Top positive review
This is the best book I have come across for learning about Transformers in NLP
Reviewed in the United States on February 11, 2021
This is a great book for anyone new to the subject of deep learning applied to AI. The author goes to great lengths to explain each topic in sufficient detail to understand it. He then follows it up with a python program to illustrate the key aspect of the topic. The Transformers model is explained in detail with the simplified attention getting as their key to encoding and decoding. BERT is then one of the metric programs often used for measuring the performance of the particular NLP app, Transformers in this case. The author goes further in explaining how Bert does it. This opens the door to using it for other mappings. Thus the book handles roBERTa, GLUE, SuperGlue and etc.
The chapters on translations and generation are particularly interesting because it leads to a discussion of generation and GPT-2 and GPT-3. This is a topic that requires massive computing because of the number of words involved in their data, (540 M) and (1.75 Billions, 8 10000 PCs) respectively. Three of the later chapters are devoted to word extraction. The final three chapters are devoted to Language understanding.
I have no hesitation in recommending this book to any student of modern AI.