GPT-3: NLP with LLMs is a unique, pragmatic take on Generative Pre-trained Transformer 3, the famous AI language model launched by OpenAI in 2020. This model is capable of tackling a wide array of tasks, like conversation, text completion, and even coding with stunningly good performance. Since its launch, the API has powered a staggering number of applications that have now grown into full-fledged startups generating business value. This book will be a deep dive into what GPT-3 is, why it is important, what it can do, what has already been done with it, how to get access to it, and how one can build a GPT-3 powered product from scratch.
This book is for anyone who wants to understand the scope and nature of GPT-3. The book will evaluate the GPT-3 API from multiple perspectives and discuss the various components of the new, burgeoning economy enabled by GPT-3. This book will look at the influence of GPT-3 on important AI trends like creator economy, no-code, and Artificial General Intelligence and will equip the readers to structure their imaginative ideas and convert them from mere concepts to reality.
Author(s): Sandra Kublik, Shubham Saboo
Edition: 1
Publisher: O'Reilly Media
Year: 2022
Language: English
Commentary: Vector PDF
Pages: 148
City: Sebastopol, CA
Tags: Machine Learning; Natural Language Processing; Python; Java; Go; Startups; GPT-3
Cover
Copyright
Table of Contents
Preface
Conventions Used in This Book
Using Code Examples
O’Reilly Online Learning
How to Contact Us
Acknowledgments
From Sandra
From Shubham
Chapter 1. The Era of Large Language Models
Natural Language Processing: Under the Hood
Language Models: Bigger and Better
The Generative Pre-Trained Transformer: GPT-3
Generative Models
Pre-trained Models
Transformer Models
A Brief History of GPT-3
GPT-1
GPT-2
GPT-3
Accessing the OpenAI API
Chapter 2. Using the OpenAI API
Navigating the OpenAI Playground
Prompt Engineering and Design
How the OpenAI API Works
Execution Engine
Response Length
Temperature and Top P
Frequency and Presence Penalties
Best Of
Stop Sequence
Inject Start Text and Inject Restart Text
Show Probabilities
Execution Engines
Davinci
Curie
Babbage
Ada
Instruct Series
Endpoints
List Engines
Retrieve Engine
Completions
Semantic Search
Files
Classification (Beta)
Answers (Beta)
Embeddings
Customizing GPT-3
Apps Powered by Customized GPT-3 Models
How to Customize GPT-3 for Your Application
Tokens
Pricing
GPT-3’s Performance on Conventional NLP Tasks
Text Classification
Named Entity Recognition
Text Summarization
Text Generation
Conclusion
Chapter 3. Programming with GPT-3
Using the OpenAI API with Python
Using the OpenAI API with Go
Using the OpenAI API with Java
GPT-3 Sandbox Powered by Streamlit
Going Live with GPT-3-Powered Applications
Conclusion
Chapter 4. GPT-3 as a Launchpad for Next-Generation Start-ups
Model-as-a-Service
The New Start-up Ecosystem: Case Studies
Creative Applications of GPT-3: Fable Studio
Data Analysis Applications of GPT-3: Viable
Chatbot Applications of GPT-3: Quickchat
Marketing Applications of GPT-3: Copysmith
Coding Applications of GPT-3: Stenography
An Investor’s Perspective on the GPT-3 Start-up Ecosystem
Conclusion
Chapter 5. GPT-3 for Corporations
Case Study: GitHub Copilot
How It Works
Developing Copilot
No-Code/Low-Code: Simplifying Software Development?
Scaling with the API
What’s Next for GitHub Copilot?
Case Study: Algolia Answers
Evaluating NLP Options
Data Privacy
Cost
Speed and Latency
Lessons Learned
Case Study: Microsoft Azure OpenAI Service
A Partnership That Was Meant to Be
An Azure-Native OpenAI API
Resource Management
Security and Data Privacy
Model-as-a-Service at the Enterprise Level
Other Microsoft AI and ML Services
Advice for Enterprises
OpenAI or Azure OpenAI Service: Which Should You Use?
Conclusion
Chapter 6. Challenges, Controversies, and Shortcomings
The Challenge of AI Bias
Anti-Bias Countermeasures
Low-Quality Content and the Spread of Misinformation
The Environmental Impact of LLMs
Proceeding with Caution
Conclusion
Chapter 7. Democratizing Access to AI
No Code? No Problem!
Access and Model-as-a-Service
Conclusion
Index
About the Authors
Colophon
