Construcción de grandes modelos de lenguaje desde cero

The physicist Richard P. Feynman famously said, "I cannot understand what I cannot create." Inspired by this principle, Sebastian Raschka offers a practical and detailed guide in this book to building a large language model that can run even on a laptop. This book analyzes every stage of the process, from planning and coding the model to training and fine-tuning it. Through a hands-on and deeply satisfying approach, you will explore the fundamentals of generative artificial intelligence. Without relying on existing language model libraries, you will program a base model from scratch, turn it into a text classification system, and finally create a chatbot capable of following instructions in natural conversations. You will understand it thoroughly because you will be the one building it. With this book you will learn to: * Plan and code a GPT-2-like language model. * Load and use pre-trained weights. * Design a complete training flow. * Tune a language model to classify text. * Develop models that can respond to human instructions. Aimed at readers with intermediate Python skills and basic knowledge of machine learning (ML), this book will enable you to create and run a large language model on a modern laptop, with the option to take advantage of GPU acceleration.