Illustrating Generative AI
A Series of Fun Examples to Help Learn Generative Engineering Interactively
Buch in deiner Nähe kaufen
Beschreibung
Every model response printed in this book is real output you can regenerate yourself, on a laptop, for free, using the same small, open-weights models the author ran.
Illustrating Generative AI is an example-driven introduction to building software with free and fully local AI models and tools. It works through the fundamentals of modern AI-assisted engineering, one demo at a time, showing the instructive failures beside the successes. Along the way, the reader gradually builds the discretion, judgment, and critical thinking that working well with these tools demands. The book assumes no mathematics background and keeps every code example simple and close to plain English.
Topics and features:
· Covers foundational concepts such as prompting, tokenization, embeddings, and model efficiency.
· Builds up to tool-using agents, step-by-step reasoning, retrieval, multimodal input, security concerns, and more.
· Illustrates how our choice of model can affect the outcomes of a range of software engineering tasks.
· All examples, demos, and experiments can run entirely on the reader's own hardware with free, open-weights models.
· Every chapter is a runnable Jupyter notebook, all powered by a small companion library, with every reproduction script available in the book's open code repository.
· Uses only free, open-source software: Ollama, Jupyter, Hugging Face Transformers, PyTorch, and Chroma, alongside open models such as FLUX and Whisper.
Illustrating Generative AI is written for undergraduate students, their instructors, and developers building on their own hardware. It's a book for any curious tinkerer who wants to build with private and local AI systems.
Chiké Abuah is an Assistant Professor in Computer Science at Walla Walla University, Washington, USA, where he teaches various Software Engineering and Artificial Intelligence courses. His work spans system verification, machine learning, and privacy-preserving technology.
Illustrating Generative AI is an example-driven introduction to building software with free and fully local AI models and tools. It works through the fundamentals of modern AI-assisted engineering, one demo at a time, showing the instructive failures beside the successes. Along the way, the reader gradually builds the discretion, judgment, and critical thinking that working well with these tools demands. The book assumes no mathematics background and keeps every code example simple and close to plain English.
Builds up to tool-using agents, step-by-step reasoning, retrieval, multimodal input, security concerns, and more Illustrates how our choice of model can affect the outcomes of a range of software engineering tasks Covers foundational concepts such as prompting, tokenization, embeddings, and model efficiency
Autor*in
Chiké Abuah
Themen in »Illustrating Generative AI«
Generative AI programming with Python Large language model applications Retrieval augmented generation (RAG) systems Ollama local LLM programming LLM fine-tuning with LoRA