Everyone, here comes a heavyweight new book from Xiaoyi: O’Reilly’s animal book “Introduction to Generative AI and Practical AWS”!This book has a rating of 4.6 stars on Amazon and has received strong recommendations from numerous industry leaders such as Jia Yangqing, Wang Xiaochuan, and Zhou Ming!
“Introduction to Generative AI and Practical AWS” is your first practical guide to large models, allowing you to easily master the core points of generative AI and ride the wave of future technology!
Part.1
What is Generative AI?
“All products deserve to be redone with large models.” This is a very popular viewpoint in the AI community in recent years. While everyone is discussing large models and generative AI, how to quickly land these cool technologies and truly help businesses and society has become a major challenge. However, AWS has significantly lowered the threshold for large models and generative AI.
What is Generative AI?
Generative AI is a general technology that uses machine learning techniques (especially deep learning) to create new, original content. It can generate various forms of content such as text, images, sounds, videos, code, and web pages based on patterns and rules learned from large amounts of data, using large language models (LLM) and foundational models (FM).
(Generative AI project lifecycle framework) Currently, there are various types of multimodal generative AI tasks, such as:
· Text Summarization
Generating a shorter piece of text while retaining the main ideas. Typically used in customer support to quickly summarize interactions between customers and representatives.
· Rewriting
Modifying text wording to suit different audiences, levels of formality, or tones.
· Information Extraction
Extracting information such as names, addresses, events, data, or numbers from documents.
· Question Answering (QA) and Visual Question Answering (VQA)
Directly asking questions about a set of documents, images, videos, or audio.
· Harmful Content Detection
As an extension of the question-answering task, you can ask generative models whether a set of text, images, videos, or audio contains harmful content.
· Classification and Content Review
Assigning a category to given content (such as documents, images, videos, or audio clips).
· Conversational Interfaces
Handling multi-turn conversations in a chat-like interface to complete corresponding tasks.
· Translation
Language translation is one of the earliest use cases of generative AI.
· Source Code Generation
Generating source code based on natural language comments or hand-drawn sketches.
· Reasoning
Discovering potential new solutions, trade-offs, or hidden details through question reasoning.
· Masking Personally Identifiable Information Generative models can be used to mask personally identifiable information from a given text corpus. This is very useful for many use cases that need to handle sensitive data and wish to remove PII data from workflows.
· Personalized Marketing and Advertising
Generating personalized product descriptions, videos, or advertisements based on user profile characteristics.
In these use cases and tasks of generative AI, the content created by the model is approaching human understanding of language, which is truly amazing.This is achieved through the Transformer neural network architecture.So how does AWS implement the last mile of generative AI?AWS, short for Amazon Web Services, is a cloud computing platform provided by Amazon that is committed to providing enterprise-level security and privacy while offering tools and infrastructure to build and scale generative AI applications.
(AWS services supporting generative AI and corresponding functions)
We typically start generative AI projects from the model hub, for example, in this book, we will use resources from Hugging Face Model Hub and Amazon SageMaker JumpStart to explore foundational models like Meta’s Llama 2, TII’s Falcon, and Google’s FLAN-T5.However, building generative AI applications is not limited to the generative models themselves. To create a reliable and scalable application while ensuring security, and ultimately deliver it to users or other systems as a service, multiple components need to work closely together and integrate. And AWS can perfectly combine them, providing packaged generative AI services like Amazon CodeWhisperer as well as a wide range of services needed to build end-to-end generative AI applications.
(AWS provides a wide range of services needed to build end-to-end generative AI applications)Increasing choice flexibility, enterprise-level security and governance capabilities, advanced generative AI capabilities, reducing operational costs through fully managed services, and the ability to continuously innovate are all important reasons we choose AWS to build generative AI projects.
Part.2
Why Choose This Book to Learn About Large Models?
If you are a programmer, you will be familiar with O’Reilly’s animal books, and this book continues the style of O’Reilly’s animal books, with a cute thrush on the cover. For over 40 years, O’Reilly Media has focused on cutting-edge technology, helping professionals master best practices and lead technological innovation.This book has a rating of 4.6 stars on Amazon, ensuring its quality!
Written by three AWS generative AI experts, it is easy to understand and leads you in learning about large models.
Chris Fregly
AWS Generative AI Chief Solutions Architect and co-author of the O’Reilly book Data Science on AWS.
Antje Barth
AWS Generative AI Chief Developer Advocate and co-author of the O’Reilly book Data Science on AWS.
Shelbee Eigenbrode
AWS Generative AI Chief Solutions Architect, she has over 35 patents in various technical fields.This book, which meets the needs of the times, has naturally received joint recommendations from industry giants and experts from AWS, Meta, Snowflake, Lepton AI, CSDN, Baichuan Intelligence, and others.
Part.3
What Can This Book Help You Learn?
Theory
In this book, you will first explore the concepts of generative AI and its potential applications in products and services, understanding the complete lifecycle of generative AI projects. Then, you will learn about topics such as prompt engineering, few-shot context learning, pre-training of generative models, domain adaptation, model evaluation, parameter-efficient fine-tuning (PEFT), and reinforcement learning from human feedback (RLHF).
Practice
Opening this book, you will understand the commonly used generative AI use cases and tasks in today’s industry and academia. First, you will explore various model types, such as large language models and multimodal models, and learn practical techniques for optimizing these models through prompt engineering and context learning. After gaining practical experience in building these cutting-edge generative models, you can choose to reuse existing generative models or build a new model from scratch. Then, you will learn how to adapt these generative models to specific domain datasets, tasks, and use cases to support commercial applications.
The key generative AI technologies you want to learn, such as LoRA, RLHF, LangChain, and ReAct, can all be found in this book. How to fine-tune models using LoRA technology? How to align models with human values through RLHF? How to develop agents using LangChain and ReAct? Detailed explanations of such technologies and practical application cases are all included.This book also comes with rich code examples, helping you quickly get started with key generative AI technologies!
(Accompanying code for this book: https://github.com/generative-ai-on-aws )Finally, you will learn how to build applications based on generative AI using Amazon Bedrock.
In short, this book deeply analyzes the core points and development difficulties of generative AI from theory to practice! It not only provides a clear learning path for beginners to help them fully understand the depth and breadth of this technical field but also offers in-depth industry insights and specific operational guidelines for professionals who want to build powerful and flexible generative AI applications on AWS.With this book, you will transform from a novice in AI application development into an expert proficient in advanced techniques such as training, tuning, and development. For newly entering AI engineers, product managers, marketers, or business leaders, this book is definitely your strong assistant!With a 4.6-star rating on Amazon, this is your first practical guide to large models, allowing you to easily master the core points of generative AI and ride the wave of future technology!
