Custom cover image
Custom cover image

Building and Training Generative AI Models : A Practical Guide to Generative AI Development and Scaling / by Irena Cronin

By: Resource type: Ressourcentyp: Buch (Online)Book (Online)Language: English Publisher: Berkeley, CA : Apress, 2026Publisher: Berkeley, CA : Imprint: Apress, 2026Edition: 1st ed. 2026Description: 1 Online-Ressource(XXXVIII, 625 p. 3 illus.)ISBN:
  • 9798868823329
Subject(s): Additional physical formats: 9798868823312 | 9798868823336 | Erscheint auch als: 9798868823312 Druck-Ausgabe | Erscheint auch als: 9798868823336 Druck-AusgabeDDC classification:
  • 006.3 23
DOI: DOI: 10.1007/979-8-8688-2332-9Online resources: Summary: Chapter 1: Introduction to Generative AI Systems -- Chapter 2: Choosing the Right Architecture -- Chapter 3: Data Collection and Preparation -- Chapter 4: Training Fundamentals and Self-Supervised Learning -- Chapter 5: Optimization and Learning Strategies -- Chapter 6: Scaling Training with Infrastructure and Distributed Systems -- Chapter 7: Fine-Tuning and Domain Adaptation -- Chapter 8: Reinforcement Learning with Human Feedback (RLHF) -- Chapter 9: Model Compression and Inference Optimization -- Chapter 10: Addressing Bias, Hallucinations, and Failure Modes -- Chapter 11: Evaluation and Benchmarking of Generative Models -- Chapter 12: Future Directions and Deployment in Production.Summary: This book is a hands-on, technical guide to building and deploying generative AI models using advanced deep learning architectures like transformers, GANs, VAEs, and diffusion models. Designed for AI engineers, data scientists, and ML practitioners, it offers a practical roadmap from data ingestion to real-world deployment and evaluation. The book starts by guiding readers on selecting the right model architecture for their application, be it text generation, image synthesis, or multimodal tasks. It then walks through essential components of model training, including dataset handling, self-supervised learning, and core optimisation techniques such as backpropagation, gradient descent, and learning rate scheduling. It also delves into large-scale training infrastructure, covering GPU/TPU usage, distributed computing frameworks, and system-level strategies for scaling performance. Practical guidance is provided on fine-tuning models with domain-specific data and applying reinforcement learning from human feedback (RLHF), model quantisation, and pruning to improve efficiency. Key challenges in generative AI—such as overfitting, bias, hallucination, and data efficiency—are addressed through proven techniques and emerging best practices. Readers will also gain insight into model interpretability and generalisation, ensuring robust and trustworthy outputs. The book demonstrates how to build scalable, production-ready generative systems across domains like media, healthcare, scientific simulation, and design through real-world examples and applied case studies. By the end, readers will gain an understanding of how to architect, optimise, and apply generative models across diverse domains such as media creation, healthcare, design, scientific simulation, and beyond. What you will learn; Learn how to choose and implement generative models—VAEs, GANs, transformers, and diffusion models—for specific use cases. Master training optimization techniques such as backpropagation, gradient descent, adaptive learning rates, and regularization. Apply best practices for large-scale training using GPUs, TPUs, and distributed computing frameworks for performance scaling. Boost model efficiency through quantization, pruning, fine-tuning, and RLHF to enhance output quality and reduce overhead. Who this book is for: AI Engineers and Machine Learning Practitioners looking to build and deploy generative models in real-world applications. Data Scientists working on deep learning projects involving text, vision, audio, or multimodal generation.PPN: PPN: 1967983585Package identifier: Produktsigel: ZDB-2-SEB | ZDB-2-CWD | ZDB-2-SXPC
No physical items for this record

Barrierefreier Inhalt: PDF/UA-1. Table of contents navigation. Single logical reading order. Short alternative textual descriptions. Use of color is not sole means of conveying information. Use of high contrast between text and background color. Next / Previous structural navigation. All non-decorative content supports reading without sight

Anmerkungen zur Barrierefreiheit: This PDF has been created in accordance with the PDF/UA-1 standard to enhance accessibility, including screen reader support, described non-text content (images, graphs), bookmarks for easy navigation, keyboard-friendly links and forms and searchable, selectable text. We recognize the importance of accessibility, and we welcome queries about accessibility for any of our products. If you have a question or an access need, please get in touch with us at accessibilitysupport@springernature.com. Please note that a more accessible version of this eBook is available as ePub.. No reading system accessibility options actively disabled. Publisher contact for further accessibility information: accessibilitysupport@springernature.com

KIT – The University in the Helmholtz Association                         Home | Impressum | Datenschutz | Barrierefreiheit | KIT