What this book covers
Chapter 1, Understanding Generative AI: An Introduction, lays the conceptual groundwork, broadening the reader’s fundamental understanding of what this technology does, how it was derived, and how it can be used. It establishes how generative models differ from classical machine learning paradigms and elucidates how they discern complex relationships and idiosyncrasies in data to synthesize human-like text, audio, and video.
Chapter 2, Surveying GenAI Types and Modes: An Overview of GANs, Diffusers, and Transformers, explores the theoretical foundations and real-world applications of these techniques in greater depth. It dissects the architectural innovations and enhancements that improved training stability and output quality over time, bringing us to state-of-the-art LLMs.
Chapter 3, Tracing the Foundations of Natural Language Processing and the Impact of the Transformer, covers the evolution of natural language processing (NLP) that ultimately led to the advent of the Transformer architecture. It introduces the Transformer—its basis in deep learning, its self-attention architecture, and its rapid evolution, which has led to the generative AI phenomenon.
Chapter 4, Applying Pretrained Generative Models: From Prototype to Production, outlines the process of transitioning a generative AI prototype to a production-ready deployment. It walks through setting up a robust Python environment using Docker, GitHub, and CI/CD pipelines, then presents considerations for selecting and deploying a suitable pre-trained model for the project at hand, emphasizing computational considerations, proper evaluation, monitoring, and responsible AI practices.
Chapter 5, Fine-Tuning Generative Models for Specific Tasks, examines how Parameter-Efficient Fine-Tuning (PEFT) facilitates approachable continued training for specific tasks such as question-answering. It explores and defines a range of scalable fine-tuning techniques, comparing them with other approaches such as in-context learning.
Chapter 6, Understanding Domain Adaptation for Large Language Models, introduces domain adaptation, a unique fine-tuning approach that equips models to interpret language unique to specific industries or domains, addressing the gap in LLMs’ understanding of specialized language.
Chapter 7, Mastering the Fundamentals of Prompt Engineering, explores prompting techniques to examine how to adapt a general-purpose LLM without fine-tuning. It explores various prompting strategies that leverage the model’s inherent capabilities to produce targeted and contextually relevant outputs. It explores a simple approach to RAG and provides techniques to understand and measure performance.
Chapter 8, Addressing Ethical Considerations and Charting a Path Toward Trustworthy Generative AI, recognizes the increasing prominence of generative AI and explores the ethical considerations that should guide its progress. It outlines key concepts such as transparency, fairness, accountability, respect for privacy, informed consent, security, and inclusivity, which are essential for the responsible development and use of these technologies.
