Skip to main content

Some fun with Apache Wayang and Spark / Tensorflow

Image
By
Listen:

Apache Wayang is an open-source Federated Learning (FL) framework developed by the Apache Software Foundation. It provides a platform for distributed machine learning, with a focus on ease of use and flexibility. It supports multiple FL scenarios and provides a variety of tools and components for building FL systems. It also includes support for various communication protocols and data formats, as well as integration with other Apache projects such as Apache Kafka and Apache Pulsar for data streaming. The project aims to make it easier to develop and deploy machine learning models in decentralized environments.

It's important to note that this are just examples and they may not be the way for your project to interact with Apache Wayang, you may need to check the documentation of the Apache Wayang project (https://wayang.apache.org) to see how to interact with it. I just point out how easy it is to use different languages to interact between Wayang and Spark.

Also, you need to make sure that you have the correct permissions and credentials to interact with the Wayang API and make changes to the Spark cluster.

Wayang - Scala - Spark:

import org.apache.wayang.{Wayang, WayangClient}

class SparkScaler(wayangUrl: String) {
    val wayang = new WayangClient(wayangUrl)

    def scaleUp(numWorkers: Int): Unit = {
        wayang.addWorkers(numWorkers)
    }

    def scaleDown(numWorkers: Int): Unit = {
        wayang.removeWorkers(numWorkers)
    }
}

The SparkScaler class takes a single parameter, the URL of the Wayang API endpoint, when it is initialized. The scaleUp() method can be called to add a specified number of workers to the Spark cluster, and the scaleDown() method can be called to remove a specified number of workers.

Wayang - Python - Spark

from apache_wayang import Wayang

class SparkScaler:
    def __init__(self, wayang_url):
        self.wayang = Wayang(wayang_url)

    def scale_up(self, num_workers):
        self.wayang.add_workers(num_workers)

    def scale_down(self, num_workers):
        self.wayang.remove_workers(num_workers)

The SparkScaler class takes a single parameter, the URL of the Wayang API endpoint, when it is initialized. The scale_up() method can be called to add a specified number of workers to the Spark cluster, and the scale_down() method can be called to remove a specified number of workers.

Wayang - Java Streams - Spark

import org.apache.wayang.WayangClient;
import java.util.stream.IntStream;

public class SparkScaler {
    private WayangClient wayang;

    public SparkScaler(String wayangUrl) {
        wayang = new WayangClient(wayangUrl);
    }

    public void scaleUp(int numWorkers) {
        IntStream.range(0, numWorkers).forEach(i -> wayang.addWorker());
    }

    public void scaleDown(int numWorkers) {
        IntStream.range(0, numWorkers).forEach(i -> wayang.removeWorker());
    }
}

The SparkScaler class takes a single parameter, the URL of the Wayang API endpoint, when it is initialized. The scaleUp() method can be called to add a specified number of workers to the Spark cluster, and the scaleDown() method can be called to remove a specified number of workers.

Iterate the K-Means clustering algorithm from Apache Wayang to TensorFlow

import org.apache.wayang.WayangClient;
import org.tensorflow.Graph;
import org.tensorflow.Session;
import org.tensorflow.Tensor;

public class KMeansIteration {
    private WayangClient wayang;
    private Graph graph;
    private Session session;

    public KMeansIteration(String wayangUrl, String modelPath) {
        wayang = new WayangClient(wayangUrl);
        graph = new Graph();
        graph.importGraphDef(modelPath);
        session = new Session(graph);
    }

    public void iterate(Tensor input) {
        Tensor wayangOutput = wayang.runKMeans(input);
        Tensor tfOutput = session.runner().feed("input", wayangOutput).fetch("output").run().get(0);
        // Perform further processing on tfOutput
    }
}

That's are only examples to show how easy it can be to get started with FL and also get involved into Wayang as a developer. Also consider to contribute to the project, check the project under wayang.apache.org 

The KMeansIteration class takes two parameters, the URL of the Wayang API endpoint and the path of the TensorFlow model, when it is initialized. The iterate() method can be called with an input Tensor, it will pass it to the Wayang's K-Means clustering algorithm, it will receive the output, and then will pass it to the TensorFlow's model as an input.

Labels:

Comments

  1. Alexander Alten15 June, 2023

    I like also to point to https://wayang.apache.org as open source alternative. Another good one, commercial wise, is Blossom Sky (https://www.databloom.ai/blossom-sky).

    ReplyDelete

Post a Comment

Popular posts from this blog

Deal with corrupted messages in Apache Kafka

By
Under some strange circumstances, it can happen that a message in a Kafka topic is corrupted. This often happens when using 3rd party frameworks with Kafka. In addition, Kafka < 0.9 does not have a lock on Log.read() at the consumer read level, but does have a lock on Log.write(). This can lead to a rare race condition as described in KAKFA-2477 [1]. A likely log entry looks like this: ERROR Error processing message, stopping consumer: (kafka.tools.ConsoleConsumer$) kafka.message.InvalidMessageException: Message is corrupt (stored crc = xxxxxxxxxx, computed crc = yyyyyyyyyy Kafka-Tools Kafka stores the offset of each consumer in Zookeeper. To read the offsets, Kafka provides handy tools [2]. But you can also use zkCli.sh, at least to display the consumer and the stored offsets. First we need to find the consumer for a topic (> Kafka 0.9): bin/kafka-consumer-groups.sh --zookeeper management01:2181 --describe --group test Prior to Kafka 0.9, the only way to get this in...
Post a Comment
Read more

Beyond Ctrl+F - Use LLM's For PDF Analysis

By
PDFs are everywhere, seemingly indestructible, and present in our daily lives at all thinkable and unthinkable positions. We've all got mountains of them, and even companies shouting about "digital transformation" haven't managed to escape their clutches. Now, I'm a product guy, not a document management guru. But I started thinking: if PDFs are omnipresent in our existence, why not throw some cutting-edge AI at the problem? Maybe Large Language Models (LLMs) and Retrieval Augmented Generation (RAG) could be the answer. Don't get me wrong, PDF search indexes like Solr exist, but they're basically glorified Ctrl+F. They point you to the right file, but don't actually help you understand what's in it. And sure, Microsoft Fabric's got some fancy PDF Q&A stuff, but it's a complex beast with a hefty price tag. That's why I decided to experiment with LLMs and RAG. My idea? An intelligent knowledge base built on top of our existing P...
Post a Comment
Read more

Run Llama3 (or any LLM / SLM) on Your MacBook in 2024

By
I'm gonna be real with you: the Cloud and SaaS / PaaS is great... until it isn't. When you're elbow-deep in doing something with the likes of ChatGPT or Gemini or whatever, the last thing you need is your AI assistant starts choking (It seems that upper network connection was reset) because 5G or the local WiFi crapped out or some server halfway across the world is having a meltdown(s). That's why I'm all about running large language models (LLMs) like Llama3 locally. Yep, right on your trusty MacBook. Sure, the cloud's got its perks, but here's why local is the way to go, especially for me: Privacy:  When you're brainstorming the next big thing, you don't want your ideas floating around on some random server. Keeping your data local means it's  yours , and that's a level of control I can get behind. Offline = Uninterrupted Flow:  Whether you're on a plane, at a coffee shop with spotty wifi, or jus...
Post a Comment
Read more