acknowledgments
We deeply thank the contributing authors who created and sustained the collective intelligence that went into this book, often putting into words the ideas we could not.
Hannes Hapke and Cole Howard were crucial in creating the first edition of this book and fostering our mutual learning and growth as NLP engineers.
When we set out to write the second edition, we were fortunate to tap into the collective intelligence of the San Diego Machine Learning community, and it amazed us how many people chose to generously give their time and mind to cocreate with us.
Brian Cox took on the daunting task of rewriting the entire vector and linear algebra appendix.
Geoffrey Marshall valiantly drafted all of chapter 9, which Hobson then mangled, trying to get it up to speed with PyTorch’s evolution—as with the other chapters, all bugs and mistakes are Hobson’s. Geoffrey’s writing discipline inspired us throughout the entire process of writing this book.
John Sundin enriched chapter 6 with network diagrams that connect sentences and concepts.
Ted Kye contributed paragraphs about byte pair encoding as well as subword tokenization.
Vishvesh Bhat contributed large parts of chapter 11 and continues to share his groundbreaking research into grounding LLMs at the startups he has cofounded.