Shipping Open Models with Hugging Face
An Independent Guide to Transformers, Accelerate and Serving Models in Production
-
- $17.99
-
- $17.99
Publisher Description
Shipping Open Models with Hugging Face is an independent, practical guide for engineers who need to take an open model from a repository on the Hugging Face Hub to a service other people depend on. It uses Transformers 5 and Accelerate as they stand in late 2026. It teaches how to read a repository before depending on it, check its license, pin a commit and load weights safely; how to test with pipelines, then control the details with Auto classes, tokenizers and chat templates; how to set generation options, run embeddings, reranking and classification, and handle images; how to place models on devices, batch correctly, quantize and evaluate; how to package a service, when to use transformers serve, and when to move to an engine such as vLLM or SGLang; how Inference Endpoints and Inference Providers work and what they cost; and how to upgrade and monitor without surprises. One fictional running example, a university library catalog assistant, connects every chapter with Python, with the core listings run on a CPU and their real output printed, along with checklists, a lab and a glossary.
This book is an independent publication and is not authorized, sponsored or endorsed by any company whose products it mentions. All trademarks belong to their respective owners and are used only for identification. This book was drafted with generative AI tools and checked with automated tools. Tools, features and prices change often; confirm them in current documentation.