The Developer's Guide to AI is out everywhere today. It starts with how LLMs are trained and works up through RAG, fine-tuning, and building agents with real code you can run.
Get it at nostarch.com or wherever you get books
ALT The opening page of Chapter 1 of The Developer's Guide to AI, titled "Understanding Large Language Models," on a yellow background. A large numeral 1 sits above the chapter title, with a small circular illustration of a robot beside introductory text explaining that LLMs are deep learning models pretrained on vast amounts of data. The facing page is partly visible at right, showing the edge of a figure and a table.
ALT A page from Chapter 1 of The Developer's Guide to AI on a yellow background. It includes Figure 1-1, a diagram showing data sources (web pages, books, articles, research papers, and code) feeding into LLM training with a September 30, 2024 cutoff date, and Table 1-1, listing the training data size and sources for BERT, GPT-3, and LLaMA. A shaded sidebar titled "The BERT Breakthrough" explains how BERT advanced natural language processing.
Physical AI without screens. Ira Renfrew and Familiar Machines & Magic are developing AI companions that use touch, movement, and expressiveness instead of displays encouraging users to reconnect with the physical world rather than adding another source of digital distraction.