Transformers for Natural Language Processing: Build, train, and fine-tune deep neural network architectures for NLP with Python, PyTorch, TensorFlow, BERT, and GPT-3, 2nd Edition Paperback – 25 Mar. 2022

for NLP practitioners who do not want to skim through the entire internet to look for any concept, this is a very good book as it is concise and good in both theory as well as code

Transformer models have powered recent NLP developments and have completely changed the way NLP problems are now approached. Rothman believes that Industry 4.0 professionals need to be aware of multiple approaches and understand that each has its own pros and cons. However, this book is not designed to explain every single transformer model out there. Instead, it tries to explain enough so that readers have enough knowledge to know how to tackle an NLP problem.There is much to like about this book. The book has 16 chapters and begins by explaining transformers, and exploring interesting ideas such as whether programming is now becoming a sub-domain of NLP. There are also fantastic, and very useful, practical examples on how to work with a Bert tokeniser, conditioning a GPT-2 model, question-answering, pre-training RoBERTa models from scratch, and training a tokeniser. The running NLP tasks online section is also useful. Given this readers own background, the chapter on detecting customer emotions to make predictions was fascinating, and frankly left this reader wanting more!. In todays world where XAI is vitally important, the chapter on interpreting black box transformer models, and the section on using BertViz to show visualisations of the activity of transformer models are key to understanding how models work and to interpret model behaviour.This is one of those books where reading is fine, and the questions section at end of each chapter is useful, but one must really do to gain maximum benefit. The book contain lots of Python code to follow but, to be clear, this book is not geared towards Python beginners - there are plenty of great Packt Python beginners books out there already - hence knowledge of NLP and Python is mandatory.In summary, this book cements Rothman's position as the #1 authority on Transformers, and is absolutely the go-to book for Transformers. Highly recommended and a must-buy for any serious NLP practitioner.

The book is great and have all recent advance development in Transformers and has practical codes for each concept. Unfortunately the delivery was very poos and package was half open, I really wanted to have this book so I didn't return it but I expected more from delivery and packaging. Perhaps if the book was in another pack even the outside pack if is broken the book can be protected from any damages.

The book has a shameful amount of repetition of words and statements. No serious editor would let this pass. Moreover, it does not go into a proper explanation of the concepts at all, which is what you expect if you aim to write a book about it. Disappointing.