Large Language Models in Practice

The book is dedicated to the practical use of modern language models and methods for working with textual data. It examines the key principles underlying transformer architecture, as well as ways to apply pre-trained models for text generation, annotation, classification, and semantic search. Techniques for building search systems based on vector representations, re-ranking methods, approaches to document analysis and clustering, as well as principles for utilizing generative models – from query handling techniques to solutions with retrieval-augmented generation (RAG) – are thoroughly discussed. Special attention is given to practical recommendations for creating efficient LLM pipelines and optimizing models for specific tasks using modern training methods. For AI specialists, data engineers, developers, technical managers