Book Review: Exploring GPT-3


Exploring GPT-3 is a book written for those aiming to gain practical experience with Artificial Intelligence for Natural Language Processing tasks using the Generative Pre-trained Transformer 3 (GPT-3) model. In my view, the book is aimed at those who have a machine learning, Python and JavaScript programming background. The book begins its first chapter by introducing GPT-3 and OpenAI API and very briefly covers NLP concepts.

The GPT-3 Applications and Use Cases chapter (Chapter 2) provides a comprehensive introduction to the OpenAI Playground and how to handle text generation and classification tasks. Having an OpenAI account and access to GPT-3 will make it easier for the reader to follow the book and to try out the different examples found in the Playground. The book provides a really clear and concise explanation of the concepts and steps needed for getting started with the Playground web-based tool that makes it easy to get familiar with how the API works.

This book speeds up the learning curve of undertanding the OpenAI API and GPT3 model as it contains all the necessary information to get started with basic as well as more advanced tasks. The book also provides a comprehensive understanding of the Playground settings. Chapter 4 provides a step-by-step guide on how to use OpenAI API directly by making HTTP requests, and this is particularly useful to those aiming to include the GPT-3 API into their applications. Chapter 4 also includes a comprehensive introduction to APIs and the HTTP protocol, and this again saves the reader time from having to look elsewhere for the information. Chapter 5 describes how to call the OpenAI API with code.

Importantly, Chapter 6 of the book acknowledges that GPT-3 can generate text that could be considered inappropriate to some users, particularly since GPT-3 was trained using data from the internet which often contains inappropriate content. The chapter demonstrates how to use content filtering for blocking such content. Content filtering is the task of preventing inappropriate and offensive results, by blocking or hiding certain content to prevent users from seeing such results upon application use. Chapter 6 explains Content Filtering and provides step by step instructions with code on how to implement content filtering.

The remaining chapters are focused on applications of GPT-3 such as text translation, chatbots, sentiment analysis, text classification and question answering. The last chapter of the book explains how to go live with OpenAI-Powered Apps. The book clearly explains that before going live with OpenAI-Powered Apps, they must be approved for publishing by OpenAI. The reason behind the approval process is to help prevent the OpenAI from being misused and to also help app providers, and OpenAI, plan for resource requirements and to monitor usage.

Overall, this book is nicely structured and written, it contains all the necessary information the reader needs for gathering a good understanding and practical experience on how to use the GPT-3 model. This 264 page book is easy to read and follow, and I like the fact that in each chapter of the book the reader gets an explanation of concepts and step by step instructions with code snippets and images. The example code files for the book can be downloaded from GiHub.

Details of the book can be found here

Thanks for reading this review, I hope you found it helpful!