Book Image

Distributed Machine Learning with Python

By : Guanhua Wang
Book Image

Distributed Machine Learning with Python

By: Guanhua Wang

Overview of this book

Reducing time cost in machine learning leads to a shorter waiting time for model training and a faster model updating cycle. Distributed machine learning enables machine learning practitioners to shorten model training and inference time by orders of magnitude. With the help of this practical guide, you'll be able to put your Python development knowledge to work to get up and running with the implementation of distributed machine learning, including multi-node machine learning systems, in no time. You'll begin by exploring how distributed systems work in the machine learning area and how distributed machine learning is applied to state-of-the-art deep learning models. As you advance, you'll see how to use distributed systems to enhance machine learning model training and serving speed. You'll also get to grips with applying data parallel and model parallel approaches before optimizing the in-parallel model training and serving pipeline in local clusters or cloud environments. By the end of this book, you'll have gained the knowledge and skills needed to build and deploy an efficient data processing pipeline for machine learning model training and inference in a distributed manner.
Table of Contents (17 chapters)
Section 1 – Data Parallelism
Section 2 – Model Parallelism
Section 3 – Advanced Parallelism Paradigms

Case study of Mesh-TensorFlow

We discussed Megatron-LM in detail due to its popularity. Now, we will briefly discuss Mesh-TensorFlow in this section.

This approach is quite easy to understand. Basically, Mesh-TensorFlow combines data and model parallelism by allowing users to configure two dimensions—that is, batch and model dimensions—as shown in the following diagram:

Figure 9.13 – Mesh-TensorFlow's two-dimensional (2D) parallelism

As shown in the preceding diagram, mesh-tensorflow allows users to set parallelism levels in two dimensions, as follows:

  • Batch dimension: How many concurrent batches to train (data parallelism)
  • Model dimension: How many splits over the model (model parallelism)

As shown in Figure 9.13, let's assume the user sets both batch dimension as 2 and model dimension as 2. This means that we use two GPUs to do model-parallel training, and we have two groups of this two-GPU model parallelism...