chore: update to spark 3.3.1

build.sbt

@@ -7,10 +7,10 @@ import scala.xml.transform.{RewriteRule, RuleTransformer}
 import scala.xml.{Node => XmlNode, NodeSeq => XmlNodeSeq, _}
 
 val condaEnvName = "synapseml"
-val sparkVersion = "3.4.1"
+val sparkVersion = "3.3.2"
 name := "synapseml"
 ThisBuild / organization := "com.microsoft.azure"
-ThisBuild / scalaVersion := "2.12.17"
+ThisBuild / scalaVersion := "2.12.15"
 
 val scalaMajorVersion = 2.12
 
@@ -20,24 +20,23 @@ val excludes = Seq(
 )
 
 val coreDependencies = Seq(
-  // Excluding protobuf-java, as spark-core is bringing the older version transitively.
-  "org.apache.spark" %% "spark-core" % sparkVersion % "compile" exclude("com.google.protobuf", "protobuf-java"),
+  "org.apache.spark" %% "spark-core" % sparkVersion % "compile",
   "org.apache.spark" %% "spark-mllib" % sparkVersion % "compile",
-  "org.apache.spark" %% "spark-avro" % sparkVersion % "compile",
+  "org.apache.spark" %% "spark-avro" % sparkVersion % "provided",
   "org.apache.spark" %% "spark-tags" % sparkVersion % "test",
   "com.globalmentor" % "hadoop-bare-naked-local-fs" % "0.1.0" % "test",
   "org.scalatest" %% "scalatest" % "3.2.14" % "test")
 val extraDependencies = Seq(
-  "commons-lang" % "commons-lang" % "2.6",
   "org.scalactic" %% "scalactic" % "3.2.14",
   "io.spray" %% "spray-json" % "1.3.5",
   "com.jcraft" % "jsch" % "0.1.54",
   "org.apache.httpcomponents.client5" % "httpclient5" % "5.1.3",
   "org.apache.httpcomponents" % "httpmime" % "4.5.13",
-  "com.linkedin.isolation-forest" %% "isolation-forest_3.4.1" % "3.0.3",
-  // Although breeze 2.1.0 is already provided by Spark, this is needed for Azure Synapse Spark 3.4 pools.
-  // Otherwise a NoSuchMethodError will be thrown by interpretability code.
-  "org.scalanlp" %% "breeze" % "2.1.0"
+  "com.linkedin.isolation-forest" %% "isolation-forest_3.3.3" % "3.0.3",
+  // Although breeze 1.2 is already provided by Spark, this is needed for Azure Synapse Spark 3.2 pools.
+  // Otherwise a NoSuchMethodError will be thrown by interpretability code. This problem only happens
+  // to Azure Synapse Spark 3.2 pools.
+  "org.scalanlp" %% "breeze" % "1.2"
 ).map(d => d excludeAll (excludes: _*))
 val dependencies = coreDependencies ++ extraDependencies
 

core/src/main/scala/com/microsoft/azure/synapse/ml/codegen/PyCodegen.scala

@@ -70,7 +70,7 @@ object PyCodegen {
       // There's `Already borrowed` error found in transformers 4.16.2 when using tokenizers
       s"""extras_require={"extras": [
          |    "cmake",
-         |    "horovod==0.28.1",
+         |    "horovod==0.27.0",
          |    "pytorch_lightning>=1.5.0,<1.5.10",
          |    "torch==1.13.1",
          |    "torchvision>=0.14.1",

core/src/test/scala/com/microsoft/azure/synapse/ml/codegen/RTestGen.scala

@@ -101,7 +101,7 @@ object RTestGen {
          |  "spark.sql.shuffle.partitions=10",
          |  "spark.sql.crossJoin.enabled=true")
          |
-         |sc <- spark_connect(master = "local", version = "3.4.1", config = conf)
+         |sc <- spark_connect(master = "local", version = "3.3.2", config = conf)
          |
          |""".stripMargin, StandardOpenOption.CREATE)
 

core/src/test/scala/com/microsoft/azure/synapse/ml/nbtest/DatabricksUtilities.scala

@@ -30,11 +30,11 @@ object DatabricksUtilities {
 
   // ADB Info
   val Region = "eastus"
-  val PoolName = "synapseml-build-13.3"
-  val GpuPoolName = "synapseml-build-13.3-gpu"
-  val AdbRuntime = "13.3.x-scala2.12"
-  // https://docs.databricks.com/en/release-notes/runtime/13.3lts-ml.html
-  val AdbGpuRuntime = "13.3.x-gpu-ml-scala2.12"
+  val PoolName = "synapseml-build-12.2"
+  val GpuPoolName = "synapseml-build-12.2-gpu"
+  val AdbRuntime = "12.2.x-scala2.12"
+  // https://learn.microsoft.com/en-us/azure/databricks/release-notes/runtime/
+  val AdbGpuRuntime = "12.2.x-gpu-ml-scala2.12"
   val NumWorkers = 5
   val AutoTerminationMinutes = 15
 
@@ -82,9 +82,9 @@ object DatabricksUtilities {
     Map("maven" -> Map("coordinates" -> PackageMavenCoordinate, "repo" -> PackageRepository)),
     Map("pypi" -> Map("package" -> "pytorch-lightning==1.5.0")),
     Map("pypi" -> Map("package" -> "torchvision==0.14.1")),
-    Map("pypi" -> Map("package" -> "transformers==4.32.1")),
-    Map("pypi" -> Map("package" -> "petastorm==0.12.0")),
-    Map("pypi" -> Map("package" -> "protobuf==3.20.3"))
+    Map("pypi" -> Map("package" -> "transformers==4.25.1")),
+    Map("pypi" -> Map("package" -> "petastorm==0.12.1")),
+    Map("pypi" -> Map("package" -> "protobuf==3.19.4"))
   ).toJson.compactPrint
 
   val GPUInitScripts: String = List(

core/src/test/scala/com/microsoft/azure/synapse/ml/nbtest/SynapseExtension/SynapseExtensionUtilities.scala

@@ -83,7 +83,7 @@ object SynapseExtensionUtilities {
          |"{
          |  'Default${store}ArtifactId': '$storeId',
          |  'ExecutableFile': '$path',
-         |  'SparkVersion':'3.4',
+         |  'SparkVersion':'3.3',
          |  'SparkSettings': {
          |    'spark.jars.packages' : '$SparkMavenPackageList',
          |    'spark.jars.repositories' : '$SparkMavenRepositoryList',

core/src/test/scala/com/microsoft/azure/synapse/ml/nbtest/SynapseUtilities.scala

@@ -254,7 +254,7 @@ object SynapseUtilities {
        |    "nodeSizeFamily": "MemoryOptimized",
        |    "provisioningState": "Succeeded",
        |    "sessionLevelPackagesEnabled": "true",
-       |    "sparkVersion": "3.4"
+       |    "sparkVersion": "3.3"
        |  }
        |}
        |""".stripMargin

deep-learning/src/main/python/horovod_installation.sh

@@ -8,9 +8,9 @@ set -eu
 # Install prerequisite libraries that horovod depends on
 pip install pytorch-lightning==1.5.0
 pip install torchvision==0.14.1
-pip install transformers==4.32.1
+pip install transformers==4.25.1
 pip install petastorm>=0.12.0
-pip install protobuf==3.20.3
+pip install protobuf==3.19.1
 
 # Remove Outdated Signing Key:
 sudo apt-key del 7fa2af80
@@ -35,13 +35,13 @@ libcusparse-dev-11-0=11.1.1.245-1
 
 git clone --recursive https://github.com/horovod/horovod.git
 cd horovod
-# git fetch origin refs/tags/v0.28.1:tags/v0.28.1
-git checkout 1d217b59949986d025f6db93c49943fb6b6cc78f
+# git fetch origin refs/tags/v0.27.0:tags/v0.27.0
+git checkout bfaca90d5cf66780a97d8799d4e1573855b64560
 git checkout -b tmp-branch
 rm -rf build/ dist/
 HOROVOD_GPU_ALLREDUCE=NCCL HOROVOD_CUDA_HOME=/usr/local/cuda-11/ HOROVOD_WITH_PYTORCH=1 HOROVOD_WITHOUT_MXNET=1 \
 /databricks/python3/bin/python setup.py bdist_wheel
 
 readlink -f dist/horovod-*.whl
 
-pip install --no-cache-dir dist/horovod-0.28.1-cp38-cp38-linux_x86_64.whl --force-reinstall --no-deps
+pip install --no-cache-dir dist/horovod-0.27.0-cp38-cp38-linux_x86_64.whl --force-reinstall --no-deps

deep-learning/src/main/python/synapse/ml/dl/DeepTextClassifier.py

@@ -11,12 +11,12 @@
 if _TRANSFORMERS_AVAILABLE:
     import transformers
 
-    _TRANSFORMERS_EQUAL_4_32_1 = transformers.__version__ == "4.32.1"
-    if _TRANSFORMERS_EQUAL_4_32_1:
+    _TRANSFORMERS_EQUAL_4_25_1 = transformers.__version__ == "4.25.1"
+    if _TRANSFORMERS_EQUAL_4_25_1:
         from transformers import AutoTokenizer
     else:
         raise RuntimeError(
-            "transformers should be == 4.32.1, found: {}".format(
+            "transformers should be == 4.25.1, found: {}".format(
                 transformers.__version__
             )
         )

deep-learning/src/main/python/synapse/ml/dl/DeepVisionClassifier.py

@@ -19,10 +19,10 @@
 if _HOROVOD_AVAILABLE:
     import horovod
 
-    _HOROVOD_EQUAL_0_28_1 = horovod.__version__ == "0.28.1"
-    if not _HOROVOD_EQUAL_0_28_1:
+    _HOROVOD_EQUAL_0_27_0 = horovod.__version__ == "0.27.0"
+    if not _HOROVOD_EQUAL_0_27_0:
         raise RuntimeError(
-            "horovod should be of version 0.28.1, found: {}".format(horovod.__version__)
+            "horovod should be of version 0.27.0, found: {}".format(horovod.__version__)
         )
 else:
     raise ModuleNotFoundError("module not found: horovod")

deep-learning/src/main/python/synapse/ml/dl/LitDeepTextModel.py

@@ -13,12 +13,12 @@
 if _TRANSFORMERS_AVAILABLE:
     import transformers
 
-    _TRANSFORMERS_EQUAL_4_32_1 = transformers.__version__ == "4.32.1"
-    if _TRANSFORMERS_EQUAL_4_32_1:
+    _TRANSFORMERS_EQUAL_4_25_1 = transformers.__version__ == "4.25.1"
+    if _TRANSFORMERS_EQUAL_4_25_1:
         from transformers import AutoModelForSequenceClassification
     else:
         raise RuntimeError(
-            "transformers should be == 4.32.1, found: {}".format(
+            "transformers should be == 4.25.1, found: {}".format(
                 transformers.__version__
             )
         )

docs/Explore Algorithms/Deep Learning/Quickstart - Fine-tune a Text Classifier.ipynb

@@ -16,24 +16,33 @@
   },
   {
    "cell_type": "markdown",
-   "source": [
-    "### Environment Setup on databricks"
-   ],
    "metadata": {
     "collapsed": false
-   }
+   },
+   "source": [
+    "### Environment Setup on databricks"
+   ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
+   "metadata": {
+    "collapsed": false
+   },
    "outputs": [],
    "source": [
     "# install cloudpickle 2.0.0 to add synapse module for usage of horovod\n",
     "%pip install cloudpickle==2.0.0 --force-reinstall --no-deps"
-   ],
-   "metadata": {
-    "collapsed": false
-   }
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%pip install protobuf==3.20.1 --force-reinstall"
+   ]
   },
   {
    "cell_type": "code",

docs/Explore Algorithms/Deep Learning/Quickstart - Fine-tune a Vision Classifier.ipynb

@@ -25,6 +25,15 @@
     "%pip install cloudpickle==2.0.0 --force-reinstall --no-deps"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%pip install protobuf==3.20.1 --force-reinstall"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,

docs/Explore Algorithms/Hyperparameter Tuning/HyperOpt.ipynb

@@ -31,7 +31,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "%pip install hyperopt mlflow"
+    "%pip install hyperopt mlflow==2.5.0"
    ]
   },
   {

environment.yml

@@ -11,7 +11,7 @@ dependencies:
   - r-devtools=2.4.2
   - pip:
     - pyarrow>=0.15.0
-    - pyspark==3.4.1
+    - pyspark==3.3.2
     - pandas==1.2.5
     - wheel
     - sphinx==4.2.0
@@ -34,13 +34,13 @@ dependencies:
     - numpy
     - torch==1.13.1
     - torchvision==0.14.1
-    - horovod==0.28.1
+    - horovod==0.27.0
     - petastorm>=0.11.0
     - pytorch_lightning==1.5.0
     - onnxmltools==1.7.0
     - matplotlib
     - Pillow
-    - transformers==4.32.1
+    - transformers==4.25.1
     - huggingface-hub>=0.8.1
     - langchain==0.0.151
     - openai==0.27.5

pipeline.yaml

@@ -527,7 +527,7 @@ jobs:
           fi
           sbt publishM2
 
-          SPARK_VERSION=3.4.1
+          SPARK_VERSION=3.3.2
           HADOOP_VERSION=3
           wget https://archive.apache.org/dist/spark/spark-${SPARK_VERSION}/spark-${SPARK_VERSION}-bin-hadoop${HADOOP_VERSION}.tgz
           (timeout 20m sbt "project $(PACKAGE)" coverage testR) || (echo "retrying" && timeout 20m sbt "project $(PACKAGE)" coverage testR) || (echo "retrying" && timeout 20m sbt "project $(PACKAGE)" coverage testR)

start

@@ -1,8 +1,7 @@
 #!/bin/bash
 
 export OPENMPI_VERSION="3.1.2"
-
-export SPARK_VERSION="3.4.1"
+export SPARK_VERSION="3.3.2"
 export HADOOP_VERSION="3.3"
 export SYNAPSEML_VERSION="0.11.4" # Binder compatibility version
 

tools/docker/demo/Dockerfile

@@ -3,7 +3,7 @@ FROM mcr.microsoft.com/oss/mirror/docker.io/library/ubuntu:20.04
 ARG SYNAPSEML_VERSION=0.11.4
 ARG DEBIAN_FRONTEND=noninteractive
 
-ENV SPARK_VERSION=3.4.1
+ENV SPARK_VERSION=3.3.2
 ENV HADOOP_VERSION=3
 ENV SYNAPSEML_VERSION=${SYNAPSEML_VERSION}
 ENV JAVA_HOME /usr/lib/jvm/java-1.11.0-openjdk-amd64

tools/docker/minimal/Dockerfile

@@ -3,7 +3,7 @@ FROM mcr.microsoft.com/oss/mirror/docker.io/library/ubuntu:20.04
 ARG SYNAPSEML_VERSION=0.11.4
 ARG DEBIAN_FRONTEND=noninteractive
 
-ENV SPARK_VERSION=3.4.1
+ENV SPARK_VERSION=3.3.2
 ENV HADOOP_VERSION=3
 ENV SYNAPSEML_VERSION=${SYNAPSEML_VERSION}
 ENV JAVA_HOME /usr/lib/jvm/java-1.11.0-openjdk-amd64

tools/dotnet/dotnetSetup.sh

@@ -20,11 +20,11 @@ echo "##vso[task.setvariable variable=DOTNET_WORKER_DIR]$DOTNET_WORKER_DIR"
 # Install Sleet
 dotnet tool install -g sleet
 
-# Install Apache Spark-3.4.1
-curl https://archive.apache.org/dist/spark/spark-3.4.1/spark-3.4.1-bin-hadoop3.tgz -o spark-3.4.1-bin-hadoop3.tgz
+# Install Apache Spark-3.3
+curl https://archive.apache.org/dist/spark/spark-3.3.2/spark-3.3.2-bin-hadoop3.tgz -o spark-3.3.2-bin-hadoop3.tgz
 mkdir ~/bin
-tar -xzvf spark-3.4.1-bin-hadoop3.tgz -C ~/bin
-export SPARK_HOME=~/bin/spark-3.4.1-bin-hadoop3/
+tar -xzvf spark-3.3.2-bin-hadoop3.tgz -C ~/bin
+export SPARK_HOME=~/bin/spark-3.3.2-bin-hadoop3/
 export PATH=$SPARK_HOME/bin:$PATH
 echo "##vso[task.setvariable variable=SPARK_HOME]$SPARK_HOME"
 echo "##vso[task.setvariable variable=PATH]$SPARK_HOME/bin:$PATH"

tools/tests/run_r_tests.R

@@ -3,7 +3,7 @@ if (!require("sparklyr")) {
   library("sparklyr")
 }
 
-spark_install_tar(paste(getwd(), "/../../../../../../spark-3.4.1-bin-hadoop3.tgz", sep = ""))
+spark_install_tar(paste(getwd(), "/../../../../../../spark-3.3.2-bin-hadoop3.tgz", sep = ""))
 
 options("testthat.output_file" = "../../../../r-test-results.xml")
 devtools::test(reporter = JunitReporter$new())