Book Image

Mastering PyTorch - Second Edition

By : Ashish Ranjan Jha

4 (1)

Book Image

Mastering PyTorch - Second Edition

4 (1)

By: Ashish Ranjan Jha

Overview of this book

PyTorch is making it easier than ever before for anyone to build deep learning applications. This PyTorch deep learning book will help you uncover expert techniques to get the most out of your data and build complex neural network models. You’ll build convolutional neural networks for image classification and recurrent neural networks and transformers for sentiment analysis. As you advance, you'll apply deep learning across different domains, such as music, text, and image generation, using generative models, including diffusion models. You'll not only build and train your own deep reinforcement learning models in PyTorch but also learn to optimize model training using multiple CPUs, GPUs, and mixed-precision training. You’ll deploy PyTorch models to production, including mobile devices. Finally, you’ll discover the PyTorch ecosystem and its rich set of libraries. These libraries will add another set of tools to your deep learning toolbelt, teaching you how to use fastai to prototype models and PyTorch Lightning to train models. You’ll discover libraries for AutoML and explainable AI (XAI), create recommendation systems, and build language and vision transformers with Hugging Face. By the end of this book, you'll be able to perform complex deep learning tasks using PyTorch to build smart artificial intelligence models.

Preface

Who this book is for

What this book covers

To get the most out of this book

Overview of Deep Learning Using PyTorch

Overview of Deep Learning Using PyTorch

A refresher on deep learning

Optimization schedule

Exploring the PyTorch library in contrast to TensorFlow

Free Chapter

Deep CNN Architectures

Deep CNN Architectures

Why are CNNs so powerful?

Evolution of CNN architectures

Developing LeNet from scratch

Fine-tuning the AlexNet model

Running a pretrained VGG model

Exploring GoogLeNet and Inception v3

Discussing ResNet and DenseNet architectures

Understanding EfficientNets and the future of CNN architectures

Combining CNNs and LSTMs

Combining CNNs and LSTMs

Building a neural network with CNNs and LSTMs

Building an image caption generator using PyTorch

Deep Recurrent Model Architectures

Deep Recurrent Model Architectures

Exploring the evolution of recurrent networks

Training RNNs for sentiment analysis

Building a bidirectional LSTM

Discussing GRUs and attention-based models

Advanced Hybrid Models

Advanced Hybrid Models

Building a transformer model for language modeling

Developing a RandWireNN model from scratch

Graph Neural Networks

Graph Neural Networks

Introduction to GNNs

Types of graph learning tasks

Reviewing prominent GNN models

Training a GAT model with PyTorch Geometric

Music and Text Generation with PyTorch

Music and Text Generation with PyTorch

Building a transformer-based text generator with PyTorch

Using GPT models as text generators

Generating MIDI music with LSTMs using PyTorch

Neural Style Transfer

Neural Style Transfer

Understanding how to transfer style between images

Implementing neural style transfer using PyTorch

Deep Convolutional GANs

Deep Convolutional GANs

Defining the generator and discriminator networks

Training a DCGAN using PyTorch

Using GANs for style transfer

Image Generation Using Diffusion

Image Generation Using Diffusion

Understanding image generation using diffusion

Training a diffusion model for image generation

Understanding text-to-image generation using diffusion

Using the Stable Diffusion model to generate images from text

Deep Reinforcement Learning

Deep Reinforcement Learning

Reviewing RL concepts

Discussing Q-learning

Understanding deep Q-learning

Building a DQN model in PyTorch

Model Training Optimizations

Model Training Optimizations

Distributed training with PyTorch

Distributed training on GPUs with CUDA

Operationalizing PyTorch Models into Production

Operationalizing PyTorch Models into Production

Model serving in PyTorch

Building a basic model server

Creating a model microservice

Serving a PyTorch model using TorchServe

Exporting universal PyTorch models using TorchScript and ONNX

Running a PyTorch model in C++

Using ONNX to export PyTorch models

Serving PyTorch models in the cloud

PyTorch on Mobile Devices

PyTorch on Mobile Devices

Deploying a PyTorch model on Android

Using the phone camera in the Android app to capture images

Building PyTorch apps on iOS

Rapid Prototyping with PyTorch

Rapid Prototyping with PyTorch

Using fastai to set up model training in a few minutes

Training models on any hardware using PyTorch Lightning

Profiling MNIST model inference using PyTorch Profiler

PyTorch and AutoML

PyTorch and AutoML

Finding the best neural architectures with AutoML

Using Optuna for hyperparameter search

PyTorch and Explainable AI

PyTorch and Explainable AI

Model interpretability in PyTorch

Using Captum to interpret models

Recommendation Systems with PyTorch

Recommendation Systems with PyTorch

Using deep learning for recommendation systems

Understanding and processing the MovieLens dataset

Training and evaluating a recommendation system model

Building a recommendation system using the trained model

PyTorch and Hugging Face

PyTorch and Hugging Face

Understanding Hugging Face within the PyTorch context

Using the Hugging Face Hub for pre-trained models

Using the Hugging Face Datasets library with PyTorch

Using Accelerate to speed up PyTorch model training

Using Optimum to optimize PyTorch model deployment

Index

Customer Reviews

4 (1)

5 star

0

4 star

100%

3 star

0

2 star

0

1 star

0

Model Training Optimizations

Before serving pre-trained machine learning models, which we will discuss extensively in Chapter 13, Operationalizing PyTorch Models into Production, we need to train them. In Chapters 2 to 6, we saw the vast expanse of increasingly complex deep learning model architectures. Such gigantic models often have millions and even billions of parameters. The recent (at the time of writing) Pathways Language Model (PaLM) can have up to 540 billion parameters, for example using backpropagation to tune these many parameters requires enormous amounts of memory and compute power. And even then, model training can take days to finish.

In this chapter, we will explore ways of speeding up the model training process by distributing the training task across machines and processes within machines. We will learn about the distributed training APIs offered by PyTorch – torch.distributed, torch.multiprocessing and torch.utils.data.distributed.DistributedSampler...