalAttrs())) {
+
+      ClassName optionsClassName;
+      if (isStateSubclass) {
+        optionsClassName = ClassName.get(fullPackage, statefulPair.selectorClassName, "Options");
+      } else {
+        optionsClassName = ClassName.get(fullPackage, className, "Options");
+      }
+
       factoryBuilder.addParameter(
-          ParameterSpec.builder(
-                  ArrayTypeName.of(ClassName.get(fullPackage, className, "Options")), "options")
-              .build());
+          ParameterSpec.builder(ArrayTypeName.of(optionsClassName), "options").build());
       paramTags.put("options", CodeBlock.of("$L", "carries optional attribute values"));
       factoryBuilder.varargs();
 
       body.beginControlFlow("if (options != null)");
 
-      body.beginControlFlow("for (Options opts : options)");
+      body.beginControlFlow("for ($T opts : options)", optionsClassName);
       for (AttrDef attr : optionalAttributes) {
         String name = getJavaName(attr);
         body.beginControlFlow("if (opts.$L != null)", name);
@@ -557,7 +630,40 @@ private void buildFactoryMethods() {
     body.addStatement(
         "return new $L(opBuilder.build())", typeParams.isEmpty() ? className : (className + "<>"));
 
+    if (isStateSelector) {
+      body.clear();
+
+      body.addStatement("boolean isStateful = false");
+      functionArgs.forEach(
+          arg -> {
+            body.beginControlFlow("if ($L.isStateful())", arg)
+                .addStatement("isStateful = true")
+                .endControlFlow();
+          });
+      iterableFunctionArgs.forEach(
+          arg -> {
+            body.beginControlFlow("if ($L.stream().anyMatch(x -> x.isStateful()))", arg)
+                .addStatement("isStateful = true")
+                .endControlFlow();
+          });
+
+      StringJoiner argList = new StringJoiner(", ");
+      factoryBuilder.parameters.forEach(x -> argList.add(x.name));
+
+      body.beginControlFlow("if (isStateful)")
+          .addStatement(
+              "return $T.create($L)",
+              ClassName.get(fullPackage, statefulPair.statefulClassName),
+              argList.toString())
+          .nextControlFlow("else")
+          .addStatement(
+              "return $T.create($L)",
+              ClassName.get(fullPackage, statefulPair.statelessClassName),
+              argList.toString())
+          .endControlFlow();
+    }
     factoryBuilder.addCode(body.build());
+
     paramTags.forEach(
         (param, doc) -> {
           String description = doc.toString();
@@ -651,7 +757,10 @@ private void buildSecondaryFactory(
 
     factoryBuilder.addJavadoc(
         "\n@return a new instance of $L, with default output types", className);
-    factoryBuilder.addCode(body.build());
+
+    if (!isStateSelector) {
+      factoryBuilder.addCode(body.build());
+    }
     factoryBuilder.addTypeVariables(typeVars);
 
     builder.addMethod(factoryBuilder.build());
@@ -683,15 +792,25 @@ private void buildGettersAndSetters() {
     for (ArgDef output : op.getOutputArgList()) {
       String name = getJavaName(output);
       ApiDef.Arg argDef = argApis.get(output);
-      builder.addMethod(
+
+      MethodSpec.Builder method =
           MethodSpec.methodBuilder(name)
               .addModifiers(Modifier.PUBLIC)
               .returns(resolver.typeOf(output).listIfIterable().javaType)
               .addJavadoc("Gets $L.\n", name)
               .addJavadoc("$L", parseDocumentation(argDef.getDescription()))
-              .addJavadoc("\n@return $L.", name)
-              .addCode("return $L;", name)
-              .build());
+              .addJavadoc("\n@return $L.", name);
+
+      if (isStateSelector) {
+        method.addModifiers(Modifier.ABSTRACT);
+      } else {
+        if (isStateSubclass) {
+          method.addAnnotation(Override.class);
+        }
+        method.addCode("return $L;", name);
+      }
+
+      builder.addMethod(method.build());
     }
   }
 
@@ -711,14 +830,18 @@ private void buildInterfaceImpl() {
               .returns(outputType)
               .addAnnotation(Override.class);
 
-      if (uncheckedCast) {
-        asOutput.addAnnotation(
-            AnnotationSpec.builder(SuppressWarnings.class)
-                .addMember("value", "$S", "unchecked")
-                .build());
-        asOutput.addCode("return ($T) $L;", outputType, getJavaName(output));
+      if (isStateSelector) {
+        asOutput.addModifiers(Modifier.ABSTRACT);
       } else {
-        asOutput.addCode("return $L;", getJavaName(output));
+        if (uncheckedCast) {
+          asOutput.addAnnotation(
+              AnnotationSpec.builder(SuppressWarnings.class)
+                  .addMember("value", "$S", "unchecked")
+                  .build());
+          asOutput.addCode("return ($T) $L;", outputType, getJavaName(output));
+        } else {
+          asOutput.addCode("return $L;", getJavaName(output));
+        }
       }
 
       builder.addMethod(asOutput.build());
@@ -736,7 +859,11 @@ private void buildInterfaceImpl() {
                       .addMember("value", "{$S, $S}", "rawtypes", "unchecked")
                       .build());
 
-      iterator.addCode("return ($T) $L.iterator();", Iterator.class, getJavaName(output));
+      if (isStateSelector) {
+        iterator.addModifiers(Modifier.ABSTRACT);
+      } else {
+        iterator.addCode("return ($T) $L.iterator();", Iterator.class, getJavaName(output));
+      }
 
       builder.addMethod(iterator.build());
     }
diff --git a/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/FullOpDef.java b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/FullOpDef.java
new file mode 100644
index 00000000000..5e7775194d7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/FullOpDef.java
@@ -0,0 +1,144 @@
+/* Copyright 2021 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================
+*/
+package org.tensorflow.op.generator;
+
+import com.squareup.javapoet.TypeSpec;
+import java.util.StringJoiner;
+import org.tensorflow.proto.framework.ApiDef;
+import org.tensorflow.proto.framework.ApiDef.Endpoint;
+import org.tensorflow.proto.framework.OpDef;
+
+public final class FullOpDef {
+  public final OpDef opDef;
+  public final ApiDef apiDef;
+  public final String basePackage;
+  public final String packageName;
+  public final String group;
+  public final String className;
+  public final Endpoint endpoint;
+
+  public FullOpDef(
+      OpDef opDef,
+      ApiDef apiDef,
+      String basePackage,
+      String packageName,
+      String group,
+      String className,
+      Endpoint endpoint) {
+    this.group = group;
+    this.endpoint = endpoint;
+    if (opDef == null) {
+      throw new IllegalArgumentException("Can't have a null OpDef");
+    }
+    if (apiDef == null) {
+      throw new IllegalArgumentException("Can't have a null ApiDef");
+    }
+    this.opDef = opDef;
+    this.apiDef = apiDef;
+    this.basePackage = basePackage;
+    this.packageName = packageName;
+    this.className = className;
+  }
+
+  public boolean isStateful() {
+    return opDef.getIsStateful();
+  }
+
+  public boolean isStateVariant(FullOpDef other) {
+    if (this.equals(other)) return false;
+
+    if (this.isStateful() == other.isStateful()) return false;
+
+    OpDef copy =
+        opDef.toBuilder().setName(other.opDef.getName()).setIsStateful(other.isStateful()).build();
+    return copy.equals(other.opDef)
+        && packageName.equals(other.packageName)
+        && group.equals(other.group);
+  }
+
+  public TypeSpec buildOpClass() {
+    return buildOpClass(className, null);
+  }
+
+  TypeSpec buildOpClass(String className, StatefulPair pair) {
+    TypeSpec.Builder cls = TypeSpec.classBuilder(className);
+    try {
+      new ClassGenerator(
+              cls, opDef, apiDef, basePackage, packageName, group, className, endpoint, pair)
+          .buildClass();
+    } catch (Exception e) {
+      throw new IllegalStateException("Failed to generate class for op " + opDef.getName(), e);
+    }
+    return cls.build();
+  }
+
+  @Override
+  public boolean equals(Object o) {
+    if (this == o) {
+      return true;
+    }
+    if (o == null || getClass() != o.getClass()) {
+      return false;
+    }
+
+    FullOpDef fullOpDef = (FullOpDef) o;
+
+    if (!opDef.equals(fullOpDef.opDef)) {
+      return false;
+    }
+    if (!apiDef.equals(fullOpDef.apiDef)) {
+      return false;
+    }
+    if (!basePackage.equals(fullOpDef.basePackage)) {
+      return false;
+    }
+    if (!packageName.equals(fullOpDef.packageName)) {
+      return false;
+    }
+    if (!group.equals(fullOpDef.group)) {
+      return false;
+    }
+    if (!className.equals(fullOpDef.className)) {
+      return false;
+    }
+    return endpoint.equals(fullOpDef.endpoint);
+  }
+
+  @Override
+  public int hashCode() {
+    int result = opDef.hashCode();
+    result = 31 * result + apiDef.hashCode();
+    result = 31 * result + basePackage.hashCode();
+    result = 31 * result + packageName.hashCode();
+    result = 31 * result + group.hashCode();
+    result = 31 * result + className.hashCode();
+    result = 31 * result + endpoint.hashCode();
+    return result;
+  }
+
+  @Override
+  public String toString() {
+    return new StringJoiner(", ", FullOpDef.class.getSimpleName() + "(", ")")
+        .add("opDef=" + opDef)
+        .add("apiDef=" + apiDef)
+        .add("basePackage='" + basePackage + "'")
+        .add("packageName='" + packageName + "'")
+        .add("group='" + group + "'")
+        .add("className='" + className + "'")
+        .add("endpoint=" + endpoint)
+        .toString();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/GeneratorUtils.java b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/GeneratorUtils.java
index 80d6698fe36..0aa1638a861 100644
--- a/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/GeneratorUtils.java
+++ b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/GeneratorUtils.java
@@ -1,19 +1,18 @@
-/*
-  Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+/* Copyright 2021 The TensorFlow Authors. 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.
- You may obtain a copy of the License at
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
 
-     http://www.apache.org/licenses/LICENSE-2.0
+    http://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.
- ==============================================================================
- */
+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.
+=======================================================================
+*/
 package org.tensorflow.op.generator;
 
 import org.commonmark.node.Node;
@@ -21,9 +20,7 @@
 import org.tensorflow.op.generator.javadoc.JavaDocRenderer;
 import org.tensorflow.proto.framework.OpDef.ArgDef;
 
-/**
- * Utilities for op generation
- */
+/** Utilities for op generation */
 final class GeneratorUtils {
 
   private static final Parser parser = Parser.builder().build();
@@ -31,11 +28,11 @@ final class GeneratorUtils {
   /**
    * Convert a Python style name to a Java style name.
    *
-   * Does snake_case -> camelCase and handles keywords.
+   * <p>Does snake_case -> camelCase and handles keywords.
    *
-   * Not valid for class names, meant for fields and methods.
+   * <p>Not valid for class names, meant for fields and methods.
    *
-   * Generally you should use {@link ClassGenerator#getJavaName(ArgDef)}.
+   * <p>Generally you should use {@link ClassGenerator#getJavaName(ArgDef)}.
    */
   static String javaizeMemberName(String name) {
     StringBuilder result = new StringBuilder();
@@ -68,13 +65,28 @@ static String javaizeMemberName(String name) {
   }
 
   /**
-   * Convert markdown descriptions to JavaDocs.
+   * Get the name of the Ops method, or null to not specify one (the decapitalized class name will
+   * be used).
    */
+  static String getOpMethodName(String className) {
+    switch (className) {
+      case "If":
+        return "ifOp";
+      case "While":
+        return "whileOp";
+      case "For":
+        return "forOp";
+      case "Case":
+        return "caseOp";
+      default:
+        return null;
+    }
+  }
+
+  /** Convert markdown descriptions to JavaDocs. */
   static String parseDocumentation(String docs) {
     Node document = parser.parse(docs);
     JavaDocRenderer renderer = JavaDocRenderer.builder().build();
     return renderer.render(document).trim();
   }
-
-
 }
diff --git a/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/OpGenerator.java b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/OpGenerator.java
index da4f63405cc..9dc597c7f74 100644
--- a/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/OpGenerator.java
+++ b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/OpGenerator.java
@@ -1,5 +1,4 @@
-/*
- Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+/* Copyright 2021 The TensorFlow Authors. 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.
@@ -12,17 +11,14 @@
 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 org.tensorflow.op.generator;
 
 import com.google.protobuf.InvalidProtocolBufferException;
+import com.google.protobuf.UnknownFieldSet;
 import com.squareup.javapoet.JavaFile;
 import com.squareup.javapoet.TypeSpec;
-import org.tensorflow.proto.framework.ApiDef;
-import org.tensorflow.proto.framework.OpDef;
-import org.tensorflow.proto.framework.OpList;
-
 import java.io.File;
 import java.io.FileInputStream;
 import java.io.FileNotFoundException;
@@ -35,7 +31,12 @@
 import java.nio.file.StandardOpenOption;
 import java.nio.file.attribute.BasicFileAttributes;
 import java.util.LinkedHashMap;
+import java.util.List;
 import java.util.Map;
+import java.util.stream.Collectors;
+import org.tensorflow.proto.framework.ApiDef;
+import org.tensorflow.proto.framework.OpDef;
+import org.tensorflow.proto.framework.OpList;
 
 public final class OpGenerator {
 
@@ -169,7 +170,8 @@ private static void generate(File outputDir, String packageName, File opDefs) {
                     .getField(API_DEF_FIELD_NUMBER)
                     .getLengthDelimitedList()
                     .get(0));
-        defs.put(op, api);
+        defs.put(
+            op.toBuilder().setUnknownFields(UnknownFieldSet.newBuilder().build()).build(), api);
       } catch (InvalidProtocolBufferException e) {
         throw new RuntimeException(
             "Could not parse attached ApiDef for op "
@@ -184,80 +186,76 @@ private static void generate(File outputDir, String packageName, File opDefs) {
     generate(outputDir, packageName, defs);
   }
 
+  private static void writeToFile(TypeSpec spec, File outputDir, String packageName) {
+    JavaFile file =
+        JavaFile.builder(packageName, spec).indent("  ").skipJavaLangImports(true).build();
+
+    File outputFile =
+        new File(outputDir, packageName.replace('.', '/') + '/' + spec.name + ".java");
+    outputFile.getParentFile().mkdirs();
+    try {
+      StringBuilder builder = new StringBuilder();
+      builder.append(LICENSE + '\n');
+      builder.append("// This class has been generated, DO NOT EDIT!\n\n");
+      file.writeTo(builder);
+
+      Files.write(
+          outputFile.toPath(),
+          builder.toString().getBytes(StandardCharsets.UTF_8),
+          StandardOpenOption.WRITE,
+          StandardOpenOption.CREATE,
+          StandardOpenOption.TRUNCATE_EXISTING);
+    } catch (IOException ioException) {
+      throw new IllegalStateException("Failed to write file " + outputFile, ioException);
+    }
+  }
+
   /** Generate all the ops that pass {@link ClassGenerator#canGenerateOp(OpDef, ApiDef)}. */
   private static void generate(File outputDir, String basePackage, Map<OpDef, ApiDef> ops) {
-    ops.entrySet().stream()
-        .filter(e -> ClassGenerator.canGenerateOp(e.getKey(), e.getValue()))
-        .forEach(
-            (entry) -> {
-              entry
-                  .getValue()
-                  .getEndpointList()
-                  .forEach(
-                      (endpoint) -> {
-                        String name;
-                        String pack;
-
-                        int pos = endpoint.getName().lastIndexOf('.');
-                        if (pos > -1) {
-                          pack = endpoint.getName().substring(0, pos);
-                          name = endpoint.getName().substring(pos + 1);
-                        } else {
-                          pack = "core";
-                          name = endpoint.getName();
-                        }
-
-                        TypeSpec.Builder cls = TypeSpec.classBuilder(name);
-                        try {
-                          new ClassGenerator(
-                                  cls,
+    List<FullOpDef> fullOps =
+        ops.entrySet().stream()
+            .filter(e -> ClassGenerator.canGenerateOp(e.getKey(), e.getValue()))
+            .flatMap(
+                (entry) ->
+                    entry.getValue().getEndpointList().stream()
+                        .map(
+                            (endpoint) -> {
+                              String name;
+                              String pack;
+
+                              int pos = endpoint.getName().lastIndexOf('.');
+                              if (pos > -1) {
+                                pack = endpoint.getName().substring(0, pos);
+                                name = endpoint.getName().substring(pos + 1);
+                              } else {
+                                pack = "core";
+                                name = endpoint.getName();
+                              }
+
+                              return new FullOpDef(
                                   entry.getKey(),
                                   entry.getValue(),
                                   basePackage,
                                   basePackage + "." + pack,
                                   pack,
                                   name,
-                                  endpoint)
-                              .buildClass();
-                        } catch (Exception e) {
-                          throw new IllegalStateException(
-                              "Failed to generate class for op " + entry.getKey().getName(), e);
-                        }
-                        TypeSpec spec = cls.build();
-
-                        JavaFile file =
-                            JavaFile.builder(basePackage + "." + pack, spec)
-                                .indent("  ")
-                                .skipJavaLangImports(true)
-                                .build();
-
-                        File outputFile =
-                            new File(
-                                outputDir,
-                                basePackage.replace('.', '/')
-                                    + '/'
-                                    + pack.replace('.', '/')
-                                    + '/'
-                                    + spec.name
-                                    + ".java");
-                        outputFile.getParentFile().mkdirs();
-                        try {
-                          StringBuilder builder = new StringBuilder();
-                          builder.append(LICENSE + '\n');
-                          builder.append("// This class has been generated, DO NOT EDIT!\n\n");
-                          file.writeTo(builder);
-
-                          Files.write(
-                              outputFile.toPath(),
-                              builder.toString().getBytes(StandardCharsets.UTF_8),
-                              StandardOpenOption.WRITE,
-                              StandardOpenOption.CREATE,
-                              StandardOpenOption.TRUNCATE_EXISTING);
-                        } catch (IOException ioException) {
-                          throw new IllegalStateException(
-                              "Failed to write file " + outputFile, ioException);
-                        }
-                      });
-            });
+                                  endpoint);
+                            }))
+            .collect(Collectors.toList());
+
+    List<StatefulPair> statefulPairs = StatefulPair.extractStatefulPairs(fullOps);
+
+    fullOps.forEach(
+        (def) -> {
+          TypeSpec spec = def.buildOpClass();
+
+          writeToFile(spec, outputDir, def.packageName);
+        });
+
+    statefulPairs.forEach(
+        (pair) -> {
+          pair.buildOpClasses()
+              .forEach((spec) -> writeToFile(spec, outputDir, pair.getPackageName()));
+        });
   }
 }
diff --git a/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/StatefulPair.java b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/StatefulPair.java
new file mode 100644
index 00000000000..246eb3878f6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-generator/src/main/java/org/tensorflow/op/generator/StatefulPair.java
@@ -0,0 +1,164 @@
+/* Copyright 2021 The TensorFlow Authors. 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.
+You may obtain a copy of the License at
+
+    http://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.
+=======================================================================
+*/
+package org.tensorflow.op.generator;
+
+import com.squareup.javapoet.TypeSpec;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
+import java.util.StringJoiner;
+
+public final class StatefulPair {
+  public final FullOpDef statefulOp;
+  public final FullOpDef statelessOp;
+  public final String selectorClassName;
+  public final String statefulClassName;
+  public final String statelessClassName;
+
+  public StatefulPair(FullOpDef statefulOp, FullOpDef statelessOp) {
+    this.statefulOp = statefulOp;
+    this.statelessOp = statelessOp;
+
+    this.selectorClassName = statefulOp.className.replace("stateful", "").replace("Stateful", "");
+
+    if (statefulOp.className.toLowerCase().contains("stateful")) {
+      statefulClassName = statefulOp.className;
+    } else {
+      statefulClassName = "Stateful" + statefulOp.className;
+    }
+
+    if (statelessOp.className.toLowerCase().contains("stateless")) {
+      statelessClassName = statelessOp.className;
+    } else {
+      statelessClassName = "Stateless" + statelessOp.className;
+    }
+  }
+
+  public static List<StatefulPair> extractStatefulPairs(List<FullOpDef> ops) {
+    List<StatefulPair> pairs = new ArrayList<>(10);
+    for (FullOpDef stateful : ops) {
+      if (!stateful.isStateful()) {
+        continue;
+      }
+
+      for (FullOpDef stateless : ops) {
+        if (stateful.isStateVariant(stateless) && !stateful.equals(stateless)) {
+          if (stateful.opDef.getName().toLowerCase().contains("stateful")
+              || stateless.opDef.getName().toLowerCase().contains("stateless")) {
+            pairs.add(new StatefulPair(stateful, stateless));
+          }
+        }
+      }
+    }
+    for (StatefulPair pair : pairs) {
+      ops.remove(pair.statefulOp);
+      ops.remove(pair.statelessOp);
+    }
+    return pairs;
+  }
+
+  public String getPackageName() {
+    // guaranteed to be the same in FullOpDef.isStateVariant
+    return statefulOp.packageName;
+  }
+
+  public String getGroup() {
+    // guaranteed to be the same in FullOpDef.isStateVariant
+    return statefulOp.group;
+  }
+
+  public boolean hasOptionalAttrs() {
+    return statefulOp.opDef.getAttrList().stream()
+        .anyMatch(attr -> attr.hasDefaultValue() && !attr.getType().contains("type"));
+  }
+
+  public List<TypeSpec> buildOpClasses() {
+
+    TypeSpec.Builder selector = TypeSpec.interfaceBuilder(selectorClassName);
+    try {
+      new ClassGenerator(
+              selector,
+              statefulOp.opDef,
+              statefulOp.apiDef,
+              statefulOp.basePackage,
+              getPackageName(),
+              getGroup(),
+              selectorClassName,
+              statefulOp.endpoint,
+              this)
+          .buildClass();
+    } catch (Exception e) {
+      throw new IllegalStateException(
+          "Failed to generate statefulness selector class for ops "
+              + statefulOp.opDef.getName()
+              + " and "
+              + statelessOp.opDef.getName(),
+          e);
+    }
+
+    TypeSpec stateful = statefulOp.buildOpClass(statefulClassName, this);
+    TypeSpec stateless = statelessOp.buildOpClass(statelessClassName, this);
+
+    return Arrays.asList(stateful, stateless, selector.build());
+  }
+
+  @Override
+  public boolean equals(Object o) {
+    if (this == o) {
+      return true;
+    }
+    if (o == null || getClass() != o.getClass()) {
+      return false;
+    }
+
+    StatefulPair that = (StatefulPair) o;
+
+    if (!statefulOp.equals(that.statefulOp)) {
+      return false;
+    }
+    if (!statelessOp.equals(that.statelessOp)) {
+      return false;
+    }
+    if (!selectorClassName.equals(that.selectorClassName)) {
+      return false;
+    }
+    if (!statefulClassName.equals(that.statefulClassName)) {
+      return false;
+    }
+    return statelessClassName.equals(that.statelessClassName);
+  }
+
+  @Override
+  public int hashCode() {
+    int result = statefulOp.hashCode();
+    result = 31 * result + statelessOp.hashCode();
+    result = 31 * result + selectorClassName.hashCode();
+    result = 31 * result + statefulClassName.hashCode();
+    result = 31 * result + statelessClassName.hashCode();
+    return result;
+  }
+
+  @Override
+  public String toString() {
+    return new StringJoiner(", ", StatefulPair.class.getSimpleName() + "(", ")")
+        .add("statefulOp=" + statefulOp)
+        .add("statelessOp=" + statelessOp)
+        .add("selectorClassName='" + selectorClassName + "'")
+        .add("statefulClassName='" + statefulClassName + "'")
+        .add("statelessClassName='" + statelessClassName + "'")
+        .toString();
+  }
+}