book

Accelerate Deep Learning Workloads with Amazon SageMaker

Name: Accelerate Deep Learning Workloads with Amazon SageMaker
Author: Vadim Dabravolski
ISBN: 9781801816441

by Vadim Dabravolski

October 2022

Intermediate to advanced

278 pages

6h 51m

English

Packt Publishing

Read now

Unlock full access

Accelerate Deep Learning Workloads with  Amazon SageMaker
ContributorsAbout the AuthorAbout the reviewer
Preface
Who this book is forWhat this book coversTo get the most out of this bookDownload the example code filesDownload the color imagesConventions usedGet in touchShare Your ThoughtsDownload a free PDF copy of this book
Part 1: Introduction to Deep Learning on Amazon SageMaker
Chapter 1: Introducing Deep Learning with Amazon SageMaker
Technical requirementsExploring DL with Amazon SageMakerUsing SageMakerChoosing Amazon SageMaker for DL workloadsManaged compute and storage infrastructureManaged DL software stacksExploring SageMaker’s managed training stack Step 1 – configuring and creating a training jobStep 2 – provisioning the SageMaker training clusterStep 3 – SageMaker accesses the training dataStep 4 – SageMaker deploys the training containerStep 5 – SageMaker starts and monitors the training jobStep 6 – SageMaker persists the training artifactsUsing SageMaker’s managed hosting stackReal-time inference endpointsCreating and using your SageMaker endpointSageMaker asynchronous endpointsSageMaker Batch TransformIntegration with AWS servicesData storage servicesAmazon EFSAmazon FSx for LustreOrchestration servicesSecurity servicesMonitoring servicesSummary
Chapter 2: Deep Learning Frameworks and Containers on SageMaker
Technical requirementsExploring DL frameworks on SageMakerTensorFlow containers PyTorch containers Hugging Face containers Using SageMaker Python SDKUsing SageMaker DL containersContainer usage patternsSageMaker toolkitsDeveloping for script modeDeveloping an inference script for script modeExtending the prebuilt containersDeveloping a BYO container for inferenceProblem overviewDeveloping the serving containerSummary
Chapter 3: Managing SageMaker Development Environment
Technical requirementsSelecting a development environment for SageMakerSetting up a local environment for SageMakerUsing SageMaker StudioDebugging SageMaker code locallySummary
Chapter 4: Managing Deep Learning Datasets
Technical requirementsSelecting storage solutions for ML datasetsAmazon EBS – high-performance block storageAmazon S3 – industry-standard object storageAmazon EFS – general-purpose shared filesystemAmazon FSx for Lustre – high-performance filesystemSageMaker Feature Store – purpose-built ML storageProcessing data at scaleAugmenting image data using SageMaker ProcessingOptimizing data storage and retrievalChoosing a storage solutionStreaming datasets Summary
Part 2: Building and Training Deep Learning Models
Chapter 5: Considering Hardware for Deep Learning Training
Technical requirementsSelecting optimal compute instancesReviewing specialized DL hardwareChoosing optimal instance typesImproving network throughput with EFAIntroducing EFAUsing EFA with custom training containersCompiling models for GPU devices with Training CompilerIntroducing the XLA optimization libraryUsing SageMaker Training CompilerUsing Training CompilerSummary
Chapter 6: Engineering Distributed Training
Technical requirementsEngineering data parallel training Coordination patterns – Parameter Server versus AllreduceEngineering TensorFlow data parallel trainingEngineering PyTorch data parallel trainingEngineering SageMaker’s DDP jobsEngineering model parallel training jobsEngineering training with SDMPOptimizing distributed training jobsCluster layout and computation affinitySummary

Chapter 7: Operationalizing Deep Learning Training
Technical requirementsDebugging training jobsUsing TensorBoard with SageMakerMonitoring training with SageMaker DebuggerProfiling your DL trainingHyperparameter optimizationUsing EC2 Spot InstancesSummary
Part 3: Serving Deep Learning Models
Chapter 8: Considering Hardware for Inference
Technical requirementsSelecting hardware accelerators in AWS CloudLatency-throughput trade-offsCostSupported frameworks and operatorsG4 instance family – best price and performance ratio for inferenceP3 instance family – performant and expensive for inference AWS InferentiaAmazon Elastic InferenceCompiling models for inferenceUsing TensorRTUsing Neuron SDKUsing SageMaker NeoSummary
Chapter 9: Implementing Model Servers
Technical requirementsUsing TFSReviewing TFS conceptsIntegrating TFS with SageMakerOptimizing TFS Implementing TFS servingUsing PTS Integration with SageMaker Optimizing PTS on SageMakerServing models with PTSUsing NVIDIA Triton Integration with SageMakerOptimizing Triton inference Serving models with Triton on SageMakerSummary
Chapter 10: Operationalizing Inference Workloads
Technical requirementsManaging inference deploymentsConsidering model deployment optionsAdvanced model deployment techniquesMonitoring inference workloadsUsing Amazon CloudWatch Monitoring inference workload qualitySelecting your workload configurationUsing SageMaker Inference RecommenderSummary
Index
Why subscribe?
Other Books You May EnjoyPackt is searching for authors like youShare Your ThoughtsDownload a free PDF copy of this book

Content preview from Accelerate Deep Learning Workloads with Amazon SageMaker

Part 2: Building and Training Deep Learning Models

In this part, we will learn how to train DL models using SageMaker-managed capabilities, outlining available software frameworks to distribute training processes across many nodes, optimizing hardware utilization, and monitoring your training jobs.

This section comprises the following chapters:

Chapter 5, Considering Hardware for Deep Learning Training
Chapter 6, Engineering Distributed Training
Chapter 7, Operationalizing Deep Learning Training

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Machine Learning with Amazon SageMaker Cookbook

Publisher Resources

ISBN: 9781801816441

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Accelerate Deep Learning Workloads with Amazon SageMaker

by Vadim Dabravolski

Part 2: Building and Training Deep Learning Models

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.