Authors: Denis Rothman
Publication Date: January 29, 2021
Prerequisites: Intermediate Python, some knowledge of NLP
Disclaimer: The publisher sent me a copy of this book for review. I promise that everything said here is my own opinion regardless. All reviews at the Cross Trained Mind are open and honest.
About This Book
This is the first book available about the use of transformers in natural language processing. It’s the technique that the top NLP models use as of this writing. It goes into detail about the architecture of transformers before looking at some of the NLP models. I would say that this is THE book to read to go deep into NLP for research or development.
Who Is This For?
This book is for the developer or researcher with a basic knowledge of natural language processing as well as a good handle on Python programming. I would recommend you wait to read this if you don’t know anything about NLP or are still learning Python. This book also makes my required reading list for AI.
The author sets up the book in three overall sections, as per the preface, though the table of contents makes no reference to this macrostructure. I feel this is in error, as a strong macrostructure makes the book more readable. The microstructure is more like what you typically see from Packt, though they add in a Questions section at the end of each chapter. The questions seem to all be true/false, which aids a bit with recall. With how deep the author goes, I would have liked to see deeper questions that the reader can follow up with to gain a better understanding of the content.
Did This Book Succeed?
This book fully succeeds in teaching the reading about the use of transformers, particularly in NLP. A chapter on the use of transformers outside of NLP would have been nice, as well, but I understand that the author kept in their domain so that they could go and stay deep.
Rating and Final Thoughts
Overall, I give this book a 5 out of 5.
This is the second book you should read if you are just starting out in natural language processing, but very much required. It is also useful for an AI generalist since the transformer seems to have applications outside of pure NLP, as we will likely see in the near future. Make sure this is part of your AI library.