Getting Started with MLeap

The MLeap runtime itself provides everything needed to execute andserialize entire ML pipelines. It does not include anything required fortraining ML pipelines. In order to start with MLeap, you will need toadd it to your project.

Adding MLeap to Your Project

MLeap and its snapshots are hosted on Maven Central and so should beeasily accessible via a maven build file or SBT. MLeap is currentlycross-compiled for Scala versions 2.10 and 2.11. We try to maintainScala compatibility with Spark.

Using SBT

  1. libraryDependencies += "ml.combust.mleap" %% "mleap-runtime" % "0.14.0"

Using Maven

  1. <dependency>
  2. <groupId>ml.combust.mleap</groupId>
  3. <artifactId>mleap-runtime_2.11</artifactId>
  4. <version>0.14.0</version>
  5. </dependency>

If you are packaging libraries into a single JAR, you need to use the Maven Shade plugin with the following transformer to ensure reference.conf files are merged instead of being overwritten:

  1. <transformer implementation="org.apache.maven.plugins.shade.resource.AppendingTransformer">
  2. <resource>reference.conf</resource>
  3. </transformer>
  1. See build instructions to build MLeap from source.
  2. See core concepts for an overview of ML pipelines.
  3. See basic usage of MLeap to start transforming leap frames.